BECKHOFF EtherCAT Technology Section I User Manual

Download

Hardware Data Sheet Section I

Slave Controller

Section I – Technology EtherCAT Protocol, Physical Layer,

EtherCAT Processing Unit, FMMU, SyncManager, SII EEPROM, Distributed Clocks

Section II – Register Description (Online at http://www.beckhoff.com)

Section III – Hardware Description

(Online at http://www.beckhoff.com)

Version 2.2 Date: 2014-07-07

DOCUMENT ORGANIZATION

Trademarks

Beckhoff®, TwinCAT®, EtherCAT®, Safety over EtherCAT®, TwinSAFE® and XFC® are registered trademarks of and licensed by Beckhoff Automation GmbH. Other designations used in this publication may be trademarks whose use by third parties for their own purposes could violate the rights of the owners.

Patent Pending

The EtherCAT Technology is covered, including but not limited to the following German patent applications and patents: DE10304637, DE102004044764, DE102005009224, DE102007017835 with corresponding applications or registrations in various other countries.

Disclaimer

The documentation has been prepared with care. The products described are, however, constantly under development. For that reason the documentation is not in every case checked for consistency with performance data, standards or other characteristics. In the event that it contains technical or editorial errors, we retain the right to make alterations at any time and without warning. No claims for the modification of products that have already been supplied may be made on the basis of the data, diagrams and descriptions in this documentation.

© Beckhoff Automation GmbH 07/2014. The reproduction, distribution and utilization of this document as well as the communication of its contents to others without express authorization are prohibited. Offenders will be held liable for the payment of damages. All rights reserved in the event of the grant of a patent, utility model or design.

DOCUMENT ORGANIZATION

The Beckhoff EtherCAT Slave Controller (ESC) documentation covers the following Beckhoff ESCs:

 ET1200  ET1100  EtherCAT IP Core for Altera® FPGAs  EtherCAT IP Core for Xilinx® FPGAs  ESC20

The documentation is organized in three sections. Section I and section II are common for all Beckhoff ESCs, Section III is specific for each ESC variant.

The latest documentation is available at the Beckhoff homepage (http://www.beckhoff.com).

Section I – Technology (All ESCs)

Section I deals with the basic EtherCAT technology. Starting with the EtherCAT protocol itself, the frame processing inside EtherCAT slaves is described. The features and interfaces of the physical layer with its two alternatives Ethernet and EBUS are explained afterwards. Finally, the details of the functional units of an ESC like FMMU, SyncManager, Distributed Clocks, Slave Information Interface, Interrupts, Watchdogs, and so on, are described.

Since Section I is common for all Beckhoff ESCs, it might describe features which are not available in a specific ESC. Refer to the feature details overview in Section III of a specific ESC to find out which features are available.

Section II – Register Description (All ESCs)

Section II contains detailed information about all ESC registers. This section is also common for all Beckhoff ESCs, thus registers, register bits, or features are described which might not be available in a specific ESC. Refer to the register overview and to the feature details overview in Section III of a specific ESC to find out which registers and features are available.

Section III – Hardware Description (Specific ESC)

Section III is ESC specific and contains detailed information about the ESC features, implemented registers, configuration, interfaces, pinout, usage, electrical and mechanical specification, and so on. Especially the Process Data Interfaces (PDI) supported by the ESC are part of this section.

Additional Documentation

Application notes and utilities like pinout configuration tools for ET1100/ET1200 can also be found at the Beckhoff homepage.

I-II Slave Controller – Technology

DOCUMENT HISTORY

Version

Comment

1.0

Initial release

1.1

 Chapter Interrupts – AL Event Request: corrected AL Event Mask register

address to 0x0204:0x0207

 EtherCAT Datagram: Circulating Frame bit has position 14 (not 13)  PHY addressing configuration changed  Loop control: a port using Auto close mode is automatically opened if a valid

Ethernet frame is received at this port

 EEPROM read/write/reload example: steps 1 and 2 swapped  EEPROM: Configured Station Alias (0x0012:0x0013) is only taken over at first

EEPROM load after power-on or reset

 SyncManager: Watchdog trigger and interrupt generation in mailbox mode with

single byte buffers requires alternating write and read accesses for some ESCs, thus buffered mode is required for Digital I/O watchdog trigger generation

 National Semiconductor DP83849I Ethernet PHY deprecated because of large

link loss reaction time and delay

 Added distinction between permanent ports and Bridge port (frame processing)  Added PDI chapter  PDI and DC Sync/Latch signals are high impedance until the SII EEPROM is

successfully loaded

 Editorial changes

1.2

 PHY address configuration revised. Refer to Section III for ESC supported

configurations

 Added Ethernet Link detection chapter  Added MI Link Detection and Configuration, link detection descriptions updated  Added EEPROM Emulation for EtherCAT IP Core  Added General Purpose Input chapter  Corrected minimum datagram sizes in EtherCAT header figure  Editorial changes

1.2.1

 Chapter 5.1.1: incompatible PHYs in footnote 1 deleted

1.3

 Added advisory for unused MII/RMII/EBUS ports  Ethernet PHY requirements revised: e.g., configuration by strapping options,

recommendations enhanced. Footnote about compatible PHYs removed, information has moved to the EtherCAT Slave Controller application note “PHY Selection Guide”.

 Frame Error detection chapter enhanced  FIFO size reduction chapter enhanced  EBUS enhanced link detection chapter enhanced  Ethernet PHY link loss reaction time must be faster than 15 µs, otherwise use

Enhanced link detection

 Enhanced link detection description corrected. Enhanced link detection does

not remain active if it is disabled by EEPROM and EBUS handshake frames are received

 ARMW/FRWM commands increase the working counter by 1  Editorial changes

1.4

 Update to EtherCAT IP Core Release 2.1.0/2.01a  Added restriction to enhanced link configuration: RX_ER has to be asserted

outside of frames (IEEE802 optional feature)

 ESC power-on sequence for IP Core corrected  Removed footnote on t

Diff

figures, refer to Section III for actual figures

 Editorial changes

DOCUMENT HISTORY

Slave Controller – Technology I-III

DOCUMENT HISTORY

Version

Comment

1.5

 EEPROM Read/Write/Reload example: corrected register addresses  Updated/clarified PHY requirements, PHY link loss reaction time is mandatory  Enhanced Link Detection can be configured port-wise depending on ESC  Added DC Activation and DC Activation State features for some ESCs  ESC10 removed  Editorial changes

1.6

 Fill reserved EEPROM words of the ESC Configuration Area with 0  Interrupt chapter: example for proper interrupt handling added  Use Position Addressing only for bus scanning at startup and to detect newly

attached devices

 System Time PDI controlled: detailed description added  Added MII back-to-back connection example  Renamed Err(x) LED to PERR(x)  Editorial changes

1.7

 Link status description enhanced  Clarifications for DC System Time and reference between clocks and registers  Chapter on avoiding unconnected Port 0 configurations added  Direct ESC to standard Ethernet MAC MII connection added  MI link detection and configuration must not be used without LINK_MII signals  Added criteria for detecting when DC synchronization is established  SII EEPROM interface is a point-to-point connection  PHY requirements: PHY startup should not rely on MDC clocking, ESD

tolerance and baseline wander compensation recommendations added

 Editorial changes

1.8

 Update to EtherCAT IP Core Release 2.3.0/2.03a  EEPROM acknowledge error (0x0502[13]) can also occur for a read access  ERR and STATE LED updated  Editorial changes

1.9

 EtherCAT state machine: additional AL status codes defined  EtherCAT protocol: LRD/LRW read data depends on bit mask  Updated EBUS Enhanced Link Detection  Updated FMMU description  Loop control description updated  EtherCAT frame format (VLAN tag) description enhanced  Update to EtherCAT IP Core Release 2.3.2/2.03c

2.0

 Update to EtherCAT IP Core Release 2.4.0/2.04a  SII/ESI denotation now consistent with ETG  Updated AL Status codes  Editorial changes

2.1

 Update to EtherCAT IP Core Release 3.0.0/3.00a  Update to ET1100-0003 and ET1200-0003  RUN/ERR LED description enhanced  Added RGMII and FX operation  Added Gigabit Ethernet PHY chapter  Updated FIFO size configuration (default from SII)  Updated PHY address configuration  Added PDI register function acknowledge by write  Added propagation delay measurement in reverse mode (especially ET1200)  Enhanced ERR_LED description  Editorial changes

2.2

 Update to EtherCAT IP Core Release 3.0.6/3.00g  Added resetting Distributed Clocks Time Loop Control filters to the

synchronization steps

 Extended Back-to-Back MII connection schematic  Clarified EBUS standard link detection restrictions  Editorial changes

I-IV Slave Controller – Technology

DOCUMENT HISTORY

Slave Controller – Technology I-V

CONTENTS

1 EtherCAT Slave Controller Overview 1

1.1 EtherCAT Slave Controller Function Blocks 2

1.2 Further Reading on EtherCAT and ESCs 3

1.3 Scope of Section I 3

2 EtherCAT Protocol 4

2.1 EtherCAT Header 4

2.2 EtherCAT Datagram 5

2.3 EtherCAT Addressing Modes 6

2.3.1 Device Addressing 7

2.3.2 Logical Addressing 7

2.4 Working Counter 8

2.5 EtherCAT Command Types 9

3 Frame Processing 12

3.1 Loop Control and Loop State 12

3.2 Frame Processing Order 14

3.3 Permanent Ports and Bridge Port 15

3.4 Shadow Buffer for Register Write Operations 15

3.5 Circulating Frames 15

3.5.1 Unconnected Port 0 16

3.6 Non-EtherCAT Protocols 16

3.7 Special Functions of Port 0 16

4 Physical Layer Common Features 17

4.1 Link Status 17

4.2 Selecting Standard/Enhanced Link Detection 18

4.3 FIFO Size Reduction 19

4.4 Frame Error Detection 19

5 Ethernet Physical Layer 20

5.1 Requirements to Ethernet PHYs 20

5.2 PHY reset and Link partner notification/loop closing 20

5.3 MII Interface 21

5.4 RMII Interface 22

5.5 RGMII Interface 22

5.5.1 RGMII In-Band Link Status 22

5.6 Link Detection 22

5.6.1 LINK_MII Signal 22

5.6.2 MI Link Detection and Configuration 23

5.7 Standard and Enhanced MII Link Detection 23

5.8 EtherCAT over Optical Links (FX) 24

5.8.1 Link partner notification and loop closing 24

I-VI Slave Controller – Technology

CONTENTS

5.8.2 Far-End-Fault (FEF) 24

5.8.3 ESCs with native FX support 25

5.8.4 ESCs without native FX support 25

5.9 Gigabit Ethernet PHYs 25

5.10 MII Management Interface (MI) 26

5.10.1 PHY Addressing/PHY Address Offset 26

5.10.2 Logical Interface 28

5.10.3 MI Protocol 29

5.10.4 Timing specifications 29

5.11 MII management example schematic 30

5.12 Ethernet Termination and Grounding Recommendation 31

5.13 Ethernet Connector (RJ45 / M12) 32

5.14 Back-to-Back MII Connection 33

5.14.1 ESC to ESC Connection 33

5.14.2 ESC to Standard Ethernet MAC 34

6 EBUS/LVDS Physical Layer 35

6.1 Interface 35

6.2 EBUS Protocol 36

6.3 Timing Characteristics 36

6.4 Standard EBUS Link Detection 37

6.5 Enhanced EBUS Link Detection 37

6.6 EBUS RX Errors 38

6.7 EBUS Low Jitter 38

6.8 EBUS Connection 38 7 FMMU 39 8 SyncManager 41

8.1 Buffered Mode 42

8.2 Mailbox Mode 43

8.2.1 Mailbox Communication Protocols 43

8.3 PDI register function acknowledge by Write 44

8.4 Interrupt and Watchdog Trigger Generation, Latch Event Generation 44

8.5 Single Byte Buffer Length / Watchdog Trigger for Digital Output PDI 45

8.6 Repeating Mailbox Communication 45

8.7 SyncManager Deactivation by the PDI 46 9 Distributed Clocks 47

9.1 Clock Synchronization 47

9.1.1 Clock Synchronization Process 49

9.1.2 Propagation Delay Measurement 50

9.1.3 Offset Compensation 55

9.1.4 Resetting the Time Control Loop 56

9.1.5 Drift Compensation 56

Slave Controller – Technology I-VII

CONTENTS

9.1.6 Reference between DC Registers/Functions and Clocks 58

9.1.7 When is Synchronization established? 59

9.1.8 Clock Synchronization Initialization Example 59

9.2 SyncSignals and LatchSignals 60

9.2.1 Interface 60

9.2.2 Configuration 60

9.2.3 SyncSignal Generation 61

9.2.4 LatchSignals 64

9.2.5 ECAT or PDI Control 65

9.3 System Time PDI Controlled 66

9.4 Communication Timing 68 10 EtherCAT State Machine 70

10.1 EtherCAT State Machine Registers 71

10.1.1 AL Control and AL Status Register 71

10.1.2 Device Emulation 71

10.1.3 Error Indication and AL Status Code Register 71

11 SII EEPROM 72

11.1 SII EEPROM Content 73

11.2 SII EEPROM Logical Interface 74

11.2.1 SII EEPROM Errors 75

11.2.2 SII EEPROM Interface Assignment to ECAT/PDI 76

11.2.3 Read/Write/Reload Example 77

11.2.4 EEPROM Emulation 77

11.3 SII EEPROM Electrical Interface (I2C) 78

11.3.1 Addressing 78

11.3.2 EEPROM Size 78

11.3.3 I²C Access Protocol 79

11.3.4 Timing specifications 80

12 Interrupts 82

12.1 AL Event Request (PDI Interrupt) 82

12.2 ECAT Event Request (ECAT Interrupt) 83

12.3 Clearing Interrupts Accidentally 83 13 Watchdogs 84

13.1 Process Data Watchdog 84

13.2 PDI Watchdog 84 14 Error Counters 85

14.1 Frame error detection 86

14.2 Errors and Forwarded Errors 86 15 LED Signals (Indicators) 87

15.1 RUN LED 87

15.1.1 RUN LED override 87

I-VIII Slave Controller – Technology

CONTENTS

15.2 ERR LED 88

15.2.1 ERR LED override 88

15.3 STATE LED and STATE_RUN LED Signal 89

15.4 LINKACT LED 89

15.5 Port Error LED (PERR) 90 16 Process Data Interface (PDI) 91

16.1 PDI Selection and Configuration 91

16.2 PDI register function acknowledge by write 92

16.3 General Purpose I/O 93

16.3.1 General Purpose Inputs 93

16.3.2 General Purpose Output 93

17 Additional Information 94

17.1 ESC Clock Source 94

17.2 Power-on Sequence 94

17.3 Write Protection 95

17.3.1 Register Write Protection 95

17.3.2 ESC Write Protection 95

17.4 ESC Reset 95 18 Appendix 96

18.1 Support and Service 96

18.1.1 Beckhoff’s branch offices and representatives 96

18.2 Beckhoff Headquarters 96

Slave Controller – Technology I-IX

TABLES

Table 1: ESC Main Features ................................................................................................................... 1

Table 2: EtherCAT Frame Header........................................................................................................... 4

Table 3: EtherCAT Datagram .................................................................................................................. 6

Table 4: EtherCAT Addressing Modes .................................................................................................... 6

Table 5: Working Counter Increment ...................................................................................................... 8

Table 6: EtherCAT Command Types .................................................................................................... 10

Table 7: EtherCAT Command Details ................................................................................................... 11

Table 8: Registers for Loop Control and Loop/Link Status ................................................................... 13

Table 9: Frame Processing Order ......................................................................................................... 14

Table 10: Link Status Description .......................................................................................................... 17

Table 11: Registers for Enhanced Link Detection ................................................................................. 18

Table 12: Registers for FIFO Size Reduction ........................................................................................ 19

Table 13: Special/Unused MII Interface signals .................................................................................... 21

Table 14: Registers used for Ethernet Link Detection ........................................................................... 22

Table 15: PHY Address configuration matches PHY address settings ................................................. 27

Table 16: PHY Address configuration does not match actual PHY address settings ........................... 27

Table 17: MII Management Interface Register Overview ...................................................................... 28

Table 18: MII Management Interface timing characteristics .................................................................. 29

Table 19: Signals used for Fast Ethernet .............................................................................................. 32

Table 20: EBUS Interface signals ......................................................................................................... 35

Table 21: EBUS timing characteristics .................................................................................................. 36

Table 22: Example FMMU Configuration .............................................................................................. 39

Table 23: SyncManager Register overview ........................................................................................... 41

Table 24: EtherCAT Mailbox Header .................................................................................................... 44

Table 25: Registers for Propagation Delay Measurement .................................................................... 50

Table 26: Parameters for Propagation Delay Calculation ..................................................................... 53

Table 27: Registers for Offset Compensation ....................................................................................... 55

Table 28: Registers for Resetting the Time Control Loop ..................................................................... 56

Table 29: Registers for Drift Compensation .......................................................................................... 57

Table 30: Reference between DC Registers/Functions and Clocks ..................................................... 58

Table 31: Distributed Clocks signals ..................................................................................................... 60

Table 32: SyncSignal Generation Mode Selection ................................................................................ 61

Table 33: Registers for SyncSignal Generation .................................................................................... 62

Table 34: Registers for Latch Input Events ........................................................................................... 65

Table 35: Registers for the EtherCAT State Machine ........................................................................... 71

Table 36: AL Control and AL Status Register Values ........................................................................... 71

Table 37: ESC Configuration Area ........................................................................................................ 73

Table 38: SII EEPROM Content Excerpt ............................................................................................... 74

Table 39: SII EEPROM Interface Register Overview ............................................................................ 74

Table 40: SII EEPROM Interface Errors ................................................................................................ 75

Table 41: I²C EEPROM signals ............................................................................................................. 78

Table 42: EEPROM Size ....................................................................................................................... 78

Table 43: I²C Control Byte ..................................................................................................................... 79

Table 44: I²C Write Access .................................................................................................................... 79

Table 45: I²C Read Access .................................................................................................................... 80

Table 46: EEPROM timing characteristics ............................................................................................ 80

Table 47: Registers for AL Event Request Configuration ..................................................................... 82

Table 48: Registers for ECAT Event Request Configuration ................................................................ 83

Table 49: Registers for Watchdogs ....................................................................................................... 84

Table 50: Error Counter Overview ......................................................................................................... 85

Table 51: Errors Detected by Physical Layer, Auto-Forwarder, and EtherCAT Processing Unit ......... 86

Table 52: RUN LED state indication ...................................................................................................... 87

Table 53: Registers for RUN LED control ............................................................................................. 87

Table 54: Automatic ESC ERR LED state indication ............................................................................ 88

Table 55: Registers for ERR LED control .............................................................................................. 88

Table 56: LINKACT LED States ............................................................................................................ 89

Table 57: Available PDIs depending on ESC ........................................................................................ 91

Table 58: Functions/registers affected by PDI register function acknowledge by write ........................ 92

Table 59: ESC Power-On Sequence ..................................................................................................... 94

Table 60: Registers for Write Protection ............................................................................................... 95

I-X Slave Controller – Technology

TABLES

Slave Controller – Technology I-XI

FIGURES

Figure 1: EtherCAT Slave Controller Block Diagram .............................................................................. 1

Figure 2: Ethernet Frame with EtherCAT Data ....................................................................................... 4

Figure 3: EtherCAT Datagram ................................................................................................................. 5

Figure 4: Auto close loop state transitions ............................................................................................ 13

Figure 5: Frame Processing .................................................................................................................. 14

Figure 6: Circulating Frames ................................................................................................................. 15

Figure 7: All frames are dropped because of Circulating Frame Prevention ........................................ 16

Figure 8: Write access ........................................................................................................................... 29

Figure 9: Read access ........................................................................................................................... 29

Figure 10: MII management example schematic .................................................................................. 30

Figure 11: Termination and Grounding Recommendation .................................................................... 31

Figure 12: RJ45 Connector ................................................................................................................... 32

Figure 13: M12 D-code Connector ........................................................................................................ 32

Figure 14: Back-to-Back MII Connection (two ESCs) ........................................................................... 33

Figure 15: Back-to-Back MII Connection (ESC and standard MAC)..................................................... 34

Figure 16: EBUS Interface Signals ........................................................................................................ 35

Figure 17: EBUS Protocol ..................................................................................................................... 36

Figure 18: Example EtherCAT Network ................................................................................................ 37

Figure 19: EBUS Connection ................................................................................................................ 38

Figure 20: FMMU Mapping Principle ..................................................................................................... 39

Figure 21: FMMU Mapping Example ..................................................................................................... 40

Figure 22: SyncManager Buffer allocation ............................................................................................ 42

Figure 23: SyncManager Buffered Mode Interaction............................................................................. 42

Figure 24: SyncManager Mailbox Interaction ........................................................................................ 43

Figure 25: EtherCAT Mailbox Header (for all Types) ............................................................................ 44

Figure 26: Handling of a Repeat Request with Read Mailbox .............................................................. 46

Figure 27: Propagation Delay, Offset, and Drift Compensation ............................................................ 49

Figure 28: Propagation Delay Calculation ............................................................................................. 52

Figure 29: Distributed Clocks signals .................................................................................................... 60

Figure 30: SyncSignal Generation Modes ............................................................................................. 61

Figure 31: SYNC0/1 Cycle Time Examples .......................................................................................... 63

Figure 32: System Time PDI Controlled with three steps ..................................................................... 66

Figure 33: System Time PDI Controlled with two steps ........................................................................ 67

Figure 34: DC Timing Signals in relation to Communication ................................................................. 68

Figure 35: EtherCAT State Machine ..................................................................................................... 70

Figure 36: SII EEPROM Layout............................................................................................................. 72

Figure 37: I²C EEPROM signals ............................................................................................................ 78

Figure 38: Write access (1 address byte, up to 16 Kbit EEPROMs) ..................................................... 80

Figure 39: Write access (2 address bytes, 32 Kbit - 4 Mbit EEPROMs) ............................................... 81

Figure 40: Read access (1 address byte, up to 16 Kbit EEPROMs) .................................................... 81

Figure 41: PDI Interrupt Masking and interrupt signals ......................................................................... 82

Figure 42: ECAT Interrupt Masking ....................................................................................................... 83

I-XII Slave Controller – Technology

ABBREVIATIONS

µC

Microcontroller

ADR

Address

ADS

Automation Device Specification (Beckhoff)

Application Layer

AMBA®

Advanced Microcontroller Bus Architecture from ARM® APRD

Auto Increment Physical Read

APWR

Auto Increment Physical Write

APRW

Auto Increment Physical ReadWrite

ARMW

Auto Increment Physical Read Multiple Write

AoE

ADS over EtherCAT

ASIC

Application Specific Integrated Chip

Auto Crossover

Automatic detection of whether or not the send and receive lines are crossed.

Auto Negotiation

Automatic negotiation of transmission speeds between two stations.

Avalon®

On-chip bus for Altera® FPGAs

AXITM

Advanced eXtensible Interface Bus, an AMBA interconnect. Used as On-Chipbus

Big Endian

Data format (also Motorola format). The more significant byte is transferred first when a word is transferred. However, for EtherCAT the least significant bit is the first on the wire.

BOOT

BOOT state of EtherCAT state machine

Boundary Clock

A station that is synchronized by another station and then passes this information on.

Bridge

A term for switches used in standards. Bridges are devices that pass on messages based on address information.

Broadcast

An unacknowledged transmission to an unspecified number of receivers. BRD

Broadcast Read

BWR

Broadcast Write

BRW

Broadcast ReadWrite

Cat

Category – classification for cables that is also used in Ethernet. Cat 5 is the minimum required category for EtherCAT. However, Cat 6 and Cat 7 cables are available.

CoE

CAN® application layer over EtherCAT

Communication Stack

A communication software package that is generally divided into successive layers, which is why it is referred to as a stack.

Confirmed

Means that the initiator of a service receives a response.

