Full-speed Universal Serial Bus single-chip host and device
controller
Rev. 03 — 23 December 2004Product data
1.General description
The ISP1161A is a single-chip Universal Serial Bus (USB) Host Controller (HC) and
Device Controller (DC). The Host Controller portion of the ISP1161A complies with
Universal Serial Bus Specification Rev. 2.0
(12 Mbit/s) and low-speed (1.5 Mbit/s). The Device Controller portion of the
ISP1161A also complies with
data rates at full-speed (12 Mbit/s). These two USB controllers, the HC and the DC,
share the same microprocessor bus interface. They have the same data bus, but
different I/O locations. They also have separate interrupt request output pins,
separate DMA channels that include separate DMA request output pins and DMA
acknowledge input pins. This makes it possible for a microprocessor to control both
the USB HC and the USB DC at the same time.
ISP1161A provides two downstream ports forthe USB HC and one upstream port for
the USB DC. Each downstream port has an overcurrent (OC) detection input pin and
power supply switching control output pin. The upstream port has a V
input pin. ISP1161A also provides separate wake-up input pins and suspended status
output pins for the USB HC and the USB DC, respectively. This makes power
management flexible. The downstream ports for the HC can be connected with any
USB compliant devices and hubs that have USB upstream ports. The upstream port
for the DC can be connected to any USB compliant USB host and USB hubs that
have USB downstream ports.
, supporting data rates at full-speed
Universal Serial Bus Specification Rev. 2.0
BUS
, supporting
detection
The HC is adapted from the
Release 1.0a
The DC is compliant with most USB device class specifications such as Imaging
Class, Mass Storage Devices, Communication Devices, Printing Devices and Human
Interface Devices.
ISP1161A is well suited for embedded systems and portable devices that require a
USB host only, a USB device only, or a combination of a configurable USB host and
USB device. ISP1161A brings high flexibility to the systems that have it built-in. For
example, a system that uses an ISP1161A allows it not only to be connected to a PC
or USB hub with a USB downstream port, but also to be connected to a device that
has a USB upstream port such as a USB printer, USB camera, USB keyboard or a
USB mouse. Therefore, the ISP1161A enables peer-to-peer connectivity between
embedded systems. An interesting application example is to connect an ISP1161A
HC with an ISP1161A DC.
Consider an example of an ISP1161A being used in a Digital Still Camera (DSC)
design. Figure 1 shows an ISP1161A being used as a USB DC. Figure 2 shows an
ISP1161A being used as a USB HC. Figure 3 shows an ISP1161A being used as a
USB HC and a USB DC at the same time.
, referred to as OHCI in the rest of this document.
Open Host Controller Interface Specification for USB
Philips Semiconductors
ISP1161A
Full-speed USB single-chip host and device controller
EMBEDDED SYSTEM
PC
(host)
USB I/F
Fig 1. ISP1161A operating as a USB device.
EMBEDDED SYSTEM
µP
ISP1161A
HOST/
DEVICE
DSC
USB cable
µP SYSTEM
MEMORY
µP bus I/F
USB host
USB I/F
USB I/F
µP
ISP1161A
USB device
USB cable
µP SYSTEM
MEMORY
µP bus I/F
HOST/
DEVICE
DSC
004aaa080
PRINTER
(device)
USB I/F
004aaa081
Fig 2. ISP1161A operating as a stand-alone USB host.
EMBEDDED SYSTEM
ISP1161A
HOST/
DEVICE
µP SYSTEM
MEMORY
µP bus I/F
USB host
PRINTER
(device)
USB I/FUSB I/F
USB I/F
004aaa082
PC
(host)
µP
DSC
USB cableUSB cable
USB I/F
USB device
Fig 3. ISP1161A operating as both USB host and device simultaneously.
and polarity; see Section 10.4.1
INT230ODC interrupt output; programmable level, edge triggered
and polarity; see Section 13.1.4
TEST31Otest output; used for test purposes only; this pin is not
connected during normal operation
RESET32Ireset input (Schmitt trigger); a LOW level produces an
asynchronous reset (internal pull-up resistor)
NDP_SEL33Iindicates to the HC software the Number of Downstream
Ports (NDP) present:
0 — select 1 downstream port
1 — select 2 downstream ports
only changes the value of the NDP field in the
HcRhDescriptorA register; both ports will always be
enabled; see Section 10.3.1
(internal pull-up resistor)
EOT34IDMA master device to inform the ISP1161A of end of DMA
transfer; active level is programmable; see Section 10.4.1
DGND35-digital ground
D_SUSPEND 36ODC ‘suspend’ state indicator output; active HIGH
D_WAKEUP37IDC wake-up input; generates a remote wake-up from
‘suspend’ state (active HIGH); when not in use, this pin
must be connected to DGND via an external 10 kΩ resistor
(internal pull-down resistor)
GL38OGoodLink LED indicator output (open-drain, 8 mA); the
LED is default ON, blinks OFF upon USB traffic; to connect
a LED use a series resistor of 470 Ω (V
330 Ω (V
CC
= 3.3 V)
D_VBUS39IDC USB upstream port V
sensing input; when not in
BUS
= 5.0 V) or
CC
use, this pin must be connected to DGND via a 1 MΩ
resistor
H_WAKEUP40IHC wake-up input; generates a remote wake-up from
‘suspend’ state (active HIGH); when not in use, this pin
must be connected to DGND via an external 10 kΩ resistor
Full-speed USB single-chip host and device controller
Table 2:Pin description for LQFP64
Symbol
[1]
PinTypeDescription
…continued
DGND45-digital ground
H_PSW146Opower switching control output for downstream port 1;
open-drain output
H_PSW247Opower switching control output for downstream port 2;
open-drain output
D_DM48AI/OUSB D− data line for DC upstream port; when not in use,
this pin must be left open
D_DP49AI/OUSB D+ data line for DC upstream port; when not in use,
this pin must be left open
H_DM150AI/OUSB D− data line for HC downstream port 1
H_DP151AI/OUSB D+ data line for HC downstream port 1
H_DM252AI/OUSB D− data line for HC downstream port 2; when not in
use, this pin must be left open
H_DP253AI/OUSB D+ data line for HC downstream port 2; when not in
use, this pin must be left open
H_OC154Iovercurrent sensing input for HC downstream port 1
H_OC255Iovercurrent sensing input for HC downstream port 2
V
CC
56-power supply voltage input (3.0 V to 3.6 V or
4.75 V to 5.25 V). This pin connects to the internal 3.3 V
regulator input. When connected to 5 V, the internal
regulator will output 3.3 V to pins V
reg(3.3),Vhold1
and V
hold2
When connected to 3.3 V, it will bypass the internal
regulator.
AGND57-analog ground
V
reg(3.3)
58-internal 3.3 V regulator output; when the VCC pin is
connected to 5 V, this pin outputs 3.3 V. When the V
CC
pin
is connected to 3.3 V, connect this pin to 3.3 V.
A059Iaddress input; selects command (A0 = 1) or data (A0 = 0)
A160Iaddress input; selects AutoMux switching to DC (A1 = 1) or
AutoMux switching to HC (A1 = 0); see Table 3
n.c.61-no connection
DGND62-digital ground
D063I/Obit 0 of bidirectional data; slew-rate controlled; TTL input;
A 6 MHz to 48 MHz clock multiplier Phase-Locked Loop (PLL) is integrated on-chip.
This allows for the use of a low-cost 6 MHz crystal, which also minimizes EMI. No
external components are required for the operation of the PLL.
7.2 Bit clock recovery
The bit clock recovery circuit recovers the clock from the incoming USB data stream
using a 4 times over-sampling principle. It is able to track jitter and frequency drift as
specified in the
7.3 Analog transceivers
Three sets of transceivers are embedded in the chip: two are used for downstream
ports with USB connector type A; one is used for upstream port with USB connector
type B. The integrated transceivers are compliant with the
Specification Rev. 2.0
through external termination resistors.
ISP1161A
Full-speed USB single-chip host and device controller
Universal Serial Bus Specification Rev. 2.0
. They interface directly with the USB connectors and cables
.
Universal Serial Bus
7.4 Philips Serial Interface Engine (SIE)
The Philips SIE implements the full USB protocol layer. It is completely hardwired for
speed and needs no firmware intervention. The functions of this block include:
synchronization pattern recognition, parallel to serial conversion, bit (de)stuffing,
CRC checking and generation, Packet IDentifier (PID) verification and generation,
address recognition, handshake evaluation and generation. There are separate SIEs
in both the HC and the DC.
7.5 SoftConnect
The connection to the USB is accomplished by bringing D+ (for full-speed USB
devices) HIGH through a 1.5 kΩ pull-up resistor. In the ISP1161A DC, the 1.5 kΩ
pull-up resistor is integrated on-chip and is not connected to VCC by default. The
connection is established through a command sent by the external or system
microcontroller. This allows the system microcontroller to complete its initialization
sequence before deciding to establish connection with the USB. Re-initialization of
the USB connection can also be performed without disconnecting the cable.
The ISP1161A DC will check for USB V
established. V
Remark: The tolerance of the internal resistors is 25 %. This is higher than the 5 %
tolerance specified by the USB specification. However, the overall voltage
specification for the connection can still be met with a good margin. The decision to
make use of this feature lies with the USB equipment designer.
Indication of a good USB connection is provided at pin GL through GoodLink
technology. During enumeration, the LED indicator will blink on momentarily. When
the DC has been successfully enumerated (the device address is set), the LED
indicator will remain permanently on. Upon each successful packet transfer (with
ACK) to and from the ISP1161A the LED will blink off for 100 ms. During ‘suspend’
state the LED will remain off.
This feature provides a user-friendly indication of the status of the USB device, the
connected hub and the USB traffic. It is a useful field diagnostics tool for isolating
faulty equipment. It can therefore help to reduce field support and hotline overhead.
8.Microprocessor bus interface
8.1 Programmed I/O (PIO) addressing mode
A generic PIO interface is defined for speed and ease-of-use. It also allows direct
interfacing to most microcontrollers. To a microcontroller, the ISP1161A appears as a
memory device with a 16-bit data bus and uses only two address lines: A1 and A0 to
access the internal control registers and FIFO buffer RAM. Therefore, the ISP1161A
occupies only four I/O ports or four memory locations of a microprocessor. External
microprocessors can read from or write to the ISP1161A internal control registers and
FIFO bufferRAM through the Programmed I/O (PIO) operating mode. Figure 8 shows
the Programmed I/O interface between a microprocessor and an ISP1161A.
ISP1161A
Full-speed USB single-chip host and device controller
MICRO-
PROCESSOR
D[15:0
WR
IRQ1
IRQ2
µP bus I/F
]
RD
CS
A2
A1
D[15:0
RD
WR
CS
A1
A0
INT1
INT2
]
ISP1161A
004aaa086
Fig 8. Programmed I/O interface between a microprocessor and an ISP1161A.
8.2 DMA mode
The ISP1161A also provides DMA mode for external microprocessors to access its
internal FIFO buffer RAM. Data can be transferred by DMA operation between a
microprocessor’s system memory and the ISP1161A internal FIFO buffer RAM.
Remark: The DMA operation must be controlled by the external microprocessor
system DMA controller (Master).
Figure 9 shows the DMA interface between a microprocessor system and the
ISP1161A. The ISP1161A provides two DMA channels:
• DMA channel 1 (controlled by DREQ1, DACK1 signals) is for the DMA transfer
• DMA channel 2 (controlled by DREQ2, DACK2 signals) is for the DMA transfer
The EOT signal is an external end-of-transfer signal used to terminate the DMA
transfer. Some microprocessors may not have this signal. In this case, the ISP1161A
provides an internal EOT signal to terminate the DMA transfer as well. Setting the
HcDMAConfiguration register (21H - read, A1H - write) enables the ISP1161A HC
internal DMA counter for DMA transfer. When the DMA counter reaches the value set
in the HcTransferCounter register (22H - read, A2H - write), an internal EOT signal
will be generated to terminate the DMA transfer.
ISP1161A
Full-speed USB single-chip host and device controller
between a microprocessor’s system memory and ISP1161A HC internal FIFO
buffer RAM
between a microprocessor system memory and the ISP1161A DC internal FIFO
buffer RAM.
µP bus I/F
]
D[15:0
RD
WR
MICRO-
PROCESSOR
DACK1
DREQ1
DACK2
DREQ2
EOT
Fig 9. DMA interface between a microprocessor and an ISP1161A.
8.3 Control register access by PIO mode
8.3.1 I/O port addressing
Table 3 shows the ISP1161A I/O port addressing. Complete decoding of the I/O port
address should include the chip select signal CS and the address lines A1 and A0.
However, the direction of the access of the I/O ports is controlled by the RD and WR
signals. When RD is LOW, the microprocessor reads data from the ISP1161A data
port. When WR is LOW, the microprocessor writes a command to the command port,
or writes data to the data port.
D[15:0
RD
WR
DACK1
DREQ1
DACK2
DREQ2
EOT
]
ISP1161A
004aaa087
Table 3:I/O port addressing
PortPin CSPin A1Pin A0AccessData bus widthDescription
0LOWLOWLOWR/W16 bitsHC data port
1LOWLOWHIGHW16 bitsHC command port
2LOWHIGHLOWR/W16 bitsDC data port
3LOWHIGHHIGHW16 bitsDC command port
Figure 10 and Figure 11 illustrate how an external microprocessor accesses the
ISP1161A internal control registers.
Fig 10. A microprocessor accessing an HC or a DC via an automux switch.
Full-speed USB single-chip host and device controller
AUTOMUX
DC/HC
0
µP bus I/F
1
A1
When A1 = 0, the microprocessor accesses the HC.
When A1 = 1, the microprocessor accesses the DC.
CMD/DATA
SWITCH
command port
Host or Device
bus I/F
A0
1
data port
0
Host bus I/F
Device bus I/F
MGT935
.
.
.
ISP1161A
Commands
Command register
When A0 = 0, the microprocessor accesses the data port.
When A0 = 1, the microprocessor accesses the command port.
Fig 11. A microprocessor accessing internal control registers.
8.3.2 Register access phases
The ISP1161A register structure is a command-data register pair structure. A
complete register access cycle comprises a command phase followed by a data
phase. The command (also known as the index of a register) points the ISP1161A to
the next register to be accessed. A command is 8 bits long. On a microprocessor’s
16-bit data bus, a command occupies the lower byte, with the upper byte filled with
zeros.
Figure 12 shows a complete 16-bit register access cycle for the ISP1161A. The
microprocessor writes a command code to the command port, and then reads or
writes the data word from or to the data port. Take the example of a microprocessor
attempting to read the ISP1161A’s ID. The ID is kept in the HC’s HcChipID register
(index 27H, read only). The 16-bit register access cycle is therefore:
1. Microprocessor writes the command code of 27H (0027H in 16-bit width) to the
HC command port
2. Microprocessor reads the data word of the chip’s ID from the HC data port.
Full-speed USB single-chip host and device controller
16-bit register access cycle
write command
(16 bits)
read/write data
(16 bits)
t
MGT937
Fig 12. 16-bit register access cycle.
Most of the ISP1161A internal control registers are 16 bits wide. Some of the internal
control registers have 32-bit width. Figure 13 shows how the 32-bit internal control
register is accessed. The complete cycle of accessing a 32-bit register consists of a
command phase followed by two data phases. In the two data phases, the
microprocessor first reads or writes the lower 16-bits, followed by the upper 16-bits.
32-bit register access cycle
write command
(16 bits)
read/write data
(lower 16 bits)
read/write data
(upper 16 bits)
t
MGT938
Fig 13. 32-bit register access cycle.
To further describe the complete access cycles of the internal control registers, the
status of some pins of the microprocessor bus interface are shown in Figure 14 and
Full-speed USB single-chip host and device controller
CS
A1, A0
WR
RD
D[15:0
]
111010
DC command
code
DC register data
(lower word)
readread
writewritewrite
writewrite
readread
DC register data
(upper word)
Fig 15. Accessing DC control registers.
8.4 FIFO buffer RAM access by PIO mode
Since the ISP1161A internal memory is structured as a FIFO buffer RAM, the FIFO
buffer RAM is mapped to dedicated register fields. Therefore, accessing the internal
FIFO buffer RAM is similar to accessing the internal control registers in multiple data
phases.
