Publication Number 23569 Revision A Amendment +4 Issue Date August 7, 2002
Page 2
ADVANCE INFORMATION
Am29PDS322D
32 Megabit (2 M x 16-Bit) CMOS 1.8 Volt-only (1.8 V to 2.2 V)
Simultaneous Read/Write Page-Mode Boot Sector Flash Memory
DISTINCTIVE CHARACTERISTICS
ARCHITECTURAL ADVANTAGES
■ Simultaneous Read/Write operations
— Data can be continuously read from one bank while
executing erase/program functions in other bank.
— Zero latency between read and write operations
■ Page Mode Operation
— 4 word page allows fast asynchronous reads
■ Dual Bank architecture
— One 4 Mbit bank and one 28 Mbit bank
■ SecSi (Secured Silicon) Sector: Extra 64 KByte
sector
— Factory locked and identifiab le: 16 byte Electronic
Serial Number available for factory secure, random
ID; verifiable as factory locked through autoselect
function. ExpressFlash option allows entire sector to
be available for factory-secured data
— Customer lockable: Can be read, programmed, or
erased just like other sectors. Once locked, data
cannot be changed
■ Zero Power Operation
— Sophisticated power management circuits reduce
power consumed during inactive periods to nearly
zero.
■ Package options
— 48-ball FBGA
■ Top or bottom boot block
■ Manufactured on 0.23 µm process technology
■ Compatible with JEDEC standards
— Pinout and software compatible with
single-power-supply flash standard
PERFORMANCE CHA RA CT ER IST ICS
■ High performance
— Access time as fast 40 ns (100 ns random access
time) at 1.8 V to 2.2 V V
— Random access time of 100 ns at 1.8 V to 2.2 V VCC
will be required as customers migrate downward in
voltage
■ Ultra low power consumption (typical values)
— 2.5 mA active read current at 1 MHz for initial page
read
— 24 mA active read current at 10 MHz for initial page
read
— 0.5 mA active read current at 10 MHz for intra-page
read
CC
— 1 mA active read current at 20 MHz for intra-page
read
— 200 nA in standby or automatic sleep mode
■ Minimum 1 million write cycles guaranteed per
sector
■ 20 year data retention at 125°C
— Reliable operation for the life of the system
SOFTWARE FEATURES
■ Data Management Software (DMS)
— AMD-supplied software manages data programming,
enabling EEPROM emulation
— Eases historical sector erase flash limitations
■ Erase Suspend/Erase Resume
— Suspends erase operations to allow programming in
same bank
■ Data# Polling and Toggle Bits
— Provides a software method of detecting the status of
program or erase cycles
■ Unlock Bypass Program command
— Reduces overall programming time when issuing
multiple program command sequences
HARDWARE FEATURES
■ Any combination of sectors can be erased
■ Ready/Busy# output (RY/BY#)
— Hardware method for detecting program or erase
cycle completion
■ Hardware reset pin (RESET#)
— Hardware method of resetting the internal state
machine to the read mode
■ WP#/ACC input pin
— Write protect (WP#) function allows protection of two
outermost boot sectors, regardless of sector protect
status
— Acceleration (ACC) function accelerates program
timing
— ACC voltage is 8.5 V to 12.5 V
■ Sector protection
— Hardware method of locking a sector, either
in-system or using programming equipment, to
prevent any program or erase operation within that
sector
— Temporary Sector Unprotect allows changing data in
protected sectors in-system
This document contains information on a product under development at Advanced Micro Devices. The information
is intended to help you evaluate this product. AMD reserves the right to change or discontinue work on this proposed
product without notice.
Refer to AMD’s Website (www.amd.com) for the latest information.
Publication# 23569 Rev: A Amendment/+4
Issue Date: August 7, 2002
Page 3
ADVANCE INFORMATION
GENERAL DESCRIPTION
The Am29PDS322D is a 32 Mbit, 1.8 V-only Flash
memory organized as 2,097,152 words of 16 bits
each. This device is offered in a 48-ball FBGA package. The device is designed to be programmed in system with standard system 1.8 V V
supply. This
CC
device can also be reprogrammed in standard
EPROM programmers.
The Am29PDS3 22D offers fast pag e access ti me of
40 ns with ra ndom a ccess t ime of 10 0 ns ( at 1.8 V to
2.2 V V
), allowing operation of high-speed micropro-
CC
cessors without wait states. To eliminate bus contention the device has sepa rate chip enable (C E), write
enable (WE), and output enable (OE) controls. The
page size is 4 words.
The device requires only a single 1.8 volt power sup-ply for both read and write functions. Internally
generated and regulated voltages are provided for the
program and erase operations.
Simultaneous Read/Write Operations with
Zero Latency
The Simultaneous Read/Write architecture provides
simultaneous operation by dividing the me mory
space into two banks. The device can improve overall
system performance by allowing a host system to program or erase in one bank, then immediately and simultaneously read from the other bank, with zero
latency. This releases the system from waiting for the
completion of program or erase operations.
The device is divided as shown in the following table:
Bank 1 SectorsBank 2 Sectors
Quantity SizeQuantitySize
84 Kwords
732 Kwords
4 Mbits total28 Mbits total
5632 Kwords
Am29PDS322D Features
The SecSi (Se cured Silicon) S ector is an extra 64
KByte sector capable of being perm anently lo cked by
AMD or customers. The SecSi Indicator Bit (DQ7) is
permanently set to a 1 if the part is factory locked,
and set to a 0 if customer lockable . This way, customer lockable parts can never be used to replace a
factory locked part.