CRC

Cyclic Redundancy Check, used for FCS

ABBREVIATIONS

Slave Controller – Technology I-XIII

ABBREVIATIONS

Cut Through

Procedure for cutting directly through an Ethernet frame by a switch before the complete message is received.

Cycle

Cycle in which data is to be exchanged in a system operating on a periodical basis.

Distributed Clocks Mechanism to synchronize EtherCAT slaves and master

Delay

Delays can be caused by run-times during transfer or internal delays of a network component.

Dest Addr

Destination address of a message (the destination can be an individual network station or a group (multicast).

DHCP

Dynamic Host Configuration Protocol, used to assign IP addresses (and other important startup parameter in the Internet context).

Data Link Layer, also known as Layer 2. EtherCAT uses the Data Link Layer of Ethernet, which is standardized as IEEE 802.3.

DNS

Domain Name Service, a protocol for domain name to IP addresses resolution.

EBUS

Based on LVDS (Low Voltage Differential Signaling) standard specified in ANSI/TIA/EIA-644-1995

ECAT

EtherCAT

EEPROM

Electrically Erasable Programmable Read Only Memory. Non-volatile memory used to store EtherCAT Slave Information (ESI). Connected to the SII.

EMC

Electromagnetic Compatibility, describes the robustness of a device with regard to electrical interference from the environment.

EMI

Electromagnetic Interference

Engineering

Here: All applications required to configure and program a machine.

EoE

Ethernet over EtherCAT

EOF

End of Frame

ERR

Error indicator for AL state

Err(x)

Physical Layer RX Error LED for debugging purposes

ESC

EtherCAT Slave Controller ESI

EtherCAT Slave Information, stored in SII EEPROM

ESM

EtherCAT State Machine

ETG

EtherCAT Technology Group (http://www.ethercat.org)

EtherCAT

Real-time Standard for Industrial Ethernet Control Automation Technology (Ethernet for Control Automation Technology)

EtherType

Identification of an Ethernet frame with a 16-bit number assigned by IEEE. For example, IP uses EtherType 0x0800 (hexadecimal) and the EtherCAT protocol uses 0x88A4.

I-XIV Slave Controller – Technology

ABBREVIATIONS

EPU

EtherCAT Processing Unit. The logic core of an ESC containing e.g. registers, memory, and processing elements.

Fast Ethernet

Ethernet with a transmission speed of 100 Mbit/s.

FCC

Federal Communications Commission

FCS

Frame Check Sequence

FIFO

First In First Out

Firewall

Routers or other network component that acts as a gateway to the Internet and enables protection from unauthorized access.

FMMU

Fieldbus Memory Management Unit

FoE

File access over EtherCAT

Follow Up

Message that follows Sync and indicates when the Sync frame was sent from the last node (defined in IEEE 1588).

FPGA

Field Programmable Gate Array FPRD

Configured Address Physical Read

FPWR

Configured Address Physical Write

FPRW

Configured Address Physical ReadWrite

FRMW

Configured Address Physical Read Multiple Write

Frame

See PDU

FTP

File Transfer Protocol

Get

Access method used by a client to read data from a device.

GND

Ground GPI

General Purpose Input

GPO

General Purpose Output

Hardware

I²C

Inter-Integrated Circuit, serial bus used for SII EEPROM connection

ICMP

Internet Control Message Protocol: Mechanisms for signaling IP errors.

IEC

International Electrotechnical Commission

IEEE

Institute of Electrical and Electronics Engineers

INIT

INIT state of EtherCAT state machine

Interval

Time span

Internet Protocol: Ensures transfer of data on the Internet from end node to end node. Intellectual Property

IRQ

Interrupt Request

ISO

International Standard Organization

Slave Controller – Technology I-XV

ABBREVIATIONS

ISO/OSI Model

ISO Open Systems Interconnection Basic Reference Model (ISO 7498): describes the division of communication into 7 layers.

Information Technology: Devices and methods required for computer-aided information processing.

LatchSignal

Signal for Distributed Clocks time stamping

LED

Light Emitting Diode, used as an indicator

Link/Act

Link/Activity Indicator (LED)

Little Endian

Data format (also Intel format). The less significant byte is transferred first when a word is transferred. With EtherCAT, the least significant bit is the first on the wire.

LLDP

Lower Layer Discovery Protocol – provides the basis for topology discovery and configuration definition (see IEEE802.1ab) LRD

Logical Read

LWR

Logical Write

LRW

Logical ReadWrite

LVDS

Low Voltage Differential Signaling

M12

Connector used for industrial Ethernet

MAC

Media Access Control: Specifies station access to a communication medium. With full duplex Ethernet, any station can send data at any time; the order of access and the response to overload are defined at the network component level (switches).

MAC Address

Media Access Control Address: Also known as Ethernet address; used to identify an Ethernet node. The Ethernet address is 6 bytes long and is assigned by the IEEE.

Mandatory Services

Mandatory services, parameters, objects, or attributes. These must be implemented by every station.

MBX

Mailbox

MDI

Media Dependent Interface: Use of connector Pins and Signaling (PC side)

MDI-X

Media Dependent Interface (crossed): Use of connector Pins and Signaling with crossed lines (Switch/hub side)

(PHY) Management Interface

MII

Media Independent Interface: Standardized interface between the Ethernet MAC and PHY.

Multicast

Transmission to multiple destination stations with a frame – generally uses a special address.

NOP

No Operation

NVRAM

Non-volatile random access memory, e.g. EEPROM or Flash.

Octet

Term from IEC 61158 – one octet comprises exactly 8 bits. OP

Operational state of EtherCAT state machine

I-XVI Slave Controller – Technology

ABBREVIATIONS

OPB

On-Chip Peripheral Bus

Optional Service

Optional services can be fulfilled by a PROFINET station in addition to the mandatory services.

OSI

Open System Interconnect

OUI

Organizationally Unique Identifier –the first 3 Bytes of an Ethernet-Address that will be assign to companies or organizations and can be used for protocol identifiers as well (e.g. LLDP)

PDI

Process Data Interface or Physical Device Interface: an interface that allows access to ESC from the process side.

PDO

Process Data Object

PDU

Protocol Data Unit: Contains protocol information (Src Addr, Dest Addr, Checksum and service parameter information) transferred from a protocol instance of transparent data to a subordinate level (the lower level contains the information being transferred).

Protection Earth

PHY

Physical layer device that converts data from the Ethernet controller to electric or optical signals.

Ping

Frame that verifies whether the partner device is still available. PLB

Processor Local Bus

PLL

Phase Locked Loop

PREOP

Pre-Operational state of EtherCAT state machine

Priority Tagging

Priority field inserted in an Ethernet frame.

Protocol

Rules for sequences – here, also the sequences (defined in state machines) and frame structures (described in encoding) of communication processes.

Provider

Device that sends data to other consumers in the form of a broadcast message.

PTP

Precision Time Protocol in accordance with IEEE 1588: Precise time synchronization procedures.

PTP Master

Indicates time in a segment.

PTP Slave

Station synchronized by a PTP master.

Quad Cable

Cable type in which the two cable pairs are twisted together. This strengthens the electromagnetic resistance.

RAM

Random Access Memory. ESC have User RAM and Process Data RAM.

Read

Service enabling read access to an I/O device.

Real-Time

Real-time capability of a system to perform a task within a specific time.

Request

Call of a service in the sender/client.

Response

Response to a service on the client side.

Slave Controller – Technology I-XVII

ABBREVIATIONS

RJ45

FCC Registered Jack, standard Ethernet connector (8P8C)

RMII

Reduced Media Independent Interface

Router

Network component acting as a gateway based on the interpretation of the IP address.

RSTP

Rapid Spanning Tree Protocol: Prevents packet from looping infinitely between switches; RSTP is specified in IEEE 802.1 D (Edition 2004)

Real-time. Name for a real-time protocol that can be run in Ethernet controllers without special support.

RTC

Real-time Clock chip of PCs

RT Frames

EtherCAT Messages with EtherType 0x88A4.

Receive

RXPDO

Receive PDO, i.e. Process Data that will be received by ESC20

RUN

RUN indicator (LED) for application state

SAFEOP

Safe-Operational state of EtherCAT state machine

Safety

Safety function, implemented by an electric, electronic programmable fail-safe system that maintains the equipment in a safe state, even during certain critical external events.

Schedule

Determines what should be transferred and when.

Services

Interaction between two components to fulfill a specific task.

Set

Access method used by a client to write data to a server.

SII

Slave Information Interface

SyncManager

SNMP

Simple Network Management Protocol: SNMP is the standard Internet protocol for management and diagnostics of network components (see also RFC 1157 and RFC 1156 at www.ietf.org ).

SoE

Servo Profile over EtherCAT

SOF

Start of Frame Ethernet SOF delimiter at the end of the preamble of Ethernet frames

SPI

Serial Peripheral Interface

Src Addr

Source Address: Source address of a message.

Store and Forward

Currently the common operating mode in switches. Frames are first received in their entirety, the addresses are evaluated, and then they are forwarded. This result in considerable delays, but guarantees that defective frames are not forwarded, causing an unnecessary increase in the bus load.

STP

Shielded Twisted Pair: Shielded cable with at least 2 core pairs to be used as the standard EtherCAT cable.

I-XVIII Slave Controller – Technology

ABBREVIATIONS

Subnet Mask

Divides the IP address into two parts: a subnet address (in an area separated from the rest by routers) and a network address.

Switch

Also known as Bridge. Active network component to connect different EtherCAT participants with each other. A switch only forwards the frames to the addressed participants.

SyncManager

ESC unit for coordinated data exchange between master and slave µController

SyncSignal

Signal generated by the Distributed Clocks unit

TCP

Transmission Control Protocol: Higher-level IP protocol that ensures secure data exchange and flow control.

Transmit

TXPDO

Transmit PDO, i.e. Process Data that will be transmitted by ESC20

UDP

User Datagram Protocol: Non-secure multicast/broadcast frame.

UTP

Unshielded Twisted Pair: Unshielded cable with at least 2 core pairs are not recommended for industrial purpose but are commonly used in areas with low electro-magnetic interference.

VLAN

Virtual LAN

VoE

Vendor specific profile over EtherCAT

Watchdog

WKC

Working Counter

XML

Extensible Markup Language: Standardized definition language that can be interpreted by nearly all parsers.

XML Parser

Program for checking XML schemas.

Slave Controller – Technology I-XIX

EtherCAT Slave Controller Overview

Feature

ET1200

ET1100

IP Core

ESC20

Ports

2-3 (each EBUS/MII, max. 1xMII)

2-4 (each EBUS/MII)

1-3 MII/ 1-3 RGMII/ 1-2 RMII

2 MII FMMUs

3 8 0-8 4 SyncManagers

4 8 0-8 4 RAM [Kbyte]

1 8 0-60

Distributed Clocks

64 bit

32/64 bit

32 bit

Process Data Interfaces

Digital I/O

16 bit

32 bit

8-32 bit

32 bit

SPI Slave

Yes

8/16 bit µController

Async/Sync

Async

On-chip bus

- - Yes

ECAT Processing Unit

AutoForwarder +

Loopback

SyncManager

FMMU

ESC address space

User RAMRegisters Process RAM

EEPROM

Distributed

Clocks

Monitoring Status

Reset

PHY

Management

Reset

SYNC LEDsI²C EEPROM

PHY MI

SPI / µC parallel /

Digital I/O / On-chip bus

0 1 2 3

Ports (Ethernet/EBUS)

LATCH

PDI

ECAT Interface PDI Interface

1 EtherCAT Slave Controller Overview

An EtherCAT Slave Controller (ESC) takes care of the EtherCAT communication as an interface between the EtherCAT fieldbus and the slave application. This document covers the following Beckhoff ESCs: ASIC implementations (ET1100, ET1200), functionally fixed binary configurations for FPGAs (ESC20), and configurable IP Cores for FPGAs (ET1810/ET1815).

Table 1: ESC Main Features

The general functionality of an ESC is shown in Figure 1:

Figure 1: EtherCAT Slave Controller Block Diagram

Slave Controller – Technology I-1

EtherCAT Slave Controller Overview

1.1 EtherCAT Slave Controller Function Blocks EtherCAT Interfaces (Ethernet/EBUS)

The EtherCAT interfaces or ports connect the ESC to other EtherCAT slaves and the master. The MAC layer is integral part of the ESC. The physical layer may be Ethernet or EBUS. The physical layer for EBUS is fully integrated into the ASICs. For Ethernet ports, external Ethernet PHYs connect to the MII/RGMII/RMII ports of the ESC. Transmission speed for EtherCAT is fixed to 100 Mbit/s with Full Duplex communication. Link state and communication status are reported to the Monitoring device. EtherCAT slaves support 2-4 ports, the logical ports are numbered 0-1-2-3, formerly they were denoted by A-B-C-D.

EtherCAT Processing Unit

The EtherCAT Processing Unit (EPU) receives, analyses, and processes the EtherCAT data stream. It is logically located between port 0 and port 3. The main purpose of the EtherCAT Processing unit is to enable and coordinate access to the internal registers and the memory space of the ESC, which can be addressed both from the EtherCAT master and from the local application via the PDI. Data exchange between master and slave application is comparable to a dual-ported memory (process memory), enhanced by special functions e.g. for consistency checking (SyncManager) and data mapping (FMMU). The EtherCAT Processing Units contains the main function blocks of EtherCAT slaves besides Auto-Forwarding, Loop-back function, and PDI.

Auto-Forwarder

The Auto-Forwarder receives the Ethernet frames, performs frame checking and forwards it to the Loop-back function. Time stamps of received frames are generated by the Auto-Forwarder.

Loop-back function

The Loop-back function forwards Ethernet frames to the next logical port if there is either no link at a port, or if the port is not available, or if the loop is closed for that port. The Loop-back function of port 0 forwards the frames to the EtherCAT Processing Unit. The loop settings can be controlled by the EtherCAT master.

FMMU

Fieldbus Memory Management Units are used for bitwise mapping of logical addresses to physical addresses of the ESC.

SyncManager

SyncManagers are responsible for consistent data exchange and mailbox communication between EtherCAT master and slaves. The communication direction can be configured for each SyncManager. Read or write transactions may generate events for the EtherCAT master and an attached µController respectively. The SyncManagers are responsible for the main difference between and ESC and a dual-ported memory, because they map addresses to different buffers and block accesses depending on the SyncManager state. This is also a fundamental reason for bandwidth restrictions of the PDI.

Monitoring

The Monitoring unit contains error counters and watchdogs. The watchdogs are used for observing communication and returning to a safe state in case of an error. Error counters are used for error detection and analysis.

Reset

The integrated reset controller observes the supply voltage and controls external and internal resets (ET1100 and ET1200 ASICs only).

PHY Management

The PHY Management unit communicates with Ethernet PHYs via the MII management interface. This is either used by the master or by the slave. The MII management interface is used by the ESC itself for optionally restarting auto negotiation after receive errors with the enhanced link detection mechanism, and for the optional MI link detection and configuration feature.

Distributed Clock

Distributed Clocks (DC) allow for precisely synchronized generation of output signals and input sampling, as well as time stamp generation of events. The synchronization may span the entire EtherCAT network.

I-2 Slave Controller – Technology

EtherCAT Slave Controller Overview

Memory

An EtherCAT slave can have an address space of up to 64Kbyte. The first block of 4 Kbyte (0x00000x0FFF) is used for registers and user memory. The memory space from address 0x1000 onwards is used as the process memory (up to 60 Kbyte). The size of process memory depends on the device. The ESC address range is directly addressable by the EtherCAT master and an attached µController.

Process Data Interface (PDI) or Application Interface

There are several types of PDIs available, depending on the ESC:

 Digital I/O (8-32 bit, unidirectional/bidirectional, with DC support)  SPI slave  8/16 bit µController (asynchronous or synchronous)  On-chip bus (e.g., Avalon® , PLB®, or AXI®, depending on target FPGA type and selection)  General purpose I/O

The PDIs are described in Section III of the particular ESC, since the PDI functions are highly depending on the ESC type.

SII EEPROM

One non-volatile memory is needed for EtherCAT Slave Information (ESI) storage, typically an I²C EEPROM. If the ESC is implemented as an FPGA, a second non-volatile memory is necessary for the FPGA configuration code.

Status / LEDs

The Status block provides ESC and application status information. It controls external LEDs like the application RUN LED/ERR LED and port Link/Activity LEDs.

1.2 Further Reading on EtherCAT and ESCs

For further information on EtherCAT, refer to the EtherCAT specification ETG.1000, available from the EtherCAT Technology Group (ETG, http://www.ethercat.org), and the IEC standard “Digital data communications for measurement and control – Fieldbus for use in industrial control systems”, IEC 61158 Type 12: EtherCAT, available from the IEC (http://www.iec.ch).

Additional documents on EtherCAT can be found on the EtherCAT Technology Group website (http://www.ethercat.org).

Documentation on Beckhoff Automation EtherCAT Slave Controllers is available at the Beckhoff website (http://www.beckhoff.com), e.g., data sheets, application notes, and ASIC pinout configuration tools.

1.3 Scope of Section I

Since Section I is common for all Beckhoff ESCs, it contains features which might not be available in every individual ESC. Refer to the feature details overview in Section III of a specific ESC to find out which features are actually available.

The following Beckhoff ESCs are covered by Section I:

 ET1200-0003  ET1100-0003  EtherCAT IP Core for Altera® FPGAs (V3.0.6)  EtherCAT IP Core for Xilinx® FPGAs (V3.00g)  ESC20 (Build 22)

Slave Controller – Technology I-3

EtherCAT Protocol

Ethernet Header

Destination Source

EtherType

0x88A4

Destination Source

EtherType

0x0800

Destination Source

EtherType

0x0800

IP Header

UDP Header

dest. port 0x88A4

EtherType

0x88A4

Ethernet Data

EtherCAT Data

EtherCAT Header

FCS

FCSIP HeaderVLAN Tag

Datagrams

EtherCAT Header

Length

Res. Type

48 bit 48 bit 16 bit 32 bit

11 bit 1 bit 4 bit

160 bit 64 bit

32 bit

46-1500 Byte

16-1478 Byte

Datagrams

44-1498 Byte

16-1478 Byte

64-1522 Byte

UDP Header

dest. port 0x88A4

Destination Source

EtherType

0x88A4

FCSEtherCAT Header Datagrams

16 bit 44-1498 Byte

Basic EtherCAT frame

EtherCAT in UDP/IP frame

Basic EtherCAT frame with VLAN tag

EtherCAT in UDP/IP frame with VLAN tag

Basic EtherCAT frame

Ethernet frame

EtherCAT frame header

VLAN Tag

Field

Data Type

Value/Description

Length

11 bit

Length of the EtherCAT datagrams (excl. FCS)

Reserved

1 bit

Reserved, 0

Type

4 bit

Protocol type. Only EtherCAT commands (Type = 0x1) are supported by ESCs.

2 EtherCAT Protocol

EtherCAT uses standard IEEE 802.3 Ethernet frames, thus a standard network controller can be used and no special hardware is required on master side.

EtherCAT has a reserved EtherType of 0x88A4 that distinguishes it from other Ethernet frames. Thus, EtherCAT can run in parallel to other Ethernet protocols1.

EtherCAT does not require the IP protocol, however it can be encapsulated in IP/UDP. The EtherCAT Slave Controller processes the frame in hardware. Thus, communication performance is independent from processor power.

An EtherCAT frame is subdivided into the EtherCAT frame header followed by one or more EtherCAT datagrams. At least one EtherCAT datagram has to be in the frame. Only EtherCAT frames with Type 1 in the EtherCAT Header are currently processed by the ESCs. The ESCs also support IEEE802.1Q VLAN Tags, although the VLAN Tag contents are not evaluated by the ESC.

If the minimum Ethernet frame size requirement is not fulfilled, padding bytes have to be added. Otherwise the EtherCAT frame is exactly as large as the sum of all EtherCAT datagrams plus EtherCAT frame header.

2.1 EtherCAT Header

Figure 2 shows how an Ethernet frame containing EtherCAT data is assembled.

NOTE: The EtherCAT header length field is ignored by ESCs, they rely on the datagram length fields.

ESCs have to be configured to forward non-EtherCAT frames via DL Control register 0x0100[0].

I-4 Slave Controller – Technology

Figure 2: Ethernet Frame with EtherCAT Data

Table 2: EtherCAT Frame Header

EtherCAT Protocol

Datagram Header

FCS

Data

4 Byte

Position Addressing

1...n Datagrams

44*-1498 Byte

Ethernet header

Length

Res. Type

11 bit 1 bit 4 bit14 Byte

1st EtherCAT Datagram 2

... n

EtherCAT Datagram...

Working Counter

(WKC)

Cmd Idx

Position Offset

Address Offset

Logical Address

Address Len R C M IRQ

10 Byte 0-1486 Byte 2 Byte

8 Bit 8 Bit 32 Bit 11 Bit 16 Bit3 Bit 1 1

16 Bit 16 Bit

Node Addressing

Logical Addressing

* add 1-32 padding bytes if Ethernet frame is shorter than

64 Bytes (Ethernet Header+Ethernet Data+FCS)

Ethernet Data

EtherCAT header

168 6248 590 63 64 79

More EtherCAT Datagrams

2.2 EtherCAT Datagram

Figure 3 shows the structure of an EtherCAT frame.

Figure 3: EtherCAT Datagram

Slave Controller – Technology I-5

EtherCAT Protocol

Field

Data Type

Value/Description

Cmd

BYTE

EtherCAT Command Type (see 2.5)

Idx

BYTE

The index is a numeric identifier used by the master for identification of duplicates/lost datagrams. It shall not be changed by EtherCAT slaves

Address

BYTE[4]

Address (Auto Increment, Configured Station Address, or Logical Address, see 2.3)

Len

11 bit

Length of the following data within this datagram

3 bit

Reserved, 0

1 bit

Circulating frame (see 3.5): 0: Frame is not circulating 1: Frame has circulated once

1 bit

More EtherCAT datagrams 0: Last EtherCAT datagram 1: More EtherCAT datagrams will follow

IRQ

WORD

EtherCAT Event Request registers of all slaves combined with a logical OR

Data

BYTE[n]

Read/Write Data

WKC

WORD

Working Counter (see 2.4)

Mode

Field

Data Type

Value/Description

Auto Increment

Address

Position

WORD

Each slave increments Position. Slave is addressed if Position = 0.

Offset

WORD

Local register or memory address of the ESC

Configured

Station Address

Address

WORD

Slave is addressed if Address matches Configured Station Address or Configured Station Alias (if enabled).

Offset

WORD

Local register or memory address of the ESC

Broadcast

Position

WORD

Each slave increments Position (not used for addressing)

Offset

WORD

Local register or memory address of the ESC

Logical Address

Address

DWORD

Logical Address (configured by FMMUs) Slave is addressed if FMMU configuration matches Address.

Table 3: EtherCAT Datagram

2.3 EtherCAT Addressing Modes

Two addressing modes of EtherCAT devices are supported within one segment: device addressing and logical addressing. Three device addressing modes are available: auto increment addressing, configured station address, and broadcast. EtherCAT devices can have up to two configured station addresses, one is assigned by the master (Configured Station Address), the other one is stored in the SII EEPROM and can be changed by the slave application (Configured Station Alias address). The EEPROM setting for the Configured Station Alias address is only taken over at the first EEPROM loading after power-on or reset.

Table 4: EtherCAT Addressing Modes

I-6 Slave Controller – Technology

EtherCAT Protocol

2.3.1 Device Addressing

The device can be addressed via Device Position Address (Auto Increment address), by Node Address (Configured Station Address/Configured Station Alias), or by a Broadcast.

 Position Address / Auto Increment Address:

The datagram holds the position address of the addressed slave as a negative value. Each slave increments the address. The slave which reads the address equal zero is addressed and will execute the appropriate command at receive. Position Addressing should only be used during start-up of the EtherCAT system to scan the fieldbus and later only occasionally to detect newly attached slaves. Using Position addressing is problematic if loops are closed temporarily due to hot connecting or link problems. Position addresses are shifted in this case, and e.g., a mapping of error register values to devices becomes impossible, thus the faulty link cannot be localized.