Figure 16 shows a complete access cycle of the HC internal FIFO buffer RAM. For a
write cycle, the microprocessor first writes the FIFO buffer RAM’s command code to
the command port, and then writes the data words one by one to the data port until
half of the transfer’s byte count is reached. The HcTransferCounter register (22H read, A2H - write) is used to specify the byte count of a FIFO buffer RAM’s read cycle
or write cycle. Every access cycle must be in the same access direction. The read
cycle procedure is similar to the write cycle.
For access to the DC FIFO buffer RAM access, see Section 13.
The DMA interface between a microprocessor and the ISP1161A is shown in
Figure 9.
When doing a DMA transfer, at the beginning of every burst the ISP1161A outputs a
DMA request to the microprocessor via the DREQ pin (DREQ1 for HC, DREQ2 for
DC). After receiving this signal, the microprocessor will reply with a DMA
acknowledge via the DACK pin (DACK1 for HC, DACK2 for DC), and at the same
time, execute the DMA transfer through the data bus. In the DMA mode, the
microprocessor must issue a read or write signal to the ISP1161A RD or WR pin. The
ISP1161A will repeat the DMA cycles until it receives an EOT signal to terminate the
DMA transfer.
ISP1161A supports both external and internal EOT signals. The external EOT signal
is received as input on pin EOT, and generally comes from the external
microprocessor. The internal EOT signal is generated by the ISP1161A internally.
To select either EOT method, set the appropriate DMA configuration register (see
Section 10.4.2 and Section 13.1.6). For example, for the HC, setting
DMACounterSelect bit of the HcDMAConfiguration register (21H - read, A1H - write)
to logic 1 will enable the DMA counter for DMA transfer. When the DMA counter
reaches the value of the HcTransferCounter register, the internal EOT signal will be
generated to terminate the DMA transfer.
ISP1161A
Full-speed USB single-chip host and device controller
ISP1161A supports either single-cycle DMA operation or burst mode DMA operation.
Full-speed USB single-chip host and device controller
INT
Mode 0 level triggered, active LOW
INT
Mode 1 level triggered, active HIGH
INT
Mode 2 edge triggered, active LOW
INT
Mode 3 edge triggered, active HIGH
Fig 19. Interrupt pin operating modes.
8.6.2 HC’s interrupt output pin (INT1)
To program the four configuration modes of the HC’s interrupt output signal (INT1),
set bits InterruptPinTrigger and InterruptOutputPolarity of the
HcHardwareConfiguration register (20H - read, A0H - write). Bit InterruptPinEnableis
used as the master enable setting for pin INT1.
INT active
INT active
INT active
166 ns
INT active
166 ns
clear or disable INT
clear or disable INT
MGT944
INT1 has many associated interrupt events, as shown as in Figure 20.
The interrupt events of the HcµPInterrupt register (24H - read, A4H - write) changes
the status of pin INT1 when the corresponding bits of the HcµPInterruptEnable
register (25H - read, A5H - write) and pin INT1’s global enable bit (InterruptPinEnable
of the HcHardwareConfiguration register) are all set to enable status.
However, events that come from the HcInterruptStatus register (03H - read, 83H write) affect only the OPR_Reg bit of the HcµPInterrupt register. They cannot directly
change the status of pin INT1.
Full-speed USB single-chip host and device controller
HcInterruptEnable
register
MIE
RHSC
FNO
UE
RD
SF
SO
RHSC
FNO
UE
RD
SF
SO
HcInterruptStatus
register
group 2
OR
HcµPInterrupt
register
ATLInt
SOFITLInt
AllEOTInterrupt
INT1
OPR_Reg
HCSuspended
group 1
ClkReady
OR
LATCH
HcµPInterruptEnable
SOFITLInt
LE
register
ATLInt
AllEOTInterrupt
HcHardwareConfiguration
InterruptPinEnable
OPR_Reg
HCSuspended
register
ClkReady
MGT945
Fig 20. HC interrupt logic.
There are two groups of interrupts represented by group 1 and group 2 in Figure 20.
A pair of registers control each group.
Group 2 contains six possible interrupt events (recorded in the HcInterruptStatus
register). On occurrence of any of these events, the corresponding bit would be set to
logic 1; and if the corresponding bit in the HcInterruptEnable register is also logic 1,
the 6-input OR gate would output logic 1. This output is ANDed with the value of MIE
(bit 31 of HcInterruptEnable). Logic 1 at the AND gate will cause bit OPR in the
HcµPInterrupt register to be set to logic 1.
Group 1 contains six possible interrupt events, one of which is the output of group 2
interrupt sources. The HcµPInterrupt and HcµPInterruptEnable registers work in the
same way as the HcInterruptStatus and HcInterruptEnable registers in the interrupt
group 2. The output from the 6-input OR gate is connected to a latch, which is
controlled by InterruptPinEnable (bit 0 of the HcHardwareConfiguration register).
In the eventin which the software wishes to temporarily disable the interrupt output of
the ISP1161A Host Controller, the following procedure should be followed:
1. Make sure that bit InterruptPinEnable in the HcHardwareConfiguration register is
set to logic 1.
Remark: Bit InterruptPinEnable in the HcHardwareConfiguration register latches the
interrupt output. When this bit is set to logic 0, the interrupt output will remain
unchanged, regardless of any operations on the interrupt control registers.
If INT1 is asserted, and the HCD wishes to temporarily mask off the INT signal
without clearing the HcµPInterrupt register, the following procedure should be
followed:
1. Make sure that bit InterruptPinEnable is set to logic 1.
2. Clear all bits in the HcµPInterruptEnable register.
3. Set bit InterruptPinEnable to logic 0.
To re-enable the interrupt generation:
1. Set all bits in the HcµPInterruptEnable register according to the HCD
2. Set bit InterruptPinEnable to logic 1.
ISP1161A
Full-speed USB single-chip host and device controller
requirements.
8.6.3 DC interrupt output pin (INT2)
The four configuration modes of DC’s interrupt output pin INT2 can also be
programmed by setting bits INTPOL and INTLVL of the DcHardwareConfiguration
register (BBH - read, BAH - write). Bit INTENA of the DcMode register (B9H - read,
B8H - write)is used to enable pin INT2. Figure 21 shows the relationship between the
interrupt events and pin INT2.
Each of the indicated USB events is logged in a status bit of the DcInterrupt register.
Corresponding bits in the Interrupt Enable register determine whether or not an event
will generate an interrupt.
Interrupts can be masked globally by means of the INTENA bit of the DcMode
register (see Table 81).
The active level and signalling mode of the INT output is controlled by the INTPOL
and INTLVL bits of the DcHardwareConfiguration register (see Table 83). Default
settings after reset are active LOW and level mode. When pulse mode is selected, a
pulse of 166 ns is generated when the OR-ed combination of all interrupt bits
changes from logic 0 to logic 1.
Bits RESET, RESUME, SP_EOT, EOT and SOF are cleared upon reading the
DcInterrupt register. The endpoint bits (EP0OUT to EP14) are cleared by reading the
associated DcEndpointStatus register.
Bit BUSTATUS follows the USB bus status exactly, allowing the firmware to get the
current bus status when reading the DcInterrupt register.
SETUP and OUT token interrupts are generated after the DC has acknowledged the
associated data packet. In bulk transfer mode, the DC will issue interrupts for every
ACK received for an OUT token or transmitted for an IN token.
In isochronous mode, an interrupt is issued upon each packet transaction. The
firmware must take care of timing synchronization with the host. This can be done via
the Pseudo Start-Of-Frame (PSOF) interrupt, enabled via bit IEPSOF in the Interrupt
Enable register. If a Start-Of-Frame is lost, PSOF interrupts are generated every
1 ms. This allows the firmware to keep data transfer synchronized with the host. After
3 missed SOF events, the DC will enter ‘suspend’ state.
An alternative way of handling isochronous data transfer is to enable both the SOF
and the PSOF interrupts and disable the interrupt for each isochronous endpoint.
DcInterrupt register
RESET
SUSPND
RESUME
EP14
EP0IN
EP0OUT
IERST
IESUSP
IERESM
IESOF
IEP14
IEP0IN
IEP0OUT
IEEOT
DcInterruptEnable register
SOF
...
EOT
...
.
.
.
.
.
.
MGT946
ISP1161A
Full-speed USB single-chip host and device controller
.
.
.
.
.
.
DcMode register
INTENA
LE
LATCH
INT2
Fig 21. DC interrupt logic.
Interrupt control: Bit INTENA in the DcMode register is a global enable/disable bit.
The behavior of this bit is given in Figure 22.
A
INT2 pin
INTENA = 0
(during this time,
an interrupt event
occurs. For example,
SOF asserted.)
Pin INT2: HIGH = de-assert; LOW= assert (individual interrupts are enabled).
Event A (see Figure 22): When an interrupt eventoccurs (for example, SOF interrupt)
with bit INTENA set to logic 0, an interrupt will not be generated at pin INT2.
However, it will be registered in the corresponding DcInterrupt register bit.
Event B (see Figure 22): When bit INTENA is set to logic 1, pin INT2 is asserted
because bit SOF in the DcInterrupt register is already asserted.
Event C (see Figure 22): If the firmware sets bit INTENA to logic 0, pin INT2 will still
be asserted. The bold dashed line shows the desired behavior of pin INT2.
De-assertion of pin INT2 can be achieved in the following manner. Bits[23:8] of the
DcInterrupt register are endpoint interrupts. These interrupts are cleared on reading
their respective DcEndpointStatus register. Bits[7:0] of the DcInterrupt register are
bus status and EOT interrupts that are cleared on reading the DcInterrupt register.
Make sure that bit INTENA is set to logic 1 when you perform the clear interrupt
commands.
For more information on interrupt control, see Section 13.1.3, Section 13.1.5 and
Section 13.3.6.
ISP1161A
Full-speed USB single-chip host and device controller
The ISP1161A USB HC has four USB states − USBOperational, USBReset,
USBSuspend, and USBResume − that define the HC’s USB signaling and bus states
responsibilities.
ISP1161A
Full-speed USB single-chip host and device controller
USBOperational write
USBSuspend write
Fig 23. ISP1161A HC USB states.
The USB states are reflected in the HostControllerFunctionalState field of the
HcControl register (01H - read, 81H - write), which is located at bits 7 and 6 of the
register.
USBOperational
USBSuspend
USBOperational write
USBResume
USBResume write
or
remote wake-up
USBReset write
USBReset write
USBReset
hardware or software
reset
USBReset write
MGT947
The Host Controller Driver (HCD) can perform only the USB state transitions shown
in Figure 23.
Remark: The Software Reset in Figure 23 is not caused by the HcSoftwareReset
command. It is caused by the HostControllerReset field of the HcCommandStatus
register (02H - read, 82H - write).
9.2 Generating USB traffic
USB traffic can be generated only when the ISP1161A USB HC is in the
USBOperational state. Therefore, the HCD must set the
HostControllerFunctionalState field of the HcControl register before generating USB
traffic.
A simplistic flow diagram showing when and how to generate USB traffic is shown in
Full-speed USB single-chip host and device controller
Reset
HC state =
Initialize
HC
Entry
USBOperational
USB traffic?
HC informs HCD of
USB traffic results
Fig 24. ISP1161A HC USB transaction loop
The USB traffic blocks are:
• Reset
This includes hardware reset by pin RESET and software reset by the
HcSoftwareReset command (A9H). The reset function will clear all the HC’s
internal control registers to their reset status. After reset, the HCD must initialize
the ISP1161A USB HC by setting some registers.
• Initialize HC
It includes:
– Setting the physical size for the HC’s internal FIFO buffer RAM by setting the
HcITLBufferLength register (2AH - read, AAH - write) and the
HcATLBufferLength register (2BH - read, ABH - write)
– Setting the HcHardwareConfiguration register according to requirements
– Clearing interrupt events, if required
– Enabling interrupt events, if required
– Setting the HcFmInterval register (0DH - read, 8DH - write)
– Setting the HC’s Root Hub registers
– Setting the HcControl register to move the HC into USBOperational state
See also Section 9.5.
• Entry
The normal entry point. The microprocessor returns to this point when there are
HC requests.
• Need USB Traffic
USB devices need the HC to generate USB traffic when they have USB traffic
requests such as:
– Connecting to or disconnecting from the downstream ports
– Issuing the Resume signal to the HC
To generate USB traffic, the HCD must enter the USB transaction loop.
Exit
Need
no
yes
Prepare PTD data in
µP system RAM
HC performs USB transactions
via USB bus I/F
Transfer PTD data into
HC FIFO buffer RAM
HC interprets
PTD data
MGT948
Prepare PTD data in µP System RAM
The communication between the HCD and the ISP1161A HC is in the form of
Philips Transfer Descriptor (PTD) data. The PTD data provides USB traffic
information about the commands, status, and USB data packets.
• HC performs USB transactions via USB Bus interface
• HC informs HCD the USB traffic results
ISP1161A
Full-speed USB single-chip host and device controller
The physical storage media of PTD data for the HCD is the microprocessor’s
system RAM. For the ISP1161A HC, the storage media is the internal FIFO buffer
RAM.
The HCD prepares PTD data in the microprocessor system RAM for transferto the
ISP1161A HC internal FIFO buffer RAM.
When PTD data is ready in the microprocessor’s system RAM, the HCD must
transfer the PTD data from the microprocessor’s system RAM into the ISP1161A
internal FIFO buffer RAM.
The HC determines what USB transactions are required based on the PTD data
that has been transferred into the internal FIFO buffer RAM.
The HC performs the USB transactions with the specified USB device endpoint
through the USB bus interface.
The USB transaction status and the feedback from the specified USB device
endpoint will be put back into the ISP1161A HC internal FIFO buffer RAM in PTD
data format. The HCD can read back the PTD data from the internal FIFO buffer
RAM.
9.3 PTD data structure
The Philips Transfer Descriptor (PTD) data structure provides communication
between the HCD and the ISP1161A USB HC. The PTD data contains information
required by the USB traffic. PTD data consists of a PTD followed by its payload data,
as shown in Figure 25.
top
bottom
Fig 25. PTD data in FIFO buffer RAM.
FIFO buffer RAM
PTD
payload data
PTD
payload data
PTD
payload data
PTD data #1
PTD data #2
PTD data #N
MGT949
The PTD data structure is used by the HC to define a buffer of data that will be moved
to or from an endpoint in the USB device. This data buffer is set up for the current
frame (1 ms frame) by the HCD. The payload data forevery transferin the frame must
have a PTD as a header to describe the characteristics of the transfer. PTD data is
DWORD aligned.
Full-speed USB single-chip host and device controller
9.3.1 PTD data header definition
The PTD forms the header of the PTD data. It tells the HC the transfer type, where
the payload data goes, and the payload data’s actual size. A PTD is an 8 byte data
structure that is very important for HCD programming.
Table 4:Philips Transfer Descriptor (PTD): bit allocation
Full-speed USB single-chip host and device controller
Table 5:Philips Transfer Descriptor (PTD): bit description
SymbolAccessDescription
ActualBytes[9:0]R/WContains the number of bytes that were transferred for this PTD
CompletionCode[3:0]R/W0000NoErrorGeneral TD or isochronous data packet processing
completed with no detected errors.
0001CRCLast data packet from endpoint contained a CRC error.
0010BitStuffingLast data packet from endpoint contained a bit stuffing
violation.
0011DataToggleMismatchLast packet from endpoint had data toggle PID that did
not match the expected value.
0100StallTD was moved to the Done queue because the
endpoint returned a STALL PID.
0101DeviceNotResponding Device did not respond to token (IN) or did not provide a
handshake (OUT).
0110PIDCheckFailureCheck bits on PID from endpoint failed on data PID (IN)
or handshake (OUT)
0111UnexpectedPIDReceived PID was not valid when encountered or PID
value is not defined.
1000DataOverrunThe amount of data returned by the endpoint exceeded
either the size of the maximum data packet allowed
from the endpoint (foundinMaximumPacketSizefieldof
ED) or the remaining buffer size.
1001DataUnderrunThe endpoint returned is less than MaximumPacketSize
and that amount was not sufficient to fill the specified
buffer.
1010reserved1011reserved1100BufferOverrunDuring an IN, the HC received data from an endpoint
faster than it could be written to system memory.
1101BufferUnderrunDuring an OUT, the HC could not retrieve data from the
system memory fast enough to keep up with the USB
data rate.
ActiveR/WSet to logic 1 by firmware to enable the execution of transactions by the HC. When the
transaction associated with this descriptor is completed, the HC sets this bit to logic 0,
indicating that a transaction for this element will not be executed when it is next
encountered in the schedule.