Factory locked parts pro vide several options. The
SecSi Sector may store a secure, random 16 byte
ESN (Electronic Serial Numb er), cust omer co de (programmed through AMD’s ExpressFlash service), or
both. Customer Lockable parts may utilize the SecSi
Sector as bonus space, reading and writing like any
other flash sector, or may permanen tly lock their ow n
code there.
DMS (Data Management Software) allows systems
to easily take advantage of the advanced architecture
of the simultaneous read/write product line by allowing
removal of EEPROM devices. DMS will also allow the
system software to be simplified, as it will perform all
functions necessary to modify data in file structures,
as opposed to single-byte modifications. To write or
update a particular piece of data (a phone number or
configuration data, for example), the user only needs
to state which piece of data is to be updated, and
where the updated data is located in the system. This
is an advantage compared to systems where
user-written software must keep track of the old data
location, status, logical to physical translation o f the
data onto the Flash memory device (or memory devices), and more. Using DMS, user-written software
does not need to interface with the Flash memory directly. Instead, the user's software accesses the Flash
memory by calling one of only six functions. AMD provides this software to simplify system design and software integration efforts.
The device offers complete compatibility with the
JEDEC single-power-supply Flash command set
standard. Commands are written to the command
register using standard microprocessor write timings.
Reading data out of the device is similar to reading
from other Flash or EPROM devices.
The host system can detect whether a program or
erase operation is complete by using the device sta-tus bits: RY/BY# pin, DQ7 (Data# Polling) and
DQ6/DQ2 (toggle bits). After a program or erase cycle
has been completed, the device automatically re turns
to the read mode.
The sector erase archite cture allow s memo ry sectors to be erased and reprogrammed without affecting
the data conten ts of oth er sec tors. Th e devi ce is fully
erased when shipped from the factory.
Hardware data protection measures include a low
detector that automatically inhibits write opera-
V
CC
tions during power transitions. The hardware sectorprotection feature disables both program and erase
operations in any com bination of the secto rs of memory. This can be achieved in-system or via programming equipment.
The device offers two power-saving features. When
addresses have been stable for a specified amount of
time, the device enters the automatic sleep mode.
The system can also place the device into the
standby mode. Power consumption is greatly r educed in both modes.
Max Random Address Acce ss Time (ns)100120
Max Page Address Access Time (ns)4045
CE# Access Time (ns)100120
OE# Access Time (ns)3540
Note: See “AC Characteristics” for full specifications.
BLOCK DIAGRAM
V
V
CC
SS
A0–A20
A0–A20
RESET#
WE#
CE#
WP#/ACC
Mux
RY/BY#
A0–A20A0–A20
STATE
CONTROL
&
COMMAND
REGISTER
Upper Bank Address
Upper Bank
Y-Decoder
X-Decoder
Status
Control
OE#
Latches and Control Logic
DQ0–DQ15
Mux
DQ0–DQ15
A0–A20
Lower Bank Address
Mux
X-Decoder
Lower Bank
Y-Decoder
Latches and
Control Logic
DQ0–DQ15DQ0–DQ15
4Am29PDS322DAugust 7, 2002
Page 6
CONNECTION DIAGRAMS
A6B6C6D6E6F6G6H6
A5B5C5D5E5F5G5H5
A4B4C4D4E4F4G4H4
A3B3C3D3E3F3G3H3
ADVANCE INFORMATION
48-Ball FBGA
Top View, Balls Facing Down
DQ15
DQ13DQ6DQ14DQ7A11A10A8A9
V
CCDQ4DQ12DQ5A19NCRESET#WE#
DQ11DQ3DQ10DQ2A20A18WP#/ACCRY/BY#
V
SSNCA16A15A14A12A13
A2B2C2D2E2F2G2H2
A1B1C1D1E1F1G1H1
Special Handling Instructions for FBGA
Package
Special handling is required for Flash Memory products
in FBGA packages.
DQ9DQ1DQ8DQ0A5A6A17A7
OE#
V
SSCE#A0A1A2A4A3
Flash memory devices in FBGA packages may be
damaged if exposed to ultrasonic cleaning methods.
The package and/or data integrity may be compromised
if the package body is exposed to temperatures above
150°C for prolonged periods of time.
AMD standard products are available in several packages and operating ranges. The order number (Valid Combination) is
formed by a combination of the following:
Am29PDS322DB10WMIN
OPTIONAL PROCESSING
Blank =Standard Processing
N=16-byte ESN devices
(Contact an AMD representative fo r more information)
TEMPERATURE RANGE
I = Industrial (–40
PACKAGE TYPE
WM=48-Ball Fine-Pitch Ball Grid Array (FBGA)
0.80 mm pitch, 6 x 12 mm package (FBD048)
SPEED OPTION
See Product Selector Guide and Valid Combinations
BOOT CODE SE CTOR ARCHITECTURE
T=Top sector
B=Bottom sector
°C to +85°C)
Valid Combinations for FBGA Package
Order NumberPackage Marking
Am29PDS322DT10,
Am29PDS322DB10
Am29PDS322DT12,
Am29PDS322DB12
DEVICE NUMBER/DESCRIPTION
Am29PDS322D
32 Megabit (2 M x 16-Bit) CMOS Boot Sector Page Mode Flash Memory
1.8 Volt-only Read, Program, and Erase
Valid Combinations
Valid Combinations list configurations planned to be supported in volume for this device. Consult the local AMD sales
WMI
WMI
P322DT10U,
P322DB10U
P322DT12U,
P322DB12U
I
I
office to confirm availability of specific valid combinations and
to check on newly released combinations.