 Node Address / Configured Station Address and Configured Station Alias:

The configured Station Address is assigned by the master during start up and cannot be changed by the EtherCAT slave. The Configured Station Alias address is stored in the SII EEPROM and can be changed by the EtherCAT slave. The Configured Station Alias has to be enabled by the master. The appropriate command action will be executed if Node Address matches with either Configured Station Address or Configured Station Alias. Node addressing is typically used for register access to individual and already identified devices.

 Broadcast:

Each EtherCAT slave is addressed. Broadcast addressing is used e.g. for initialization of all slaves and for checking the status of all slaves if they are expected to be identical.

Each slave device has a 16 bit local address space (address range 0x0000:0x0FFF is dedicated for EtherCAT registers, address range 0x1000:0xFFFF is used as process memory) which is addressed via the Offset field of the EtherCAT datagram. The process memory address space is used for application communication (e.g. mailbox access).

2.3.2 Logical Addressing

All devices read from and write to the same logical 4 Gbyte address space (32 bit address field within the EtherCAT datagram). A slave uses a mapping unit (FMMU, Fieldbus Memory Management Unit) to map data from the logical process data image to its local address space. During start up the master configures the FMMUs of each slave. The slave knows which parts of the logical process data image have to be mapped to which local address space using the configuration information of the FMMUs.

Logical Addressing supports bit wise mapping. Logical Addressing is a powerful mechanism to reduce the overhead of process data communication, thus it is typically used for accessing process data.

Slave Controller – Technology I-7

EtherCAT Protocol

Command

Data Type

Increment

Read command

No success

no change

Successful read

Write command

No success

no change

Successful write

ReadWrite command

No success

no change

Successful read

Successful write

Successful read and write

2.4 Working Counter

Every EtherCAT datagram ends with a 16 Bit Working Counter (WKC). The Working Counter counts the number of devices that were successfully addressed by this EtherCAT datagram. Successfully means that the ESC is addressed and the addressed memory is accessible (e.g., protected SyncManager buffer). EtherCAT Slave Controllers increment the Working Counter in hardware. Each datagram should have an expected Working Counter value calculated by the master. The master can check the valid processing of EtherCAT datagrams by comparing the Working Counter with the expected value.

The Working Counter is increased if at least one byte/one bit of the whole multi-byte datagram was successfully read and/or written. For a multi-byte datagram, you cannot tell from the Working Counter value if all or only one byte was successfully read and/or written. This allows reading separated register areas using a single datagram by ignoring unused bytes.

The Read-Multiple-Write commands ARMW and FRMW are either treated like a read command or like a write command, depending on the address match.

Table 5: Working Counter Increment

I-8 Slave Controller – Technology

EtherCAT Protocol

2.5 EtherCAT Command Types

All supported EtherCAT Command types are listed in Table 6. For ReadWrite operations, the Read operation is performed before the Write operation.

Slave Controller – Technology I-9

EtherCAT Protocol

CMD

Abbr.

Name

Description

NOP

No Operation

Slave ignores command

APRD

Auto Increment Read

Slave increments address. Slave puts read data into the EtherCAT datagram if received address is zero.

APWR

Auto Increment Write

Slave increments address. Slave writes data into memory location if received address is zero.

APRW

Auto Increment Read Write

Slave increments address. Slave puts read data into the EtherCAT datagram and writes the data into the same memory location if received address is zero.

FPRD

Configured Address Read

Slave puts read data into the EtherCAT datagram if address matches with one of its configured addresses

FPWR

Configured Address Write

Slave writes data into memory location if address matches with one of its configured addresses

FPRW

Configured Address Read Write

Slave puts read data into the EtherCAT datagram and writes data into the same memory location if address matches with one of its configured addresses

BRD

Broadcast Read

All slaves put logical OR of data of the memory area and data of the EtherCAT datagram into the EtherCAT datagram. All slaves increment position field.

BWR

Broadcast Write

All slaves write data into memory location. All slaves increment position field.

BRW

Broadcast Read Write

All slaves put logical OR of data of the memory area and data of the EtherCAT datagram into the EtherCAT datagram, and write data into memory location. BRW is typically not used. All slaves increment position field.

LRD

Logical Memory Read

Slave puts read data into the EtherCAT datagram if received address matches with one of the configured FMMU areas for reading.

LWR

Logical Memory Write

Slaves writes data to into memory location if received address matches with one of the configured FMMU areas for writing.

LRW

Logical Memory Read Write

Slave puts read data into the EtherCAT datagram if received address matches with one of the configured FMMU areas for reading. Slaves writes data to into memory location if received address matches with one of the configured FMMU areas for writing.

ARMW

Auto Increment Read Multiple Write

Slave increments address. Slave puts read data into the EtherCAT datagram if received address is zero, otherwise slave writes the data into memory location.

FRMW

Configured Read Multiple Write

Slave puts read data into the EtherCAT datagram if address matches with one of its configured addresses, otherwise slave writes the data into memory location.

15-255

reserved

Table 6: EtherCAT Command Types

I-10 Slave Controller – Technology

EtherCAT Protocol

CMD

High Addr. In

High Addr. Out

Low Addr.

Address Match

Data In

Data Out

WKC

NOP

untouched

none

untouched

APRD

Position

Pos.+1

Offset

ADP=0

Read

+0/1

APWR

Position

Pos.+1

Offset

ADP=0

Write

+0/1

APRW

Position

Pos.+1

Offset

ADP=0

Write

Read

+0/1/2/3

FPRD

Address

Offset

ADP=conf. station addr.

Read

+0/1

FPWR

Address

Offset

ADP=conf. station addr.

Write

+0/1

FPRW

Address

Offset

ADP=conf. station addr.

Write

Read

+0/1/2/3

BRD

High Add. In+1

Offset

all

Data In OR Read

+0/1

BWR

High Add. In+1

Offset

all

Write

+0/1

BRW

High Add. In+1

Offset

all

Write

Data In OR Read

+0/1/2/3

LRD

Logical address

FMMU

(Read AND bitmask1)

(Data In AND NOT bit_mask1)

+0/1

LWR

Logical address

FMMU

Write

+0/1

LRW

Logical address

FMMU

Write

(Read AND bit_mask1) OR (Data In AND NOT bit_mask1)

+0/1/2/3

ARMW

Position

Pos.+1

Offset Read: ADP=0

Read

+0/1

Write: ADP/=0

Write

+0/1

FRMW

Address

Offset

Read: ADP= conf. station address./alias

Read

+0/1

Write: ADP/= conf. station address./alias

Write

+0/1

Table 7: EtherCAT Command Details

NOTE: Working Counter (WKC) increment depends on address match

by the logical read/write command.

bit_mask depends on FMMU configuration if bit-wise mapping is used: only masked bits are actually addressed

Slave Controller – Technology I-11

Frame Processing

3 Frame Processing

The ET1100, ET1200, IP Core, and ESC20 slave controllers only support Direct Mode addressing: neither a MAC address nor an IP address is assigned to the ESC, they process EtherCAT frames with any MAC or IP address.

It is not possible to use unmanaged switches between these ESCs or between master and the first slave, because source and destination MAC addresses are not evaluated or exchanged by the ESCs. Only the source MAC address is modified when using the default settings, so outgoing and incoming frames can be distinguished by the master.

NOTE: Attaching an ESC directly to an office network will result in network flooding, since the ESC will reflect any frame – especially broadcast frames – back into the network (broadcast storm).

The frames are processed by the ESC on the fly, i.e., they are not stored inside the ESC. Data is read and written as the bits are passing the ESC. The forwarding delay is minimized to achieve fast cycle times. The forwarding delay is defined by the receive FIFO size and the EtherCAT Processing Unit delay. A transmit FIFO is omitted to reduce delay times.

The ESCs support EtherCAT, UDP/IP, and VLAN tags. EtherCAT frames and UDP/IP frames containing EtherCAT datagrams are processed. Frames with VLAN tags are processed by the ESCs, the VLAN settings are ignored and the VLAN tag is not modified.

The source MAC address is changed for every frame passing the EtherCAT Processing Unit (SOURCE_MAC[1] is set to 1 – locally administered address). This helps to distinguish between frames transmitted by the master and frames received by the master.

3.1 Loop Control and Loop State

Each port of an ESC can be in one of two states: open or closed. If a port is open, frames are transmitted to other ESCs at this port, and frames from other ESCs are received. A port which is closed will not exchange frames with other ESCs, instead, the frames are forwarded internally to the next logical port, until an open port is reached.

The loop state of each port can be controlled by the master (ESC DL Control register 0x0100). The ESCs supports four loop control settings, two manual configurations, and two automatic modes:

Manual open

The port is open regardless of the link state. If there is no link, outgoing frames will be lost.

Manual close

The port is closed regardless of the link state. No frames will be sent out or received at this port, even if there is a link with incoming frames.

Auto

The loop state of each port is determined by the link state of the port. The loop is open if there is a link, and it is closed without a link.

I-12 Slave Controller – Technology

Frame Processing

open

closed

wait

link lost

a) master writes loop configuration again to AL Control register (via another open port)

b) a valid Ethernet frame is received at this port

link lost

link establishd

Name

Description

0x0100[15:8]

ESC DL Control

Loop control/loop setting

0x0110[15:4]

ESC DL Status

Loop and link status

0x0518:0x051B

PHY Port status

PHY Management link status

Auto close (manual open)

The port is closed depending on the link state, i.e., if the link is lost, the loop will be closed (auto close). If the link is established, the loop will not be automatically opened, instead, it will remain closed (closed wait state). Typically, the port has to be opened by the master explicitly by writing the loop configuration again to the ESC DL Control register 0x0100. This write access has to enter the ESC via a different open port. There is an additional fallback option for opening the port: if a valid Ethernet frame is received at the closed port in Auto close mode, it will also be opened after the CRC is received correctly, without explicit master interaction.

Figure 4: Auto close loop state transitions

A port is considered open if the port is available, i.e., it is enabled in the configuration, and one of the following conditions is met:

 The loop setting in the DL Control register is Auto and there is an active link at the port.  The loop setting in the DL Control register is Auto close and there is an active link at the port and

the DL Control register was written again after the link was established.

 The loop setting in the DL Control register is Auto close and there is an active link at the port and

a valid frame was received at this port after the link was established.

 The loop setting in the DL control register is Always open A port is considered closed if one of the following conditions is met:

 The port is not available or not enabled in the configuration.  The loop setting in the DL Control register is Auto and there is no active link at the port.  The loop setting in the DL Control register is Auto close and there is no active link at the port or

the DL Control register was not written again after the link was established

 The loop setting in the DL Control register is Always closed

NOTE: If all ports are closed (either manually or automatically), port 0 will be opened as the recovery port. Reading and writing via this port is possible, although the DL status register reflects the correct status. This can be used to correct DL control register settings.

Registers used for loop control and loop/link status are listed in Table 8.

Slave Controller – Technology I-13

Table 8: Registers for Loop Control and Loop/Link Status

Frame Processing

Number of Ports

Frame processing order

0→EtherCAT Processing Unit→0

0→EtherCAT Processing Unit→1 / 1→0

0→EtherCAT Processing Unit→1 / 1→2 / 2→0 (log. ports 0,1, and 2) or 0→EtherCAT Processing Unit→3 / 3→1 / 1→0 (log. ports 0,1, and 3)

0→EtherCAT Processing Unit→3 / 3→1 / 1→2 / 2→0

Port 2

Auto-

Forwarder

Port 1

Auto-

Forwarder

Port 3

Auto-

Forwarder

Port 0

Auto-

Forwarder

Loopback function

EtherCAT

Processing Unit

Loopback function

EtherCAT

Slave Controller

port 2 closed

port 2 open

port 1 closed

port 1 open

port 3 open

port 3 closed

port 0 open

or all ports

closed

port 0 closed

3.2 Frame Processing Order

The frame processing order of EtherCAT Slave Controllers depends on the number of ports (logical port numbers are used):

Table 9: Frame Processing Order

The direction through an ESC including the EtherCAT Processing Unit is called “processing” direction,

other directions without passing the EtherCAT Processing Unit are called “forwarding” direction. Ports which are not implemented behave similar to closed ports, the frame is forwarded to the next

port. Figure 5 shows the frame processing in general:

Figure 5: Frame Processing

I-14 Slave Controller – Technology

Frame Processing

EtherCAT

master

EtherCAT

slave 1

Port 0

Port 1

EtherCAT

slave 2

Port 0

Port 1

EtherCAT

slave 3

Port 0

EtherCAT

slave N

Port 0

Port 1

Example Port Configuration with Ports 0, 1, and 2

If there are only ports 0, 1, and 2, a frame received at port 0 goes via the Auto-Forwarder and the Loopback function to the EtherCAT Processing Unit which processes it. Then, the frame is sent to logical port 3 which is not configured, so the Loopback function of port 3 forwards it to port 1. If port 1 is closed, the frame is forwarded by the Loopback function to port 2. If port 1 is open, the frame is sent out at port 1. When the frame comes back into port 1, it is handled by the Auto-Forwarder and sent to port 2. Again, if port 2 is closed, the frame is forwarded to port 0, otherwise, it is sent out at port 2. When the frame comes back into port 2, it is handled by the Auto-Forwarder and then sent to the Loopback function of port 0. Then it is handled by the Loopback function and sent out at port 0 – back to the master.

3.3 Permanent Ports and Bridge Port

The EtherCAT ports of an ESC are typically permanent ports, which are directly available after PowerOn. Permanent ports are initially configured for Auto mode, i.e., they are opened after the link is established. Additionally, some ESCs support an EtherCAT Bridge port (port 3), which is configured in the SII EEPROM like PDI interfaces. This Bridge port becomes available if the EEPROM is loaded successfully, and it is closed initially, i.e., it has to be opened (or set to Auto mode) explicitly by the EtherCAT master.

3.4 Shadow Buffer for Register Write Operations

The ESCs have shadow buffers for write operations to registers (0x0000 to 0x0F7F). During a frame, write data is stored in the shadow buffers. If the frame is received correctly, the values of the shadow buffers are transferred into the effective registers. Otherwise, the values of the shadow buffers are not taken over. As a consequence of this behavior, registers take their new value shortly after the FCS of an EtherCAT frame is received. SyncManagers also change the buffers after the frame was received correctly.

User and Process Memory do not have shadow buffers. Accesses to these areas are taking effect directly. If a SyncManager is configured to User Memory or Process Memory, write data will be placed in the memory, but the buffer will not change in case of an error.

3.5 Circulating Frames

The ESCs incorporate a mechanism for prevention of circulating frames. This mechanism is very important for proper watchdog functionality.

This is an example network with a link failure between slave 1 and slave 2:

Figure 6: Circulating Frames

Both slave 1 and slave 2 detect the link failure and close their ports (port 1 at slave 1 and port 0 at slave 2). A frame currently traveling through the ring at the right side of slave 2 might start circulating. If such a frame contains output data, it might trigger the built-in watchdog of the ESCs, so the watchdog never expires, although the EtherCAT master cannot update the outputs anymore.

To prevent this, a slave with no link at port 0 and loop control for port 0 set to Auto or Auto close (ESC DL Control register 0x0100) will do the following inside the EtherCAT Processing Unit:

 If the Circulating bit of the EtherCAT datagram is 0, set the Circulating bit to 1  If the Circulating bit is 1, do not process the frame and destroy it

The result is that circulating frames are detected and destroyed. Since the ESCs do not store the frames for processing, a fragment of the frame will still circulate triggering the Link/Activity LEDs. Nevertheless, the fragment is not processed.

Slave Controller – Technology I-15

Frame Processing

EtherCAT

master

EtherCAT

slave 1

Port 0

Port 1

EtherCAT

slave 2

Port 0Port 0

EtherCAT

slave 3

Port 0

Port 1

EtherCAT

slave 4

Port 0Port 0

Port 3

Port 2

Port 3

3.5.1 Unconnected Port 0

Port 0 must not be left intentionally unconnected (slave hardware or topology) because of the circulating frame prevention. All frames will be dropped after they have passed an automatically closed Port 0 for the second time, and this can prohibit any EtherCAT communication.

Example: Port 0 of slave 1 and 3 are automatically closed because nothing is connected. The Circulating bit of each frame is set at slave 3. Slave 1 detects this and destroys the frames.

Figure 7: All frames are dropped because of Circulating Frame Prevention

In redundancy operation, only one Port 0 is automatically closed, so the communication remains active.

3.6 Non-EtherCAT Protocols

If non-EtherCAT protocols are used, the forwarding rule in the ESC DL Control register (0x0100[0]) has to be set to forward non-EtherCAT protocols. Otherwise they are destroyed by the ESC.

3.7 Special Functions of Port 0

Port 0 of each EtherCAT is characterized by some special functions in contrast to ports 1, 2, and 3:  Port 0 leads to the master, i.e., port 0 is the upstream port, all other ports (1-3) are downstream

ports (unless an error has occurred and the network is in redundancy mode).

 The link state of Port 0 influences the Circulating Frame bit, and frames are dropped at port 0 if

the bit is set and the link is automatically closed.

 Port 0 loop state is open if all ports are closed (either automatically or manually).  Port 0 has a special behavior when using standard EBUS link detection.

I-16 Slave Controller – Technology

Physical Layer Common Features

Status register

MII

EBUS

LINK_MII signal

Management Interface

Standard link detection

Enhanced link detection

Standard link detection

Enhanced link detection

Standard link detection

Enhanced link detection

ESC DL Status: Physical link 0x0110[7:4]

LINK_MII signal state

LINK_MII signal combined with

MI Link Detection and

Configuration result

Result of the standard link

detection mechanism

ESC DL Status: Communication established 0x0110[15,13,11,9]

LINK_MII signal state

LINK_MII signal state combined with RX_ERR threshold state

LINK_MII signal combined with MI Link Detection and Configuration result

LINK_MII signal combined with RX_ERR threshold state and MI Link Detection and Configuration result

Result of the standard link

detection

mechanism.

Result of the

enhanced link

detection

mechanism. PHY port status: physical link status 0x0518[0], 0x0519[0], 0x051A[0], 0x051B[0]

n.a.

PHY has detected link

(PHY Status register 1[2])

n.a.

PHY port status: Link status 0x0518[1], 0x0519[1], 0x051A[1], 0x051B[1]

PHY has detected link,

link is suitable for ECAT

4 Physical Layer Common Features

EtherCAT supports two types of Physical Layers, Ethernet and EBUS. The Ethernet interface of ESCs is MII, RGMII, or RMII, connecting to an external Ethernet PHY according to IEEE 802.3 100BaseTX or FX. For EBUS, the physical layer is integrated into the EtherCAT ASICs. EtherCAT requires 100 Mbit/s links with full duplex communication.

The MII interface of Beckhoff ESCs is optimized e.g. for low processing/forwarding delay. The resulting additional requirements to Ethernet PHYs are described in the corresponding chapters.

4.1 Link Status

The link status of each port is available in the ESC DL Status register (0x0110:0x0111), most important are the “Communication established” bits 15, 13, 11, and 9. Additional link information is available in the PHY Port status register (0x0518:0x051B) if MI link detection and configuration is used. All other status bits are mainly for debugging purposes. The link status bits are described in the following table.

Table 10: Link Status Description

If all ports are closed (either manually or automatically, e.g., because no port has a communication link), port 0 is automatically opened as the recovery port. Reading and writing via this port is possible, although the DL status register reflects the correct status. This can be used to correct erroneous DL control register settings or to fix LINK_MII polarity configuration.

Slave Controller – Technology I-17

Physical Layer Common Features

Name

Description

0x0141[1]

ESC Configuration

Enable/disable Enhanced link detection for all ports

0x0141[7:4]

ESC Configuration

Enable/disable Enhanced link detection portwise

0x0110[2]

ESC DL Status

Enhanced link detection status

4.2 Selecting Standard/Enhanced Link Detection

Some ESCs distinguish between standard and enhanced link detection. Enhanced link detection provides additional security mechanisms regarding link establishment and surveillance. Using enhanced link detection is recommended for Ethernet PHY ports (refer to chapter 6.5 for compatibility issues with EBUS enhanced link detection). Some ESCs only support global Enhanced Link Detection configuration for all ports, some support port-wise configuration.

After power-on, enhanced link detection is enabled by default. It is disabled or remains enabled after the SII EEPROM is loaded according to the EEPROM setting (register 0x0141). An invalid EEPROM content will also disable enhanced link detection.

The EEPROM setting for enhanced Link detection is only taken over at the first EEPROM loading after power-on or reset. Changing the EEPROM and manually reloading it will not affect the enhanced link detection enable status (register 0x0110[2]), even if the EEPROM could not be read initially.

Registers used for Enhanced link detection are listed in Table 11.

Table 11: Registers for Enhanced Link Detection

NOTE: Some of these register bits are set via SII EEPROM/IP Core configuration. Some of the registers are not available in specific ESCs. Refer to Section II and III for details.

I-18 Slave Controller – Technology

Physical Layer Common Features

Name

Description

0x0100[18:16]

ESC DL Control

Current FIFO size setting

4.3 FIFO Size Reduction

The ESCs incorporate a receive FIFO (RX FIFO) for decoupling receive clock and processing clock. The FIFO size is programmable by the EtherCAT master (ESC DL Control register 0x0100). Some ESCs support a default value for the FIFO size loaded from the SII EEPROM.

The FIFO size values determine a reduction of the FIFO size, the FIFO cannot be disabled completely. The FIFO size can be reduced considering these three factors:

 Accuracy of the receiver’s clock source  Accuracy of the sender’s clock source  Maximum frame size

The default FIFO size is sufficient for maximum Ethernet Frames and default Ethernet clock source accuracy (100 ppm). If the clock accuracy is 25 ppm or better, the FIFO size can be reduced to the minimum. If the FIFO size was accidentally reduced too much, a short 64 Byte frame should be sent for resetting the FIFO size to the default value, since a smaller frame is not utilizing the FIFO as much as a larger frame.

The FIFO size can be reduced to minimum if both sender and receiver have 25 ppm accuracy of their clock sources, even with maximum frame size.

Since 25 ppm clock accuracy can typically not be guaranteed for the entire life-time of a clock source, the actual clock deviation has to be measured on a regular basis for FIFO size reduction. If a slave does not support Distributed Clocks or the actual deviation is larger than 25 ppm, the FIFO size of all neighbors and the slave itself cannot be reduced. The actual deviation can be measured using Distributed Clocks:

 Compare DC Receive Times over a period of time for slaves which only support DC Receive

Times. Do not use this method if both slaves which are compared support DC Time Loop, since the measured deviation will approximate zero if the DC control loop has settled, but the actual deviation determining the FIFO size might be larger than 25 ppm.

 Compare calculated deviation from register Speed Counter Diff (0x0932:0x0933) for adjacent

slaves with DC Time Loop support after the DC control loop has settled (i.e., System Time Difference 0x092C:0x092F is at its minimum).

NOTE: Be careful with FIFO size reduction at the first slave if off-the-shelf network interface cards without 25 ppm accuracy are used by master.

Table 12: Registers for FIFO Size Reduction

NOTE: Some of these register bits are set via SII EEPROM. Some of the registers are not available in specific ESCs. Refer to Section II and III for details.

4.4 Frame Error Detection

Refer to chapter 14 (Error Counters) for details on frame error detection.

Slave Controller – Technology I-19

Ethernet Physical Layer

5 Ethernet Physical Layer

ESCs with Ethernet Physical Layer support use the MII interface, some do also support the RMII or RGMII interface. Since RMII/RGMII PHYs include FIFOs, they increase the forwarding delay of an EtherCAT slave device as well as the jitter. MII is recommended due to these reasons.

5.1 Requirements to Ethernet PHYs

EtherCAT and Beckhoff ESCs have some general requirements to Ethernet PHYs, which are typically fulfilled by state-of-the-art Ethernet PHYs.