ToggleR/WUsed to generate or compare the data PID value (DATA0 or DATA1). It is updated after
each successful transmission or reception of a data packet.
MaxPacketSize[9:0]RThe maximum number of bytes that can be sent to or received from the endpoint in a
single data packet.
EndpointNumber[3:0]RUSB address of the endpoint within the function.
LastRLast PTD of a list (ITL or ATL). Logic 1 indicates that the PTD is the last PTD.
SpeedRSpeed of the endpoint:
0 — full speed
1 — low speed
TotalBytes[9:0]RSpecifies the total number of bytes to be transferred with this data structure.ForBulk and
Control only, this can be greater than MaximumPacketSize.
Full-speed USB single-chip host and device controller
Table 5:Philips Transfer Descriptor (PTD): bit description
SymbolAccessDescription
DirectionPID[1:0]R00SETUP
01OUT
10IN
11reserved
B5_5R/WThis bit is logic 0 at power-on reset. When this feature is not used, software used for
ISP1161A is the same for ISP1160 and ISP1161. When this bit is set to logic 1 in this
PTD for interrupt endpoint transfer, only 1 PTD USB transaction will be sent out in 1 ms.
FormatRThe format of this data structure. If this is a Control, Bulk or Interrupt endpoint, then
Format = 0. If this is an Isochronous endpoint, then Format = 1.
FunctionAddress[6:0]RThis is the USB address of the function containing the endpoint that this PTD refers to.
…continued
9.4 HC internal FIFO buffer RAM structure
9.4.1 Partitions
According to the
USB data transfers: Control, Bulk, Interrupt and Isochronous.
The HC’s internal FIFO bufferRAM has a physical size of 4 kbytes. This internal FIFO
buffer RAM is used for transferring data between the microprocessor and USB
peripheral devices. This on-chip buffer RAM can be partitioned into two areas:
Acknowledged Transfer List (ATL) buffer and Isochronous (ISO)Transfer List (ITL)
buffer. The ITL buffer is a Ping-Pong structured FIFO buffer RAM that is used to keep
the payload data and their PTD header for Isochronous transfers. The ATL buffer is a
non Ping-Pong structured FIFO buffer RAM that is used for the other three types of
transfers.
Universal Serial Bus Specification Rev. 2.0
, there are four types of
The ITL buffer can be further partitioned into ITL0 and ITL1 for the Ping-Pong
structure. The ITL0 buffer and ITL1 buffer always have the same size. The
microprocessor can put ISO data into either the ITL0 buffer or the ITL1 buffer. When
the microprocessor accesses an ITL buffer, the HC can take over the other ITL buffer
at the same time. This architecture improves the ISO transfer performance.
The HCD can assign the logical size for the ATL bufferand ITL buffers at any time, but
normally at initialization after power-on reset. This is done by setting the
HcATLBufferLength register (2BH - read, ABH - write) and HcITLBufferLength
register (2AH - read, AAH - write). The total buffer length cannot exceed the
maximum RAM size of 4 kbytes (ATL buffer + ITL buffer). Figure 26 shows the
partitions of the internal FIFO buffer RAM. When assigning buffer RAM sizes, follow
this formula:
ATL buffer length + 2 × (ITL buffer size) ≤ 1000H (that is, 4 kbytes)
where: ITL buffer size = ITL0 buffer length = ITL1 buffer length
The following assignments are examples of legal uses of the internal FIFO buffer
Full-speed USB single-chip host and device controller
This is insufficient use of the internal FIFO buffer RAM.
This will use the internal FIFO buffer RAM for only ATL transfers.
FIFO buffer RAM
top
ITL0
ITL buffer
ITL1
ISO_A
ISO_B
programmable
sizes
ATL buffer
Fig 26. HC internal FIFO buffer RAM partitions.
ATL
bottom
control/bulk/interrupt
data
not used
MGT950
4 kbytes
The actual requirement for the buffer RAM need not reach the maximum size. You
can make your selection based on your application. The following are some
calculations of the ISO_A or ISO_B space for a frame of data:
• Maximum number of useful data sent during one USB frame is 1280 bytes (20
ISO packets of 64 bytes). The total RAM size needed is:
20 × 8 + 1280 = 1440 bytes.
• Maximum number of packets for different endpoints sent during one USB frame is
150 (150 ISO packets of 1 byte). The total RAM size needed is:
150 × 8 + 150 × 1 = 1350 bytes.
• The Ping buffer RAM (ITL0) and the Pong bufferRAM (ITL1) have a maximum size
of 2 kbytes each. All data needed for one frame can be stored in the Ping or the
Pong buffer RAM.
When the embedded system wants to initiate a transfer to the USB bus, the data
needed for one frame is transferred to the ATL buffer or ITL buffer. The
microprocessor detects the buffer status through the interrupt routines. When the
HcBufferStatus register (2CH - read only) indicates that the buffer is empty, then the
microprocessor writes data into the buffer. When the HcBufferStatus register
indicates that the buffer is full, the data is ready on the buffer,and the microprocessor
needs to read data from the buffer.
During every 1 ms, there might be many events to generate interrupt requests to the
microprocessor for data transfer or status retrieval. However, each of the interrupt
types defined in this specification can be enabled or disabled by setting the
HcµPInterruptEnable register bits accordingly.
The data transfer can be done via the PIO mode or the DMA mode. The data transfer
rate can go up to 15 Mbyte/s. In DMA operation, single-cycle or multi-cycle burst
modes are supported. Multi-cycle burst modes of 1, 4, or 8 cycles per burst is
supported for ISP1161A.
9.4.2 Data organization
PTD data is used for every data transfer between a microprocessor and the USB bus,
and the PTD data resides in the buffer RAM. For an OUT or SETUP transfer, the
payload data is placed just after the PTD, after which the next PTD is placed. For an
IN transfer,RAMspace is reservedforreceiving a number of bytes that is equal to the
total bytes of the transfer. After this, the next PTD and its payload data are placed
(see Figure 27).
Remark: The PTD is defined for both ATL and ITL type data transfers. For ITL, the
PTD data is put into ITL buffer RAM, and the ISP1161A takes care of the Ping-Pong
action for the ITL buffer RAM access.
ISP1161A
Full-speed USB single-chip host and device controller
RAM buffer
top
PTD of OUT transfer
000H
payload data of OUT transfer
PTD of IN transfer
empty space for IN total data
PTD of OUT transfer
payload data of OUT transfer
bottom
Fig 27. Buffer RAM data organization.
7FFH
MGT952
The PTD data (PTD header and its payload data) is a structure of DWORD (doubleword or 4-byte) alignment. This means that the memory address is organized in
blocks of 4 bytes. Therefore, the first byte of every PTD and the first byte of every
payload data are located at an address which is a multiple of 4. Figure 28 illustrates
an example in which the first payload data is 14 bytes long, meaning that the last byte
of the payload data is at the location 15H. The next addresses (16H and 17H) are not
multiples of 4. Therefore, the first byte of the next PTD will be located at the next
multiple-of-four address, 18H.
Full-speed USB single-chip host and device controller
RAM buffer
PTD
(8 bytes)
payload data
(14 bytes)
PTD
(8 bytes)
payload data
00Htop
08H
15H
18H
20H
MGT953
Fig 28. PTD data with DWORD alignment in buffer RAM.
9.4.3 Operation and C program example
Figure 29 shows the block diagram for internal FIFO buffer RAM operations in PIO
mode. The ISP1161A provides one register as the access port for each buffer RAM.
Forthe ITL buffer RAM, the access port is the ITLBufferPort register (40H - read, C0H
- write). For the ATL buffer RAM, the access port is the ATLBufferPort register (41H read, C1H - write). The buffer RAM is an array of bytes (8 bits) while the access port
is a 16-bit register. Therefore, each read/write operation on the port accesses two
consecutive memory locations, incrementing the pointer of the internal bufferRAM by
two. The lower byte of the access port register corresponds to the data byte at the
even location of the buffer RAM, and the upper byte corresponds to the next data
byte at the odd location of the buffer RAM. Regardless of the number of data bytes to
be transferred, the command code must be issued merely once, and it will be
followed by a number of accesses of the data port (see Section 8.4).
When the pointer of the buffer RAM reaches the value of the HcTransferCounter
register, an internal EOT signal will be generated to set bit 2, AllEOTInterrupt, of the
HcµPinterrupt register and update the HcBufferStatus register, to indicate that the
whole data transfer has been completed.
For ITL buffer RAM, every Start Of Frame (SOF) signal (1 ms) will cause toggling
between ITL0 and ITL1, but this depends on the buffer status. If both ITL0BufferFull
and ITL1BufferFull of the HcBufferStatus register are already logic 1, meaning that
both ITL0 and ITL1 buffer RAMs are full, the toggling will not happen. In this case, the
microprocessor will always have access to ITL1.
Full-speed USB single-chip host and device controller
Control registers
TransferCounter
µPInterrupt
BufferStatus
ITLBufferPort
ATLBufferPort
(16-bit width)
Commands
Command register
=
toggle
T
EOT
12
internal EOT
ISP1161A
0
SOF
BufferStatus
000H
001H
3FFH
ITL0 buffer RAM
(8-bit width)
000H
001H
3FFH
Fig 29. PIO access to internal FIFO buffer RAM.
Following is an example of a C program that shows how to write data into the ATL
buffer RAM. The total number of data bytes to be transferred is 80 (decimal) which
will be set into the HcTransferCounterregister as 50H. The data consists of four types
of PTD data:
1. The first PTD header (IN) is 8 bytes, followed by 16 bytes of space reserved for
its payload data;
2. The second PTD header (IN) is also 8 bytes, followed by 8 bytes of space
reserved for its payload data;
3. The third PTD header (OUT) is 8 bytes, followed by 16 bytes of payload data with
values beginning from 0H to FH incrementing by 1;
4. The fourth PTD header (OUT) is also 8 bytes, followed by 8 bytes of payload data
with values beginning from 0H to EH incrementing by 2.
ITL1 buffer RAM
(8-bit width)
000H
001H
7FFH
ATL buffer RAM
(8-bit width)
Pointer
automatically
increments by 2
MGT951
In all PTDs, we have assigned device address as 5 and endpoint as 1. ActualBytes is
always zero (0). TotalBytes equals the number of payload data bytes transferred,
however, note that for bulk and control transfers, TotalBytes can be greater than
MaxPacketSize.
Table 6 shows the results after running this program.
// Define I/O Port Address for HC
#define HcDataPort 0x290
#define HcCmdPort 0x292
// Declare external functions to be used
unsigned int HcRegRead(unsigned int wIndex);
void HcRegWrite(unsigned int wIndex,unsigned int wValue);
ISP1161A
Full-speed USB single-chip host and device controller
void main(void)
{
unsigned int i;
unsigned int wCount,wData;
// Prepare PTD data to be written into HC ATL buffer RAM:
unsigned int PTDData[0x28]=
{
0x0800,0x1010,0x0810,0x0005, // PTD header for IN token #1
// Reserved space for payload data of IN token #1
0x0000,0x0000,0x0000,0x0000, 0x0000,0x0000,0x0000,0x0000,
0x0800,0x1008,0x0808,0x0005, // PTD header for IN token #2
// Reserved space for payload data of IN token #2
0x0000,0x0000,0x0000,0x0000,
0x0800,0x1010,0x0410,0x0005, // PTD header for OUT token #1
0x0100,0x0302,0x0504,0x0706, // Payload data for OUT token #1
0x0908,0x0b0a,0x0d0c,0x0f0e,
0x0800,0x1808,0x0408,0x0005, // PTD header for OUT token #2
0x0200,0x0604,0x0a08,0x0e0c // Payload data for OUT token #2
};
HcRegWrite(wHcuPInterrupt,0x04); // Clear EOT interrupt bit
// HcRegWrite(wHcITLBufferLength,0x0);
HcRegWrite(wHcATLBufferLength,0x1000); // RAM full use for ATL
// Set the number of bytes to be transferred
HcRegWrite(wHcTransferCounter,0x50);
wCount = 0x28; // Get word count outport
(HcCmdPort,0x00c1); // Command for ATL buffer write
}
//
// Write HC 16-bit registers
//
void HcRegWrite(unsigned int wIndex,unsigned int wValue)
{
outport(HcCmdPort,wIndex | 0x80);
outport(HcDataPort,wValue);
}
Table 6:Run results of the C program example
Observed itemsHC notinitializedandnot in
USBOperational state
HcµPinterrupt register
Bit 1 (ATLInt)01microprocessor must read ATL
Bit 2 (AllEOTInterrupt)11transfer completed
HcBufferStatus register
Bit 2 (ATLBufferFull)11transfer completed
Bit 5 (ATLBufferDone)01PTD data processed by HC
USB Traffic on USB BusNoYesOUT packets can be seen
HC initialized and in
USBOperational state
Comments
9.5 HC operational model
Upon power-up, the HCD initializes all operational registers (32-bit). The
FSLargestDataPacket field (bits 30 to 16) of the HcFmInterval register (0DH - read,
8DH - write) and the HcLSThreshold register (11H - read, 91H - write) determine the
end of the frame for full-speed and low-speed packets. By programming these fields,
the effectiveUSB bus usage can be changed. Furthermore,the size of the ITL buffers
(HcITLBufferLength, 2AH - read, AAH - write) is programmed.
In the case when a USB frame contains both ISO and AT packets, two interrupts will
be generated per frame.
One interrupt is issued concurrently with the SOF.This interrupt (the ITLint bit is set in
the HcµPInterrupt register) triggers reading and writing of the ITL buffer by the
microprocessor, after which the interrupt is cleared by the microprocessor.
Next the programmable ATL Interrupt (the ATLint bit is set in the HcµPInterrupt
register) is issued, which triggers reading and writing of the ATL buffer by the
microprocessor, after which the interrupt is cleared by the microprocessor. If the
microprocessor cannot handle the ISO interrupt before the next ISO interrupt,
disrupted ISO traffic can result.
To be able to send more than one packet to the same Control or Bulk endpoint in the
same frame, an Active bit and a TotalBytes field are introduced (see Table 5). The
Active bit is cleared only if all data of the Philips Transfer Descriptor (PTD) has been
transferredor if a transaction at that endpoint contained a fatal error. If all PTDs of the
ATL are serviced, and the frame is not over yet, the HC starts looking for a PTD with
the Active bit still set. If such a PTD is found and there is still enough time in this
frame, another transaction is started on the USB bus for this endpoint.
ISP1161A
Full-speed USB single-chip host and device controller
For ISO processing, the HCD also has to take care of the HCBufferStatus register
(2CH, read only) for the ITL buffer RAM operations. After the HCD writes ISO data
into ITL buffer RAM, the ITL0BufferFull or ITL1BufferFull bit (depends if it is ITL0 or
ITL1) will be set to logic 1.
After the HC processes the ISO data in the ITL buffer RAM, the corresponding
ITL0BufferDone or ITL1BufferDone bit will automatically be set to logic 1.The HCD
can clear the buffer status bits by a read of the ITL buffer RAM. This must be done
within the 1 ms frame from which the ITL0BufferDone or ITL1BufferDone was set.
For example, the HCD writes ISO_A data into the ITL0 buffer in the first frame. This
will cause the HCBufferStatus register to show that the ITL0 buffer is full by setting
the ITL0BufferFull bit to logic 1. At this stage the HCD cannot write ISO data into the
ITL0 buffer RAM again.
In the second frame, the HC will process the ISO-A data in the ITL0 buffer. At the
same time, the HCD can write ISO-B data into ITL1 buffer. When the next SOF
comes (the beginning of the third frame), both the ITL1BufferFull and ITL0BufferDone
are automatically set to logic 1.
In the third frame the HCD has to read at least two bytes (one word) of the ITL0 buffer
to clear both the ITL0BufferFull and ITL0BufferDone bits. If both are not cleared,
when the next SOF comes (the beginning of the fourth frame) the ITL0BufferDone
and ITL0BufferFull bits will be cleared automatically. This also applies to the ITL1
buffer because the ITL0 and ITL1 are Ping-Pong structured buffers. To recover from
this state, a power-on reset or software reset will have to be applied.
In example 1 (Figure 30), the microprocessor is fast enough to read back and
download a scenario before the next interrupt. Note that on the ISO interrupt of
frame N:
• The ISO packet for frame N + 1 will be written
• The AT packet for frame N + 1 will be written.