August 7, 2002Am29PDS322D7
Page 9
ADVANCE INFORMATION
DEVICE BUS OPERATIONS
This section describes the requirements and use of
the device bus operations, which are in itiated through
the internal command register. The command register
itself does not occupy any addressabl e memory l ocation. The register is a latch used to store the commands, along with the ad dress and da ta information
needed to execute the command. The contents of the
Table 1. Am29PDS322D Device Bus Operations
OperationCE#OE#WE#RESET#WP#/ACC
register serve as inputs to the intern al state machine.
The state machine outputs dictate the function of the
device. Table 1 lists the device bus operations, the inputs and control l evels they requir e, and the resultin g
output. The following subsections de scribe each of
these operations in further detail.
Legend: L = Logic Low = VIL, H = Logic High = VIH, VID = 9.0–11.0 V, VHH = 9.0 ± 0.5 V, X = Don’t Care, SA = Sector Address,
= Address In, DIN = Data In, D
A
IN
Notes:
1. The sect or protect and sector unprotect funct ions may also be implemented via programming equipment . See the “Sector/Sect or
Block Protection and Unprotection” section.
2. If WP#/ACC = V
protection depends on whether they were last protected or unprotected using the method described in “Sector/S ec t or Bl o ck
Protection and Unprotection”. If WP#/ACC = V
, the two outermost boot sectors remain protected. If WP#/ACC = VIH, the two outermost boot sector
IL
= Data Out
OUT
all sectors will be unprotected.
HH,
Requirements for Reading Array Data
To read array data from the outputs, the system must
drive the CE # and OE # pins to V
control and selects the device. OE# is the output control and gates array data to the output pins. WE#
should remain at V
.
IH
The internal state machine is set for reading array data
upon device power-up, or after a hardware reset. This
ensures that no spurious alteration of the memory
content occurs durin g the power transition. No command is necessary in this mode to obtain array data.
Standard microprocessor read cycles that assert valid
addresses on the device address inputs produce valid
data on the device data outputs. The device remains
enabled for read access until the command register
contents are altered.
See “Requirements for Reading Array Data” for more
information. Refer to the AC Read-Only Operations
table for timing specifications and to F igure 15 for the
. CE# is the power
IL
ID
ID
ID
timing diagram. I
represents the activ e curren t specific ation for r eading
array data.
Read Mode
Random Read (Non-Page Mode Read)
The device has two c ontrol functions which mu st be
satisfied in order to obtain data at the outputs. CE# is
the power control and should be used for device selection. OE# is the output control and should be used to
gate data to the output pins if the device is selected.
Address access t ime (t
stable addresses to valid output data. The chip enable
access time (t
dresses and stable CE# to valid da ta at the output
pins. The output enable access time is the delay from
the falling edge of OE# to valid data at the output pins
(assuming the addresses have been stable for at least
t
ACC–tOE
L/H
(Note 2)
(Note 2)A
CC1
) is the delay from the stable ad-
CE
time).
in the DC Characteristics table
IN
IN
SA, A6 = L,
A1 = H, A0 = L
SA, A6 = H,
A1 = H, A0 = L
IN
) is equal to the delay from
ACC
D
OUT
D
IN
D
IN
D
IN
D
IN
8Am29PDS322DAugust 7, 2002
Page 10
ADVANCE INFORMATION
Page Mode Read
The device is capable of fast Page mode read and is
compatible with the Page mode Mask ROM read operation. This mode provides fast er read access speed
for random locations within a page. The Page size of
the device is 4 words. T he appropriate Page is selected by the higher address bits A20–A2 and the LSB
bits A1–A0 determine the specific word within that
page. This is an asynchronous operation with the microprocessor supplying the specific word location.
The random or initial page access is equal to t
and subsequent Page read accesses (as long as
t
CE
the locations specified by the microprocessor falls
within that Page) are equivalent to t
. When CE# is
PACC
deasserted and reasserted for a subsequent access,
the access time is t
or tCE. Here again, CE# selects
ACC
the device and OE# is the output c ontrol and s hould be
used to gate data to the output pins if the device is selected. Fast Pag e mode accesses ar e obtained by
keeping A2–A20 co nstant and changing A0 to A1 to
select the specific word within that page. See Figure
16 for timing specifications.
The following table determines the specific word within
the selected page:
ACC
or
Accelerated Program Operation
The device offers accelerated program operations
through the ACC function. This is one of two functions
provided by the WP#/ACC pin. This function is primarily intended to allow faster manu facturing throu ghput
at the factory.
If the system asserts V
on this pin, the device auto-
HH
matically enters the aforementioned Unlock Bypass
mode, temporarily unprotects any protec ted sectors,
and uses the higher voltage on the pin to reduce the
time required for program operations. The system
would use a two-cycle progra m command sequence
as required by the Unlock Bypass mode. Removing
from the ACC pin returns the device to normal op-
V
HH
eration.
Autoselect Functions
If the system writes the autoselect command s equence, the device enters the autoselect mo de. The
system can then read autoselect codes from the internal register (which is separate from the memory array)
on DQ7–DQ0. Standard read cycle timings apply in
this mode. Refer to the Autoselect M ode and Au toselect Command Sequence sections for more inform ation.