The MII interfaces of Beckhoff ESCs are optimized for low processing/forwarding delays by omitting a transmit FIFO. To allow this, the Beckhoff ESCs have additional requirements to Ethernet PHYs, which are easily accomplished by several PHY vendors.

Refer to the EtherCAT Slave Controller application note “PHY Selection Guide” for Ethernet PHY

requirements and example Ethernet PHYs. Refer to Section III for ESC specific information about supported features.

5.2 PHY reset and Link partner notification/loop closing

The main principle of EtherCAT operation in case of link errors is disabling unreliable links by closing loops. This is automatically performed by the ESCs. The ESCs rely on the LINK_MII signal from the PHYs for detecting the link state.

It is crucial that a PHY does not establish a link while the ESC is not operating. Otherwise, the communication partner would also detect a physical link, causing him to open the communication link. Subsequently, all frames will get lost because the ESC is not operating.

So at least the following requirements have to be fulfilled, otherwise frames will be lost:  ESC in reset state → PHY disabled The recommended solution for this issue is to enable the PHY together with the ESC by using the

ESC’s reset signal for the PHY, too. If the ESC has individual PHY reset outputs, they should be used instead. This solution prevents the PHYs from establishing a link while the ESCs are not operating.

I-20 Slave Controller – Technology

Ethernet Physical Layer

Signal

Direction at PHY

Description

TX_CLK

OUT

ESC20:

TX_CLK of one PHY is used as clock source, TX_CLK of other

PHY is unused, leave open. ESC dependent (e.g., IP Core): TX_CLK is optionally used for automatic TX Shift compensation. Other Beckhoff ESCs:

Unused, leave unconnected.

COL

OUT

Collision detected. ESC20:

Connected, but not used. Other Beckhoff ESCs:

Unused. Leave unconnected.

CRS

OUT

Carrier sense. ESC20:

Connected, but not used. Other Beckhoff ESCs:

Unused. Leave unconnected

TX_ER

Transmit error. ESC20:

Connected, always driven low. Other Beckhoff ESCs:

Connect to GND.

5.3 MII Interface

Refer to Section III for ESC specific MII information. If an ESC MII interface is not used, LINK_MII has to be tied to a logic value which indicates no link,

and RX_CLK, RXD, RX_ER, and especially RX_DV have to be tied to GND. The TX outputs can be left unconnected, unless they are used for ESC configuration.

Table 13: Special/Unused MII Interface signals

For more details about the MII interface, refer to IEEE Standard 802.3 (Clause 22), available from the IEEE.

Slave Controller – Technology I-21

Ethernet Physical Layer

Name

Description

0x0110:0x0111

ESC DL Status

Link Status (Link MII, Communication established)

0x0518:0x051B

PHY Port Status

MI Link Detection results if available

5.4 RMII Interface

Refer to Section III for ESC specific RMII information. If an ESC RMII interface is not used, LINK_MII has to be tied to a logic value which indicates no link,

and RXD, RX_ER, and especially CRS_DV have to be tied to GND. The TX signals can be left unconnected, unless they are used for ESC configuration.

For more details about the RMII interface, refer to the RMII Specification, available from the RMII consortium.

5.5 RGMII Interface

Refer to Section III for ESC specific RGMII information. If an ESC RGMII interface is not used, LINK_MII has to be tied to a logic value which indicates no link,

and RX_CLK, RX_CTL, and RXD have to be tied to GND. The TX signals can be left unconnected, unless they are used for ESC configuration.

For more details about the RGMII interface, refer to the “Reduced Gigabit Media Independent Interface (RGMII)” specification version 2.0.

5.5.1 RGMII In-Band Link Status

Some ESCs support RGMII PHYs. With these PHYs, it is possible to use the RGMII In-Band Status instead of the LINK_MII signal. In this case, the LINK_MII input of the ESC has to be tied to a value which indicates “no link”. The RGMII In-Band status is checked to indicate a 100 Mbit/s Full-Duplex link. The In-Band status is equivalent to the LINK_MII signal.

NOTE: Some PHYs require that RGMII In-Band Status is enabled by writing to the PHY management registers. This cannot be done by the ESC itself.

5.6 Link Detection

All ESCs support a LINK_MII signal for fast link detection at each Ethernet MII port. Some ESCs (e.g., EtherCAT IP Core) additionally support link detection and configuration via the MII management interface. Both the LINK_MII signals and the MI Link Detection and Configuration results (if available) are combined to determine the link state of each port, which is reflected in the ESC DL Status register (0x0110[15,13,11,9] – Communication established). Using the LINK_MII signal is mandatory since it is the only way to achieve fast link loss reaction times.

Table 14: Registers used for Ethernet Link Detection

5.6.1 LINK_MII Signal

The LINK_MII signal used for link detection is typically an LED output signal of the Ethernet PHY. If available, LINK_MII should be connected to a combined signal indicating a 100 Mbit/s Full Duplex link. If such a signal is not available, a signal indicating a 100 Mbit/s link (speed LED) might be used. If only a Link signal is available (link LED), this might be used. Never use (combined) activity signals, e.g., Link/Act LED outputs, because the link state will toggle upon activity.

The main advantage of using a dedicated link signal instead of reading out MII management interface registers is the fast reaction time in case of a link loss. This is crucial for redundancy operation, since only one lost frame is tolerated. The EtherCAT port of an ESC which loses a link has to be closed as fast as possible to maintain EtherCAT communication at the other ports and to reduce the number of lost frames.

The LINK_MII signal state is reflected in the ESC DL Status register (0x0110[7:4]).

I-22 Slave Controller – Technology

Ethernet Physical Layer

5.6.2 MI Link Detection and Configuration

The EtherCAT IP Core supports link detection and PHY configuration by using the MII management interface. Initially, the PHY configuration is checked and updated if necessary. Afterwards, the link status of each Ethernet port is cyclically polled. PHY accesses of the EtherCAT master are inferred upon request.

The MI Link Configuration mechanism configures the Ethernet PHYs to use Auto negotiation and advertise only 100BASE-TX Full-Duplex connections. Some ESCs support using Gigabit Ethernet PHYs, which are restricted to 100 Mbit/s links by the MI Link Configuration. ESCs which support FX operation natively are configuring the FX ports to use fixed 100BASE-TX Full-Duplex connections without Auto negotiation.

Depending on the physical layer configuration and the supported ESC features (e.g., TX vs. FX), the MI Link Detection will check if the link characteristics fulfill EtherCAT requirements (Auto negotiation, Speed, Duplex, etc.). If all conditions are met, an MI Link is detected.

Since the MI Link Detection does not solely rely on the PHY link status bit (register 1[2]), the local PHY and the remote PHY may indicate a link, but the ESC refuses it because it does not fulfill EtherCAT requirements. The current MI Link Detection state is reflected in the MI Interface registers (PHY Port Status 0x0518:0x051B).

MI Link Detection and Configuration must not be used without link detection via LINK_MII signals, because link loss reaction time would otherwise be to slow for redundancy operation. Enhanced Link Detection might not be a suitable solution in this case if too few RX_ERR are issued to the ESC before the PHY takes down the link.

NOTE: For testing it is possible to use MI Link Detection without LINK_MII signals: If the LINK_MII signals are statically set to “no link”, only MI Link Detection is used.

The MI Link Detection and Configuration checks the management communication with Ethernet PHYs. If communication is not possible – e.g. because no PHY is configured for the expected PHY address – the results are ignored. Take care of proper PHY address configuration to prevent erroneous behavior.

NOTE: Proper PHY address settings and PHY address offset configuration is crucial for MI Link Detection and Configuration.

5.7 Standard and Enhanced MII Link Detection

For Ethernet, the standard or enhanced MII link detection feature is a feature of link error detection and reaction. This has to be distinguished from the actual link detection, which tells the ESC if a physical link is available (i.e., the LINK_MII signal or the MI link detection and configuration mechanism).

Enhanced MII link detection, in contrast to standard MII link detection, will additionally disconnect a link if at least 32 RX errors (RX_ER) occur in a fixed interval of time (~10 µs). The local loop is closed and the link partner is informed by restarting the Auto-Negotiation mechanism via the MII Management Interface. This informs the link partner of the error condition, and the link partner will close the loop.

The ESC keeps the port closed until the link goes down during Auto-Negotiation and comes up again (the port remains closed if the link does not go down).

The availability of Enhanced MII Link Detection depends on a supported PHY address configuration, otherwise it has to be disabled.

Slave Controller – Technology I-23

Ethernet Physical Layer

5.8 EtherCAT over Optical Links (FX)

EtherCAT communication over optical links using Ethernet PHYs is possible, but some requirements of EtherCAT have to be respected, and some characteristics of EtherCAT slave controllers have to be considered.

Some ESCs are prepared for FX operation, others require external logic to achieve EtherCAT compatibility.

Refer to the EtherCAT Slave Controller application note “PHY Selection Guide” for details on FX PHY connection and example schematics. The application note is available at the Beckhoff website (http://www.beckhoff.com).

5.8.1 Link partner notification and loop closing

With FX connections, it could happen that the Transceiver device is powered, while the PHY (and/or the ESC) is not active. The communication partner would detect a signal, causing him to open the link. All frames will get lost because the PHY (and/or the ESC) is not operating.

So at least the following two requirements have to be fulfilled, otherwise frames will be lost:

 ESC in reset state → Transceiver disabled  PHY in reset state → Transceiver disabled

The recommended solution for this issue is to enable the transceiver together with the PHY by using the PHY’s reset signal for the transceiver, too. If the transceiver has no suitable input, the power supply of the transceiver can be switched off. Since the PHY’s reset should be controlled by the ESC’s reset output (either main reset output or individual PHY reset output), the transceiver will power down while the PHY is in the reset state and also while the ESC is in the reset state. Thus, the ESC and the PHY will be active when the transceiver gets active, and no frames are lost.

5.8.2 Far-End-Fault (FEF)

Some FX PHYs offer a Far-End-Fault (FEF) generation/detection feature. The intention is to inform the link partner of a bad link.

FEF Generation

If an FEF-supporting PHY receives a signal with a quality which is not sufficient, the PHY will transmit a special FEF pattern to the link partner.

FEF Detection

If an FEF-supporting PHY receives the FEF pattern with good signal quality, it will continue transmitting regularly, but it will indicate “no link” locally to the ESC, until the FEF pattern ends.

Conclusion

The FEF feature is advantageous for EtherCAT, because the PHYs will only indicate a link when the signal quality is high enough. Without FEF, the EtherCAT slave controllers have to rely on the Enhanced Link detection feature for detecting a low quality link. Nevertheless, Enhanced Link detection becomes active only after the link is already established, thus, in case of a low quality link, the link status will be toggling on/off (link up → Enhanced link detection

tears down link → link up …). This is sufficient to locate an issue in the network, but it might disturb

operation of the remaining network. So, it is highly recommended to use PHYs which fully implement FEF generation and detection.

NOTE: Some PHYs are claiming FEF support, but they are either not supporting FEF generation or detection, or they require configuration commands via MI management interface, which cannot be issued by the ESCs automatically.

I-24 Slave Controller – Technology

Ethernet Physical Layer

5.8.3 ESCs with native FX support

ESCs with native FX support have individual PHY reset outputs for each port. This PHY reset output is intended to hold the PHY and the transceiver in reset state while the ESC is in reset state, and additionally, to issue a reset cycle when a link failure is detected by the enhanced link detection mechanism.

If at least one port is configured for FX operation, all ports have to use the individual PHY reset outputs. This is especially important for enhanced link detection, since all the PHY reset outputs are used for link down signaling instead of auto-negotiation restart, which is not used anymore – regardless of the port using FX or TX.

5.8.4 ESCs without native FX support

ESCs without native FX support require special attention regarding reset connection, link down signaling and enhanced link detection. External logic might be required.

5.9 Gigabit Ethernet PHYs

Gigabit Ethernet PHYs can generally be used for EtherCAT, as long as the link speed is restricted to 100 Mbit/s, either by strapping options of the PHY or by using the auto negotiation advertisement.

Some ESCs are capable of restricting the auto negotiation advertisement of Gigabit Ethernet PHYs to 100 Mbit/s full-duplex if MI link detection and configuration is enabled.

Nevertheless, all other requirements of EtherCAT have to be fulfilled – especially the link loss reaction time (Enhanced Link Detection might be required).

Slave Controller – Technology I-25

Ethernet Physical Layer

5.10 MII Management Interface (MI)

Most EtherCAT slave controllers with MII/RMII/RGMII ports use the MII management interface for communication with the Ethernet PHYs. The MII management interface can be used by the EtherCAT master – or the local µController if supported by the ESC. Enhanced MII link detection uses the management interface for configuration and restarting auto negotiation after communication errors occurred (TX PHYs only). Some ESCs can make use of the MII management interface for link detection and PHY configuration. For fast link detection, the ESCs require to use a separate signal (LINK_MII).

Refer to chapter 5.6 for details about link detection with Ethernet PHYs. For more details about the MII management interface, refer to IEEE Standard 802.3 (Clause 22), available from the IEEE.

The ESCs support a shared MII Management Interface for all PHYs, i.e., MCLK and MDIO are connected to the ESC and to all PHYs.

5.10.1 PHY Addressing/PHY Address Offset

Proper PHY address configuration is crucial for Enhanced Link Detection and MI Link Detection and configuration, because the ESC itself needs to relate logical ports to the corresponding PHY addresses. The EtherCAT master can access the Ethernet PHYs using any address by means of the PHY address register 0x0513.

Typically, the logical port numbers match with the PHY addresses, i.e. the EtherCAT master and the ESC itself use PHY address 0 for accessing the PHY at port 0 (PHY address 1 for the PHY at port 1 and so on).

Depending on the ESC, there are two options for configuring the PHY addresses:  PHY address offset:

The ESC accesses a PHY at logical port x with the address “x + PHY address offset”. Depending on the ESC only some or all possible offsets are configurable. The EtherCAT master uses the PHY addresses 0-3 to access the logical ports 0-3. These addresses are incremented by the PHY address offset inside the ESC. If the master uses PHY addresses 4-31, these addresses are also incremented by the PHY address offset inside the ESC.

 Individual PHY address:

The ESC accesses a PHY at each logical port with an individual PHY address. The EtherCAT master uses the PHY addresses 0-3 to access the logical ports 0-3. These addresses are replaced with the individual addresses inside the ESC. If the master uses PHY addresses 4-31, these addresses are not translated.

Depending on the ESC the PHY address configuration is a strapped at power-up, configured in advance (IP Core) or configured by signals.

Typically, the PHY address offset should be 0, and the logical port numbers match with the PHY addresses. Some Ethernet PHYs associate a special function with PHY address 0, e.g., address 0 is a broadcast PHY address. In these cases, PHY address 0 cannot be used. Instead, a PHY address offset different from 0 should be selected, preferably an offset which is supported by the ESC. If PHY addresses are chosen which are not supported by the ESC, Enhanced Link Detection and MI Link Detection and Configuration cannot be used and have to be disabled (the PHY address offset should be 0 in these cases). Nevertheless, the EtherCAT master can communicate with the PHYs using the actual PHY addresses, and EtherCAT communication is possible anyway – using the LINK_MII signal. It is recommended that the PHY addresses are selected to be equal to the logical port number plus 1 in this case. If port 0 is EBUS, ports 1-3 should have PHY addresses 1-3, i.e., PHY address offset is 0.

If the PHY address offset configuration of an ESC reflects the actual PHY address settings, the EtherCAT master can use addresses 0-3 in PHY address register 0x0513 for accessing the PHYs of logical ports 0-3, regardless of the PHY address offset.

I-26 Slave Controller – Technology

Ethernet Physical Layer

Logical Port

Configured address of the PHY

PHY address register value

used by EtherCAT master

PHY address

offset = 0

PHY address

offset = 16

0 0 16 0 1 1 17 1 2 2 18 2 3 3 19

none

4-15

20-31

4-15

none

16-31

0-15

16-31

Logical Port

Configured

address of the

PHY

PHY address register value

used by EtherCAT master

1 1 1

3 3 3

none

0, 5-31

0, 5-13

Table 15: PHY Address configuration matches PHY address settings

If the actual PHY address settings differ from the PHY address configuration of the ESC, the EtherCAT master has to use the actual PHY address mapping, i.e., PHY addresses 1-4 for accessing the PHYs of logical ports 0-3.

Table 16: PHY Address configuration does not match actual PHY address settings

NOTE: PHY address offset is 0 in this case (recommended).

Slave Controller – Technology I-27

Ethernet Physical Layer

Description

0x0510:0x0511

MII Management Control/Status

0x0512

PHY Address

0x0513

PHY Register Address

0x0514:0x0515

PHY Data

5.10.2 Logical Interface

The MI of the ESC is typically controlled by EtherCAT via the MI registers1.

Table 17: MII Management Interface Register Overview

The MI supports two commands: write to one PHY register or read one PHY register.

5.10.2.1 MI read/write example

The following steps have to be performed for a PHY register access:

1. Check if the Busy bit of the MI Status register is cleared and the MI is not busy.

2. Write PHY address to PHY Address register.

3. Write PHY register number to be accessed into PHY Register Address register (0-31).

4. Write command only: put write data into PHY Data register (1 word/2 byte).

5. Issue command by writing to Control register. For read commands, write 1 into Command Register Read 0x0510[8]. For write commands, write 1 into Write Enable bit 0x0510[0] and also 1 into Command Register Write 0x0510[9]. Both bits have to be written in one frame. The Write enable bit realizes a write protection mechanism. It is valid for subsequent MI commands issued in the same frame and selfclearing afterwards.

6. The command is executed after the EOF, if the EtherCAT frame had no errors.

7. Wait until the Busy bit of the MI Status register is cleared.

8. Check the Error bits of the MI Status register. The command error bit is cleared with a valid command or by clearing the command register. The read error bit indicates a read error, e.g., a wrong PHY address. It is cleared by writing to the register.

9. Read command only: Read data is available in PHY Data register.

NOTE: The Command register bits are self-clearing. Manually clearing the command register will also clear the status information.

5.10.2.2 MI Interface Assignment to ECAT/PDI

The EtherCAT master controls the MI Interface (default) if the MII Management PDI Access State register 0x0517[0] is not set. The EtherCAT master can prevent PDI control over the MI Interface, and it can force the PDI to release the MI Interface control. After power-on, the PDI can take over MI Interface control without any master transactions.

ET1100 only: MI Control is transferred to PDI if the Transparent Mode is enabled. IP Core: MI Control by PDI is

possible.

I-28 Slave Controller – Technology

Ethernet Physical Layer

Parameter

Comment

Clk

MDC period

Write

Write access time

Read

Read access time

1 0 1

1 0

MDC

MDIO

Clk

A4 A2A4 A1 R4A0 R3 R2

Preamble PHY Address

D7 D5D6 D4 D2D3 D1 D0

High Data Byte Low Data Byte

MDC

MDIO

Write

Start OP code

R1 R0

PHY Register Address

Turnaround

D15 D13D14 D12 D10D11 D9 D8

Write

IDLE

1 1 00

MDC

MDIO

Clk

A4 A2A4 A1 R4A0 R3 R2

Preamble PHY Address

D7 D5D6 D4 D2D3 D1 D0

High Data Byte Low Data Byte

MDC

MDIO

Read

Start OP code

R1 R0

PHY Register Address

Turn-

around

D15 D13D14 D12 D10D11 D9 D8

Read

IDLE

5.10.3 MI Protocol

Each MI access begins with a Preamble of “Ones“(32 without preamble suppression, less if both ESC and PHY support preamble suppression), followed by a Start-of-Frame (01) and the Operation Code (01 for write and 10 for read operations). Then the PHY address (5 bits) and the PHY register address (5 bits) are transmitted to the PHY. After a Turnaround (10 for write and Z0 for read operations – Z means MDIO is high impedance), two bytes of data follow. The transfer finishes after the second data byte and at least one IDLE cycle.

5.10.4 Timing specifications

Table 18: MII Management Interface timing characteristics

Figure 8: Write access

Figure 9: Read access

Slave Controller – Technology I-29

Ethernet Physical Layer

ESC

Ethernet PHY

MDIO

MCLK

MDIO

MDC

4K7

CC I/O

Ethernet PHY

MDIO

MDC

Ethernet PHY

MDIO

MDC

5.11 MII management example schematic

The MII management interface is a shared bus for all PHYs. The example schematic shows the connection between ESC and PHYs. Take care of proper PHY address configuration.

Figure 10: MII management example schematic

I-30 Slave Controller – Technology

Ethernet Physical Layer

Protection Earth (PE)

Virtual Ground

Protection Earth (PE)

Magnetics

Shield

RJ-45

2 3 4 5 6 7 8

1 M

10 nF/500 V

> 2mm

gap

> 1mm gap between

PE and Virtual Ground

> 1mm

gap

plane connection

75R

TX+

TX-

RX+

RX-

System Ground

5.12 Ethernet Termination and Grounding Recommendation

This termination and grounding design recommendation may help to meet the overall requirements for industrial communication. Nevertheless, implementation may vary depending on other requirements like board layout, other capacities, common ground interconnection, and shield grounding.

Unused RJ-45 pins are terminated by 75Ω resistors which will be connected to virtual ground. Virtual GND is connected to Protection Earth (PE) by a 10nF/500V capacitor in parallel to a 1MΩ resistor. Shield is also connected to PE by a 10nF/500V capacitor in parallel to a 1MΩ resistor. Especially the values for the connection between Virtual GND and PE are subject to change to meet industrial requirements (EMC). Shield is not directly connected with PE to avoid ground loops across large industrial plants with different PE potentials.

This design recommendation of termination and grounding is shown in Figure 11.

Figure 11: Termination and Grounding Recommendation

Slave Controller – Technology I-31

Ethernet Physical Layer

Signal

Name

Core Color

Contact Assignment

MDI

(may change depending on cable)

RJ45

M12

D-code

TX+

Transmission Data +

Yellow (light orange)

TX-

Transmission Data -

Orange 2 3

RX+

Receive Data +

Light green

RX-

Receive Data -

Green 6 4

4 1

5.13 Ethernet Connector (RJ45 / M12)

Fast Ethernet (100BASE-TX) uses two pairs/four pins. An RJ45 connector (8P8C) or an M12 D-code connector can be used. The RJ45 connector is recommended to use MDI pinout (PC side) for all ports for uniformity reasons. Standard Ethernet cables are used, not crossover cables. PHYs have to support MDI/MDI-X auto-crossover.

Table 19: Signals used for Fast Ethernet

NOTE: MDI-X (Switches/Hubs) uses same outline as MDI but TX+ is RX+ and TX- is RX- and vice versa.

Figure 12: RJ45 Connector

Figure 13: M12 D-code Connector

I-32 Slave Controller – Technology

Ethernet Physical Layer

ESC

REF_CLK

LINK_MII*

RX_CLK

RX_DV

RXD[3:0]

TX_CLK

TX_EN

TXD[3:0]

RX_ER

MDC

MDIO

ESC

REF_CLK

RESET*

RX_CLK

RX_DV

RXD[3:0]

TX_CLK

TX_EN

TXD[3:0]

RX_ER

MDC

MDIO

25 MHz

4K7

* check polarity: active Reset = no link

CC I/O

4K7

CC I/O

Configure TX Shift and

check RX timing

(setup/hold>10 ns)

Configure TX Shift and

check RX timing

(setup/hold>10 ns)

CLK25_OUT

25 MHz

CLK25_OUT

LINK_MII*RESET*

Single clock source is possible, too.

5.14 Back-to-Back MII Connection

5.14.1 ESC to ESC Connection

Two EtherCAT slave controllers can be connected back-to-back using MII as shown in the figure below. The timing of RX_DV and RXD with respect to RX_CLK has to be checked at both ESCs to be compliant with the IEEE 802.3 requirements of min. 10 ns setup time and min. 10 ns hold time. The timing can be adjusted by configuring the TX Shift settings of each ESC.

Enhanced Link Detection cannot be enabled for the back-to-back ports on both ESCs. Other clocking options besides using quartz oscillators are also possible. If CLK25_OUT is not

available, REF_CLK can be used instead.