SOF
(frame N)(frame N + 1)(frame N + 2)(frame N + 3)
traffic
on USB
AT
interrupt
ISP1161A
Full-speed USB single-chip host and device controller
ISO
interrupt
read ISO_A(N − 1) write ISO_A(N + 1)
Fig 30. HC time domain behavior: example 1.
In example 2 (Figure 31), the microprocessor is still busy transferring the AT data
when the ISO interrupt of the next frame (N + 1) is raised. As a result, there will be no
AT traffic in frame N + 1. The HC does not raise an AT interrupt in frame N + 1. The
AT part is simply postponed until frame N + 2. On the AT N + 2 interrupt, the transfer
mechanism is back to normal operation. This simple mechanism ensures, among
other things, that Control transfers are not dropped systematically from the USB in
case of an overloaded microprocessor.
(frame N)(frame N + 1)(frame N + 2)(frame N + 3)
MGT954
read AT(N)write AT(N + 1)
MGT955
Fig 31. HC time domain behavior: example 2.
In example 3 (Figure 32), the ISO part is still being written while the Start of Frame
(SOF) of the next frame has occurred. This will result in undefined behavior for the
ISO data on the USB bus in frame N + 1 (depending on if the exact timing data is
corrupted or not). The HC should not raise an AT interrupt in frame N + 1.
The different phases of a Control transfer (SETUP,Data and Status) should never be
put in the same ATL.
9.6 Microprocessor loading
The maximum amount of data that can be transferred for an endpoint during one
frame is 1023 bytes. The number of USB packets that are needed for this batch of
data depends on the maximum packet size that is specified.
ISP1161A
Full-speed USB single-chip host and device controller
MGT956
The HCD has to schedule the transactions in a frame. On the other hand, the HCD
must have the ability to handle the interrupts coming from the HC every 1 ms. It must
also be able to do the scheduling for the next frame, reading the frame information
from and writing the next frame information to the bufferRAM in the time between the
end of the current frame and the start of the next frame.
9.7 Internal pull-down resistors for downstream ports
There are four internal 15 kΩ pull-down resistors built into the ISP1161A for the two
downstream ports: two resistors for each port. These resistors are software
selectable by programming bit 12 (2_DownstreamPort15K resistorsel) of the
HcHardwareConfiguration register (20H - read, A0H - write). When bit 12 is logic 0,
external 15 kΩ pull-down resistors are used. If bit 12 is set to logic 1, the internal
15 kΩ pull-down resistors are used. See Figure 33.
This feature is a cost-saving option. However, the power-on reset default value of
bit 12 is logic 0. If using the internal resistors, the HCD must set this bit status after
every reset, because a reset action (hardware or software) will clear this bit.
Full-speed USB single-chip host and device controller
HcHardware
Configuration
ISP1161A
bit 12
internal
15 kΩ
(2×)
D−
D+
004aaa088
22 Ω
22 Ω
external
15 kΩ
(2×)
V
BUS
connector
47 pF
(2×)
USB
Using either internal or external 15 kΩ resistors.
Fig 33. Use of 15 kΩ pull-down resistors on downstream ports.
9.8 OC detection and power switching control
A downstream port provides 5 V power supply to V
hardware functions to monitor the downstream ports loading conditions and control
their power switching. These hardware functions are implemented by the internal
power switching control circuit and overcurrent detection circuit. H_PSW1 and
H_PSW2 are power switching control output pins (active LOW, open drain) for
downstream port 1 and 2, respectively. H_OC1 and H_OC2 are overcurrent detection
input pins for downstream ports 1 and 2, respectively.
. The ISP1161A has built-in
BUS
Figure 34 shows the ISP1161A downstream port power management scheme
(‘n’ represents the downstream port numbers, n=1or2).
‘n’ represents the downstream port number (n=1or2)
Fig 34. Downstream port power management scheme.
9397 750 13962
Product dataRev. 03 — 23 December 200439 of 134
Philips Semiconductors
9.8.1 Using an internal OC detection circuit
The internal OC detection circuit can be used only when VCC(pin 56) is connected to
a 5 V power supply. The HCD must set AnalogOCEnable, bit 10 of the
HcHardwareConfiguration register, to logic 1.
An application using the internal OC detection circuit and internal 15 kΩ pull-down
resistors is shown in Figure 35. In this example, the HCD must set both
AnalogOCEnable and DownstreamPort15KresistorSel to logic 1. They are bit 10 and
bit 12 of the HcHardwareConfiguration register, respectively.
When H_OCn detects an overcurrent status on a downstream port, H_PSWn will
output HIGH, logic 1 to turn off the 5 V power supply to the downstream port V
When there is no such condition, H_PSWn will output LOW,logic 0 to turn on the 5 V
power supply to the downstream port V
In general applications, a P-channel MOSFET can be used as the power switch for
V
BUS
the drain, and H_PSWn to the gate. Call the voltage drop across the drain and source
the overcurrent detection voltage (VOC). For the internal overcurrent detection circuit,
a voltage comparator has been incorporated with a nominal voltage threshold (∆V
of 75 mV. When VOC exceeds V
off the P-channel MOSFET. If the P-channel MOSFET has a R
overcurrent threshold will be 500 mA. The selection of a P-channel MOSFET with a
different R
ISP1161A
Full-speed USB single-chip host and device controller
.
BUS
. Connect the 5 V power supply to the source of the P-channel MOSFET, V
, H_PSWn will output a HIGH level, logic 1 to turn
trip
of 150 mΩ, the
DSon
will result in a different overcurrent threshold.
DSon
BUS
BUS
trip
.
to
)
P-Ch
MOSFET
+
5 V
∆V = +5 V − V
V
BUS
‘n’ represents the downstream port number (n=1or2)
When VCC (pin 56) is connected to a 3.3 V instead of the 5 V power supply, the
internal OC detection circuit cannot be used. An external OC detection circuit must be
used instead. Regardless of the VCC value, an external OC detection circuit can
always be used. To use an external OC detection circuit, AnalogOCEnable, bit 10 of
the HcHardwareConfiguration register, must be logic 0. By default after reset, this bit
is already logic 0; therefore, the HCD does not need to clear this bit.
Figure 36 shows how to use an external OC detection circuit.
V
BUS
downstream
connector
external
OC detect
VoV
OC
EN
USB
port
ISP1161A
Full-speed USB single-chip host and device controller
+
3.3 V or +5 V
V
CC
+
5 V
22 Ω
H_OCn
H_PSWn
H_DMn
i
regulator
OC detect
≥
OC select
1
0
Reg
PSW
C/L
HC CORE
HcHardware
Configuration
bit 10
22 Ω
47 pF
(2×)
‘n’ represents the downstream port number (n=1or2)
Fig 36. Using an external OC detection circuit.
9.9 Suspend and wake-up
9.9.1 HC suspended state
The HC can be put into suspended state by setting the HcControl register (01H read, 81H - write). See Figure 23 for the HC’s flow of USB state changes.
Full-speed USB single-chip host and device controller
XOSC
On
On
VOLTAGE
REGULATOR
XOSC_6MHz
(to DC PLL)
DC_EnableClock
On
PLL_Lock
HC PLL
PLL_ClkOut
HC_EnableClock
Fig 37. ISP1161A suspend and resume clock scheme.
In suspended state, the device will consume considerably less power by turning off
the internal 48 MHz clock, PLL and crystal, and setting the internal regulator to
power-down mode. The ISP1161A suspend and resume clock scheme is shown in
Figure 37.
Remark: The ISP1161A can only be put into a fully suspended state only after both
the HC and the DC go into suspend state. At this point, the crystal can be turned off
and the internal regulator can be put into power-down mode.
HC_ClkOk
HC_RawClk48M
HcHardware Configuration
bit 11 (SuspendClkNotStop)
H_Wakeup (pin)
DIGITAL
CLOCK
SWITCH
On
HC_Clk48MOut
HC_NeedClock
CS (pin)
HC
CORE
MGT958
Pin H_SUSPEND is the sensing output pin for HC’s suspended state. When the HC
goes into USBSuspend state, this pin will output a HIGH level (logic 1). This pin is
cleared to LOW (logic 0) level only when the HC is put into a USBReset state or
USBOperational state (refer to the HcControl register bits 7 to 6, 01H - read, 81H write). Bit 11, SuspendClkNotStop, of the HcHardwareConfiguration register (20H read, A0H - write), defines if the HC internal clock is stopped or kept running when
the HC goes into USBSuspend state. After the HC enters the USBSuspend state for
1.3 ms, the internal clock will be stopped if bit SuspendClkNotStop is logic 0.
9.9.2 HC wake-up from suspended state
There are three methods to wake up the HC from the USBSuspend state: hardware
wake-up, software wake-up, and USB bus resume.
They are described as follows:
• Wake-up by pin H_WAKEUP
Pins H_SUSPEND and H_WAKEUP provide a method of remote wake-up control
for the HC without the need to access the HC internal registers. H_WAKEUP is an
external wake-up control input pin for the HC. After the HC goes into USBSuspend
state, it can be woken up by sending a HIGH level pulse to pin H_WAKEUP. This
will turn on the HC’s internal clock, and set bit 6, ClkReady, of the HcµPInterrupt
register (24H - read, A4H - write).
No matter which method is used to wake up the HC from USBSuspend state, the
corresponding interrupt bits must be enabled before the HC goes into USBSuspend
state so that the microprocessor can receive the correct interrupt request to wake up
the HC.
ISP1161A
Full-speed USB single-chip host and device controller
Under the USBSuspend state, once pin H_WAKEUP goes HIGH, after 160 µs, the
internal clock will be up. If pin H_WAKEUP continues to be HIGH, then the internal
clock will be kept running, and the microprocessor can set the HC into
USBOperational state during this time.
If H_WAKEUP goes LOW for more than 1.14 ms, the internal clock stops, and the
HC goes back into USBSuspend state.
During the USBSuspend state, an external microprocessor issues a chip select
signal through pin CS. This method of access to ISP1161A internal registers is a
software wake-up.
For a USB bus resume, a USB device attached to the root hub port issues a
resume signal to the HC through the USB bus, switching the HC from
USBSuspend state to USBResume state. This will also set the ResumeDetected
bit of the HcInterruptStatus register (03H - read, 83H - write).
10. HC registers
The HC contains a set of on-chip control registers. These registers can be read or
written by the Host Controller Driver (HCD). The Control and Status register sets,
Frame Counter register sets, and Root Hub register sets are grouped under the
category of HC Operational registers (32 bits). These operational registers are made
compatible to OpenHCI (Host Controller Interface) Operational registers. This allows
the OpenHCI HCD to be easily ported to ISP1161A.
Reserved bits may be defined in future releases of this specification. To ensure
interoperability, the HCD must not assume that a reserved field contains logic 0.
Furthermore, the HCD must always preserve the values of the reserved field. When a
R/W register is modified, the HCD must first read the register, modify the bits desired,
and then write the register with the reserved bits still containing the original value.
Alternatively, the HCD can maintain an in-memory copy of previously written values
that can be modified and then written to the HC register. When a ‘write to set’ or ‘clear
the register’ is performed, bits written to reserved fields must be logic 0.
As shown in Table 7, the addresses (the commands for accessing registers) of these
32-bit Operational registers are similar the offsets defined in the OHCI specification
with the addresses being equal to offset divided by 4.
Full-speed USB single-chip host and device controller
Table 7:HC registers summary
Address (Hex)RegisterWidth ReferenceFunctionality
readwrite
00-HcRevision32Section 10.1.1 on page 44HC Control and Status registers
0181HcControl32Section 10.1.2 on page 45
0282HcCommandStatus32Section 10.1.3 on page 46
0383HcInterruptStatus32Section 10.1.4 on page 48
0484HcInterruptEnable32Section 10.1.5 on page 49
0585HcInterruptDisable32Section 10.1.6 on page 50
0D8DHcFmInterval32Section 10.2.1 on page 52HC Frame Counter registers
0E-HcFmRemaining32Section 10.2.2 on page 53
0F-HcFmNumber32Section 10.2.3 on page 53
1191HcLSThreshold32Section 10.2.4 on page 54
1292HcRhDescriptorA32Section 10.3.1 on page 56HC Root Hub registers
1393HcRhDescriptorB32Section 10.3.2 on page 57
1494HcRhStatus32Section 10.3.3 on page 59
1595HcRhPortStatus[1]32Section 10.3.4 on page 61
1696HcRhPortStatus[2]32Section 10.3.4 on page 61
20A0HcHardwareConfiguration16Section 10.4.1 on page 65HC DMA and Interrupt Control
21A1HcDMAConfiguration16Section 10.4.2 on page 66
22A2HcTransferCounter16Section 10.4.3 on page 67
24A4HcµPInterrupt16Section 10.4.4 on page 68
25A5HcµPInterruptEnable16Section 10.4.5 on page 69
27-HcChipID16Section 10.5.1 on page 70HC Miscellaneous registers
28A8HcScratch16Section 10.5.2 on page 71
-A9HcSoftwareReset16Section 10.5.3 on page 71
2AAAHcITLBufferLength16Section 10.6.1 on page 72HC Buffer RAM Control registers
2BABHcATLBufferLength16Section 10.6.2 on page 72
2C-HcBufferStatus16Section 10.6.3 on page 73
2D-HcReadBackITL0Length16Section 10.6.4 on page 74
2E-HcReadBackITL1Length16Section 10.6.5 on page 74
40C0HcITLBufferPort16Section 10.6.6 on page 75
41C1HcATLBufferPort16Section 10.6.7 on page 75
31 to 8−Reserved
7 to 0REV[7:0]Revision: This read-only field contains the BCD representation of
the version of the HCI specification that is implemented by this HC.
For example, a value of 11H corresponds to version 1.1. All HC
implementations that are compliant with this specification will have
a value of 10H.
10.1.2 HcControl register (R/W: 01H/81H)
The HcControl register defines the operating modes for the HC.
RemoteWakeupEnable (RWE) is modified only by the HCD.
31 to 11-reserved
10RWERemoteWakeupEnable: This bit is used by the HCD to enable or
disable the remote wake-up feature upon the detection of
upstream resume signaling. When this bit is set and the
ResumeDetected bit in HcInterruptStatus is set, a remote wake-up
is signaled to the host system. Setting this bit has no impact on the
generation of hardware interrupt.
9RWCRemoteWakeupConnected: This bit indicates whether the HC
supports remote wake-up signaling. If remote wake-up is
supported and used by the system, it is the responsibility of
system firmware to set this bit during POST. The HC clears the bit
upon a hardware reset but does not alter it upon a software reset.
Remote wake-up signaling of the host system is host-bus-specific,
and is not described in this specification.
8-reserved
7 to 6HCFSHostControllerFunctionalState for USB:
00B — USBReset
01B — USBResume
10B — USBOperational
11B — USBSuspend
A transition to USBOperational from another state causes
start-of-frame (SOF) generation to begin 1 ms later. The HCD
determines whether the HC has begun sending SOFs by reading
the StartofFrame field of HcInterruptStatus.
This field can be changed by the HC only when in the
USBSuspend state. The HC can move from the USBSuspend
state to the USBResumestateafterdetectingthe resume signaling
from a downstream port.
The HC enters USBReset after a software reset and a hardware
reset. The latter also resets the Root Hub and asserts subsequent
reset signaling to downstream ports.
5 to 0-reserved
10.1.3 HcCommandStatus register (R/W: 02H/82H)
The HcCommandStatus register is used by the HC to receive commands issued by
the HCD, and it also reflects the HC’s current status. To the HCD, it appears to be a
‘write to set’ register. The HC must ensure that bits written as logic 1 become set in
Full-speed USB single-chip host and device controller
the register while bits written as logic 0 remain unchanged in the register. The HCD
may issue multiple distinct commands to the HC without concern for corrupting
previously issued commands. The HCD has normal read access to all bits.
The SchedulingOverrunCount field indicates the number of frames with which the HC
has detected the scheduling overrun error. This occurs when the Periodic list does
not complete before EOF. When a scheduling overrun error is detected, the HC
increments the counter and sets the SchedulingOverrun field in the HcInterruptStatus
register.