Table 2. Page Word Mode
Word A1A0
Word 000
Word 101
Word 210
Word 311
Writing Commands/Command Sequences
To write a command or command sequence (which includes programming data to the device and erasing
sectors of memory), the system must drive WE# and
CE# to V
The device features an Unlock Bypass mode to facilitate faster programming. Once the device enters the
Unlock Bypass mo de, only two write cycles are required to program a word, instead of four. The “Word
Program Command Sequence” section has details on
programming data to the device using both standard
and Unlock Bypass command sequences.
An erase operation can erase one sector, multiple sectors, or the entire device. Table 2 indica tes the address
space that each sector occupies.
I
CC2
tive current specification for the write mode. The AC
Characteristics section contains timing specification
tables and timing diagrams for write operations.
, and OE# to VIH.
IL
in the DC Characteristics table represents the ac-
Simultaneous Read/Write Operations with
Zero Latency
This device is c apable of r eading da ta from on e bank
of memory while programming or erasing in the other
bank of memory. An erase operation may also be suspended to read from or program to another location
within the same bank (except the sector being
erased). Figure 21 shows how read and w rite cycles
may be initiated for simultaneous operation with zero
latency. I
CC6
and I
in the DC Characteristics table
CC7
represent the current specifications for read-while-program and read-while-erase, respectively.
Standby Mode
When the system is n ot reading or wri ting to the device, it can place the device in the standby mode. In
this mode, current consum ption is greatly reduc ed,
and the outputs are placed in the high impedance
state, independent of the OE# input.
The device enters the CMOS standby mode when the
CE# and RESET# pins are both held at V
(Note that this is a more restricted voltage range tha n
.) If CE# and RESET# are held at VIH, but not within
V
IH
± 0.3 V, the device will be in the s tandby mod e,
V
CC
but the standby current will be greater. The device requires standard ac cess time (t
) for read access
CE
when the device is in either of these standby modes,
before it is ready to read data.
± 0.3 V.
CC
August 7, 2002Am29PDS322D9
Page 11
ADVANCE INFORMATION
If the device is deselected during erasure or programming, the device draws active current until the
operation is completed.
in the DC Characte ristics table rep resents the
I
CC3
standby current specification.
Automatic Sleep Mode
The automatic sleep mode minimizes Flash device energy consumption. The device automatically enables
this mode when addresses remain stable for t
ACC
+
30 ns. The automatic sleep mode is independent of
the CE#, WE#, and OE# control signals. Standard address access timings provide new data when addresses are changed. While in sleep mode, output
data is latched and always available to the system.
Automatic sleep mode current is drawn w hen CE# =
± 0.3 V and all inputs are held at VCC ± 0.3 V. If
V
SS
CE# and RESET# voltages are not held within these
tolerances, the automatic sleep mode current will be
greater.
in the DC Char acteristics table r epresents the
I
CC5
automatic sleep mode current specification.
RESET#: Hardware Reset Pin
The RESET# pin provides a hardware me thod of resetting the device to reading array data. When the RESET# pin is driven low for at least a period of t
device immediately term inates any operation in
progress, tristates all output pins, and ignores all
read/write command s for the duration o f the RESE T#
pulse. The device also resets the internal state ma-
RP
, the
chine to reading array data. The operation that was interrupted should be reinitiated once the device is
ready to accept another command sequence, to ensure data integrity.
Current is reduced for the duration of the RESET#
pulse. When RESET# is held at V
vice draws CMOS standby current (I
held at V
but not within V
IL
± 0.3 V, the standby cur-
SS
± 0.3 V, the de-
SS
). If RESET# is
CC3
rent will be greater.
The RESET# pin may be tied to the system reset cir-
cuitry. A system reset would thus also reset the Flash
memory, enabling the system to read the boot-up firmware from the Flash memory.
If RESET# is a sserted during a prog ram or erase operation, the RY/BY# pin remains a “0” (busy) until the
internal reset operation is complete, which requires a
time of t
(during Embedded Algorithms). The
READY
system can thus monitor RY/BY# to determine
whether the reset operation is complete. If RESET# is
asserted when a program or erase operation is not executing (RY/BY# pin is “1”), the reset operation is completed within a time of t
Algorithms). The system can read data t
RESET# pin returns to V
(not during Embedded
READY
.
IH
after the
RH
Refer to the AC Characteristics tables for RESET# parameters and to Figure 17 for the timing diagram.
Output Disable Mode
When the OE# input is at VIH, output from the device is
disabled. The output pins are placed in the high
impedance state.
The autoselect mode provides manufacturer and device identification, and sector protection verification,
through identifier codes output on DQ15–DQ0. This
mode is prim arily intend ed for progr amming equi pment to automatically match a device to be programmed with its corresponding programming
algorithm. However, the autoselect codes can also be
accessed in-system through the command register .
When using programming equipment, the autoselect
mode requires V
Address pins A6, A1, and A0 must be as shown in
DescriptionCE# OE# WE#
Manufacturer ID:
AMD
Device ID Word 1LLHXXV
Device ID Word 2LLHXXV
Device ID Word 3:
Top or Bottom Boot
Sector Protection
Verification
SecSi Indicator Bit
(DQ7),
WP# protects
highest address
sector
(8.5 V to 12.5 V) on address pin A9.