Slave Controller – Technology I-33

Figure 14: Back-to-Back MII Connection (two ESCs)

Ethernet Physical Layer

ESC

REF_CLK

LINK_MII*

RX_CLK

RX_DV

RXD[3:0]

TX_CLK

TX_EN

TXD[3:0]

RX_ER

MDC

MDIO

Standard Ethernet

MAC (µC/FPGA)

CLK_IN

RX_CLK

RX_DV

RXD[3:0]

TX_CLK

TX_EN

TXD[3:0]

RX_ER

MDC

MDIO

25 MHz

4K7

* if LINK_MII is act. low

CC I/O

4K7

CC I/O

Configure TX Shift and check RX timing (setup/hold>10 ns)

Add delay (e.g. clock buffers) to adjust RX timing (setup/hold>10 ns)

5.14.2 ESC to Standard Ethernet MAC

If an ESC is to be connected directly to a standard Ethernet MAC (e.g. µController or FPGA), RX timing at the ESC and the MAC has to be checked. Since TX Shift configuration is not possible in the MAC, RX_CLK for the ESC has to be adjusted (delayed) to achieve proper RX timing at the ESC.

An EtherCAT slave controller can be directly connected to a standard Ethernet MAC using MII as shown in the figure below. The timing of RX_DV and RXD with respect to RX_CLK has to be checked at both ESC and MAC to be compliant with the IEEE 802.3 requirements of min. 10 ns setup time and min. 10 ns hold time. The timing can be adjusted by configuring the TX Shift setting of the ESC and the clock buffer (e.g. using one or more buffers).

If the standard Ethernet MAC completely uses the IEEE 802.3 TX signal timing window of 0 to 25 ns, standard compliant RX timing (-10..10 ns) is impossible. Since most Beckhoff ESCs do not require the entire IEEE802.3 RX timing window (check in Section III), a valid configuration is possible even in this case.

I-34 Slave Controller – Technology

Figure 15: Back-to-Back MII Connection (ESC and standard MAC)

EBUS/LVDS Physical Layer

EtherCAT

device

EBUS-TX-

EBUS-TX+

EBUS-RX+

EBUS-RX-

RBIAS

Signal

Direction

Description

EBUS-TX+ EBUS-TX-

OUT

EBUS/LVDS transmit signals

EBUS-RX+ EBUS-RX-

EBUS/LVDS receive signals RBIAS

BIAS resistor for EBUS-TX current adjustment

6 EBUS/LVDS Physical Layer

EBUS is an EtherCAT Physical Layer designed to reduce components and costs. It also reduces delay inside the ESC. The EBUS physical layer uses Low Voltage Differential Signaling (LVDS) according to the ANSI/TIA/EIA-644 “Electrical Characteristics of Low Voltage Differential Signaling (LVDS) Interface Circuits” standard.

EBUS has a data rate of 100 Mbit/s to accomplish the Fast Ethernet data rate. The EBUS protocol simply encapsulates Ethernet Frames, thus EBUS can carry any Ethernet frame – not only EtherCAT frames.

EBUS is intended to be used as a backplane bus, it is not qualified for wire connections.

6.1 Interface

Two LVDS signal pairs per EBUS link are used, one for reception and one for transmission of Ethernet/EtherCAT frames.

The EBUS interface has the following signals:

Figure 16: EBUS Interface Signals

Table 20: EBUS Interface signals

Unused EBUS ports can be left unconnected, only the LVDS termination resistor and the RBIAS resistor are mandatory.

Slave Controller – Technology I-35

EBUS/LVDS Physical Layer

EOF IDLE

Clk

Clock

Symbol

Manchester

Biphase L

IDLE

Data 0 0 N+ 1 0 1 0

Ethernet Preamble

1 N- 0 00/1

CRC

0/1

SOF

EOF

SOF

Paramet er

Min

Typ

Max

Comment

Clk

10 ns

EBUS Clock (100 Mbit/s)

SOF

15 ns

Positive level of SOF before falling edge

EOF

15 ns

Negative level of EOF after last edge

6.2 EBUS Protocol

Ethernet/EtherCAT frames are Manchester encoded (Biphase L) and encapsulated in Start-of-Frame (SOF) and End-of-Frame (EOF) identifiers. A beginning of a frame is detected if a Manchester violation with positive level (N+) followed by a ‘1’ bit occurs. The falling edge in the middle of the ‘1’ indicates the SOF to the ESC. This ‘1’ bit is also the first bit of the Ethernet preamble. Then the whole Ethernet frame is transmitted (including Ethernet SOF at the end of the preamble, up to the CRC). The frame finishes with a Manchester violation with negative level (N-), followed by a ‘0’ bit. This ‘0’ bit is also the first bit of the IDLE phase, which consists of ‘0’ bits.

Figure 17: EBUS Protocol

6.3 Timing Characteristics

Table 21: EBUS timing characteristics

NOTE: After power-on, a receiver which receives IDLE symbols cannot distinguish incoming ‘0’ bits from ‘1’ bits, because it is not synchronized to the transmitters phase. Synchronization is established at the falling edge at the end of the EBUS SOF, which indicates the center of the first preamble bit. After synchronization, idle errors can be detected by the ESC.

NOTE: SOF is detected at a falling edge following a period of at least 15 ns (nominal) of positive level, EOF is detected after a period of at least 15 ns (nominal) of negative level. I.e., the length of SOF and EOF can be even longer.

I-36 Slave Controller – Technology

EBUS/LVDS Physical Layer

EtherCAT

master

EtherCAT

slave 1

Link A

Link B

Port 0

Port 1

EtherCAT

slave 2

Port 0

Port 1

6.4 Standard EBUS Link Detection

Standard EBUS link detection is realized by counting the number of good signal events (no RX error) in a defined interval of time and comparing it to a given value. The link is established if enough events occurred, and disconnected if too few events occurred. IDLE symbols as well as any kind of EtherCAT traffic produce enough good events.

In order to handle partial link failures correctly, the following mechanism is used:  An ESC transmits at port 0 only if a link is detected (e.g., IDLE symbols are received), otherwise it

will transmit N- symbols.

 An ESC transmits at ports 1, 2, and 3 regardless of the link state (typically IDLE symbols if no

frames are pending).

Figure 18: Example EtherCAT Network

This method addresses these two cases of partial link failure (see Figure 18):  A failure on Link A will be detected by Slave 2, which will stop transmitting anything on Link B (and

close the loop at port 0). This is detected by Slave 1, which will close the loop at port 1. The master can still communicate with slave 1.

 A failure on Link B will be detected by Slave 1, which will close the loop at port 1. The master can

still communicate with slave 1. This failure cannot be detected by slave 2, which will leave port 0 open.

Do not connect any EBUS port 0 to another EBUS port 0 (same ESC or different ESCs) using standard link detection, because standard link detection will not establish a link after it was down.

Do not connect any EBUS port 1-3 to another EBUS port 1-3 (same ESC or different ESCs) using standard link detection, because partial link failures will not result in closed ports.

NOTE: Standard link detection cannot cope with a specific partial link fault (Link B failure), which affects redundancy operation (e.g., port 1 of slave 2 is connected to the master), because the master cannot communicate with slave 2 which leaves its port 0 open.

NOTE: Another advantage of this mechanism is that in case slave 2 is added to the network, at first port 0 of slave 2 is opened because there is activity on Link A, then transmission on Link B is started, and finally slave 1 opens Port 1. This assures that no frames get lost during link establishment.

6.5 Enhanced EBUS Link Detection

Enhanced EBUS link detection uses the standard link detection mechanism and adds a simple link detection handshake protocol before the link is established.

With enhanced link detection, the ESC transmits on all ports regardless of the link state (unless frames are transmitted; typically IDLE symbols are transmitted if no frames are pending).

The handshake protocol consists of three phases:

1. Each device starts transmitting a 4 nibble link request frame with 0x55C4 regularly at each port. This frame has no SOF/CRC and would be discarded if it was accidentally received by standard Ethernet devices (e.g., masters).

2. If a link request frame (0x55C4) is received at a port, the ESC will transmit link acknowledge frames (0x55CC) instead of link request frames at this port.

3. If a link acknowledge frame (0x55CC) is received at a port, the link at the port is established. Both link partners know that the link is good and establish the link. No further link detection frames are transmitted (until the link is interrupted and the link detection handshake phases are starting from the beginning).

Slave Controller – Technology I-37

EBUS/LVDS Physical Layer

EtherCAT

device

TX+

TX-

RX+

RX-

100R

EtherCAT

device

RX+

RX-

TX+

TX-

Link disconnection is signaled to the link partner by stopping transmission for a certain time. This will be detected by the default link detection mechanism. The link gets disconnected at both sides, and both sides close their loops. After that, the first phase of the handshake protocol starts again.

EBUS enhanced link detection is not compatible with older devices which forward enhanced link detection handshake frames depending on the direction (e.g. ESC20 and bus terminals without ASICs): the handshake frames are not forwarded through the EtherCAT Processing Unit, but they are forwarded without modification alongside the EtherCAT Processing Unit.

A device using enhanced link detection will stop generating handshake frames after the link is established or the enhanced link detection is disabled by SII EEPROM setting. It will restart generating handshake frames shortly after a link is lost (unless enhanced link detection is disabled).

6.6 EBUS RX Errors

The EBUS receiver detects the following RX errors:

 Missing edge in the middle of one bit (but not EBUS SOF/EOF)  EBUS SOF inside a frame (two SOFs without EOF in between)  EBUS EOF outside a frame (two EOFs without SOF in between)  IDLE violation: ‘1’ outside a frame  Too short pulses (less than ~3.5 ns)

A frame with an RX error is discarded. Too many RX errors in a defined interval of time will result in link disconnection.

Errors outside of a frame are counted only once, and errors inside a frame are also counted only once, because the Manchester decoding might have lost synchronization.

6.7 EBUS Low Jitter

In Low Jitter mode, the jitter of the forwarding delay (EBUS port to EBUS port) is reduced.

6.8 EBUS Connection

The connection of two EBUS ports is shown in Figure 19. LVDS termination with 100 Ohm impedance is necessary between each pair of receive signals – located adjacent to the receive inputs.

Figure 19: EBUS Connection

I-38 Slave Controller – Technology

FMMU

FMMU configuration register

FMMU reg. offset

Value

Logical Start Address

0x0:0x3

0x00010011

Length (Bytes)

0x4:0x5

Logical Start bit

0x6 3 Logical Stop bit

0x7

Physical Start Address

0x8:0x9

0x0F01

Physical Start bit

0xA

Type

0xB

read and/or write

Activate

0xC

1 (enabled)

Logical process image: up to 4 GByte

Telegram structure

Ethernet HDR HDR 1 PLC Data HDR 2 NC Data HDR n Data n FCS

PLC Data

Data n

NC Data

DVI

IPC

.. ..

SubTelegram 1

SubTelegram 2

SubTelegram n

7 FMMU

Fieldbus Memory Management Units (FMMU) convert logical addresses into physical addresses by the means of internal address mapping. Thus, FMMUs allow to use logical addressing for data segments that span several slave devices: one datagram addresses data within several arbitrarily distributed ESCs. Each FMMU channel maps one continuous logical address space to one continuous physical address space of the slave. The FMMUs of Beckhoff ESCs support bit wise mapping, the number of supported FMMUs depends on the ESC. The access type supported by an FMMU is configurable to be either read, write, or read/write.

Figure 20: FMMU Mapping Principle

The following example illustrates the functions of an FMMU configured to map 14 bits from logical address 0x00010011.3 to 0x00010013[0] to the physical register bits 0x0F01[1] to 0x0F02[6]. The FMMU length is 3 Byte, since the mapped bits span 3 Bytes of the logical address space. Length calculation begins with the first logical byte which contains mapped bits, and ends with the last logical byte which contains mapped bits.

Table 22: Example FMMU Configuration

NOTE: FMMU configuration registers start at address 0x0600.

Slave Controller – Technology I-39

FMMU

0 7654321 0 7654321 0 7654321 076

Byte 0x00010011

Logical Start Address

0 7654321 0 7654321 0 7654321 076 1

Byte 0x00010012 Byte 0x00010013

Byte 0x0F02Byte 0x0F02

Byte 0x0F01

Physical Start Address

Logical Address Space

Physical Address Space

Logical Start bit = 3 Logical Stop Bit = 0

Physical Start bit = 1

Configured FMMU Length = 3 Bytes

14 Bits mapped

Figure 21: FMMU Mapping Example

Attention: This drawing of the bit string shows the least significant bit first, in a hexadecimal representation of the octets the least significant value is at the right place and the most significant on the left place (00110011 is represented as octet by 0xCC).

Restrictions on FMMU Settings

The FMMUs of Beckhoff ESCs are subject to restrictions. The logical address ranges of two FMMUs of the same direction (read or write) in one ESC must be separated by at least 3 logical bytes not configured by any FMMU of the same type, if one of the FMMUs or both use bit-wise mapping (logical start bit ≠ 0, logical stop bit ≠ 7, or physical start bit ≠ 0). In the above example, the first logical address area after the one shown must have a logical start address of 0x00010017 or higher (the last byte of the example FMMU is 0x00010013, three bytes free 0x00010014-0x00010016).

If only byte-wise mapping is used (logical start bit = 0, logical stop bit = 7, or physical start bit = 0), the logical address ranges can be adjacent.

Bit-wise writing is only supported by the Digital Output register (0x0F00:0x0F03). All other registers and memories are always written byte-wise. If bit-wise mapping is used for writing into these areas, bits without mapping to logical addresses are written with undefined values (e.g., if only physical address bit 0x1000[0] is mapped by a write FMMU, the bits 1-7 are written with undefined values).

Additional FMMU Characteristics

 Each logical address byte can at most be mapped either by one FMMU(read) plus one

FMMU(write), or by one FMMU(read/write). If two or more FMMUs (with the same direction – read or write) are configured for the same logical byte, the FMMU with the lower number (lower configuration address space) is used, the other ones are ignored.

 One or more FMMUs may point to the same physical memory, all of them are used. Collisions

cannot occur.

 It is the same to use one read/write FMMU or two FMMUs – one read, the other one write – for the

same logical address.

 A read/write FMMU cannot be used together with SyncManagers, since independent read and

write SyncManagers cannot be configured to use the same (or overlapping) physical address range.

 Bit-wise reading is supported at any address. Bits which are not mapped to logical addresses are

not changed in the EtherCAT datagram. E.g., this allows for mapping bits from several ESCs into the same logical byte.

 Reading an unconfigured logical address space will not change the data.

I-40 Slave Controller – Technology

SyncManager

Description

Physical Start Address

0x0:0x1

Length

0x2:0x3

Control Register

0x4

Status Register

0x5

Activate

0x6

PDI Control

0x7

8 SyncManager

The memory of an ESC can be used for exchanging data between the EtherCAT master and a local application (on a µController attached to the PDI) without any restrictions. Using the memory for communication like this has some drawbacks which are addressed by the SyncManagers inside the ESCs:

 Data consistency is not guaranteed. Semaphores have to be implemented in software for

exchanging data in a coordinated way.

 Data security is not guaranteed. Security mechanisms have to be implemented in software.  Both EtherCAT master and application have to poll the memory in order to find out when the

access of the other side has finished.

SyncManagers enable consistent and secure data exchange between the EtherCAT master and the local application, and they generate interrupts to inform both sides of changes.

SyncManagers are configured by the EtherCAT master. The communication direction is configurable, as well as the communication mode (Buffered Mode and Mailbox Mode). SyncManagers use a buffer located in the memory area for exchanging data. Access to this buffer is controlled by the hardware of the SyncManagers.

A buffer has to be accessed beginning with the start address, otherwise the access is denied. After accessing the start address, the whole buffer can be accessed, even the start address again, either as a whole or in several strokes. A buffer access finishes by accessing the end address, the buffer state changes afterwards and an interrupt or a watchdog trigger pulse is generated (if configured). The end address cannot be accessed twice inside a frame.

Two communication modes are supported by SyncManagers:  Buffered Mode

- The buffered mode allows both sides, EtherCAT master and local application, to access the communication buffer at any time. The consumer always gets the latest consistent buffer which was written by the producer, and the producer can always update the content of the buffer. If the buffer is written faster than it is read out, old data will be dropped.

- The buffered mode is typically used for cyclic process data.

 Mailbox Mode

- The mailbox mode implements a handshake mechanism for data exchange, so that no data will be lost. Each side, EtherCAT master or local application, will get access to the buffer only after the other side has finished its access. At first, the producer writes to the buffer. Then, the buffer is locked for writing until the consumer has read it out. Afterwards, the producer has write access again, while the buffer is locked for the consumer.

- The mailbox mode is typically used for application layer protocols.

The SyncManagers accept buffer changes caused by the master only if the FCS of the frame is correct, thus, buffer changes take effect shortly after the end of the frame.

The configuration registers for SyncManagers are located beginning at register address 0x0800.

Table 23: SyncManager Register overview

Slave Controller – Technology I-41

SyncManager

0x1000

0x10FF

Buffer 0 (visible)

All buffers are controlled by the SyncManager. Only buffer 0 is configured by the SyncManager and addressed by ECAT and µController.

0x1100

0x11FF

Buffer 1 (invisible shall not be used)

0x1200

0x12FF

Buffer 2 (invisible shall not be used)

0x1300 …

Next usable RAM space

Master

ECAT

PDI

µController

Write begin

Load next buffer if new data available

 Read begin

Write end

Exchange buffer (if frame is ok)

 Read end Exchange buffer

 Write begin  Write end

Read begin 

Read end 

Load next buffer if new data available

8.1 Buffered Mode

The buffered mode allows writing and reading data simultaneously without interference. If the buffer is written faster than it is read out, old data will be dropped. The buffered mode is also known as 3buffer-mode.

Physically, 3 buffers of identical size are used for buffered mode. The start address and size of the first buffer is configured in the SyncManager configuration. The addresses of this buffer have to be used by the master and the local application for reading/writing the data. Depending on the SyncManager state, accesses to the first buffer’s (0) address range are redirected to one of the 3 buffers. The memory used for buffers 1 and 2 cannot be used and should be taken into account for configuring other SyncManagers.

One buffer of the three buffers is allocated to the producer (for writing), one buffer to the consumer (for reading), and the third buffer keeps the last consistently written data of the producer.

As an example, Figure 22 demonstrates a configuration with start address 0x1000 and Length 0x100. The other buffers shall not be read or written. Access to the buffer is always directed to addresses in the range of buffer 0.

The buffer interaction is shown in Figure 23:

I-42 Slave Controller – Technology

Figure 22: SyncManager Buffer allocation

Figure 23: SyncManager Buffered Mode Interaction

SyncManager

Master

µController

Write Mailbox

Write 

Write w. failure

(Mailbox full)

Mailbox full Mailbox empty

 Read  Read w. failure

(Mailbox empty)

Read Mailbox

Read 

Read w. failure

(Mailbox empty)

Mailbox full Mailbox empty

 Write Write w. failure

(Mailbox full)

The Status register of the SyncManager reflects the current state. The last written buffer is indicated (informative only, access redirection is performed by the ESC), as well as the interrupt states. If the SyncManager buffer was not written before, the last written buffer is indicated to be 3 (start/empty).

8.2 Mailbox Mode

The mailbox mode only allows alternating reading and writing. This assures all data from the producer reaches the consumer. The mailbox mode uses just one buffer of the configured size.

At first, after initialization/activation, the buffer (mailbox, MBX) is writeable. Once it is written completely, write access is blocked, and the buffer can be read out by the other side. After it was completely read out, it can be written again.

The time it takes to read or write the mailbox does not matter.

Figure 24: SyncManager Mailbox Interaction

8.2.1 Mailbox Communication Protocols

This is only a brief overview of the mailbox communication protocols. For details on the mailbox protocols refer to the EtherCAT specification ETG.1000.6: “Application layer protocol specification” available from the EtherCAT Technology Group (http://www.ethercat.org) or to the IEC specification

“Digital data communications for measurement and control – Fieldbus for use in industrial control systems”, IEC 61158 Type 12: EtherCAT, available from the IEC (http://www.iec.ch).

Ethernet over EtherCAT (EoE)

Defines a standard way to exchange or tunnel standard Ethernet Frames over EtherCAT.

CAN application layer over EtherCAT (CoE)

Defines a standard way to access a CAN application layer object dictionary and to exchange CAN application layer Emergency and PDO messages on an event driven path.

File Access over EtherCAT (FoE)

Defines a standard way to download and upload firmware and other ‘files’.

Servo Profile over EtherCAT (SoE)

Defines a standard way to access the IEC 61800-7 identifier.

Vendor specific Profile over EtherCAT (VoE)

A vendor specific protocol follows after a VoE header that identifies the vendor and a vendor specific type.

ADS over EtherCAT (AoE)

AoE defines a standard way to exchange Automation Device Specification (ADS) messages over EtherCAT.

Slave Controller – Technology I-43

SyncManager

Field

Data Type

Value/Description

Length

WORD

Number of Bytes of this mailbox command excluding the mailbox header

Address

WORD

Station Address of originator

Channel

6 bit

0 – reserved for future use

Priority

2 bit

Priority between 0 (lowest) and 3 (highest)

Type

4 bit

Mailbox protocol types: 0x0: Error 0x1: Vendor specific (Beckhoff: AoE – ADS over EtherCAT) 0x2: EoE (Ethernet over EtherCAT) 0x3: CoE (CAN application layer over EtherCAT) 0x4: FoE (File access over EtherCAT) 0x5: SoE (Servo profile over EtherCAT) 0xF: Vendor specific (VoE)

Ctr.

3 bit

Sequence number that is used for detection of duplicated frames

Reserved

1 bit

Ctr.

Type

Prio

Channel

Address

Length

16 Bit

6 Bit

2 Bit

4 Bit

3 Bit 1 Bit

44 047

The content of an EtherCAT mailbox header is shown in Figure 25.

Figure 25: EtherCAT Mailbox Header (for all Types)

Table 24: EtherCAT Mailbox Header

8.3 PDI register function acknowledge by Write

Refer to chapter 16.2 for the background of this function, which only affects the SyncManager operation when the PDI reads a buffer. The EtherCAT operation is not influenced by this optional feature.

Reading a SyncManager buffer consists of the following steps, if PDI register function acknowledge by Write is enabled:

 Read the first byte.

This will open the buffer. Accidentally reading subsequent bytes has no influence.

 Read any byte of the buffer.

This includes the first byte, the last byte, and any inner byte. The buffer remains open, and the buffer can even be read multiple times. Accidentally reading the last byte has no influence.

 Write to the last byte of the buffer.

Byte enable signals are used, write data is ignored (write 0). This write operation closes the buffer.

Accidentally reading the first byte of a second SyncManager buffer behind the one to be read is still possible, this will open the second SyncManager buffer. This can easily be prevented by aligning SyncManager buffer start addresses to the data width of the µController. It is recommended to align SyncManager buffer start addresses to 64 bit (8 byte) boundaries anyway.

8.4 Interrupt and Watchdog Trigger Generation, Latch Event Generation

Interrupts can be generated when a buffer was completely and successfully written or read. A watchdog trigger signal can be generated to rewind (trigger) the Process Data watchdog used for Digital I/O after a buffer was completely and successfully written. Interrupt and watchdog trigger generation are configurable. The SyncManager Status register reflects the current buffer state.

For debugging purposes it is also possible to trigger Distributed Clock Latch events upon successful buffer accesses (for some ESCs).

I-44 Slave Controller – Technology

SyncManager

8.5 Single Byte Buffer Length / Watchdog Trigger for Digital Output PDI

If a SyncManager is configured for a length of 1 byte (or even 0), the buffer mechanism is disabled, i.e., read/write accesses to the configured address will pass the SyncManager without interference. The additional buffers 1 and 2 are not used in buffered mode, and alternation of write/read accesses is not necessary for mailbox mode. Consistency is not an issue for a single byte buffer.

Nevertheless, watchdog generation is still possible if the buffer length is 1 byte (interrupt generation as well).