Code (Hex): 02 — read
Code (Hex): 82 — write
Table 12: HcCommandStatus register: bit allocation
Table 13: HcCommandStatus register: bit description
BitSymbolDescription
31 to 18-reserved
17 to 16SOC[1:0]SchedulingOverrunCount: The field is incremented on each
15 to 1-reserved
0HCRHostControllerReset: This bit is set by the HCD to initiate a
ISP1161A
Full-speed USB single-chip host and device controller
scheduling overrun error. It is initialized to 00B and wraps around
at 11B. It will be incremented when a scheduling overrun is
detected even if SchedulingOverrun in HcInterruptStatus has
already been set. This is used by HCD to monitor any persistent
scheduling problems.
software reset of the HC. Regardless of the functional state of HC,
it movestothe USBSuspend state in which most of the operational
registers are reset, except those stated otherwise, and no Host
bus accesses are allowed. This bit is cleared by HC upon the
completion of the reset operation. The reset operation must be
completed within 10 µs. This bit, when set, does not cause a reset
to the Root Hub and no subsequent reset signaling will be
asserted to its downstream ports.
10.1.4 HcInterruptStatus register (R/W: 03H/83H)
This register provides the status of the events that cause hardware interrupts. When
an event occurs, the HC sets the corresponding bit in this register. When a bit is set, a
hardware interrupt is generated if the interrupt is enabled in the HcInterruptEnable
register (see Section 10.1.5) and the MasterInterruptEnable bit is set. The HCD can
clear individual bits in this register by writing logic 1 to the bit positions to be cleared,
but cannot set any of these bits. Conversely, the HC can set bits in this register, but
cannot clear the bits.
Code (Hex): 03 — read
Code (Hex): 83 — write
Table 14: HcInteruptStatus register: bit allocation
Table 15: HcInterruptStatus register: bit description
BitSymbolDescription
31 to 7reserved
6RHSCRootHubStatusChange: This bit is set when the content of
HcRhStatus or the content of any of HcRhPortStatus[1:2] has
changed.
5FNOFrameNumberOverflow: This bit is set when the MSB of
HcFmNumber (bit 15) changes value.
4UEUnrecoverableError: This bit is set when the HC detects a
system error not related to USB. The HC does not proceed with
any processing nor signaling before the system error has been
corrected. The HCD clears this bit after the HC has been reset.
OHCI: Always set to logic 0.
3RDResumeDetected: This bit is set when the HC detects that a
device on the USB is asserting resume signaling from a state of no
resume signaling. This bit is not set when HCD enters the
USBResume state.
2SFStartofFrame: At the start of each frame, this bit is set by the HC
and an SOF is generated.
1-reserved
0SOSchedulingOverrun: This bit is set when the USB schedules for
current frame overruns. A scheduling overrun will also cause the
SchedulingOverrunCount of HcCommandStatus to be
incremented.
10.1.5 HcInterruptEnable register (R/W: 04H/84H)
Each enable bit in the HcInterruptEnable register corresponds to an associated
interrupt bit in the HcInterruptStatus register. The HcInterruptEnable register is used
to control which events generate a hardware interrupt. A hardware interrupt is
requested on the host bus when three conditions occur:
• A bit is set in the HcInterruptStatus register
• The corresponding bit in the HcInterruptEnable register is set
• The MasterInterruptEnable bit is set.
Writing logic 1 to a bit in this register sets the corresponding bit, whereas writing
logic 0 to a bit in this register leaves the corresponding bit unchanged. On a read, the
current value of this register is returned.
Table 17: HcInterruptEnable register: bit description
BitSymbolDescription
31MIEMasterInterruptEnable by the HCD: Logic 0 is ignored by the HC.
Logic 1 enables interrupt generation by events specified in other
bits of this register.
30 to 7-reserved
6RHSC0 — ignore
1 — enable interrupt generation due to Root Hub Status Change
5FNO0 — ignore
1 — enable interrupt generation due to Frame Number Overflow
4UE0 — ignore
1 — enable interrupt generation due to Unrecoverable Error
3RD0 — ignore
1 — enable interrupt generation due to Resume Detect
2SF0 — ignore
1 — enable interrupt generation due to Start of Frame
1-reserved
0SO0 — ignore
1 — enable interrupt generation due to Scheduling Overrun
10.1.6 HcInterruptDisable register (R/W: 05H/85H)
Each disable bit in the HcInterruptDisable register corresponds to an associated
interrupt bit in the HcInterruptStatus register. The HcInterruptDisable register is
coupled with the HcInterruptEnable register. Thus, writing logic 1 to a bit in this
register clears the corresponding bit in the HcInterruptEnable register, whereas
Full-speed USB single-chip host and device controller
writing logic 0 to a bit in this register leaves the corresponding bit in the
HcInterruptEnable register unchanged. On a read, the current value of the
HcInterruptEnable register is returned.
Code (Hex): 05 — read
Code (Hex): 85 — write
Table 18: HcInterruptDisable register: bit allocation
Full-speed USB single-chip host and device controller
10.2 HC frame counter registers
10.2.1 HcFmInterval register (R/W: 0DH/8DH)
The HcFmIntervalregister contains a 14-bit value which indicates the bit time interval
in a frame (that is, between two consecutive SOFs), and a 15-bit value indicating the
full-speed maximum packet size that the HC may transmit or receive without causing
a scheduling overrun. The HCD may carry out minor adjustments on the
FrameIntervalby writing a new value at each SOF. This allows the HC to synchronize
with an external clock source and to adjust any unknown clock offset.
31FRTFrameRemainingToggle: This bit is loaded from the
FrameIntervalToggle field of the HcFmInterval register whenever
FrameRemaining reaches 0. This bit is used by the HCD for
synchronization between FrameInterval and FrameRemaining.
30 to 14-reserved
13 to 0FR[13:0]FrameRemaining: This counter is decremented at each bit time.
When it reaches zero, it is reset by loading the FrameIntervalvalue
specified in the HcFmInterval register at the nextbit time boundary.
When entering the USBOperational state, the HC reloads it with
the content of the FrameInterval part of the HcFmInterval register
and uses the updated value from the next SOF.
10.2.3 HcFmNumber register (R: 0FH)
The HcFmNumber register is a 16-bit counter. It provides a timing reference for
events happening in the HC and the HCD. The HCD may use the 16-bit value
specified in this register and generate a 32-bit frame number without requiring
frequent access to the register.
31 to 16−reserved
15 to 0FN[15:0]FrameNumber: This field is incremented when HcFmRemaining
is reloaded. It rolls over to 0000H after FFFFH. When the
USBOperational state is entered, this field will be incremented
automatically. HC will set the StartofFrame bit in the
HcInterruptStatus register.
10.2.4 HcLSThreshold register (R/W: 11H/91H)
The HcLSThreshold register contains an 11-bit value used by the HC to determine
whether to commit to the transfer of a maximum of 8-byte LS packet before EOF.
Neither the HC nor the HCD is allowed to change this value.
31 to 11−reserved
10 to 0LST[10:0]LSThreshold: Contains a value that is compared to the
FrameRemaining field before a low-speed transaction is initiated.
The transaction is started only if FrameRemaining ≥ this field. The
value is calculated by the HCD, which considers transmission and
set-up overhead. Default value: 1576 (628H)
10.3 HC Root Hub registers
All registers included in this partition are dedicated to the USB Root Hub, which is an
integral part of the HC although it is functionally a separate entity. The Host Controller
Driver (HCD) emulates USBD accesses to the Root Hub via a register interface. The
HCD maintains many USB-defined hub features that are not required to be supported
in hardware. For example, the Hub’s Device, Configuration, Interface, Endpoint
Descriptors, as well as some static fields of the Class Descriptor, are maintained only
in the HCD. The HCD also maintains and decodes the Root Hub’s device address as
well as other minor operations more suited for software than for hardware.
The Root Hub registers were developed to match the bit organization and operation
of typical hubs found in the system.
Four 32-bit registers have been defined:
• HcRhDescriptorA
• HcRhDescriptorB
• HcRhStatus
• HcRhPortStatus[1:NDP]
Each register is read and written as a DWORD. These registers are only written
during initialization to correspond with the system implementation. The
HcRhDescriptorA and HcRhDescriptorB registers are writeable regardless of the
HC’s USB states. HcRhStatus and HcRhPortStatus are writeable during the
USBOperational state only.
Full-speed USB single-chip host and device controller
10.3.1 HcRhDescriptorA register (R/W: 12H/92H)
The HcRhDescriptorA register is the first register of two describing the characteristics
of the Root Hub. Reset values are Implementation-Specific (IS). The descriptorlength
(11), descriptor type and hub controller current (0) fields of the hub Class Descriptor
are emulated by the HCD. All other fields are located in the registers
HcRhDescriptorA and HcRhDescriptorB.
Remark: IS denotes an implementation-specific reset value for that field.
Code (Hex): 12 — read
Code (Hex): 92 — write
Table 28: HcRhDescriptorA register: bit description
Full-speed USB single-chip host and device controller
Table 29: HcRhDescriptorA register: bit description
BitSymbolDescription
12NOCPNoOverCurrentProtection: This bit describes how the
overcurrent status for the Root Hub ports are reported. When this
bit is cleared, the OverCurrentProtectionMode field specifies
global or per-port reporting.
0 — overcurrent status is reported collectively for all downstream
ports
1 — no overcurrent reporting supported
11OCPMOverCurrentProtectionMode: This bit describes how the
overcurrent status for the Root Hub ports is reported. At reset, this
field reflects the same mode as PowerSwitchingMode.This field is
valid only if the NoOverCurrentProtection field is cleared.
0 — overcurrent status is reported collectively for all downstream
ports.
1 — overcurrent status is reported on a per-port basis. On
power-up, clear this bit and then set it to logic 1.
10DTDeviceType: This bit specifies that the Root Hub is not a
compound device—it is not permitted. This field will always
read/write 0.
9NPSNoPowerSwitching: These bits are used to specify whether
power switching is supported or ports are always powered. It is
implementation-specific. When this bit is cleared, the bit
PowerSwitchingMode specifies global or per-port switching.
…continued
0 — ports are power switched
1 — ports are always powered on when the HC is powered on
8PSMPowerSwitchingMode: This bit is used to specify how the power
switching of the Root Hub ports is controlled. It is
implementation-specific. This field is valid only if the
NoPowerSwitching field is cleared.
0 — all ports are powered at the same time
1 — each port is powered individually. This mode allows port
power to be controlled by either the global switch or per-port
switching. If the bit PortPowerControlMask is set, the port
responds to only port power commands (Set/ClearPortPower). If
the port mask is cleared, then the port is controlled only by the
global power switch (Set/ClearGlobalPower).
7 to 2-reserved
1 to 0NDP[1:0]NumberDownstreamPorts: These bits specify the number of
downstream ports supported by the Root Hub. The maximum
number of ports supported by ISP1161A is 2.
10.3.2 HcRhDescriptorB register (R/W: 13H/93H)
The HcRhDescriptorB register is the second register of two describing the
characteristics of the Root Hub. These fields are written during initialization to
correspond with the system implementation. Reset values are
implementation-specific (IS).
Full-speed USB single-chip host and device controller
Table 31: HcRhDescriptorB register: bit description
BitSymbolDescription
15 to 3-reserved
2 to 0DR[2:0]DeviceRemovable: Each bit is dedicated to a port of the Root
Hub. When cleared, the attached device is removable. When set,
the attached device is not removable.
Bit 0 — reserved
Bit 1 — Device attached to Port #1
Bit 2 — Device attached to Port #2
…continued
10.3.3 HcRhStatus register (R/W: 14H/94H)
The HcRhStatus register is divided into two parts. The lower word of a DWORD
represents the Hub Status field and the upper word represents the Hub Status
Change field. Reserved bits should always be written as logic 0.
The HcRhPortStatus[1:2] register is used to control and report port events on a
per-port basis. NumberDownstreamPorts represents the number of HcRhPortStatus
registers that are implemented in hardware. The lower word is used to reflect the port
status, whereas the upper word reflects the status change bits. Some status bits are
implemented with special write behavior. If a transaction (token through handshake)
is in progress when a write to change port status occurs, the resulting port status
change must be postponed until the transaction completes. Reserved bits should
always be written logic 0.
1. Bit 0, InterruptPinEnable, is used as pin INT1’s master interrupt enable. This bit
should be used together with the register HcµPInterruptEnable to enable pin
INT1.
2. Bits 4 and 3 are fixed at logic 0 and logic 1 for the ISP1161A.
Code (Hex): 20 — read
Code (Hex): A0 — write
Table 36: HcHardwareConfiguration register: bit allocation
Full-speed USB single-chip host and device controller
Table 39: HcDMAConfiguration register: bit description
BitSymbolDescription
2DMACounter
Select
1ITL_ATL_
DataSelect
0DMARead
WriteSelect
0 — DMA counter not used. External EOT must be used
1 — Enables the DMA counter for DMA transfer.
HcTransferCounter register must be filled with non-zero values for
DREQ1 to be raised after bit DMA Enable is set
0 — ITL buffer RAM selected for ITL data
1 — ATL buffer RAM selected for ATL data
0 — read from the HC FIFO buffer RAM
1 — write to the HC FIFO buffer RAM
10.4.3 HcTransferCounter register (R/W: 22H/A2H)
This register holds the number of bytes of a PIO or DMA transfer. For a PIO transfer,
the number of bytes being read or written to the Isochronous Transfer List (ITL) or
Acknowledged Transfer List (ATL) buffer RAM must be written into this register. For a
DMA transfer, the number of bytes must be written into this register as well. However,
for this counter to be read into the DMA counter, the HCD must set bit 2 of the
HcDMAConfiguration register. The counter value for ATL must not be greater than
1000H, and for ITL it must not be greater than 800H. When the byte count of the data
transfer reaches this value, the HC will generate an internal EOT signal to set bit 2
(AllEOTInterrupt) of the HcµPInterrupt register, and also update the HcBufferStatus
register.
…continued
Code (Hex): 22 — read
Code (Hex): A2 — write
Table 40: HcTransferCounter register: bit allocation
Bit15141312111098
SymbolCounter value
Reset00000000
AccessR/WR/WR/WR/WR/WR/WR/WR/W
Bit76543210
SymbolCounter value
Reset00000000
AccessR/WR/WR/WR/WR/WR/WR/WR/W
Table 41: HcTransferCounter register: bit description
BitSymbolDescription
15 to 0Counter
value
The number of data bytes to be read to or written from RAM
Full-speed USB single-chip host and device controller
10.4.4 HcµPInterrupt register (R/W: 24H/A4H)
All the bits in this register will be active on power-on reset. However, none of the
active bits will cause an interrupt on the interrupt pin (INT1) unless they are set by the
respective bits in the HcµPInterruptEnable register, and together with bit 0 of the
HcHardwareConfiguration register.
After this register (24H read) is read, the bits that are active will not be reset, until
logic 1 is written to the bits in this register (A4H - write) to clear it. To clear all the
enabled bits in this register, the HCD must write FFH to this register.
1 — clock is ready. After a wake-up is sent, there is a wait for clock
ready. (Maximum is 1 ms, and typical is 160 µs)
5HC
Suspended
4OPR_Reg0 — no event
3-reserved
2AllEOT
Interrupt
0 — no event
1 — the HC has been suspended and no USB activity is sent from
the microprocessor for each ms. When the microprocessor wants
to suspend the HC, the microprocessor must write to the
HcControl register. And when all downstream devices are
suspended, then the HC stops sending SOF; the HC is suspended
by having the HcControl register written into.
1 — there are interrupts from HC side. Need to read HcControl
and HcInterrupt registers to detect type of interrupt on the HC (if
the HC requires the Operational register to be updated)
0 — no event
1 — implies that data transfer has been completed via PIO transfer
or DMA transfer. Occurrence of internal or external EOT will set
this bit.
Full-speed USB single-chip host and device controller
Table 43: HcµPInterrupt register: bit description
BitSymbolDescription
1ATLInt0 — no event
1 — implies that the microprocessor must read ATL data from the
HC. This requires that the HcBufferStatus register must first be
read. The time for this interrupt depends on the number of clocks
bit set for USB activities in each ms.
0SOFITLInt0 — no event
1 — implies that SOF indicates the 1 ms mark. The ITL buffer that
the HC has handled must be read. To know the ITL buffer status,
the HcBufferStatus register must first be read. This is for the
microprocessor to get ISO data to or from the HC. For more
information, see the 6th paragraph in Section 9.5.
The bits 6:0 in this register are the same as those in the HcµPInterrupt register. They
are used together with bit 0 of the HcHardwareConfiguration register to enable or
disable the bits in the HcµPInterrupt register.
At power-on, all bits in this register are masked with logic 0. This means no interrupt
request output on the interrupt pin INT1 can be generated.