ID
Table 7. Autoselect Codes (High Voltage Method)
A20
A11
to
to
A12
A10A9
LLH X XV
LLH X XV
LLHSAXV
LLH X XV
Table 7. In addition, when verifying sector protection,
the sector address must appear on the appropriate
highest order address bits (see Tables 3 through 6).
Table 7 shows the remaining address bits that are
don’t care. When all necessary bits have been set as
required, the programming equipment may then read
the corresponding identifier code on DQ15–DQ0.
To access the autoselect codes in-system, the host
system can issue the autoselect command via the
command register, as shown in Table 10. This method
does not require V
. Refer to the Autoselect Com-
ID
mand Sequence section for more information.
A8
to
A7 A6
XLXXXLL0001h
ID
XLXLLLH227Eh
ID
XLXHHHL2206h
ID
XLXHHHH
ID
XLXXXHL
ID
XLXXXHH
ID
A5
to
A4 A3A2 A1 A0DQ15 to DQ0
2201h (Top Boot),
2200h (Bottom Boot)
XX01h (protected),
XX00h (unprotected)
80h (factory locked),
00h(not factory locked)
Legend: L = Logic Low = VIL, H = Logic High = VIH, SA = Sector Address, X = Don’t care.
August 7, 2002Am29PDS322D15
Page 17
ADVANCE INFORMATION
Sector/Sector Block Protection and
Unprotection
(Note: For the following discussion, the term “sector”
applies to both sectors and s ector blocks. A sector
block consists of two or more adjacent sectors that are
protected or unprotected at the same time (see Tables
8 and 9).
The hardware sector protection feature disables both
program and erase operations in any sector. The hardware sector unprotection feature re-enables both
program and erase operations in previou sly protected
sectors. Sector protection and unprotection can be implemented via two methods.
The primary method requires V
on the RESET# pin
ID
only, and can be implemented either in-system or via
programming equipment. Figure 3 shows the algorithms and Figure 26 shows the timing diagram. This
method uses standard microprocessor bus cycle timing. For sector unprotect, all unprotected sectors must
first be protected prior to the first sector unprotect write
cycle.
16Am29PDS322DAugust 7, 2002
Page 18
ADVANCE INFORMATION
The alternate method intended only for programming
equipment requires V
on address pin A9 and OE#.
ID
This method is compatible with programmer routines
written for earlier AMD flash devices. Contact an AMD
representative for further details.
The device is shipped with all sectors unpr otected.
AMD offers the option of pro grammin g and protectin g
sectors at its factory prior to shipping the device
through AMD’s ExpressFlash™ Service. Contact an
AMD representative for details.
It is possible to determine whether a sector is protected or unprotected. See the Autoselect Mode
section for details.
Write Protect (WP#)
The Write Protect function provides a hardware
method of protecting certain boot sectors without
using V
WP#/ACC pin.
If the system asserts V
vice disables program and erase functions in the two
“outermost” 8 Kbyte boot sectors independently of
whether those sectors were protected or unprotected
using the method described in “Sector/Sector Block
Protection and Unprotection”. Th e two outermost 8
Kbyte boot sectors are the two sectors containing th e
lowest addresses in a bottom-b oot-configured devic e,
or the two sectors containing the highest addresses in
a top-boot-configured device.
If the system asserts V
vice reverts to whether the two oute rmost 8 Kbyte boot
sectors were last set to be protected or unprotected.
That is, sector protection or unprotection for these two
sectors depe nds on w hethe r they we re las t protec ted
or unprotected using the method described in “Sec-
tor/Sector Block Protection and Unprotection”.
Note that the WP#/ACC pin must not be left floating or
unconnected; inconsistent behavior of the device may
result.
. This function is one of two provided by the
ID
on the WP#/ACC pin, the de-
IL
on the WP#/ACC pin, the de-
IH
block consists of two or more adjacent sectors that are
protected or unprotected at the same time (see Tables
8 and 9).
This feature al lows tempor ary unprotec tion of prev iously protected sectors to change data in-system. The
Sector Unprotect mode is activated by setting the RESET# pin to V
(9.0 V – 11.0 V). During this mode,
ID
formerly protected sectors can be progr ammed or
erased by selecting the sector addresses. Once V
is
ID
removed from the RESET# pin, all the previously protected sectors are protected again. Figure 1 shows the
algorithm, and Figure 25 shows the timing diagrams,
for this feature.
START
RESET# = V
(Note 1)
Perform Erase or
Program Operations
RESET# = V
Temporary Sector
Unprotect Completed
(Note 2)
Notes:
1. All protected sectors unprotected (If WP#/ACC = V
outermost boot sectors will rem ain protec ted ).
2. All previously protected sectors are protected once
again.
ID
IH
,
IL
Temporary Sector/Sector Block Unprotect
Figure 1. Temporary Sector Unprotect Operation
(Note: For the following discussion, the term “sector”
applies to both sectors and s ector blocks. A sector
August 7, 2002Am29PDS322D17
Page 19
ADVANCE INFORMATION
START
RESET# = V
(Note 1)
ID
Perform Erase or
Program Operations
RESET# = V
IH
Temporary Sector
Group Unprotect
Completed (Note 2)
Notes:
1. All protected sector groups unprotected (If WP# = V
the first or last sector will remain protected).
2. All previou sly prot ect ed sec tor gro ups are prote cte d
once again.