NOTE: For some ESCs in Mailbox mode, watchdog and interrupt generation are depending on the alternation of write and read accesses, although the write/read accesses itself are executed without interference. I.e., Buffered mode should be used for single byte buffers for watchdog generation.

Watchdog trigger generation with single byte SyncManagers is used for Digital Outputs, because the outputs are only driven with output data if the Process Data watchdog is triggered. One SyncManager has to be configured for each byte of the Digital Output register (0x0F00:0x0F03) which is used for outputs. The SyncManagers have to be configured like this:

 Buffered Mode (otherwise the process data watchdog will not be wound up with some ESCs upon

the second and following writes, because the Digital I/O PDI does not read the addresses)

 Length of 1 byte  EtherCAT write / PDI read  Watchdog Trigger enabled

For more details refer to the Digital I/O PDI description of Section III and the chapters about Watchdog and Digital Output in this document.

NOTE: A SyncManager with length 0 behaves like a disabled SyncManager. It does not interfere accesses nor generate interrupt or watchdog trigger signals.

8.6 Repeating Mailbox Communication

A lost datagram with mailbox data is handled by the application layer. The Repeat Request/Repeat Acknowledge bits in the SyncManager Activation register (offset 0x06[1]) and the PDI Control register (offset 0x07[1]) are used in mailbox mode for retransmissions of buffers from a slave to the master. If a mailbox read frame gets lost/broken on the way back to the master, the master can toggle the Repeat Request bit. The slave polls this bit or receives an interrupt (SyncManager activation register changed, register 0x0220[4]) and writes the last buffer again to the SyncManager. Then the PDI toggles the Repeat Acknowledge bit in the PDI Control register. The master will read out this bit and read the buffer content. Communication resumes afterwards.

This mechanism is shown in Figure 26 for a mailbox write service. The Mailbox confirmation is lost on its way from the slave to the master and has to be repeated again.

Slave Controller – Technology I-45

SyncManager

Master µController

Slave writes data into mailbox

Master sends mailbox read

command

Mailbox read

Slave detects Repeat Request (interrupt or polling) Slave writes stored write data into mailbox again

Maibox read data (WKC=1)

Timeout: Master detects

lost frame

Master toggles

Repeat Request bit

Toggle Repeat Request bit

Master polls Repeat Request

Acknowledge bit

Read Repeat Request Acknowledge bit

Repeat Request Acknowledge is set

Master sends mailbox read

command again

Mailbox read

Maibox read data A (WKC=1)

Master receives mailbox read

data

ESC

Write data A

Interrupt

Write data A

Toggle Repeat

Request

Acknowledge

Read Event

Slave detects that mailbox was read and stores write data for a possible repeat request

Write data B

Slave writes next data into mailbox

Slave toggles Repeat Request Acknowledge bit

Communication proceeds with read data B

8.7 SyncManager Deactivation by the PDI

A SyncManager can be deactivated by the PDI to inform the master of local problems (typically used in buffered mode only). The master can detect SyncManager deactivation by checking the Working Counter, which is not incremented if a deactivated SyncManager buffer is accessed. If a SyncManager is deactivated by the PDI (PDI Control register 0x7[0]=1), the state of the SyncManager is reset, interrupts are cleared and the SyncManager has to be written first after re-activation. The entire SyncManager buffer area is read/write protected while the SyncManager is deactivated by the PDI.

Figure 26: Handling of a Repeat Request with Read Mailbox

I-46 Slave Controller – Technology

Distributed Clocks

9 Distributed Clocks

The Distributed Clocks (DC) unit of EtherCAT slave controllers supports the following features:

 Clock synchronization between the slaves (and the master)  Generation of synchronous output signals (SyncSignals)  Precise time stamping of input events (LatchSignals)  Generation of synchronous interrupts  Synchronous Digital Output updates  Synchronous Digital Input sampling

9.1 Clock Synchronization

DC clock synchronization enables all EtherCAT devices (master and slaves) to share the same EtherCAT System Time. The EtherCAT devices can be synchronized to each other, and consequently, the local applications are synchronized as well.

For system synchronization all slaves are synchronized to one Reference Clock. Typically, the first ESC with Distributed Clocks capability after the master within one segment holds the reference time (System Time). This System Time is used as the reference clock to synchronize the DC slave clocks of other devices and of the master. The propagation delays, local clock sources drift, and local clock offsets are taken into account for the clock synchronization.

The ESCs can generate SyncSignals for local applications to be synchronized to the EtherCAT System Time. SyncSignals can be used directly (e.g., as interrupts) or for Digital Output updating/Digital Input sampling. Additionally, LatchSignals can be time stamped with respect to the EtherCAT System Time.

Definition of the System Time

 Beginning on January, 1st 2000 at 0:00h  Base unit is 1 ns  64 bit value (enough for more than 500 years)  Lower 32 bits span over 4.2 seconds (typically enough for communication and time stamping).

Some ESCs only have 32 bit DCs, which are compatible with 64 bit DCs.

Definition of the Reference Clock

One EtherCAT device will be used as a Reference Clock. Typically, the Reference Clock is the first ESC with DC capability between master and all the slaves to be synchronized (DC slaves). The Reference Clock might be adjusted to a “global” reference clock, e.g. to an IEEE 1588 grandmaster clock. The reference clock provides the System Time.

Definition of the Local Clock

Each DC slave has a local clock, initially running independent of the Reference Clock. The difference between local clock and Reference Clock (offset) can be compensated, as well as clock drifts. The offset is compensated by adding it to the local clock value. The drift is compensated by measuring and adjusting the local clock speed.

Each DC slave holds a copy of the Reference Clock, which is calculated from the local clock and the local offset. The Reference Clock has a local clock, too.

Definition of the Master Clock

The Reference Clock is typically initialized by the EtherCAT master using the master clock to deliver the System Time according to the System Time definition. The EtherCAT master clock is typically bound to a global clock reference (RTC or the master PC, IEEE1588, GPS, etc.), which is either directly available to the master or indirectly by an EtherCAT slave providing access to the reference.

Slave Controller – Technology I-47

Distributed Clocks

Propagation Delay

The propagation delay between Reference Clock and slave clock has to be taken into account when the System Time is distributed to the slaves.

Offset

The offset between local clock and Reference Clock has two reasons: the propagation delay from the ESC holding the Reference Clock to the device with the slave clock, and initial differences of the local times resulting from different times at which the ESCs have been powered up. This offset is compensated locally in each slave.

The ESC holding the Reference Clock derives the System Time from its local time by adding a local offset. This offset represents the difference between local time (started at power-up) and master time (starting on January, 1st 2000 at 0:00h).

Drift

Since Reference Clock and DC slaves are typically not sourced by the same clock source (e.g., a quartz), their clock sources are subject to small deviations of the clock periods. The result is that one clock is running slightly faster than the other one, their local clocks are drifting apart.

ESC Classification regarding DC Support

Three classes of ESCs are distinguished regarding Distributed Clocks support:

1. Slaves supporting System Time/Time Loop Control Unit: Receive time stamps and System Time/Time Loop Control Unit available; SyncSignal generation, LatchSignal time stamping, and SyncManager Event Times are optionally supported depending on application.

2. Slaves supporting only propagation delay measurement: Mandatory for ESCs with 3 or more ports (topology devices like EK1100 and ET1100). Local clock and receive time stamps are supported.

3. Slaves without Distributed Clocks support: Slaves with max. 2 ports do not have to support DC features. Processing/forwarding delay of such slaves is treated like a “wire delay” by the surrounding DC capable slaves.

I-48 Slave Controller – Technology

Distributed Clocks

Reference

Clock

Slave Clock

System Time

Local time

Drift compensation frame

carrying current System Time

Propagation delay compensation

Offset compensation

Drift compensation

Reference

Clock

Slave Clock

System Time

Local Time

Drift compensation frame

carrying current System Time

Propagation delay compensation

Offset compensation

Drift compensation

Goal: Slave clock has

copy of System Time

System Time < Local Time

System Time > Local Time

Propagation delay

System Time

Local time

System Time

Local time

Goal: Slave clock has

copy of System Time

9.1.1 Clock Synchronization Process

The clock synchronization process consists of three steps:

1. Propagation Delay Measurement: The master initiates propagation delay measurement between all slaves in all directions. Each EtherCAT slave controller measures the receive time of the measurement frame. The master collects the time stamps afterwards and calculates the propagation delays between all slaves.

2. Offset compensation to Reference Clock (System Time): The local time of each slave clock is compared to the System Time. The difference is compensated individually by writing it to each slave. All devices get the same absolute System Time.

3. Resetting the Time Control Loop: The current filter status needs to be reset to eliminate influences of previous measurements and improve repeatability.

4. Drift compensation to Reference Clock: The drift between Reference Clock and local clock has to be compensated by regularly measuring the differences and readjusting the local clocks.

The following figure illustrates the compensation calculations for two cases, in the first case the System Time is less than the slave’s local time, in the second case, it is the other way around.

Figure 27: Propagation Delay, Offset, and Drift Compensation

Slave Controller – Technology I-49

Distributed Clocks

Name

Description

0x0900:0x0903

Receive Time Port 0

Local time when receiving frame on Port 0

0x0904:0x0907

Receive Time Port 1

Local time when receiving frame on Port 1

0x0908:0x09B

Receive Time Port 2

Local time when receiving frame on Port 2

0x090C:0x090F

Receive Time Port 3

Local time when receiving frame on Port 3

0x0918:0x091F

Receive Time ECAT Processing Unit

Local time when receiving frame at the ECAT Processing Unit

0x0936

Receive Time Latch Mode

ET1200: Receive time latching in forwarding or reverse mode

9.1.2 Propagation Delay Measurement

Since each slave introduces a small processing/forwarding delay in each direction (within the device and also in the PHY), as well as the cable between the ESCs has a delay, the propagation delay between Reference Clock and the respective slave clock has to be considered for the synchronization of the slave clocks.

1. For measuring the propagation delay, the master sends a broadcast write to register DC Receive Time Port 0 (at least the first byte).

2. Each slave device stores the time of its local clock when the first bit of the Ethernet preamble of the frame was received, separately for each port (Receive Time Port 0-3 registers).

3. The master reads all time stamps and calculates the delay times with respect to the topology. The delay time between Reference Clock and an individual slave is written to the slave’s System Time Delay register (0x0928:0x092B).

The receive time registers are used to sample the receive time of a specific frame (a broadcast write to Receive Time Port 0 register).

The clocks must not be synchronized for the delay measurement, only local clock values are used. Since the local clocks of the slaves are not synchronized, there is no relation between the Receive Times of different slaves. So the propagation delay calculation has to be based on receive time differences between the ports of a slave.

Devices with two ports do not need to support Distributed Clocks at all, their delay is treated as an additional “wire delay” between the surrounding DC-capable slaves. Devices with more than 2 ports have to support at least propagation delay measurements (DC Receive Times).

NOTE: Some ESCs use the broadcast write to Receive Time Port 0 register as an indicator to latch the receive times of the next frame at all ports other than port 0 (if port 0 is open). Thus, another frame which is still traveling around the ring might trigger the measurement, and the receive times do no correlate. For these ESCs, the ring has to be empty before the broadcast write is issued. Refer to Section II Receive Time Port x registers for further information.

Table 25: Registers for Propagation Delay Measurement

9.1.2.1 Propagation Delay Measurement in Reverse Mode

For redundancy operation, it is necessary to perform propagation delay measurements either in forwarding mode (master connected to port 0), or in reverse mode (master connected to port 1-3). For ET1200, care has to be taken because the measurement principle is slightly different for these two ESC types: DC Receive Time Latch Mode register 0x0936 has to be used. As ET1100 and IP Core do not require register 0x0936, the following rules can be used for propagation delay measurement in a mixed ET1200/ET1100/IP Core environment (ESC20 does not support propagation delay measurement in reverse mode).

 In forwarding mode, the master should write 0x00 to register 0x0936 of all slaves, and perform the

delay measurement afterwards.

 In reverse mode, the master should write 0x01 to register 0x0936 of all slaves, and perform the

delay measurement afterwards.

Please refer to section II for more details on DC Receive Time Latch Mode register 0x0936 and receive time stamps in 0x0900:0x09F.

I-50 Slave Controller – Technology

Distributed Clocks

9.1.2.2 Propagation Delay Measurement Example

The propagation delay between the local device and the Reference Clock device is calculated for the network example shown in Figure 28. The example assumes that slave A is the Reference Clock. The loops of slave D and F are closed internally. The wire delays are assumed to be symmetrical, and the processing and forwarding delays are assumed to be identical for all ESCs.

Slave Controller – Technology I-51

Distributed Clocks

Slave D

Slave F

Slave A

Slave E

Slave C

Slave B

EtherCAT

Processing

Unit

Master

EtherCAT

Processing

Unit

WAB

Port 0

Port 1

EtherCAT

Processing

Unit

WCD

Port 0

Port 1

EtherCAT

Processing

Unit

Port 0

EtherCAT

Processing

Unit

WEF

Port 0

Port 1

WBC

EtherCAT

Processing

Unit

Port 0

WBE

Port 2

Reference

Clock

Figure 28: Propagation Delay Calculation

I-52 Slave Controller – Technology

Distributed Clocks

Parameter

Description

tPx

Processing delay of slave x (through EtherCAT Processing Unit, x=A-F)

Forwarding delay of slave x (alongside EtherCAT Processing Unit, x=A-F)

Propagation delay from slave x to slave y (x/y=A-F)

Wxy

Wire propagation delay between slaves x and y (assumed to be symmetrical in both directions, x/y=A-F)

tx0, tx1, t

Receive Time Port 0/1/2 values of slave x (time when first preamble bit is detected, x=A-F), measured with a write access to DC Receive Time 0 register.

Processing delay (through EtherCAT Processing Unit) if all slaves are identical

Forwarding delay (alongside EtherCAT Processing Unit) if all slaves are identical

Diff

Difference between Processing delay and forwarding delay t

Diff

= tP – tF if all slaves are identical. ESC specific information, part of the ESI. Refer to Section III for actual figures.

Ref_x

Propagation delay from Reference Clock (slave A) to slave x

Parameters used for propagation delay calculation are listed in Table 26:

Table 26: Parameters for Propagation Delay Calculation

Propagation delay between Slave C and D

The propagation delays between slave C and D (tCD and tDC) consist of a processing delay and the wire delay:

t tDC = tPD + t Assuming that the processing delays of slave C and D are identical (tP = t tCD = tDC = tP + t

= tPC + t

WCD

= tPD):

The two Receive Times of slave C have the following relation: tC1 = tC0 + tCD + t

So the propagation delays between slave C and D are tCD = t

= (tC1 – tC0) / 2

Propagation delay between Slave B and C

The propagation delays between slave B and C (tBC and tCB) are calculated as follows: t tCB = tFC + t Assuming that the processing delays of slaves B, C and D are identical (tP = t

difference between forwarding and processing delay of slave C is t t tCB = t

= tPB + t

= tP + t

– t

WBC

Diff

= t

= tPD), and the

= tPC – t

Diff

The Receive Times (port 0 and 1) of slave B have the following relation: tB1 = tB0 + tBC + tCD + tDC + t

So the propagation delay between slave B and C is 2*tBC – t tBC = ((tB1 – tB0) – (tC1 – tC0) + t

= (tB1 – tB0) – (tC1 – tC0)

Diff

) / 2

Slave Controller – Technology I-53

Distributed Clocks

And for the other direction: tCB = ((tB1 – tB0) – (tC1 – tC0) – t

Diff

) / 2

Propagation delay between Slave E and F

The propagation delays between slave E and F are calculated like the delays between slave C and D: t tFE = tPF + t Assuming that the processing delays of slave E and F are identical (tP = t

= tPE + t

WEF

= tPF):

tEF = tFE = (tE1 – tE0) / 2

Propagation delay between Slave B and E

The propagation delays between slave B and E (tBE and tEB) are calculated as follows: t tEB = tFE + t

= tFB + t

WBE

Assuming that the processing delays of slaves B to F are identical (tP = tPx), and the difference between forwarding and processing delay of these slaves is t

= tEB = tP – t

Diff

+ t

WBE

= tPx – t

Diff

The Receive Times Port 1 and 2 of slave B have the following relation: tB2 = tB1 + tBE + tEF + tFE + tEB So the propagation delay between slave B and E is 2*tBE = (tB2 – tB1) – tEF – tFE tBE = tEB = ((tB2 – tB1) – (tE1 – tE0)) / 2

Propagation delay between Slave A and B

The propagation delays between slave A and B are calculated as follows: t tBA = tFB + t

= tPA + t

WAB

Assuming that the processing delays of all slaves are identical (tP = tPx), and the difference between forwarding and processing delay of these slaves is t

t tBA = tAB – t

= tP + t

WAB

Diff

= tPx – t

Diff

The Receive Times of slave A have the following relation: tA1 = tA0 + tAB + (tB1 – tB0) + (tB2 – tB1) + tBA So the propagation delay between slave A and B is tAB = ((tA1 – tA0) – (tB2 – tB0) + t

Diff

) / 2 And for the other direction: tBA = ((tA1 – tA0) – (tB2 – tB0) – t

Diff

) / 2

I-54 Slave Controller – Technology

Distributed Clocks

Name

Description

0x0910:0x0917

System Time

Local copy of System Time (read from PDI)

0x0918:0x091F

Receive Time ECAT Processing Unit

Local Time (t

Local time

)

0x0920:0x0927

System Time Offset

Difference between local time and System Time (t

Offset

)

Summary of Propagation Delay Calculation between Slaves

tAB = ((tA1 – tA0) – (tB2 – tB0) + t tBA = ((tA1 – tA0) – (tB2 – tB0) – t tBC = ((tB1 – tB0) – (tC1 – tC0) + t tCB = ((tB1 – tB0) – (tC1 – tC0) – t tCD = t

= (tC1 – tC0) / 2

Diff

) / 2

) / 2 ) / 2

tEF = tFE = (tE1 – tE0) / 2 tBE = tEB = ((tB2 – tB1) – (tE1 – tE0)) / 2

Propagation Delays between Reference Clock and Slave Clocks

The System Time Delay register of each slave clock takes the propagation delay from the Reference Clock to the slave. This delay is calculated like this:

t t t t t

= tAB

Ref_B

= tAB + tBC

Ref_C

= tAB + tBC + tCD

Ref_D

= tAB + tBC + tCD + tDC + tCB + tBE

Ref_E

= tAB + tBC + tCD + tDC + tCB + tBE + t

Ref_F

9.1.3 Offset Compensation

The local time of each device is a free-running clock which typically will not have the same time as the Reference Clock. To achieve the same absolute System Time in all devices, the offset between the Reference Clock and every slave device’s clock is calculated by the master. The offset time is written to register System Time Offset to adjust the local time for every individual device. Small offset errors are eliminated by the drift compensation after some time, but this time might become extremely high for large offset errors – especially for 64 bit DCs.

Each DC slave calculates its local copy of the System time using its local time and the local offset value:

Local copy of System Time

= t

Local time

+ t

Offset

This time is used for SyncSignal generation and time stamping of LatchSignals. It is also provided to the PDI for use by µControllers.

The System Time of the Reference Clock is bound to the master clock by calculating the difference and compensating it using the System Time Offset of the Reference Clock.

Registers used for Offset Compensation are listed in Table 27.

Table 27: Registers for Offset Compensation

Slave Controller – Technology I-55

Distributed Clocks

Name

Description

0x0930:0x0931

Speed Counter Start

Bandwidth for adjustment of local copy of System Time Writing a value resets the internal filters.

9.1.4 Resetting the Time Control Loop

Before starting drift compensation, the internal filters of the Time Control Loop must be reset. Their current status is typically unknown, and they can have negative impact on the settling time. The filters are reset by writing the Speed Counter Start value to the Speed Counter Start register (0x0930:0x0931). Writing the current value of the register again is sufficient to reset the filters.

Registers used for resetting the Time Control Loop filters are listed in Table 27.

Table 28: Registers for Resetting the Time Control Loop

9.1.5 Drift Compensation

After the delay time between the Reference Clock and the slave clocks has been measured, and the offset between both clocks has been compensated, the natural drift of every local clock (emerging from quartz variations between the Reference Clock’s quartz and the local quartz) is compensated by the time control loop which is integrated into each ESC.

For drift compensation, the master distributes the System Time from the Reference Clock to all slave clocks periodically. The ARMW or FRMW commands can be used for this purpose. The time control loop of each slave takes the lower 32 bit of the System Time received from the Reference Clock and compares it to its local copy of the System Time. For this difference, the propagation delay has to be taken into account:

Δt = (t

Local time

+ t

Offset

– t

Propagation delay

) – t

Received System Time

If Δt is positive, the local time is running faster than the System time, and has to be slowed down. If Δt is negative, the local time is running slower than the System time, and has to be sped up. The time control loop adjusts the speed of the local clock.

For a fast compensation of the static deviations of the clock speeds, the master should initially send many ARMW/FRMW commands (e.g. 15,000) for drift compensation in separate frames after initialization of the propagation delays and offsets. The control loops compensate the static deviations and the distributed clocks are synchronized. Afterwards, the drift compensation frames are sent periodically for compensation of dynamic clock drifts.

NOTE: The System Time Offset allows fast compensation of differences between local copy of the system time and the System Time, the drift compensation is very slow. Thus, shortly before drift compensation is started, the offset should be roughly compensated using the System Time Offset register. Otherwise settling time might become very high.

NOTE: Δt must not exceed 230 ns (~ 1 second), otherwise stability is not guaranteed. For fast settling times, Δt should be as low as possible.

Time Control Loop Configuration and Status

The time control loop has some configuration and status registers (System Time Difference, Speed Counter Start, Speed Counter Difference, System Time Difference Filter Depth, and Speed Counter Filter Depth). The default settings of these registers are sufficient for proper operation of the drift compensation. Setting the Speed Counter Filter Depth (0x0935) to 0 improves control loop behavior.

The System Time Difference register (0x092C:0x092F) contains the mean value of the difference between local copy of the System Time and the System Time (Δt). This value converges to zero when both times are identical.

The Speed Counter Start register (0x0930:0x0931) represents the bandwidth of the drift compensation. The value of the Speed Counter Difference register (0x0932:0x0933) represents the deviation between the clock periods of the Reference Clock and the local ESC.

I-56 Slave Controller – Technology

Distributed Clocks

Name

Description

0x0900:0x090F

Receive Time Port n

Local time when receiving frame on Port n

0x0918:0x091F

Receive Time ECAT Processing Unit

Local time when receiving frame for ECAT Processing Unit

0x0910:0x0917

System Time

Local copy of System Time (read from PDI) (local time if System Time Offset=0)

0x0920:0x0927

System Time Offset

Time difference between System Time and local time

0x0928:0x092B

System Time Delay

Delay between Reference Clock and the ESC

0x092C:0x092F

System Time Difference

Mean difference between local copy of System Time and received System Time values

0x0930:0x0931

Speed Counter Start

Bandwidth for adjustment of local copy of System Time

0x0932:0x0933

Speed Counter Diff

Deviation between local clock period and Reference Clock’s clock period

0x0934

System Time Difference Filter Depth

Filter depth for averaging the received System Time deviation

0x0935

Speed Counter Filter Depth

Filter depth for averaging the clock period deviation

The System Time Difference Filter Depth register (0x0934) and the Speed Counter Filter Depth register (0x0935) set filter depths for mean value calculation of the received System Times and of the calculated clock period deviations.

Registers used for Control Loop/Drift Compensation are listed in Table 29.