When the bit is set to logic 1, the interrupt for the bit is not masked but enabled.
…continued
Code (Hex): 25 — read
Code (Hex): A5 — write
Table 44: HcµPInterruptEnable register: bit allocation
Full-speed USB single-chip host and device controller
Table 45: HcµPInterruptEnable register: bit description
BitSymbolDescription
5HC
Suspended
Enable
4OPR
Interrupt
Enable
3-reserved
2EOT
Interrupt
Enable
1ATL
Interrupt
Enable
0SOF
Interrupt
Enable
0 — power-up value
1 — enables HC suspended interrupt. When the microprocessor
wants to suspend the HC, the microprocessor must write to the
HcControl register. And when all downstream devices are
suspended, then the HC stops sending SOF; the HC is suspended
by having the HcControl register written into.
0 — power-up value
1 — enables the 32-bit Operational register’s interrupt (if the HC
requires the Operational register to be updated)
0 — power-up value
1 — enables the EOT interrupt which indicates an end of a
read/write transfer
0 — power-up value
1 — enables ATL interrupt. The time for this interrupt depends on
the number of clock bits set for USB activities in each ms.
0 — power-up value
1 — enables the interrupt bit due to SOF (for the microprocessor
DMA to get ISO data from the HC by first accessing the
HcDMAConfiguration register)
…continued
10.5 HC miscellaneous registers
10.5.1 HcChipID register (R: 27H)
Read this register to get the ID of the ISP1161A silicon chip. The high byte stands for
the product name (here 61H stands for ISP1161A). The low byte indicates the
revision number of the product including engineering samples.
This register provides a means for software reset of the HC. To reset the HC, the HCD
must write a reset value of F6H to this register. Upon receiving the reset value, the
HC resets all the registers except its buffer memory.
Code (Hex): A9 — write
Table 50: HcSoftwareReset register: bit allocation
Full-speed USB single-chip host and device controller
10.6 HC buffer RAM control registers
10.6.1 HcITLBufferLength register (R/W: 2AH/AAH)
Write to this register to assign the ITL buffer size in bytes: ITL0 and ITL1 are assigned
the same value. For example, if HcITLBufferLength register is set to 2 kbytes, then
ITL0 and ITL1 would be allocated 2 kbytes each.
Table 53: HcITLBufferLength register: bit description
BitSymbolDescription
15 to 0ITLBufferLength[15:0]Assign ITL buffer length
10.6.2 HcATLBufferLength register (R/W: 2BH/ABH)
Write to this register to assign ATL buffer size.
Code (Hex): 2B — read
Code (Hex): AB — write
Remark: The maximum total RAM size is 1000H (4096 in decimal) bytes. That
means ITL0 (length) + ITL1 (length) + ATL (length) ≤ 1000H bytes. For example, if
ATL buffer length has been set to be 800H, then the maximum ITL buffer length can
only be set as 400H.
Table 54: HcATLBufferLength register: bit allocation
Table 57: HcBufferStatus register: bit description
BitSymbolDescription
15 to 6-reserved
5ATLBuffer
Done
4ITL1Buffer
Done
3ITL0Buffer
Done
2ATLBuffer
Full
1ITL1Buffer
Full
0ITL0Buffer
Full
0 — ATL Buffer not read by HC yet
1 — ATL Buffer read by HC
0 — ITL1 Buffer not read by HC yet
1 — ITL1 Buffer read by HC
0 — 1TL0 Buffer not read by HC yet
1 — 1TL0 Buffer read by HC
0 — ATL Buffer is empty
1 — ATL Buffer is full
0 — 1TL1 Buffer is empty
1 — 1TL1 Buffer is full
0 — ITL0 Buffer is empty
1 — ITL0 Buffer is full
Full-speed USB single-chip host and device controller
10.6.4 HcReadBackITL0Length register (R: 2DH)
This register’s value stands for the current number of data bytes inside an ITL0 buffer
to be read back by the microprocessor. The HCD must set the HcTransferCounter
equivalent to this value before reading back the ITL0 buffer RAM.
Code (Hex): 2D — read
Table 58: HcReadBackITL0Length register: bit allocation
Table 59: HcReadBackITL0Length register: bit description
BitSymbolDescription
15 to 0RdITL0BufferLength[15:0]The number of bytes for ITL0 data to be read back by
the microprocessor
10.6.5 HcReadBackITL1Length register (R: 2EH)
This register’s value stands for the current number of data bytes inside the ITL1 buffer
to be read back by the microprocessor. The HCD must set the HcTransferCounter
equivalent to this value before reading back the ITL1 buffer RAM.
Code (Hex): 2E — read
Table 60: HcReadBackITL1Length register: bit allocation
Full-speed USB single-chip host and device controller
10.6.6 HcITLBufferPort register (R/W: 40H/C0H)
This is the ITL buffer RAM read/write port. The bits 15:8 contain the data byte that
comes from the ITL buffer RAM’s even address. The bits 7:0 contain the data byte
that comes from the ITL buffer RAM’s odd address.
Code (Hex): 40 — read
Code (Hex): C0 — write
Table 62: HcITLBufferPort register: bit allocation
Table 63: HcITLBufferPort register: bit description
BitSymbolDescription
15 to 0DataWord[15:0]read/write ITL buffer RAM’s two data bytes.
The HCD must set the byte count into the HcTransferCounter register and check the
HcBufferStatus register before reading from or writing to the buffer. The HCD must
write the command (40H for read, C0H for write) once only, and then read or write
both bytes of the data word. After every read/write, the pointer of ITL buffer RAM will
be automatically increased by two to point to the next data word until it reaches the
value of HcTransferCounter register; otherwise, an internal EOT signal is not
generated to set the bit 2 (AllEOTInterrupt) of the HcµPInterrupt register and update
the HcBufferStatus register.
The HCD must take care of the fact that the internal buffer RAM is organized in bytes.
The HCD must write the byte count into the HcTransferCounter register, but the HCD
reads or writes the buffer RAM by 16 bits (by 1 data word).
10.6.7 HcATLBufferPort register (R/W: 41H/C1H)
This is the ATL buffer RAM read/write port. The bits 15:8 contain the data byte that
comes from the Acknowledged Transfer List (ATL)bufferRAM’s odd address. Bits 7:0
contain the data byte that comes from the ATL buffer RAM’s even address.
Code (Hex): 41 — read
Code (Hex): C1 — write
Table 64: HcATLBufferPort register: bit allocation
Table 65: HcATLBufferPort register: bit description
BitSymbolDescription
15 to 0DataWord[15:0]read/write ATL buffer RAM’s two data bytes.
The HCD must set the byte count into the HcTransferCounter register and check the
HcBufferStatus register before reading from or writing to the buffer. The HCD must
write the command (41H for read, C1H for write) once only, and then read or write
both bytes of the data word. After every read/write, the pointer of ATL buffer RAM will
be automatically increased by two to point to the next data word until it reaches the
value of HcTransferCounter register; otherwise, an internal EOT signal is not
generated to set the bit 2 (AllEOTInterrupt) of the HcµPInterrupt register and update
the HcBufferStatus register.
The HCD must take care of the difference: the internal buffer RAM is organized in
bytes, so the HCD must write the byte count into the HcTransferCounter register, but
the HCD reads or writes the buffer RAM by 16 bits (by 1 data word).
The Device Controller (DC) in the ISP1161A is based on the Philips ISP1181B USB
Full-Speed Interface Device IC. The functionality, commands, and register sets are
the same as ISP1181B in 16-bit bus mode. If there are any differences between the
ISP1181B and ISP1161A data sheets, in terms of the DC functionality,the ISP1161A
data sheet supersedes content in the ISP1181B data sheet.
In general the DC in an ISP1161A provides 16 endpoints for USB device
implementation. Each endpoint can be allocated an amount of RAM space in the
on-chip Ping-Pong buffer RAM.
Remark: The Ping-Pong buffer RAM for the DC is independent of the buffer RAM in
the HC.
When the buffer RAM is full, the DC will transfer the data in the buffer RAM to the
USB bus. When the buffer RAM is empty, an interrupt is generated to notify the
microprocessor to feed in the data. The transfer of data between the microprocessor
and the DC can be done in Programmed I/O (PIO) mode or in DMA mode.
ISP1161A
Full-speed USB single-chip host and device controller
11.1 DC data transfer operation
The following session explains how the DC of an ISP1161A handles an IN data
transfer and an OUT data transfer. In the Device mode, ISP1161A acts as a USB
device: an IN data transfer means transfer from ISP1161A to an external USB Host
(through the upstream port) and an OUT transfer means transfer from external USB
Host to ISP1161A.
11.1.1 IN data transfer
• The arrival of the IN token is detected by the SIE by decoding the PID.
• The SIE also checks for the device number and endpoint number and verifies
whether they are acceptable.
• If the endpoint is enabled, the SIE checks the contents of the DcEndpointStatus
register. If the endpoint is full, the contents of the FIFO are sent during the data
phase, otherwise a Not Acknowledge (NAK) handshake is sent.
• After the data phase, the SIE expects a handshake (ACK) from the host (except for
ISO endpoints).
• On receiving the handshake (ACK), the SIE updates the contents of the
DcEndpointStatus register and the DcInterrupt register,which in turn generates an
interrupt to the microprocessor.ForISO endpoints, the interrupt register is updated
as soon as data is sent because there is no handshake phase.
• On receiving an interrupt, the microprocessor reads the DcInterrupt register. It will
know which endpoint has generated the interrupt and reads the contents of the
corresponding DcEndpointStatus register. If the buffer is empty,it fills up the buffer,
so that data can be sent by the SIE at the next IN token phase.
11.1.2 OUT data transfer
• The arrival of the OUT token is detected by the SIE by decoding the PID.
• The SIE also checks for the device number and endpoint number and verifies
• If the endpoint is enabled, the SIE checks the contents of the DcEndpointStatus
• After the data phase, the SIE sends a handshake (ACK)to the host (except for ISO
• The SIE updates the contents of the DcEndpointStatus register and the
• On receiving interrupt, the microprocessor reads the DcInterrupt register. It will
11.2 Device DMA transfer
ISP1161A
Full-speed USB single-chip host and device controller
whether they are acceptable.
register. If the endpoint is empty, the data from USB is stored to FIFO during the
data phase, otherwise a NAK handshake is sent.
endpoints).
DcInterrupt register, which in turn generates an interrupt to the microprocessor.
For ISO endpoints, the interrupt register is updated as soon as data is received
because there is no handshake phase.
know which endpoint has generated the interrupt and reads the content of the
corresponding DcEndpointStatus register. If the buffer is full, it empties the buffer,
so that data can be received by the SIE at the next OUT token phase.
11.2.1 DMA for IN endpoint (internal DC to external USB host)
When the internal DMA handler is enabled and at least one buffer (Ping or Pong) is
free, the DREQ2 line is asserted. The external DMA controller then starts negotiating
for control of the bus. As soon as it has access, it asserts the DACK2 line and starts
writing data. The burst length is programmable. When the number of bytes equal to
the burst length has been written, the DREQ2 line is de-asserted. As a result, the
DMA controller de-asserts the DACK2 line and releases the bus. At that moment the
whole cycle restarts for the next burst.
When the buffer is full, the DREQ2 line will be de-asserted and the buffer is validated
(which means that it will be sent to the host when the next IN token comes in). When
the DMA transfer is terminated, the buffer is also validated (even if it is not full). A
DMA transfer is terminated when any of the following conditions are met:
• the DMA count is complete
• DMAEN = 0
• the DMA controller asserts EOT.
11.2.2 DMA for OUT endpoint (external USB host to internal DC)
When the internal DMA handler is enabled and at least one buffer is full, the DREQ2
line is asserted. The external DMA controller then starts negotiating for control of the
bus, and as soon as it has access, it asserts the DACK2 line and starts reading the
data. The burst length is programmable. When the number of bytes equal to the burst
length has been read, the DREQ2 line is de-asserted. As a result, the DMA controller
de-asserts the DACK2 line and releases the bus. At that moment the whole cycle
restarts for the next burst. When all data are read, the DREQ2 line will be de-asserted
and the buffer is cleared (which means that it can be overwritten when a new packet
comes in).
A DMA transfer is terminated when any of the following conditions are met:
• The DMA count is complete
• DMAEN = 0
• The DMA controller asserts EOT.
When the DMA transfer is terminated, the buffer is also cleared (even if the data is not
completely read) and the DMA handler is disabled automatically. For the next DMA
transfer, the DMA controller as well as the DMA handler must be re-enabled.
11.3 Endpoint descriptions
11.3.1 Endpoints with programmable FIFO size
Each USB device is logically composed of several independent endpoints. An
endpoint acts as a terminus of a communication flow between the host and the
device. At design time each endpoint is assigned a unique number (endpoint
identifier, see Table 66). The combination of the device address (given by the host
during enumeration), the endpoint number and the transfer direction allows each
endpoint to be uniquely referenced.
ISP1161A
Full-speed USB single-chip host and device controller
The DC has 16 endpoints: endpoint 0 (control IN and OUT) plus 14 configurable
endpoints, which can be individually defined as interrupt/bulk/isochronous, IN or OUT.
Each enabled endpoint has an associated FIFO, which can be accessed either via
the Programmed I/O interface or via DMA.
11.3.2 Endpoint access
Table 66 lists the endpoint access modes and programmability. All endpoints support
I/O mode access. Endpoints 1 to 14 also support DMA access. DC FIFO DMA
access is selected and enabled via bits EPIDX[3:0] and DMAEN of the
DcDMAConfiguration register. A detailed description of the DC DMA operation is
given in Section 12.
Table 66: Endpoint access and programmability
Endpoint
identifier
064 (fixed)noyesnocontrol OUT
064 (fixed)noyesnocontrol IN
1 to 14programmablesupportedsupportedsupportedprogrammable
[1] The total amount of FIFO storage allocated to enabled endpoints must not exceed 2462 bytes.
[2] IN: input for the USB host (ISP1161A transmits); OUT: output from the USB host (ISP1161A receives). The data flow direction is
determined by bit EPDIR in the DcEndpointConfiguration register; see Section 13.1.1
FIFO size (bytes)
[1]
Double
buffering
I/O mode
access
DMA mode
access
Endpoint type
[2]
[2]
11.3.3 Endpoint FIFO size
The size of the FIFO determines the maximum packet size that the hardware can
support for a given endpoint. Only enabled endpoints are allocated space in the
shared FIFO storage, disabled endpoints have zero bytes. Table 67 lists the
programmable FIFO sizes.
The following bits in the DcEndpointConfiguration register (ECR) affect FIFO
allocation:
• Isochronous bit of an enabled endpoint (FFOISO).
Remark: Register changes that affect the allocation of the shared FIFO storage
among endpoints must not be made while valid data is present in any FIFO of the
enabled endpoints. Such changes will render all FIFO contents undefined.
Full-speed USB single-chip host and device controller
Each programmable FIFO can be configured independently via its ECR, but the total
physical size of all enabled endpoints (IN plus OUT) must not exceed 2462 bytes
(512 bytes for non-isochronous FIFOs).
Table 68 shows an example of a configuration fitting in the maximum available space
of 2462 bytes. The total number of logical bytes in the example is 1311. The physical
storage capacity used for double buffering is managed by the device hardware and is
transparent to the user.
Table 68: Memory configuration example
Physical size
(bytes)
6464control IN (64 byte fixed)
6464control OUT (64 byte fixed)
20461023double-buffered 1023-byte isochronous endpoint
161616-byte interrupt OUT
161616-byte interrupt IN
12864double-buffered 64-byte bulk OUT
12864double-buffered 64-byte bulk IN
In response to the standard USB request, Set Interface, the firmware must program
all 16 ECRs of the ISP1161A’s DC in sequence (see Table 66), whether the
endpoints are enabled or not. The hardware will then automatically allocate FIFO
storage space.
If all endpoints have been configured successfully, the firmware must return an empty
packet to the control IN endpoint to acknowledge success to the host. If there are
errors in the endpoint configuration, the firmware must stall the control IN endpoint.
When reset by hardware or via the USB bus, the ISP1161A’s DC disables all
endpoints and clears all ECRs, except for the control endpoint which is fixed and
always enabled.
Endpoint initialization can be done at any time; however, it is valid only after
enumeration.