Figure 2. Temporary Sector Group Unprotect
Operation
,
IL
18Am29PDS322DAugust 7, 2002
Page 20
ADVANCE INFORMATION
Temporary Sector
Unprotect Mode
Increment
PLSCNT
No
PLSCNT
= 25?
Yes
Device failed
Sector Protect
Algorithm
START
PLSCNT = 1
RESET# = V
Wait 1 µs
No
First Write
Cycle = 60h?
Set up sector
address
Sector Protect:
Write 60h to sector
address with
A6 = 0, A1 = 1,
A0 = 0
Wait 150 µs
Verify Sector
Protect: Write 40h
to sector address
with A6 = 0,
A1 = 1, A0 = 0
Read from
sector address
with A6 = 0,
A1 = 1, A0 = 0
No
Data = 01h?
Protect another
sector?
Remove V
from RESET#
Write reset
command
Sector Protect
complete
Yes
Yes
No
START
Protect all sectors:
The indicated portion
of the sector protect
ID
Reset
PLSCNT = 1
Yes
ID
algorithm must be
performed for all
unprotected sectors
prior to issuing the
first sector
unprotect address
Increment
PLSCNT
No
PLSCNT
= 1000?
Yes
Device failed
Sector Unprotect
PLSCNT = 1
RESET# = V
Wait 1 µs
First Write
Cycle = 60h?
No
All sectors
protected?
Set up first sector
address
Sector Unprotect:
Write 60h to sector
address with
A6 = 1, A1 = 1,
A0 = 0
Wait 15 ms
Verify Sector
Unprotect: Write
40h to sector
address with
A6 = 1, A1 = 1,
A0 = 0
Read from
sector address
with A6 = 1,
A1 = 1, A0 = 0
No
Data = 00h?
Last sector
verified?
Remove V
from RESET#
Yes
Yes
Yes
Yes
ID
No
Temporary Sector
Unprotect Mode
Set up
next sector
address
No
ID
Algorithm
Write reset
command
Sector Unprotect
complete
Figure 3. In-System Sector Group Protect/Unprotect Algorithms
August 7, 2002Am29PDS322D19
Page 21
ADVANCE INFORMATION
SecSi (Secured Silicon) Sector Flash
Memory Region
The SecSi (Secured Silicon) Sector feature provides a
Flash memory region that enables permanent part
identification through an Electronic Serial Number
(ESN). The SecSi Sector is 64 KBytes in length, and
uses a SecSi Sector Indicator Bit (DQ7) to indicate
whether or not the SecSi Sector is locked when
shipped from the factory. This bit is permanently set at
the factory and cannot be changed, which prevents
cloning of a factory locked part. This ensures the security of the ESN once the product is shipped to the fie ld.
AMD offers the device with the SecSi Sector either
factory locked or customer lockable. The factory-locked version is alw ays protected when shipped
from the factory, and has the SecSi (S ecured Silicon )
Sector Indicator Bit permanently set to a “1.” The customer-lockable version is shipped with the SecSi Sector unprotected, allo wing customers to u tilize that
sector in any manner they choose. The customer-lockable version also has the SecSi Sector Indicator Bit
permanently set to a “0.” Thu s, the SecSi Sector Indicator Bit prevents customer-lockable devices from
being used to replace devices that are factory locked.
The system accesses the SecSi Sector through a
command sequence (see “Enter SecSi Sector/Exit
SecSi Sector Command Sequence”). After the system
has written the Enter SecSi Sector command sequence, it may read the SecSi Sector by using the addresses normally occupied by the first sector (SA0).
This mode of operation continues until the system issues the Exit SecSi Sector command sequ ence, or
until power is removed from the device. On power-up,
or following a hardware reset, the device reverts to
sending commands to the bo ot sectors ins tead of th e
SecSi sector
Factory Locked: SecSi Sector Programmed and
Protected at the Factory
In a factory locked device, the SecSi Sector is protected when the device is shipped from the factory.
The SecSi Sector cannot be modified in any way. The
device is availab le preprogra mmed with on e of the
following:
■ A random, secure ESN only
■ Customer code through the ExpressFlash service
■ Both a random, secure ESN and customer code
through the ExpressFlash service.
In devices th at have an ESN , a Bottom Bo ot device
will have the 16-byte ESN in the low est addressable
memory area at addresses 000000h–000007h. In the
Top Boo t device the starting address of the ESN will
be at the bottom of the lowest 8 Kbyte boot sector at
addresses 1F8000h–1F8007h.
Customers may opt to have their code pro grammed by
AMD through the AMD ExpressFlash service. AMD
programs the customer’s code, with or without the random ESN. The device s are then shippe d from AMD’s
factory with the permane ntly locked . Contact an A MD
representative for details on using AMD’s ExpressFlash service.
Customer Lockable: SecSi Sector NOT
Programmed or Protected at the Factory
If the security feature is not required, the SecSi Sector
can be treated as an additional Flash me mo ry s pa c e,
exp an di ng t h e si ze of the available Flash array by 64
Kbytes. The SecSi Sector can be read, programmed,
and erased as often as required. The SecSi Sector area
can be protected using one of the following procedures:
■ Write the three-cycle Enter SecSi Sector Region
command sequence, and then follow the in-system
sector protect algorithm as shown in Figure 3, except that RESET# may be at either V
or VID. This
IH
allows in-system protection of the SecSi Sector
without raising any device pin to a high voltage.
Note that this method is only applicabl e to the SecSi
Sector.