Table 29: Registers for Drift Compensation

Slave Controller – Technology I-57

Distributed Clocks

Name

Referring to clock

0x0900:0x090F

Receive Time Port n

Local Time (t

Local time

)

0x0918:0x091F

Receive Time ECAT Processing Unit

Local Time (t

Local time

)

0x0910:0x0917

System Time

ECAT: Local copy of System Time when frame passed Reference Clock (t

Local time

+ t

Offset

– t

Propagation delay)

PDI: Local copy of System Time (t

Local time

+ t

Offset

)

0x0990:0y0997

SYNC0 Start Time

Local copy of System Time (t

Local time

+ t

Offset

)

0x0998:0x099F

NEXT SYNC1 Pulse

Local copy of System Time (t

Local time

+ t

Offset

)

0x09B0:0x09B7

Latch0 Time Positive Edge

Local copy of System Time (t

Local time

+ t

Offset

)

0x09B8:0x09BF

Latch0 Time Negative Edge

Local copy of System Time (t

Local time

+ t

Offset

)

0x09C0:0x09C7

Latch1 Time Positive Edge

Local copy of System Time (t

Local time

+ t

Offset

)

0x09C8:0x09CF

Latch1 Time Negative Edge

Local copy of System Time (t

Local time

+ t

Offset

)

0x09F0:0x09F3

EtherCAT Buffer Change Event Time

Local Time (t

Local time

)

0x09F8:0x09FB

PDI Buffer Start Event Time

Local Time (t

Local time

)

0x09FC:0x09FF

PDI Buffer Change Event Time

Local Time (t

Local time

)

9.1.6 Reference between DC Registers/Functions and Clocks

Table 30: Reference between DC Registers/Functions and Clocks

I-58 Slave Controller – Technology

Distributed Clocks

9.1.7 When is Synchronization established?

There are two possibilities to detect if DC synchronization of a slave is established:  Read System Time Difference (0x92C:0x092F):

If the difference is below an application specific threshold, DC has locked. Advantage: Can be read using a single BRD command for the entire network: if the upper N bits are zero, synchronization is established. Recommended if an EtherCAT slave is the reference clock. If the master is the reference clock, the threshold has to be increased to accomplish for the master jitter, which could make this solution unusable.

 Read Speed Counter Difference (0x0932:0x0933):

If the value is stable (within an application-specific range), DC has locked. Disadvantage: Loss of lock is recognized late.

9.1.8 Clock Synchronization Initialization Example

The initialization procedure of clock synchronization including propagation delay measurement, offset compensation, filter reset, and drift compensation is shown in the following. After initialization, all DC slaves are synchronized with the Reference Clock.

1. Master reads the DL Status register of all slaves and calculates the network topology.

2. Master sends a broadcast write to Receive Time Port 0 register (at least first byte). All slaves latch

the local time of the first preamble bit of this frame at all ports and at the ECAT Processing Unit. Some ESCs need the EtherCAT network to be free of frames before the broadcast write is sent.

3. Master waits until the broadcast write frame has returned.

4. Master reads all Receive Time Port 0-3 registers (depending on the topology and the Receive

Time ECAT Processing Unit register (0x0918:0x091F) which contains the upper 32 bits of the receive times.

5. Master calculates individual propagation delays and writes them to System Time Delay registers

of the slaves. Possible overruns of the 32 bit Receive Times have to be checked and taken into account.

6. Master sets System Time Offset register of the Reference Clock so that the Reference Clock is

bound to the master time. The offset for the Reference Clock is master time minus Receive Time ECAT Processing Unit (local time) of the Reference Clock.

7. Master calculates System Time offsets for all DC slaves and writes them to the System Time

Offset registers. The offset of each slave is Receive Time ECAT Processing Unit from Reference Clock minus Receive Time ECAT Processing Unit from each DC slave.

8. Master resets all slaves’s Time Control Loop filters by writing to the Speed Counter Start register

(0x0930:0x0931).

9. For static drift compensation, the master sends many separate ARMW or FRMW drift

compensation frames (e.g., 15,000 frames) to distribute the System Time of the Reference Clock to all DC slaves.

10. For dynamic drift compensation, the master sends ARMW or FRMW commands periodically to

distribute the System Time of the Reverence Clock to all DC slaves. The rate of the drift compensation commands depends on the acceptable maximum deviation.

Slave Controller – Technology I-59

Distributed Clocks

EtherCAT

device

LATCH[1:0]

SYNC[1:0]

EtherCAT

device

SYNC/LATCH[1:0]

Signal

Direction

Description

SYNC/LATCH[1:0]

IN/OUT

Combined SyncSignals / LatchSignals

or (ESC dependent)

SYNC[1:0]

OUT

SyncSignals (also named SYNC0/SYNC1)

LATCH[1:0]

LatchSignals (also named LATCH0/LATCH1)

9.2 SyncSignals and LatchSignals

ESCs with Distributed Clocks support generation of SyncSignals and time stamping of LatchSignals. The SyncSignals can be used internally for

 Interrupt generation (mapping to AL Event Request register 0x0220:0x0223 and PDI IRQ)  PDI Digital Output Update events  PDI Digital Input Latch events

The SyncSignals can also be directly mapped to output signals (SYNC[1:0]) for use by external devices, e.g., as interrupt signals (less jitter than PDI IRQ, no interrupt source decoding).

The Latch Event unit supports time stamping of up to two LatchSignals (LATCH[1:0], rising and falling edge separately), and time stamping of SyncManager events for debugging purposes.

9.2.1 Interface

The Distributed Clocks unit has the following external signals (depending on the ESC and the ESC configuration):

Not all of these signals might be available depending on the ESC and its hardware configuration.

9.2.2 Configuration

The mapping of Distributed Clocks SyncSignals and LatchSignals to the external SYNC/LATCH[1:0] signals is controlled by the setting of the Sync/Latch PDI Configuration register 0x0151. The SYNC[1:0] driver characteristics are also selected in this register. The SyncSignals are internally available for interrupt generation and Digital I/O synchronization regardless of the Sync/Latch PDI Configuration. The mapping of SyncSignals to the AL Event Request register is also controlled by the Sync/Latch PDI Configuration register 0x0151.

The length of a SyncSignal pulse is defined in the DC Pulse Length of SYNC Signals register (0x0982:0x0983). A value of 0 selects acknowledged modes.

Some ESCs support power saving options (partly disabling DC units) controlled by two bits of the ESC Configuration register (0x0141[3:2]), others have individual configuration options for each SyncSignal/LatchSignal.

The Sync/Latch signals are not driven (high-impedance) by some ESCs until the SII EEPROM is successfully loaded. Refer to Section III for details. Take care of proper SyncSignal usage while the EEPROM is not loaded (e.g. pull-down/pull-up resistors).

Figure 29: Distributed Clocks signals

Table 31: Distributed Clocks signals

I-60 Slave Controller – Technology

Distributed Clocks

SYNC0

Acknowlege

SYNC0

Single shot

Cyclic generation

Cyclic Acknowledge mode

Activation

SYNC0 Cycle Time

Acknowlege

SYNC0

Single shot Acknowledge mode

Start Time

Pulse Length of

SyncSignals

Pulse Length of SYNC

Signals (0x0982:0x0983)

SYNC0 Cycle Time (0x09A0:0x09A3)

> 0

= 0

> 0

Cyclic Generation

Single Shot

= 0

Cyclic Acknowledge

Single Shot Acknowledge

9.2.3 SyncSignal Generation

The DC Cyclic Unit / Sync Unit supports the generation of a base SyncSignal SYNC0 and a dependent SyncSignal SYNC1. The SyncSignals can both be used internally and externally of the ESC. SyncSignals can be generated at a specific System Time. Four operation modes are supported: cyclic generation, single shot, cyclic acknowledge, and single shot acknowledge mode. The acknowledged modes are typically used for interrupt generation. The interrupts have to be acknowledged by a µController.

Figure 30: SyncSignal Generation Modes

The SyncSignal operation mode is selected by the configuration of the Pulse Length and the SYNC0 Cycle Time, according to the following table:

Table 32: SyncSignal Generation Mode Selection

The cycle time of the SYNC0 signal is configured in the SYNC0 Cycle Time register (0x09A0:0x09A3), the start time is set in the Start Time Cyclic Operation register (0x0990:0x0997). After the Sync Unit is activated and the output of the SYNC0/1 signals is enabled (DC Activation register 0x0981), the Sync Unit waits until the start time is reached and generates the first SYNC0 pulse.

Some ESCs support additional activation options like auto-activation when the Start Time is written, or 64 bit extension if only 32 bit of the Start Time is written. Other options are to detect invalid Start Times and provide debug output of SyncSignals.

Internally, the SyncSignals are generated with an update rate of 100 MHz (10 ns update cycle). The jitter of the internal SyncSignal generation in comparison to the System Time is 12 ns.

Slave Controller – Technology I-61

Distributed Clocks

Name

Description

0x0141[3:2]

ESC Configuration

Enable/Disable DC Units (power saving)

0x0151

Sync/Latch PDI Configuration

Configuration of SYNC/LATCH[1:0] pins

0x0980[0]

Cyclic Unit Control

Assignment of cyclic function to EtherCAT or PDI

0x0981

Activation

Activation of cyclic function and SYNC pins

0x0982:0x0983

Pulse Length of SYNC signals

Length of SYNC impulse length

0x0984

Activation Status

Activation status of SYNC0/SYNC1

0x098E

SYNC0 Status

Status of SYNC0 signal

0x098F

SYNC1 Status

Status of SYNC1 signal

0x0990:0y0997

SYNC0 Start Time

Start System time of cyclic operation

0x0998:0x099F

NEXT SYNC1 Pulse

System Time of next Sync1 Pulse

0x09A0:0x09A3

SYNC0 Cycle Time

Cycle Time of SYNC0

0x09A4:0x09A7

SYNC1 Cycle Time

Cycle Time of SYNC1

The registers used for SyncSignal Generation are shown in Table 33.

Table 33: Registers for SyncSignal Generation

NOTE: Some of these registers are set via SII EEPROM/IP Core configuration, or they are not available in specific ESCs. Refer to Section II for details.

9.2.3.1 Cyclic Generation

In Cyclic Generation mode, the Sync unit generates isochronous SyncSignals after the Start Time. The generation ends if the Cyclic Unit is deactivated or SYNC0/1 generation is deactivated. The Cycle times are determined by the SYNC0/1 Cycle Time registers. The Pulse Length of the SYNC signals has to be greater than 0. If the Pulse Length is greater than the Cycle Time, the SyncSignal will always be activated after the Start Time.

9.2.3.2 Single Shot Mode

In Single Shot mode (SYNC0 Cycle Time set to 0), only one SyncSignal pulse is generated after the Start Time is reached. Another pulse can only be generated by deactivating the Cyclic Unit (0x0981[0]=0), reprogramming the Start Time, and reactivation of the Cyclic Unit.

9.2.3.3 Cyclic Acknowledge Mode

The Cyclic Acknowledge mode is typically used for generation of isochronous interrupts. The acknowledged modes are selected by setting the Pulse Length of SYNC Signals to 0 (0x0982:0x0983). Each SyncSignal pulse remains active until it is acknowledged – typically by a µController – by reading the appropriate SYNC0 or SYNC1 Status register (0x098E, 0x098F). The first pulse is generated after the Start Time is reached, following pulses are generated when the next regular SYNC0/1 event would occur.

9.2.3.4 Single Shot Acknowledge Mode

In Single Shot Acknowledge mode (both Pulse Length of SYNC Signals and SYNC0 Cycle Time are

0), only one pulse is generated when the Start Time is reached. The pulse remains active until it is acknowledged by reading the appropriate SYNC0/1 Status registers. Another pulse can only be generated by deactivating the Cyclic Unit (0x0981[0]=0), reprogramming the Start Time, and reactivation of the Cyclic Unit.

I-62 Slave Controller – Technology

Distributed Clocks

SYNC0

SYNC1

SYNC1 Cycle Time

SYNC0 Cycle Time

SYNC0

SYNC1

SYNC1 Cycle Time

SYNC0 Cycle Time

SYNC1 Cycle Time < SYNC0 Cycle Time

SYNC1 Cycle Time = SYNC0 Cycle Time

SYNC0

SYNC1

SYNC1 Cycle Time

SYNC0 Cycle Time

SYNC1 Cycle Time = 2*SYNC0 Cycle Time

SYNC1 Cycle Time

SYNC0

SYNC1

SYNC1 Cycle Time

SYNC0 Cycle Time

SYNC1 Cycle Time > SYNC0 Cycle Time

and

SYNC1 Cycle Time < 2*SYNC0 Cycle Time

SYNC1 Cycle Time

Start Time

9.2.3.5 SYNC1 Generation

The second SyncSignal (SYNC1) depends on SYNC0, it can be generated with a predefined delay after SYNC0 pulses. The delay is configured in the SYNC1 Cycle Time register (0x09A4:0x09A7).

If the SYNC1 Cycle Time is larger than the SYNC0 Cycle Time, it will be generated as follows: when the Start Time Cyclic Operation is reached, a SYNC0 pulse is generated. The SYNC1 pulse is generated after the SYNC0 pulse with a delay of SYNC1 Cycle Time. The next SYNC1 pulse is generated when the next SYNC0 pulse was generated, plus the SYNC1 Cycle Time.

Some example configurations are shown in the following figure:

NOTE: If The SYNC1 Cycle Time is 0, SYNC1 reflects SYNC0.

Slave Controller – Technology I-63

Figure 31: SYNC0/1 Cycle Time Examples

Distributed Clocks

9.2.3.6 SyncSignal Initialization Example

The SyncSignal generation is initialized with the following procedure:

1. Enable DC SYNC Out Unit in ESC Configuration register (0x0141[2]=1; specific ESCs only)

2. Set SYNC/Latch PDI Configuration register (0x0151, initialized by SII EEPROM) to SYNC0/1

output with appropriate driver settings.

3. Set Pulse Length register (0x0982:0x0983, initialized by EEPROM) to pulse length of SYNC

signals. Select a value > 0 ns for cyclic repetition of the SyncSignals

4. Assign Sync Unit to ECAT or PDI (0x0980, part of ESI)

5. Set cycle time of SYNC0 signal (0x09A0:0x09A3) and for SYNC1 signal (0x09A4:0x09A7)

6. Set Start Time of Cyclic Operation (0x0990:0x0997) to a time later than the time the cyclic

generation will be activated (end of activation frame; e.g., read the System Time and add the time for writing Start Time and Activation). For 32 bit DCs, the SyncSignal generation will start at worst after a turn-over of the System Time (~ 4 s), but with 64 bit DCs, SyncSignal generation may start in hundreds of years.

7. Activate Cyclic Operation (0x0981[0]=1) to start cyclic generation of SyncSignals and activate

SYNC0/1 generation (0x0981[2:1]=0x3). The Sync Unit waits until the Start Time of Cyclic Operation is reached for the generation of the first SYNC0 pulse.

Register Start Time of Cyclic Operation and register Next SYNC1 pulse can be read to get the time of the next output event. In the acknowledged modes, the Sync0/1 Status registers (0x098E:0x098F) give the status of the SyncSignals. The SyncSignals are acknowledged by reading the SYNC0/1 Status registers.

9.2.4 LatchSignals

The DC Latch Unit enables time stamping of LatchSignal events for two external signals, LATCH0 and LATCH1. Both rising edge and falling edge time stamps are recorded. Additionally, time stamping of SyncManager events is possible with some ESCs.

LatchSignals are sampled with a sample rate of 100 MHz, the corresponding time stamp has an internal jitter of 11 ns.

The state of the LatchSignals can be read from the Latch Status registers (0x09AE:0x09AF) – if supported by the ESC.

The DC Latch Unit supports two modes: single event or continuous mode, configured in the Latch0/1 Control registers (0x09A8:0x09A8).

I-64 Slave Controller – Technology

Distributed Clocks

Name

Description

0x0141[3:2]

ESC Configuration

Enable/Disable DC Units (power saving)

0x0151

Sync/Latch PDI Configuration

Configuration of SYNC/LATCH[1:0] pins

0x0980[5:4]

Cyclic Unit Control

Assignment of cyclic function to EtherCAT or PDI

0x09A8

Latch0 Control

Latch unit configuration for Latch0

0x09A9

Latch1 Control

Latch unit configuration for Latch1

0x09AE

Latch0 Status

Latch status of Latch0

0x09AF

Latch1 Status

Latch status Latch1

0x09B0:0x09B7

Latch0 Time Positive Edge

System time at positive edge Latch0

0x09B8:0x09BF

Latch0 Time Negative Edge

System time at negative edge Latch0

0x09C0:0x09C7

Latch1 Time Positive Edge

System time at positive edge Latch1

0x09C8:0x09CF

Latch1 Time Negative Edge

System time at negative edge Latch1

0x09F0:0x09F3

EtherCAT Buffer Change Event Time

Local time at beginning of frame causing ECAT SyncManager buffer change event

0x09F8:0x09FB

PDI Buffer Start Event Time

Local time at PDI SyncManager buffer start event

0x09FC:0x09FF

PDI Buffer Change Event Time

Local time at PDI SyncManager buffer change event

The registers used for LatchSignal event time stamping are shown in Table 34:

Table 34: Registers for Latch Input Events

NOTE: Some of these registers are set via SII EEPROM/IP Core configuration, or they are not available in specific ESCs. Refer to Section II for details.

9.2.4.1 Single Event Mode

In single event mode, only the timestamps of the first rising and the first falling edge of the LatchSignals are recorded. The Latch Status registers (0x09AE:0x09AF) contain information about the events which already have occurred. The Latch Time registers (0x09B0 to 0x09CF) contain the time stamps.

Each event is acknowledged by reading the corresponding Latch Time register. After reading the Latch Time register, the Latch unit is waiting for the next event. Latch events are mapped to the AL Event Request register in single event mode.

9.2.4.2 Continuous Mode

In continuous mode, each event is stored in the Latch Time registers. At reading, the time stamp of the last event is read. The Latch Status registers (0x09AE:0x09AF) do not reflect the latch event states in continuous mode.

9.2.4.3 SyncManager Event

Some ESCs support debugging of SyncManager interactions with time stamps for buffer events. The last event can be read out at the SyncManager Event Time registers (0x09F0:0x09FF), if the SyncManager is configured appropriately.

9.2.5 ECAT or PDI Control

The SyncSignal unit and the two LatchSignal units of the Distributed Clocks entity can be assigned independently by the master to be controlled either by ECAT or a local µController (PDI) using the Cyclic Unit Control register 0x0980. With PDI control, a µController can e.g. set up cyclic interrupts for itself.

Slave Controller – Technology I-65

Distributed Clocks

Output

µC generates

rising edge for

ESC 1 & 2

µC reads Latch0 Time pos. edge

from ESC 1

µC writes Latch0 Time pos. edge

from ESC 1 to System Time

ESC 1

DC source

ESC 2

destination

µController

Latch0

9.3 System Time PDI Controlled

Sometimes Distributed Clocks of different EtherCAT networks have to be synchronized. One solution is master-master communication, the other one is based on a physical device which is present in both EtherCAT networks. One of the networks contains the DC Reference Clock (DC source), the other one – DC destination – is synchronized to the Reference Clock in the DC source network.

Some ESCs support such a synchronization by a different functionality of the System Time register (0x0910:0x0913). In normal operation mode, a write access initiated by the EtherCAT master to the System Time register triggers the synchronization: the written value is compared to the local copy of the System Time, and the difference is fed into the control loop. If the System Time is PDI controlled, the PDI writes the System Time register, and the written value is compared to the DC Latch0 Time Positive Edge register (0x09B0:0x09B3). This feature makes the accuracy of the synchronization independent of the µController/PDI response times.

The following figures illustrate how the System Time is transferred from the DC source to the DC destination. ESC 1 and ESC 2 are located in different EtherCAT networks. The EtherCAT network of ESC 1 contains the Reference Clock, the network of ESC 2 will become synchronized to this Reference Clock. ESC 2 is the “reference clock” of its EtherCAT network. There are two options for synchronization, which has to be performed on a regular basis.

The first option is to let the µController generate a trigger pulse for ESC 1 and 2. The time of the rising edge is stored in the Latch0 Time Positive Edge register both in ESC 1 and 2. Afterwards, the µController reads this time from ESC 1 and writes it into the System Time register of ESC 2. The difference of the Latch0 times is used to feed the control loop.

I-66 Slave Controller – Technology

Figure 32: System Time PDI Controlled with three steps

Distributed Clocks

ESC 1

DC source

Sync0

IRQ

ESC 1 generates SyncPulse

for ESC 2 and IRQ for µC

µC writes Sync time of ESC 1

(calculated by µC) to System

Time register of ESC 2

µController

ESC 2

destination

Latch0

The second option uses a SyncSignal output of ESC 1 to trigger Latch0 at ESC 2 and an interrupt at the µController. Upon receiving an interrupt, the µController writes the time of the SyncSignal pulse to the System Time register of ESC 2. The µController has to calculate the time of the SyncSignal based upon Start Time Cyclic Operation and SYNC Cycle Time configuration of ESC 1 from interrupt to interrupt. The advantage of the second solution is less communication, the disadvantages are more calculation overhead and error detection/troubleshooting.

Figure 33: System Time PDI Controlled with two steps

Slave Controller – Technology I-67

Distributed Clocks

I O Calc

NC Task

Frame D 10% U

I O Calc

Frame D 10% U

Sync0

Master

Slave

User Shift (Master)

Fixed Shift (precalc.)

Example: 10% of Cycle Time reserved for Jitter

Frame Delay

User Shift (Slave)

DC Base

9.4 Communication Timing

Three communication modes are possible:

1. Free Run

EtherCAT Communication and application are running independently from each other.

2. Synchronized to Output Event

The slave application is synchronized to an Output event. If no Outputs are used the Input event is used for synchronization.

3. Synchronized to SyncSignal

Application is synchronized to the SyncSignal.

For further information please refer to the corresponding section within the EtherCAT Information System.

The Communication Timing with use of Distributed Clocks is explained in Figure 34.

IO(Master)

Time to load IO Data to communication buffer and vice versa.

Calc(Master)

Processing time of the master.

Frame(Communication)

Time to transmit the IO-Data-Frame (about 5µs overhead plus 80ns per Byte of Data).

D(Communication)

Delay time of the EtherCAT-Slaves to transfer data (approx. 1 µs with 100BASE-TX, plus line delay of approx. 5ns per m).

Jitter(Communication)

Depends mostly on Master timing quality.

U(Communication-Master)

Shift time that is adjusted internally by the master to deal with delays needed by the master and adjust the cycle time.

U(Slave)

Delay time of the EtherCAT-Slaves. This can be set by each slave individually and is usually 0. There is a need to set this parameter in case of timing inaccuracies of the slave or to deal with slaves that have a slow output method compared to others with high speed output.

Figure 34: DC Timing Signals in relation to Communication

I-68 Slave Controller – Technology

Distributed Clocks

Cycle Time Jitter

Cycle Time Jitter is application-specific and depends on the jitter of the master system, the used infrastructure components and the slaves. This example assumes a time of 10% of the cycle time for jitter compensation.

Slave Controller – Technology I-69

EtherCAT State Machine

Init

(IP)

Pre-Operational

Operational

Safe-Operational

(OP)

(PI)

(PS) (SP)

(OS)(SO)

(SI)

(OI)

Bootstrap

(optional)

(IB) (BI)

10 EtherCAT State Machine

The EtherCAT State machine (ESM) is responsible for the coordination of master and slave applications at start up and during operation. State changes are typically initiated by requests of the master. They are acknowledged by the local application after the associated operations have been executed. Unsolicited state changes of the local application are also possible.

Simple devices without a µController can be configured to use EtherCAT State Machine emulation. These devices simply accept and acknowledge any state change automatically.

There are four states an EtherCAT slave shall support, plus one optional state:

 Init (state after Reset)  Pre-Operational  Safe-Operational  Operational  Bootstrap (optional)

The states and the allowed state changes are shown in Figure 35:

Figure 35: EtherCAT State Machine

NOTE: Not all state changes are possible, e.g., the transition from ‘Init’ to ‘Operational’ requires the following

sequence: Init → Pre-Operational → Save-Operational → Operational.

Each state defines required services. Before a state change is confirmed by the slave all services required for the requested state have to be provided or stopped respectively.