11.3.5 Endpoint I/O mode access
When an endpoint event occurs (a packet is transmitted or received), the associated
endpoint interrupt bits (EPn) of the DcInterrupt register will be set by the SIE. The
firmware then responds to the interrupt and selects the endpoint for processing.
ISP1161A
Full-speed USB single-chip host and device controller
The endpoint interrupt bit will be cleared by reading the DcEndpointStatus register
(ESR). The ESR also contains information on the status of the endpoint buffer.
For an OUT (= receive) endpoint, the packet length and packet data can be read from
ISP1161A’s DC using the Read Buffer command. When the whole packet has been
read, the firmware sends a Clear Buffer command to enable the reception of new
packets.
For an IN (= transmit) endpoint, the packet length and data to be sent can be written
to ISP1161A’sDC using the WriteBuffercommand. When the whole packethas been
written to the buffer, the firmware sends a Validate Buffer command to enable data
transmission to the host.
11.3.6 Special actions on control endpoints
Control endpoints require special firmware actions. The arrival of a SETUP packet
flushes the IN buffer and disables the Validate Buffer and Clear Buffer commands for
the control IN and OUT endpoints. The microcontroller needs to re-enable these
commands by sending an Acknowledge Setup command.
This ensures that the last SETUP packet stays in the buffer and that no packets can
be sent back to the host until the microcontroller has explicitly acknowledged that it
has seen the SETUP packet.
The ISP1161A DC detects a USB suspend status when a constant idle state is
present on the USB bus for more than 3 ms.
The bus-powered devices that are suspended must not consume more than 500 µA
of current. This is achieved by shutting down power to system components or
supplying them with a reduced voltage.
The steps leading up to suspend status are as follows:
1. On detecting a wakeup-to-suspend transition, the ISP1161A DC sets
2. When the firmware detects a suspend condition, it must prepare all system
3. In the interrupt service routine, the firmware must check the current status of the
4. Tomeet the suspend current requirements for a bus-powered device, the internal
5. When the firmware has set and cleared bit GOSUSP in the DcMode register, the
ISP1161A
Full-speed USB single-chip host and device controller
bit SUSPND in the DcInterrupt register. This will generate an interrupt if
bit IESUSP in the DcInterruptEnable register is set.
components for the suspend state:
a. All signals connected to the ISP1161A DC must enter appropriate states to
meet the power consumption requirements of the suspend state.
b. All input pins of the ISP1161A DC must have a CMOS LOW or HIGH level.
USB bus. When bit BUSTATUS in the DcInterrupt register is logic 0, the USB bus
has left the suspend mode and the process must be aborted. Otherwise, the next
step can be executed.
clocks must be switched off by clearing bit CLKRUN in the
DcHardwareConfiguration register.
ISP1161A enters the suspend state. In powered-off application, the ISP1161A
DC asserts output SUSPEND and switches off the internal clocks after 2 ms.
Full-speed USB single-chip host and device controller
A
USB BUS
INT_N
GOSUSP
WAKEUP
SUSPEND
Fig 38. Suspend and resume timing.
In Figure 38:
> 3 ms
suspend
interrupt
> 5 ms
idle state
D
1.8 ms to 2.2 ms
C
10 ms
K-state
resume
interrupt
B
004aaa359
0.5 ms to 3.5 ms
• A: indicates the point at which the USB bus enters the idle state.
• B: indicates resume condition, which can be a 20 ms K-state on the USB bus, a
HIGH level on pin D_WAKEUP, or a LOW level on pin CS.
• C: indicates remote wake-up. The ISP1161A will drive a K-state on the USB bus
for 10 ms after pin D_WAKEUP goes HIGH or pin CS goes LOW.
• D: after detecting the suspend interrupt, set and clear bit GOSUSP in the DcMode
register.
Powered-off application: Figure 39 shows a typical bus-powered modem
application using the ISP1161A. The SUSPEND output switches off power to the
microcontroller and other external circuits during the suspend state. The ISP1161A
DC is woken up through the USB bus (global resume) or by the ring detection circuit
on the telephone line.
V
USB
BUS
DP
DM
ISP1161A
V
BUS
SUSPEND
WAKEUP
V
CC
8031
RST
RING DETECTION
LINE
004aaa674
Fig 39. SUSPEND and WAKEUP signals in a powered-off modem application.
A wake-up from the suspend state is initiated either by the USB host or by the
application:
• USB host: drives a K-state on the USB bus (global resume)
• Application: remote wake-up through a HIGH level on input WAKEUP or a LOW
The steps of a wake-up sequence are as follows:
1. The internal oscillator and the PLL multiplier are re-enabled. When stabilized, the
2. The SUSPEND output is deasserted, and bit RESUME in the DcInterrupt register
3. Maximum 15 ms after starting the wake-up sequence, the ISP1161A DC
4. In case of a remote wake-up, the ISP1161A DC drives a K-state on the USB bus
5. Following the deassertion of output SUSPEND,the application restores itself and
6. After wake-up, the internal registers of the ISP1161A DC are write-protected to
ISP1161A
Full-speed USB single-chip host and device controller
level on input CS, if enabled using bit WKUPCS in the DcHardwareConfiguration
register. Wake-up on CS will work only if V
clock signals are routed to all internal circuits of the ISP1161A.
is set. This will generate an interrupt if bit IERESM in the DcInterruptEnable
register is set.
resumes its normal functionality.
for 10 ms.
other system components to the normal operating mode.
prevent corruption by inadvertent writing during power-up of external
components. The firmware must send an Unlock Device command to the
ISP1161A DC to restore its full functionality.
is present.
BUS
11.4.3 Control bits in suspend and resume
Table 69: Summary of control bits
RegisterBitFunction
DcInterruptSUSPNDa transition from awake to the suspend state was detected
BUSTATUSmonitors USB bus status (logic 1 = suspend); used when
interrupt is serviced
RESUMEa transition from suspend to the resume state was detected
DcInterrupt
Enable
DcModeSOFTCTenables SoftConnect pull-up resistor to USB bus
DcHardware
Configuration
DcUnlockallsending data AA37H unlocks the internal registers for
IESUSPenables output INT to signal the suspend state
IERESMenables output INT to signal the resume state
GOSUSPa HIGH-to-LOW transition enables the suspend state
EXTPULselects internal (SoftConnect) or external pull-up resistor
WKUPCSenables wake-up on LOW level of input
PWROFFselects powered-off mode during the suspend state
Direct Memory Access (DMA) is a method to transfer data from one location to
another in a computer system, without intervention of the Central Processor Unit
(CPU). Many different implementations of DMA exist. The ISP1161A DC supports
two methods:
• 8237 compatible mode: based on the DMA subsystem of the IBM personal
• DACK-only mode: based on the DMA implementation in some embedded RISC
ISP1161A’s DC supports DMA transfer for all 14 configurable endpoints (see
Table 66). Only one endpoint at a time can be selected for DMA transfer. The DMA
operation of ISP1161A’s DC can be interleaved with normal I/O mode access to other
endpoints.
The following features are supported:
ISP1161A
Full-speed USB single-chip host and device controller
computers (PC, AT and all its successors and clones); this architecture uses the
Intel 8237 DMA controller and has separate address spaces for memory and I/O
processors, which has a single address space for both memory and I/O.
• Single-cycle or burst transfers (up to 16 bytes per cycle)
• Programmable signal levels on pins DREQ2 and EOT.
12.1 Selecting an endpoint for DMA transfer
The target endpoint for DMA access is selected via bits EPDIX[3:0] in the
DcDMAConfiguration register, as shown in Table 70. The transfer direction (read or
write) is automatically set by bit EPDIR in the associated ECR, to match the selected
endpoint type (OUT endpoint: read; IN endpoint: write).
Asserting input DACK2 automatically selects the endpoint specified in the
DcDMAConfiguration register, regardless of the current endpoint used for I/O mode
access.
The 8237 compatible DMA mode is selected by clearing bit DAKOLY in the
DcHardwareConfiguration register (see Table 82). The pin functions for this mode are
shown in Table 71.
Table 71: 8237 compatible mode: pin functions
SymbolDescriptionI/OFunction
DREQ2DC’s DMA requestOISP1161A’s DC requests a DMA transfer
DACK2DC’s DMA
acknowledge
EOTend of transferIDMA controller terminates the transfer
RDread strobeIinstructs ISP1161A’s DC to put data on
WRwrite strobeIinstructs ISP1161A’s DC to get data from
…continued
EPDIR = 0EPDIR = 1
IDMA controller confirms the transfer
the bus
the bus
The DMA subsystem of an IBM compatible PC is based on the Intel 8237 DMA
controller. It operates as a ‘fly-by’ DMA controller: the data is not stored in the DMA
controller, but it is transferred between an I/O port and a memory address. A typical
example of ISP1161A’s DC in 8237 compatible DMA mode is given in Figure 40.
The 8237 has two control signals for each DMA channel: DREQ (DMA Request) and
DACK (DMA Acknowledge). General control signals are HRQ (Hold Request) and
HLDA (Hold Acknowledge). The bus operation is controlled via MEMR (Memory
Read), MEMW (Memory Write), IOR (I/O read) and IOW (I/O write).
D0 to D15
ISP1161A
DEVICE
CONTROLLER
DREQ2
DACK2
RD
WR
RAM
Fig 40. ISP1161A’s device controller in the 8237 compatible DMA mode.
The following example shows the steps which occur in a typical DMA transfer:
1. ISP1161A’s DC receives a data packet in one of its endpoint FIFOs; the packet
2. ISP1161A’s DC asserts the DREQ2 signal requesting the 8237 for a DMA
3. The 8237 asks the CPU to release the bus by asserting the HRQ signal.
4. After completing the current instruction cycle, the CPU places the bus control
5. The 8237 now sets its address lines to 1234H and activates the MEMW and IOR
6. The 8237 asserts DACK to inform ISP1161A’s DC that it will start a DMA transfer.
7. ISP1161A’s DC now places the word to be transferred on the data bus lines,
8. The 8237 waits one DMA clock period and then de-asserts MEMW and IOR. This
9. ISP1161A’s DC de-asserts the DREQ2 signal to indicate to the 8237 that DMA is
10. The 8237 de-asserts the DACK output indicating that ISP1161A’s DC must stop
11. The 8237 places the bus control signals (MEMR, MEMW, IOR and IOW) and the
12. The CPU acknowledges control of the bus by de-asserting HLDA. After activating
ISP1161A
Full-speed USB single-chip host and device controller
must be transferred to memory address 1234H.
transfer.
signals (MEMR, MEMW, IOR and IOW) and the address lines in three-state and
asserts HLDA to inform the 8237 that it has control of the bus.
control signals.
because its RD signal was asserted by the 8237.
latches and stores the word at the desired memory location. It also informs
ISP1161A’s DC that the data on the bus lines has been transferred.
no longer needed. In Single cycle mode this is done after each word, in Burstmode following the last transferred word of the DMA cycle.
placing data on the bus.
address lines in three-state and de-asserts the HRQ signal, informing the CPU
that it has released the bus.
the bus control lines (MEMR, MEMW, IOR and IOW) and the address lines, the
CPU resumes the execution of instructions.
For a typical bulk transfer the above process is repeated, once for each byte. After
each byte the address register in the DMA controller is incremented and the byte
counter is decremented. When using 16-bit DMA the number of transfers is 32, and
address incrementing and byte counter decrementing is done by 2 for each word.
12.3 DACK-only mode
The DACK-only DMA mode is selected by setting bit DAKOLY in the
DcHardwareConfiguration register (see Table 82). The pin functions for this mode are
shown in Table 72. A typical example of ISP1161A’s DC in DACK-only DMA mode is
given in Figure 41.
Table 72: DACK-only mode: pin functions
SymbolDescriptionI/OFunction
DREQ2DC’s DMA requestOISP1161A DC requests a DMA transfer
DACK2DC’s DMA
Full-speed USB single-chip host and device controller
Table 72: DACK-only mode: pin functions
SymbolDescriptionI/OFunction
EOTEnd-Of-TransferIDMA controller terminates the transfer
RDread strobeInot used
WRwrite strobeInot used
…continued
In DACK-only mode ISP1161A’s DC uses the DACK2 signal as a data strobe. Input
signals RD and WR are ignored. This mode is used in CPU systems that have a
single address space for memory and I/O access. Such systems have no separate
MEMW and MEMR signals: the RD and WR signals are also used as memory data
strobes.
ISP1161A
DEVICE
CONTROLLER
DREQ2
DACK2
D0 to D15
RAM
DMA
CONTROLLER
DREQ
DACK
RD
WR
HRQ
HLDA
CPU
HRQ
HLDA
004aaa094
Fig 41. ISP1161A’s device controller in DACK-only DMA mode.
12.4 End-Of-Transfer conditions
12.4.1 Bulk endpoints
A DMA transfer to/from a bulk endpoint can be terminated by any of the following
conditions (bit names refer to the DcDMAConfiguration register, see Table 86):
• An external End-Of-Transfer signal occurs on input EOT
• The DMA transfer completes as programmed in the DcDMACounter register
(CNTREN = 1)
• A short packet is received on an enabled OUT endpoint (SHORTP = 1)
• DMA operation is disabled by clearing bit DMAEN.
External EOT: When reading from an OUT endpoint, an external EOT will stop the
DMA operation and clear any remaining data in the current FIFO. For a doublebuffered endpoint the other (inactive) buffer is not affected.
When writing to an IN endpoint, an EOT will stop the DMA operation and the data
packet in the FIFO (even if it is smaller than the maximum packet size) will be sent to
the USB host at the next IN token.
DcDMACounter register: An EOT from the DcDMACounter register is enabled by
setting bit CNTREN in the DcDMAConfiguration register. The ISP1161A has a 16-bit
DcDMACounter register,which specifies the number of bytes to be transferred. When
DMA is enabled (DMAEN = 1), the internal DMA counter is loaded with the valuefrom
the DcDMACounter register. When the internal counter completes the transfer as
programmed in the DcDMACounter, an EOT condition is generated and the DMA
operation stops.
Short packet: Normally, the transfer byte count must be set via a control endpoint
before any DMA transfertakes place. When a short packet has been enabled as EOT
indicator (SHORTP = 1), the transfer size is determined by the presence of a short
packet in the data. This mechanism permits the use of a fully autonomous data
transfer protocol.
When reading from an OUT endpoint, reception of a short packet at an OUT token
will stop the DMA operation after transferring the data bytes of this packet.
Table 73: Summary of EOT conditions for a bulk endpoint
EOT conditionOUT endpointIN endpoint
EOT inputEOT is activeEOT is active
DcDMACounter registertransfer completes as
Short packetshort packet is received and
DMAEN bit in
DcDMAConfiguration register
ISP1161A
Full-speed USB single-chip host and device controller
transfer completes as
programmed in the
DcDMACounter register
transferred
DMAEN = 0
[1]
programmed in the
DcDMACounter register
counter reaches zero in the
middle of the buffer
DMAEN = 0
[1]
[1] The DMA transfer stops. However, no interrupt is generated.
12.4.2 Isochronous endpoints
A DMA transfer to/from an isochronous endpoint can be terminated by any of the
following conditions (bit names refer to the DcDMAConfiguration register, see
Table 86):
• An external End-Of-Transfer signal occurs on input EOT
• The DMA transfer completes as programmed in the DcDMACounter register
(CNTREN = 1)
• An End-Of-Packet (EOP) signal is detected
• DMA operation is disabled by clearing bit DMAEN.
Table 74: Recommended EOT usage for isochronous endpoints
EOT conditionOUT endpointIN endpoint
EOT input activedo not usepreferred
DMA Counter register zerodo not usepreferred
End-Of-Packetpreferreddo not use
The functions and registers of ISP1161A’s DC are accessed via commands, which
consist of a command code followed by optional data bytes (read or write action). An
overview of the available commands and registers is given in Table 75.
A complete access consists of two phases:
1. Command phase: when address bit A0 = 1, the DC interprets the data on the
lower byte of the bus (bits D7 to D0) as a command code. Commands without a
data phase are executed immediately.
2. Data phase (optional): when address bit A0 = 0, the DC transfers the data on
the bus to or from a register or endpoint FIFO. Multi-byte registers are accessed
least significant byte/word first.
As the ISP1161A DC’s data bus is 16 bits wide:
• The upper byte (bits D15 to D8) in command phase, or the undefined byte in data
phase and is ignored.
• The access of registers is word-aligned: byte access is not allowed.