■ Write the three-cycle Enter SecSi Sector Region
command sequence, and then use the alternate
method of sector protection described in the “Sec-
tor/Sector Block Protection and Unprotection” sec-
tion.
Once the SecSi Sector is locked and verified, the system must write the Exit SecSi Sector Region
command seq uence to return to reading and wr iting
the remainder of the array.
The SecSi Sector protection must be used with caution since, once protected, there is no procedure
available for unprotecting the SecSi Sector area and
none of the bits in the SecSi Sector memory space
can be modified in any way.
Hardware Data Protection
The command sequence requirement of unlock cycles
for programming or erasing provides data pro tection
against inadvertent writes (refer to Table 10 for command definitions). In addition, the following hardware
data protection measures prevent accidental erasure
or programming, which might otherwise be caused by
spurious system level signals during V
and power-down transitions, or from system noise.
Low V
When V
Write Inhibit
CC
is less than V
CC
, the device does not ac-
LKO
cept any write cycles. This protects data during V
power-up and power-down. The command register
and all internal program/erase circuits are disabled,
and the device resets to the read mode. Subsequent
power-up
CC
CC
20Am29PDS322DAugust 7, 2002
Page 22
ADVANCE INFORMATION
writes are igno red unti l VCC is greater than V
LKO
. The
system must provide the proper signals to the control
pins to prevent unintentional writes when V
greater than V
LKO
.
CC
is
Write Pulse “Glitch” Protection
Noise pulses of less than 5 ns (typical) on OE#, CE#
or WE# do not initiate a write cycle.
COMMAND DEFINITIONS
Writing specific address and data commands or sequences into the command register initiates device operations. Table 10 defines the valid regis ter command
sequences. Writing incorrect address and data val-ues or writing them in the improper sequence resets
the device to reading array data.
All addresses are latched on the falling edge of WE#
or CE#, whichever happens later. All dat a is latched on
the rising edge of WE# or CE#, whichever happens
first. Refer to the AC Characteristics section for timing
diagrams.
Reading Array Data
The device is automatically set to reading array data
after device powe r-up. No com mand s ar e requ ired t o
retrieve data. The device is ready to read array data
after completing an Embedded Program or Embedded
Erase algorithm.
After the device accepts an Erase Suspend command,
the device enters the erase-suspend-read mode, after
which the system can read data from any
non-erase-suspended sector. After completing a programming operation in the Erase Suspend mode, the
system may once again read array data with the same
exception. See the Erase Suspend/Erase Resume
Commands section for more information.
The system must issue the reset command to return
the device to the read (or erase-suspend-read) mode
if DQ5 goes high during an active program or erase
operation, or if the device i s in the autoselect mod e.
See the next section, Reset Command, for more information.
See also Requirements for Reading Array Data in the
Device Bus Operations section for more information.
The Read-Only Operations table provides the read parameters, and Figure 15 shows the timing diagram.
Reset Command
Writing the reset command resets the device to the
read or erase-sus pend-read mod e. Address bi ts are
don’t cares for this command.
Logical Inhibit
Write cycles are inhibited by holding any one of OE# =
, CE# = VIH or WE# = VIH. To initiate a write cycle,
V
IL
CE# and WE# must be a logical zero while OE# is a
logical one.
Power-Up Wri t e Inhibit
If WE# = CE# = V
and OE# = VIH during power up,
IL
the device does not accept commands on the rising
edge of WE#. The internal s tate machine is automatically reset to the read mode on power-up.
The reset command may be written between the sequence cycles in an erase command sequence before
erasing begins. This resets the device to the read
mode. Once erasure begins, however, the device ignores reset commands until the operation is complete.
The reset command may be written betwee n the
sequence cycles in a program c ommand sequence
before programming begins. This resets the device to
the read mode. If the pr ogram com mand sequ ence is
written while the device is in the Erase Suspend mode,
writing the reset command returns the device to the
erase-suspend-read mode. Once programming begins, however, the device ignores reset commands
until the operation is complete.
The reset command may be written between the sequence cycles in an autoselect command sequence.
Once in the autoselect mode, the reset command
must be wr itten to r eturn to the read mode. If the de vice entered the autosel ect mode while in the Erase
Suspend mode, writing the reset command returns the
device to the erase-suspend-read mode.
If DQ5 goes high during a program or erase operation,
writing the reset command returns the device to the
read mode (or erase-suspend-read mode if the device
was in Erase Suspend).
Autoselect Command Sequence
The autoselect command sequence allows the host
system to access the manufacturer and device codes,
and determine whether or not a sector is protected.
Tab le 10 shows the address an d data require ments.
This method is an alternative to that shown in Table 7,
which is intended for PROM programmers and requires V
mand sequence may be written to an address that is
either in the read or er ase-suspend- read mode. Th e
autoselect command may not be wr itten while the device is actively programming or erasing.
The a utoselect comman d sequen ce is initia ted by wri ting two unlock cycles, followed by the autoselect
command. The device then enters the autoselect mode,
on address pin A9. The autosel ect com-
ID
August 7, 2002Am29PDS322D21
Page 23
ADVANCE INFORMATION
and the system may read any number of autoselect
codes without reinitiating the command sequence.
Table 10 shows the address and data requirements for
the command sequence. To determine sector protection information, the system must write to the appropriate sector group address (SGA). Tables 3 and 5 show
the address range associated with each sector.