I-70 Slave Controller – Technology

EtherCAT State Machine

Name

Description

0x0120:0x0121

AL Control

Requested state by the master

0x0130:0x0131

AL Status

AL Status of the slave application

0x0134:0x0135

AL Status Code

Error codes from the slave application

0x0141[0]

ESC Configuration

Device emulation configuration

AL Control Register 0x0120

AL Status Register 0x0130

Request Init state

Init state

Request Bootstrap state (optional)

Bootstrap state (optional)

Request Pre-Operational state

Pre-Operational state

Request SAFE-Operational state

SAFE-Operational state

Request Operational state

Operational state

10.1 EtherCAT State Machine Registers

The state machine is controlled and monitored via registers within the ESC. The master requests state changes by writing to the AL Control register. The slave indicates its state in the AL Status register and puts error codes into the AL Status Code register.

Table 35: Registers for the EtherCAT State Machine

NOTE: The PDI Control register is set via SII EEPROM/IP Core configuration, other registers might not be available in specific ESCs. Refer to Section II and Section III for details.

10.1.1 AL Control and AL Status Register

Writing the AL Control register (0x0120:0x0121) initiates a state transition of the device state machine. The AL Status register (0x0130:0x0131) reflects the current state of the slave.

Table 36: AL Control and AL Status Register Values

10.1.2 Device Emulation

Simple devices (without µController) have the device emulation enabled (0x0141[0]=1). The AL Control register is directly copied into the AL Status register by the ESC. The master should not set the Error Indication Acknowledge bit for such slaves at all, because setting this bit would result in setting the Error Indication bit – although no error occurred.

10.1.3 Error Indication and AL Status Code Register

The slave indicates errors during a state transition by setting the Error Indication flag (0x0130[4]=1) and writing an error description into the AL Status Code register (0x0134:0x0135). The master acknowledges the Error Indication flag of the slave by setting the Error Indication Acknowledge flag (0x0120[4]).

For more information on defined AL Status Codes, refer to the EtherCAT Knowledge Base available at the EtherCAT Technology Group website (http://www.ethercat.org, Guidelines and Protocol Enhancements).

Slave Controller – Technology I-71

SII EEPROM

EtherCAT Slave Controller Configuration Area VendorId ProductCode RevisionNo SerialNo

Additional Information (Subdivided in Categories)

…

0 8

Word

Hardware Delays Bootstrap Mailbox Config

Reserved

Mailbox Sync Man Config

Category FMMU

Category Strings

Category Generals

Category SyncManager

Category Tx- / RxPDO for each PDO

11 SII EEPROM

EtherCAT slave controllers use a mandatory NVRAM (typically a serial EEPROM with I²C interface) to store EtherCAT Slave Information (ESI). EEPROM sizes from 1 Kbit up to 4 Mbit are supported, depending on the ESC.

The EtherCAT IP Core supports omitting the serial I²C EEPROM if a µController with read/write access to an NVRAM (e.g., the one which contains the µController’s program and data, or the FPGA configuration EPPROM) is used to emulate the EEPROM transactions. Since the logical interface is the same in this case, the EEPROM emulation is treated to be equivalent to the typical I²C EEPROM solution throughout this chapter. Refer to chapter 11.2.4 for more details about EEPROM emulation.

The EEPROM structure is shown in Figure 36. The ESI uses word addressing.

Figure 36: SII EEPROM Layout

At least the information stored in the address range from word 0 to 63 (0x00 to 0x3F) is mandatory, as well as the general category (→ absolute minimum SII EEPROM size is 2Kbit, complex devices with many categories should be equipped with 32 Kbit EEPROMs or larger). The ESC Configuration area is used by the ESC for configuration. All other parts are used by the master or the local application.

For a more detailed description of the ESI and other mandatory parts refer to the ETG.2000 EtherCAT Slave Information (ESI) Specification, available from the download section of the EtherCAT Technology Group website (http://www.ethercat.org).

I-72 Slave Controller – Technology

SII EEPROM

Word

Address

Parameter

Description

0x0

PDI Control/ ESC Configuration

Initialization value for PDI Control register (EEPROM ADR 0x0000[9] is also mapped to register 0x0110[2])

0x0140:0x0141

0x1

PDI Configuration

Initialization value for PDI Configuration register

0x0150:0x0151

0x2

Pulse Length of SYNC Signals

Initialization value for Pulse Length of SYNC Signals register

0x0982:0x0983

0x3

Extended PDI Configuration

Initialization value for extended PDI Configuration register

0x0152:0x0153

0x4

Configured Station Alias

Initialization value for Configured Station Alias Address register

0x0012:0x0013

0x5

Reserved

Reserved, shall be zero

0x6

Reserved

Reserved, shall be zero

0x7

Checksum

Low byte contains remainder of division of word 0 to word 6 as unsigned number divided by the polynomial x8+x2+x+1(initial value 0xFF).

NOTE: For debugging purposes it is possible to disable the checksum validation with a checksum value of 0x88A4. Never use this for production!

11.1 SII EEPROM Content

The ESC Configuration Area (EEPROM word addresses 0 to 7) is automatically read by the ESC after power-on or reset. It contains the PDI configuration, DC settings, and the Configured Station Alias. The consistency of the ESC Configuration data is secured with a checksum.

For SII coding refer to ETG.1000 EtherCAT Specification, Part 6, Clause 5.4, available in the download area of the EtherCAT Technology Group website (http://www.ethercat.org).

The EtherCAT master can invoke reloading the EEPROM content. In this case the Configured Station Alias register 0x0012:0x0013 and ESC Configuration register bits 0x0141[1,4,5,6,7] (enhanced link detection) are not transferred into the registers, they are only transferred at the initial EEPROM loading after power-on or reset.

The ESC Configuration Area is shown in Table 37.

Table 37: ESC Configuration Area

NOTE: Reserved words or reserved bits of the ESC Configuration Area should be filled with 0.

Slave Controller – Technology I-73

SII EEPROM

Word

Address

Parameter

Word

Address

Parameter

0x0

PDI Control

0x14

Bootstrap Receive Mailbox Offset

0x1

PDI Configuration

0x15

Bootstrap Receive Mailbox Size

0x2

Pulse Length of SYNC Signals

0x16

Bootstrap Send Mailbox Offset

0x3

Extended PDI Configuration

0x17

Bootstrap Send Mailbox Size

0x4

Configured Station Alias

0x18

Standard Receive Mailbox Offset

0x5:0x6

Reserved

0x19

Standard Receive Mailbox Size

0x7

Checksum

0x1A

Standard Send Mailbox Offset

0x8:0x9

Vendor ID

0x1B

Standard Send Mailbox Size

0xA:0xB

Product Code

0x1C

Mailbox Protocol

0xC:0xD

Revision Number

0x1D:0x3D

Reserved

0xE:0xF

Serial Number

0x3E

Size

0x10

Execution Delay

0x3F

Version

0x11

Port0 Delay

0x40

First Category Type/Vendor Specific

0x12

Port1 Delay

0x41

Following Category Word Size

0x13

Reserved

0x42

Category Data

…

Second Category …

Description

0x0500

EEPROM Configuration

0x0501

EEPROM PDI Access State

0x0502:0x0503

EEPROM Control/Status

0x0504:0x0507

EEPROM Address

0x0508:0x050F

EEPROM Data

An excerpt of the SII EEPROM content following the ESC Configuration area is shown in Table 38. For more information, refer to the ETG.2000 EtherCAT Slave Information (ESI) Specification, available from the download section of the EtherCAT Technology Group website (http://www.ethercat.org).

Table 38: SII EEPROM Content Excerpt

11.2 SII EEPROM Logical Interface

The SII EEPROM interface of the ESC is either controlled by EtherCAT or by the PDI. Initially, EtherCAT has EEPROM interface access, but it can transfer access to the PDI.

Table 39: SII EEPROM Interface Register Overview

The EEPROM interface supports three commands: write to one EEPROM address (1 Word), read from EEPROM (2 or 4 Words, depending on ESC), or reload ESC configuration from EEPROM.

I-74 Slave Controller – Technology

SII EEPROM

Bit

Name

Description

Checksum Error

ESC Configuration Area checksum is wrong (after device initialization or EEPROM reload). Registers initialized with EEPROM values keep their value.

Reason: CRC error Solution: Check CRC

Error Device Information

ESC Configuration not loaded Reasons: Checksum error, acknowledge error, EEPROM

missing

Solution: Check other error bits

Error Acknowledge/ Command

Missing Acknowledge or invalid command Reason: a) Missing acknowledge from EEPROM chip (see

below). EEPROM chip is busy performing operations internally or EEPROM chip is not

available. b) Invalid command issued Solution: a) Retry access. EEPROM device does not

acknowledge if it is internally busy. b) Use valid commands EEPROM Emulation only: Missing Acknowledge error indicates

a temporary failure. Invalid command error is automatically supported by the EEPROM interface.

Error Write Enable

Write without Write Enable (ECAT control only): Reason: ECAT issued a write command without Write Enable

bit set (0x0502[0]) Solution: Set Write Enable bit in the same frame as the write

command

11.2.1 SII EEPROM Errors

The ESC retries reading the EEPROM after power-on or reset once if an error has occurred (missing acknowledge, wrong checksum). If reading the ESC Configuration Area fails twice, the Error Device Information bit is set, and the PDI Operational bit in the ESC DL Status register (0x0110[0]) remains clear and the EEPROM_Loaded signal (if available) remains inactive. The process memory is not accessible until the ESC Configuration Area is loaded successfully.

All registers initialized by the ESC Configuration Area keep their values in case of an error. This is also true for the Error Device Information bit as well as the PDI Operational bit. Only if the EEPROM was loaded/reloaded successfully, the registers take over the new values (except for Configured Station Alias register 0x0012:0x0013 and ESC Configuration register bits 0x0141[1,4,5,6,7] – enhanced link detection).

The SII EEPROM interface has these error status bits in register EEPROM Control/Status (0x0502:0x0503):

Table 40: SII EEPROM Interface Errors

Slave Controller – Technology I-75

SII EEPROM

11.2.1.1 Missing Acknowledge

Missing acknowledges from the EEPROM chip are a common issue, especially if a fast PDI uses the EEPROM interface. E.g., a write access to the EEPROM with missing acknowledge may look like this:

1. ECAT/PDI issue write command (first command)

2. ESC is busy transferring the write data to the EEPROM chip.

3. ESC is not busy anymore. EEPROM chip is internally busy transferring data from input buffer to storage area.

4. ECAT/PDI issue a second command.

5. ESC is busy transferring the write data to the EEPROM chip. EEPROM chip does not acknowledge any access until internal transfer has finished (may take up to several ms).

6. ESC is not busy anymore. Error Acknowledge/Command bit is set. (ESC has to re-issue the second command after EEPROM chip is finished and the command is acknowledged).

7. EEPROM chip finishes internal transfer.

8. ESC re-issues the second command, the command is acknowledged and executed successful.

This is also possible for a read access, because some EEPROM chips require an idle period between any two accesses. During this idle period, they do not acknowledge any access.

11.2.2 SII EEPROM Interface Assignment to ECAT/PDI

The EtherCAT master controls the EEPROM interface (default) if EEPROM configuration register 0x0500[0]=0 and EEPROM PDI Access register 0x0501[0]=0, otherwise PDI controls the EEPROM interface. These access rights should be checked by both sides before using the EEPROM Interface.

A typical EEPROM interface control hand-over is as follows:

1. ECAT assigns EEPROM interface to PDI by writing 0x0500[0]=1

2. If PDI wishes to access EEPROM, it takes over EEPROM control by writing 0x0501[0]=1.

3. PDI issues EEPROM commands.

4. After PDI has finished EEPROM commands, PDI releases EEPROM control by writing 0x0501[0]=0.

5. ECAT may take back the EEPROM interface by writing 0x0500[0]=0

6. ECAT checks EEPROM control by reading 0x0501

7. ECAT issues EEPROM commands.

If the PDI does not release EEPROM control (e.g. because of a software failure), ECAT can force releasing the access:

1. ECAT writes 0x02 to register 0x0500 (as the result, 0x0501[0] is cleared)

2. ECAT writes 0x00 to register 0x0500

3. ECAT has control over EEPROM interface

I-76 Slave Controller – Technology

SII EEPROM

11.2.3 Read/Write/Reload Example

The following steps have to be performed for an SII EEPROM read or write access:

1. Check if the Busy bit of the EEPROM Status register is cleared (0x0502[15]=0) and the EEPROM interface is not busy, otherwise wait until the EEPROM interface is not busy anymore.

2. Check if the Error bits of the EEPROM Status register are cleared. If not, write “000” to the command register (register 0x0502 bits [10:8]).

3. Write EEPROM word address to EEPROM Address register.

4. Write command only: put write data into EEPROM Data register (1 word/2 byte only).

5. Issue command by writing to Control register. a) For a read command, write 001 into Command Register 0x0502[10:8]. b) For a write command, write 1 into Write Enable bit 0x0502[0] and 010 into Command Register 0x0502[10:8]. Both bits have to be written in one frame. The Write enable bit realizes a write protection mechanism. It is valid for subsequent EEPROM commands issued in the same frame and self-clearing afterwards. The Write enable bit needs not to be written from PDI if it controls the EEPROM interface. c) For a reload command, write 100 into Command Register 0x0502[10:8].

6. The command is executed after the EOF if the EtherCAT frame had no errors. With PDI control, the command is executed immediately.

7. Wait until the Busy bit of the EEPROM Status register is cleared.

8. Check the Error bits of the EEPROM Status register. The Error bits are cleared by clearing the command register. Retry command (back to step 5) if EEPROM acknowledge was missing. If necessary, wait some time before retrying to allow slow EEPROMs to store the data internally.

9. a) For a Read command: Read data is available in EEPROM Data registers (2 or 4 Words, depending on ESC – check register 0x0502[6]). b) For a Reload command: ESC configuration is reloaded into appropriate registers.

NOTE: The Command register bits are self-clearing. Manually clearing the command register will also clear the status information.

11.2.4 EEPROM Emulation

The EEPROM emulation mode is used in ESCs with a non-volatile memory (NVRAM) attached to a µController. The ESC configuration and the device description can be stored in the NVRAM of the µController, e.g., together with the program or FPGA configuration code. An additional I²C EEPROM chip for the ESC is not needed anymore if EEPROM emulation is used.

The µController emulates the EEPROM interface actions of the ESC and executes all EEPROM reload, read, and write requests. EEPROM write data is stored in the NVRAM of the µController, and EEPROM read data is read from the NVRAM and presented to the EEPROM interface of the ESC.

From the EtherCAT master’s point of view, EEPROM emulation mode is equivalent to an I²C EEPROM. The master issues EEPROM commands and waits until the EEPROM interface is not busy anymore.

In EEPROM emulation mode, the EEPROM interface of the ESC issues an interrupt to the µController if an EEPROM command is pending and sets the busy bit. While the busy bit is set, the µController can read out the command and the EEPROM address. For a write access, write data is present in the data register. For a read command, read data has to be stored in the data register by the µController. A reload command requires the µController to place specific reload data in the data register, refer to section II for details.

Once the µController has finished reading/writing the EEPROM data register, it acknowledges the command by writing to the EEPROM command register bits. The µController has to write the command value it has executed into the EEPROM command register. Errors can be indicated using two of the error bits. After acknowledging the command, the EEPROM state machine is not busy anymore and the interrupt is released.

NOTE: The EEPROM can be assigned to the PDI even if EEPROM Emulation is used.

Slave Controller – Technology I-77

SII EEPROM

EtherCAT

device

EEPROM_DATA

EEPROM_CLK

EEPROM_SIZE

Signal

Direction

Description

EEPROM_CLK

OUT

I²C clock

EEPROM_DATA

BIDIR

I²C data

EEPROM_SIZE

EEPROM size configuration

EEPROM Size

Address Bytes

Max. I²C Address Bits

EEPROM_SIZE signal

1 Kbit – 16 Kbit

11 0 32 Kbit – 4 Mbit

11.3 SII EEPROM Electrical Interface (I2C)

The SII EEPROM Interface is intended to be a point-to-point interface between the ESC and the I²C EEPROM. If other I²C masters are required to access the I²C bus, the ESC must be held in reset state (e.g. for in-circuit-programming of the EEPROM), otherwise access collisions are possible.

The SII EEPROM interface has the following signals1:

Figure 37: I²C EEPROM signals

Table 41: I²C EEPROM signals

Both EEPROM_CLK and EEPROM_DATA must have a pull-up resistor (4.7 kΩ recommended for ESCs), either integrated into the ESC or connected externally.

11.3.1 Addressing

EtherCAT and ESCs use word addressing when accessing the EEPROM, although the I²C interface actually uses byte addressing. The lowest address bit A[0] is added internally by the EEPROM interface controller of the ESCs. I.e., the EEPROM address register (0x0504:0x0507) reflects the physical EEPROM address bits A[18:1] (higher address bits are reserved/are zero).

SII EEPROM word 0 is located at I²C address 0, i.e., the I²C device address has to be set to 0.

11.3.2 EEPROM Size

Depending on the EEPROM size, one out of two EEPROM algorithms has to be selected with the EEPROM_SIZE configuration signal. Smaller EEPROMs need only one address byte, larger ones need two address bytes:

Table 42: EEPROM Size

I-78 Slave Controller – Technology

The availability of the EEPROM signals as well as their names depend on the specific ESC.

SII EEPROM

Bit

Description

Read/Write access: 0: Write 1: Read

[3:1]

Chip Select Bits/Highest Address Bits

[7:4]

Control Code: 1010

Step

Description

Up to 16 Kbit

32 Kbit – 4 Mbit

Start condition

Control Byte (Write)

A[10:8]

A[18:16]

High Address Byte

Not preset

A[15:8]

Low Address Byte

A[7:0]

Low Data Byte

D[7:0]

High Data Byte

D[15:8]

Stop condition

11.3.3 I²C Access Protocol

Each EEPROM access begins with a Start condition and ends with a Stop condition. Data is transferred byte-wise, and each byte is acknowledged by the recipient.

The Start condition is a falling edge on EEPROM_DATA while EEPROM_CLK is high, the Stop condition is a rising edge on EEPROM_DATA while EEPROM_CLK is high. In all other cases, EEPROM_DATA has to remain stable while EEPROM_CLK is high, as this indicates valid data. A byte transfer is acknowledged in an additional bit, which is driven low by the recipient of the byte transfer if it acknowledges the byte.

NOTE: If the EEPROM does not acknowledge an access (Ack bit=high), it might be busy internally. Especially if the EERPOM interface is handled by a µController via the PDI, this situation may come up, because many µControllers can write to the EEPROM interface much faster than many EEPROMs can transfer the data from their input registers into their NVRAM.

The first byte of an I²C access is the Control Byte (Bit 7/MSB is transferred first):

Table 43: I²C Control Byte

Depending on the access, either read data will follow or additional address bytes and write data. This is described in the following chapters.

The EEPROM has an internal byte pointer, which is incremented automatically after each data byte transfer.

For more details about the I²C protocol, refer to “The I²C-Bus Specification”, available from NXP (http://www.nxp.com, document number 39340011) and http://www.i2c-bus.org.

11.3.3.1 Write Access

An EEPROM write access always writes one word (2 bytes) to the EEPROM. In this case, page boundaries are not relevant, because they will not be violated.

The ESC will perform the following steps for a write access to the EERPOM:

Table 44: I²C Write Access

* This step is only for EEPROMs larger than 16 Kbit.

Slave Controller – Technology I-79

SII EEPROM

Step

Description

Up to 16 Kbit

32 Kbit – 4 Mbit

Start condition

Control Byte (Write)

A[10:8]

A[18:16]

High Address Byte

Not present

A[15:8]

Low Address Byte

A[7:0]

Start condition

Control Byte (Read)

A[10:8]

A[18:16]

Data Byte 0

D0 [7:0]

Data Byte 1

D1 [7:0]

…

… … N+7

Data Byte N

DN [7:0]

N+8

Stop condition

Parameter

Comment

Clk

EEPROM clock period

Write

Write access time (without errors)

Read

Read access time (without errors)

Delay

Time until configuration loading begins after Reset is gone

R/W1 001

EEPROM_CLK

A10

EEPROM_DATA

Clk

A8A9 Ack A7 A5A6 A4 A2A3 A1 Ack

Start Control Byte Low Address Byte

D3 D1D2 Ack D15 D13D14 D12 D10D11 D9 D8

Data Byte 0 Data Byte 1

D0D7 D5D6 D4

EEPROM_CLK

EEPROM_DATA

Stop

Ack

Write

11.3.3.2 Read Access

An EEPROM read access reads 2 or 4 words (4 or 8 bytes, depending on device capabilities) from the EEPROM, the load or reload EEPROM access typically reads 8 words (16 bytes). The address wraparound at the end of the EEPROM address space has to be taken into account by the application; the ESC has no knowledge about it.

The ESC will perform the following steps for a read access to the EERPOM. At first, the address is written to the EEPROM, then the data is read (N=3 or N=7):

Table 45: I²C Read Access

* This step is only for EEPROMs larger than 16 Kbit.

11.3.4 Timing specifications

Table 46: EEPROM timing characteristics

Figure 38: Write access (1 address byte, up to 16 Kbit EEPROMs)

I-80 Slave Controller – Technology

BECKHOFF EtherCAT Technology Section I User Manual

Specifications and Main Features

Frequently Asked Questions

User Manual

DOCUMENT ORGANIZATION

DOCUMENT HISTORY

CONTENTS

TABLES

FIGURES

ABBREVIATIONS

1 EtherCAT Slave Controller Overview

1.1 EtherCAT Slave Controller Function Blocks EtherCAT Interfaces (Ethernet/EBUS)

1.2 Further Reading on EtherCAT and ESCs

1.3 Scope of Section I

2 EtherCAT Protocol

2.1 EtherCAT Header

2.2 EtherCAT Datagram

2.3 EtherCAT Addressing Modes

2.3.1 Device Addressing

2.3.2 Logical Addressing

2.4 Working Counter

2.5 EtherCAT Command Types

3 Frame Processing

3.1 Loop Control and Loop State

3.2 Frame Processing Order

3.3 Permanent Ports and Bridge Port

3.4 Shadow Buffer for Register Write Operations

3.5 Circulating Frames

3.5.1 Unconnected Port 0

3.6 Non-EtherCAT Protocols

3.7 Special Functions of Port 0

4 Physical Layer Common Features

4.1 Link Status

4.2 Selecting Standard/Enhanced Link Detection

4.3 FIFO Size Reduction

4.4 Frame Error Detection

5 Ethernet Physical Layer

5.1 Requirements to Ethernet PHYs

5.2 PHY reset and Link partner notification/loop closing

5.3 MII Interface

5.4 RMII Interface

5.5 RGMII Interface

5.5.1 RGMII In-Band Link Status

5.6 Link Detection

5.6.1 LINK_MII Signal

5.6.2 MI Link Detection and Configuration

5.7 Standard and Enhanced MII Link Detection

5.8 EtherCAT over Optical Links (FX)

5.8.1 Link partner notification and loop closing

5.8.2 Far-End-Fault (FEF)

5.8.3 ESCs with native FX support

5.8.4 ESCs without native FX support

5.9 Gigabit Ethernet PHYs

5.10 MII Management Interface (MI)

5.10.1 PHY Addressing/PHY Address Offset

5.10.2 Logical Interface

5.10.2.1 MI read/write example

5.10.2.2 MI Interface Assignment to ECAT/PDI

5.10.3 MI Protocol

5.10.4 Timing specifications

5.11 MII management example schematic

5.12 Ethernet Termination and Grounding Recommendation

5.13 Ethernet Connector (RJ45 / M12)

5.14 Back-to-Back MII Connection

5.14.1 ESC to ESC Connection

5.14.2 ESC to Standard Ethernet MAC

6 EBUS/LVDS Physical Layer

6.1 Interface

6.2 EBUS Protocol

6.3 Timing Characteristics

6.4 Standard EBUS Link Detection

6.5 Enhanced EBUS Link Detection

6.6 EBUS RX Errors

6.7 EBUS Low Jitter

6.8 EBUS Connection

7 FMMU

8 SyncManager

8.1 Buffered Mode

8.2 Mailbox Mode

8.2.1 Mailbox Communication Protocols