• If the packet length is odd, the upper byte of the last word in an IN endpoint buffer
is not transmitted to the host. When reading from an OUT endpoint buffer, the
upper byte of the last word must be ignored by the firmware. The packet length is
stored in the first 2 bytes of the endpoint buffer.
ISP1161A
Full-speed USB single-chip host and device controller
Table 75: DC command and register summary
NameDestinationCode (Hex)Transaction
Initialization commands
Write Control OUT ConfigurationDcEndpointConfiguration register
endpoint 0 OUT
Write Control IN ConfigurationDcEndpointConfiguration register
endpoint 0 IN
Write Endpoint n Configuration
(n = 1 to 14)
Read Control OUT ConfigurationDcEndpointConfiguration register
Read Control IN ConfigurationDcEndpointConfiguration register
Read Endpoint n Configuration
(n = 1 to 14)
Write/Read Device AddressDcAddress registerB6/B7write/read 1 word
Write/Read DcMode registerDcMode registerB8/B9write/read 1 word
Write/Read Hardware Configuration DcHardwareConfiguration registerBA/BBwrite/read 1 word
Write/Read DcInterruptEnable
register
Write/Read DMA ConfigurationDcDMAConfiguration registerF0/F1write/read 1 word
Write/Read DMA CounterDcDMACounter registerF2/F3write/read 1 word
Reset Deviceresets all registersF6-
Full-speed USB single-chip host and device controller
Table 75: DC command and register summary
…continued
NameDestinationCode (Hex)Transaction
Data flow commands
Write Control OUT Bufferillegal: endpoint is read-only(00)Write Control IN BufferFIFO endpoint 0 IN01N ≤ 64 bytes
Write Endpoint n Buffer
(n = 1 to 14)
FIFO endpoint 1 to 14
(IN endpoints only)
02 to 0Fisochronous: N ≤ 1023 bytes
interrupt/bulk: N ≤ 64 bytes
Read Control OUT BufferFIFO endpoint 0 OUT10N ≤ 64 bytes
Read Control IN Bufferillegal: endpoint is write-only(11)Read Endpoint n Buffer
(n = 1 to 14)
FIFO endpoint 1 to 14
(OUTendpoints only)
12 to 1Fisochronous:
N ≤ 1023 bytes
interrupt/bulk: N ≤ 64 bytes
Stall Control OUT EndpointEndpoint 0 OUT40Stall Control IN EndpointEndpoint 0 IN41Stall Endpoint n
Endpoint 1 to 1442 to 4F-
(n = 1 to 14)
Read Control OUT StatusDcEndpointStatus register
50read 1 word
endpoint 0 OUT
Read Control IN StatusDcEndpointStatus register
51read 1 word
endpoint 0 IN
Read Endpoint n Status
(n = 1 to 14)
Validate Control OUT Bufferillegal: IN endpoints only
Validate Control IN BufferFIFO endpoint 0 IN
Validate Endpoint n Buffer
(n = 1 to 14)
DcEndpointStatus register n
endpoint 1 to 14
[2]
[2]
FIFO endpoint 1 to 14
(IN endpoints only)
[2]
52 to 5Fread 1 word
(60)61none
62 to 6Fnone
Clear Control OUT BufferFIFO endpoint 0 OUT70none
Clear Control IN Bufferillegal
Clear Endpoint n Buffer
(n = 1 to 14)
[3]
FIFO endpoint 1 to 14
(OUT endpoints only)
(71)72 to 7Fnone
[3]
Unstall Control OUT EndpointEndpoint 0 OUT80Unstall Control IN EndpointEndpoint 0 IN81Unstall Endpoint n
Full-speed USB single-chip host and device controller
Table 75: DC command and register summary
NameDestinationCode (Hex)Transaction
Read Endpoint n Error Code
(n = 1 to 14)
Unlock Deviceall registers with write accessB0write 1 word
Write/Read DcScratch registerDcScratch registerB2/B3write/read 1 word
Read Frame NumberDcFrameNumber registerB4read 1 word
Read Chip IDDcChipID registerB5read 1 word
Read DcInterrupt registerDcInterrupt registerC0read 2 words
[1] With N representing the number of bytes, the number of words for 16-bit bus width is: (N + 1) DIV 2.
[2] Validating an OUT endpoint buffer causes unpredictable behavior of ISP1161A’s DC.
[3] Clearing an IN endpoint buffer causes unpredictable behavior of ISP1161A’s DC.
[4] Reads a copy of the DcStatus register: executing this command does not clear any status bits or interrupt bits.
[5] When accessing an 8-bit register in 16-bit mode, the upper byte is invalid.
[6] During isochronous transfer in 16-bit mode, because N ≤ 1023, the firmware must take care of the upper byte.
DcErrorCode register
endpoint 1 to 14
…continued
A2 to AFread 1 word
[1]
[5]
13.1 Initialization commands
Initialization commands are used during the enumeration process of the USB
network. These commands are used to configure and enable the embedded
endpoints. They also serve to set the USB assigned address of ISP1161A’s DC and
to perform a device reset.
This command is used to access the DcEndpointConfiguration register (ECR) of the
target endpoint. It defines the endpoint type (isochronous or bulk/interrupt), direction
(OUT/IN), FIFO size and buffering scheme. It also enables the endpoint FIFO. The
register bit allocation is shown in Table 76. A bus reset will disable all endpoints.
The allocation of FIFO memory only takes place after all 16 endpoints have been
configured in sequence (from endpoint 0 OUT to endpoint 14). Although the control
endpoints have fixed configurations, they must be included in the initialization
sequence and be configured with their default values (see Table 66). Automatic FIFO
allocation starts when endpoint 14 has been configured.
Remark: If any change is made to an endpoint configuration which affects the
allocated memory (size, enable/disable), the FIFO memory contents of all endpoints
becomes invalid. Therefore, all valid data must be removed from enabled endpoints
before changing the configuration.
Code (Hex): 20 to 2F — write (control OUT, control IN, endpoint 1 to 14)
Code (Hex): 30 to 3F — read (control OUT, control IN, endpoint 1 to 14)
Transaction — write/read 1 word
Table 76: DcEndpointConfiguration register: bit allocation
Table 77: DcEndpointConfiguration register: bit description
BitSymbolDescription
7FIFOENLogic 1 indicates an enabled FIFO with allocated memory.
6EPDIRThis bit defines the endpoint direction (0 = OUT, 1 = IN); it also
5DBLBUFLogic 1 indicates that this endpoint has double buffering.
4FFOISOLogic 1 indicates an isochronous endpoint. Logic 0 indicates a
3 to 0FFOSZ[3:0]Selects the FIFO size according to Table 67
13.1.2 DcAddress register (R/W: B7H/B6H)
This command is used to set the USB assigned address in the DcAddress register
and enable the USB device. The DcAddress register bit allocation is shown in
Table 78.
A USB bus reset sets the device address to 00H (internally) and enables the device.
The value of the DcAddress register (accessible by the microcontroller) is not altered
by the bus reset. In response to the standard USB request, Set Address, the firmware
must issue a Write Device Address command, followed by sending an empty packet
to the host. The new device address is activated when the host acknowledges the
empty packet.
ISP1161A
Full-speed USB single-chip host and device controller
Logic 0 indicates a disabled FIFO (no bytes allocated).
determines the DMA transfer direction (0 = read, 1 = write).
7DEVENLogic 1 enables the device.
6 to 0DEVADR[6:0] This field specifies the USB device address.
13.1.3 DcMode register (R/W: B9H/B8H)
This command is used to access the ISP1161A’s DcMode register, which consists of
1 byte (for bit allocation: see Table 79). In 16-bit bus mode the upper byte is ignored.
The DcMode register controls the DMA bus width, resume and suspend modes,
interrupt activity and SoftConnect operation. It can be used to enable debug mode,
where all errors and Not Acknowledge (NAK) conditions will generate an interrupt.
7DMAWDLogic 1 selects 16-bit DMA bus width (bus configuration
modes 0 and 2). Logic 0 selects 8-bit DMA bus width. Bus reset
value: unchanged.
6-reserved
5GOSUSPWriting logic 1 followed by logic 0 will activate ‘suspend’ mode.
4-reserved
3INTENALogic 1 enables all DC interrupts. Bus reset value: unchanged;
for details, see Section 8.6.3.
2DBGMODLogic 1 enables debug mode, where all NAKs and errors will
generate an interrupt. Logic 0 selects normal operation, where
interrupts are generated on every ACK (bulk endpoints) or after
every data transfer (isochronous endpoints).
Bus reset value: unchanged.
1-reserved
0SOFTCTLogic 1 enables SoftConnect (see Section 7.5). This bit is
ignored if EXTPUL = 1 in the DcHardwareConfiguration register
This command is used to access the DcHardwareConfiguration register, which
consists of 2 bytes. The first (lower) byte contains the device configuration and
control values, the second (upper) byte holds the clock control bits and the clock
division factor. The bit allocation is given in Table 82. A bus reset will not change any
of the programmed bit values.
The DcHardwareConfiguration register controls the connection to the USB bus, clock
activity and power supply during ‘suspend’ state, output clock frequency, DMA
operating mode and pin configurations (polarity, signalling mode).
This command is used to individually enable or disable interrupts from all endpoints,
as well as interrupts caused by events on the USB bus (SOF, SOF lost, EOT,
suspend, resume, reset). That is, if an interrupt event occurs while the interrupt is not
enabled, nothing will be seen on the interrupt pin. Even if you then enable the
interrupt during the interrupt event, there will still be no interrupt seen on the interrupt
pin, see Figure 42.
The DcInterrupt register will not register any interrupt, if it is not already enabled
using the DcInterruptEnable register. The DcInterruptEnable register is not an
Interrupt Mask register.
ISP1161A
Full-speed USB single-chip host and device controller
DcInterruptEnable
register
disabled
INT2 pin
interrupt
event
occurs
Pin INT2: HIGH = de-assert; LOW= assert; INTENA = 1.
DcInterruptEnable
register
enabled
interrupt is cleared
interrupt
event
occurs
004aaa197
Fig 42. Interrupt pin waveform.
A bus reset will not change any of the programmed bit values.
The command accesses the DcInterruptEnable register, which consists of 4 bytes.
The bit allocation is given in Table 84.
Remark: For details on interrupt control, see Section 8.6.3.
Code (Hex): C2/C3 — write/read DcInterruptEnable register
Transaction — write/read 2 words
Table 84: DcInterruptEnable register: bit allocation
Table 85: DcInterruptEnable register: bit description
BitSymbolDescription
31 to 24-reserved; must write logic 0
23 to 10IEP14 to IEP1 Logic 1 enables interrupts from the indicated endpoint.
9IEP0INLogic 1 enables interrupts from the control IN endpoint.
8IEP0OUTLogic 1 enables interrupts from the control OUT endpoint.
7-reserved
6SP_IEEOTLogic 1 enables interrupt upon detection of a short packet.
5IEPSOFLogic 1 enables 1 ms interrupts upon detection of Pseudo SOF.
4IESOFLogic 1 enables interrupt upon SOF detection.
3IEEOTLogic 1 enables interrupt upon EOT detection.
2IESUSPLogic 1 enables interrupt upon detection of ‘suspend’ state.
1IERESMLogic 1 enables interrupt upon detection of a ‘resume’ state.
0IERSTLogic 1 enables interrupt upon detection of a bus reset.
13.1.6 DcDMAConfiguration register (R/W: F1H/F0H)
This command defines the DMA configuration of ISP1161A’s DC and
enables/disables DMA transfers. The command accesses the DcDMAConfiguration
register, which consists of 2 bytes. The bit allocation is given in Table 86. A bus reset
will clear bit DMAEN (DMA disabled), all other bits remain unchanged.
Table 87: DcDMAConfiguration register: bit description
BitSymbolDescription
15CNTRENLogic 1 enables the generation of an EOT condition, when the
14SHORTPLogic 1 enables short/empty packet mode. When receiving
13 to 8-reserved
7 to 4EPDIX[3:0]Indicates the destination endpoint for DMA, see Table 70.
3DMAENWriting logic 1 enables DMA transfer, logic 0 forces the end of
2-reserved
1 to 0BURSTL[1:0]Selects the DMA burst length:
ISP1161A
Full-speed USB single-chip host and device controller
DcDMACounter register reaches zero. Bus reset value:
unchanged.
(OUTendpoint) a short/empty packet an EOT condition is
generated. When transmitting (IN endpoint), this bit should be
cleared. Bus reset value: unchanged.
an ongoing DMA transfer. Reading this bit indicates whether
DMA is enabled (0 = DMA stopped, 1 = DMA enabled). This bit
is cleared by a bus reset.
00 — single-cycle mode (1 byte)
01 — burst mode (4 bytes)
10 — burst mode (8 bytes)
11 — burst mode (16 bytes).
Bus reset value: unchanged.
For selecting an endpoint for device DMA transfer, see Section 11.2.
13.1.7 DcDMACounter register (R/W: F3H/F2H)
This command accesses the DcDMACounter register. The bit allocation is given in
Table 88. Writing to the register sets the number of bytes for a DMA transfer.Reading
the register returns the number of remaining bytes in the current transfer.A bus reset
will not change the programmed bit values.
The internal DMA counter is automatically reloaded from the DcDMACounter register
when DMA is re-enabled (DMAEN = 1). See Section 13.1.6 for more details.
This command resets the ISP1161A DC in the same way as an external hardware
reset via input RESET. All registers are initialized to their ‘reset’ values.
Code (Hex): F6 — reset the device
Transaction — none
13.2 Data flow commands
Data flow commands are used to manage the data transmission between the USB
endpoints and the system microprocessor. Much of the data flow is initiated via an
interrupt to the microprocessor. The data flow commands are used to access the
endpoints and determine whether the endpoint FIFOs contain valid data.
Remark: The IN bufferof an endpoint contains input data for the host, the OUT buffer
receives output data from the host.
ISP1161A
Full-speed USB single-chip host and device controller
This command is used to access endpoint FIFO buffers for reading or writing. First,
the buffer pointer is reset to the beginning of the buffer. Following the command, a
maximum of (M + 1) words can be written or read, with M given by (N + 1) DIV 2, N
representing the size of the endpoint buffer. After each read/write action the buffer
pointer is automatically incremented by 2.
In DMA access, the first word (the packet length) is skipped: transfers start at the
second word of the endpoint buffer. When reading, the ISP1161A DC can detect the
last word via the End of Packet (EOP) condition. When writing to a bulk/interrupt
endpoint, the endpoint buffer must be completely filled before sending the data to the
host. Exception: when a DMA transfer is stopped by an external EOT condition, the
current buffer content (full or not) is sent to the host.
Remark: Reading data after a Write Endpoint Buffer command or writing data after a
Read Endpoint Buffer command will cause unpredictable behavior of the ISP1161A
DC.
Code (Hex): 01 to 0F — write (control IN, endpoint 1 to 14)
Code (Hex): 10, 12 to 1F — read (control OUT, endpoint 1 to 14)
Transaction — write/read maximum (M + 1) words (isochronous endpoint: N ≤ 1023,
bulk/interrupt endpoint: N ≤ 32)
The data in the endpoint FIFO must be organized as shown in Table 90. An example
0dataD[15:0]0packet length
0dataD[15:0]1data word 1 (data byte 2, data byte 1)
0dataD[15:0]2data word 2 (data byte 4, data byte 3)
……………
ISP1161A
Full-speed USB single-chip host and device controller
D[15:8]-ignored
Remark: There is no protection against writing or reading past a buffer’sboundary or
against writing into an OUT buffer or reading from an IN buffer. Any of these actions
could cause an incorrect operation. Data residing in an OUT buffer are only
meaningful after a successful transaction. Exception: during DMA access of a
double-buffered endpoint, the buffer pointer automatically points to the secondary
buffer after reaching the end of the primary buffer.
13.2.2 DcEndpointStatus register (R: 50H–5FH)
This command is used to read the status of an endpoint FIFO. The command
accesses the DcEndpointStatus register, the bit allocation of which is shown in
Table 92. Reading the DcEndpointStatus register will clear the interrupt bit set for the
corresponding endpoint in the DcInterrupt register (see Table 108).
All bits of the DcEndpointStatus register are read-only. Bit EPSTAL is controlled by
the Stall/Unstall commands and by the reception of a SETUP token (see
Section 13.2.3).
Code (Hex): 50 to 5F — read (control OUT, control IN, endpoint 1 to 14)
Transaction — read 1 word
Table 92: DcEndpointStatus register: bit allocation