The system must write the reset command to return to
the read mode (or erase-suspend-read mode if the device was previously in Erase Suspend).
Enter SecSi Sector/Exit SecSi Sector
Command Sequence
The SecSi Sector region provides a secured data area
containing an 16-byte random Electronic Serial Number (ESN). The system can access the SecSi Sector
region by issuing the three-cycle Enter SecSi Sector
command sequence. The device continues to access
the SecSi Sector region until the system issues the
four-cycle Exit SecSi Sector command sequence. The
Exit SecSi Sector command sequence returns the device to normal op eration. Table 10 shows the address
and data requirements for both command sequences.
See also “SecSi (Secured Silicon) Sector Flash M emory Region” for further information. Note that a hardware reset (RESET#=V
) will reset the device to
IL
reading array data.
Word Program Command Sequence
Programming is a four-bus-cycle operation. The program command sequence is initiated by writing two
unlock write cycles, followed by the program set-up
command. The program addr ess and data are wr itten
next, which in tur n initiate the Embedd ed Program algorithm. The system is not required to pro vide furthe r
controls or timings. The device automatically provides
internally generated program pulses and verifies th e
programmed cell margin. Table 10 shows the address
and data requirements for the program command sequence.
When the Emb edded P rogram algori thm is c omple te,
the device then returns to the read mode and addresses are no longer latched. The system can determine the status of the program operation by using
DQ7, DQ6, or RY/BY#. Refer to the Write Operation
Status section for information on these status bits.
Any commands wr itten to the dev ice during the Embedded Program Algorithm are ignored. Note that ahardware reset immediately terminates the program
operation. The program command sequence should
be reinitiated once the device has returned to the read
mode, to ensure data integrity.
Programming is allowed in any sequence and across
sector boundaries. A bit cannot be programmed
from “0” back to a “1.” Attempting to do so may
cause the device to set DQ5 = 1, or cause the DQ7
and DQ6 status bits to indicate the operation was successful. However, a succeeding read will show that the
data is still “0.” On ly erase operations ca n convert a
“0” to a “1.”
Unlock Bypass Command Sequence
The unlock bypass feature allows the system to program words to the device faster than using the standard program command s equence. The unlock
bypass command s equence is in itiated by first w riting
two unlock cycles. This is followed by a third write
cycle containing the unlock bypass command, 20h.
The device then enters the unlock b ypass mode. A
two-cycle unlock bypass program command sequence
is all that is required to program in this mode. The first
cycle in this sequence contains the unlock bypass program command, A0h; the second cycle contains the
program address and data. Additional data is programmed in the same manner. This mode dispenses
with the initial two unlock cycles required in the standard program command sequence, resulting in faster
total programming time. Table 10 shows the requirements for the command sequence.
During the unlock bypass mode, only the Unlock Bypass Program and Unlock Bypa ss Reset commands
are valid. To exit the unlock bypass mode, the system
must issue the two-cycle unlock bypass reset command sequence. The first cycle must contain the data
90h. The second cycle must contain the data 00h. The
device then returns to reading array data. See Figu re
4 for the unlock bypass algorithm.
The device offers accelerated program operations
through the WP#/ACC pin. When the system asserts
on the WP#/ACC pin, the device automatically en-
V
HH
ters the Unlock Bypass mode. The system may then
write the two-cycle Unlock Bypass program command
sequence. The de vice us es the hig her voltag e on the
WP#/ACC pin to acc elerate the ope ration. Note th at
the WP#/ACC pin must not be at V
any operation
HH
other than accelerated programming, or device damage may result. In addition, the WP#/ACC pin must not
be left floating or unconnected; inconsistent behavior
of the device may result.
Figure 5 illustrates the algorithm for the program operation. Refer to the Erase and Program Operations
table in the AC Characteristics section for parameters,
and Figure 18 for timing diagrams.
Chip Erase Command Sequence
Chip erase is a six bus cycle operation. The chip erase
command sequence is initiated by writing two unlock
22Am29PDS322DAugust 7, 2002
Page 24
ADVANCE INFORMATION
Increment
Address
Start
555h/AAh
2AAh/55h
555h/20h
XXXh/A0h
Program Address/Program Data
Data# Polling Device
Verify Byte?
Yes
No
Last Address
?
Yes
No
Set
Unlock
Bypass
Mode
In
Unlock
Bypass
Program
START
Write Program
Command Sequence
Data Poll
Embedded
algorithm
in progress
Increment Address
Program
No
from System
Verify Data?
Yes
Last Address?
Yes
Programming
Completed
Note: See Table 10 for program command sequence.
No
Programming Completed
(BA) XXXh/90h
XXXh/F0h
Figure 4. Unlock Bypass Algorithm
Reset
Unlock
Bypass
Mode
Figure 5. Program Operation
cycles, followed by a set-up command. Two additional
unlock write cycles are then followed by the chip erase
command, which in turn invokes the Embedded Erase
algorithm. The device does not require the system to
preprogram prior to erase. The Embedded Erase algorithm automatically preprograms and verifies the entire
memory for an all zero data pattern prior to electrical
erase. The system is not required to provide any controls or timings during these operations. Table 10
shows the address and data requirements for the chip
erase command sequence.
August 7, 2002Am29PDS322D23
Page 25
ADVANCE INFORMATION
When the Embedded Erase algorithm is complete, the
device returns to the read mode and add resses are no
longer latched. The system can determine the status
of the erase operation by using DQ7, DQ6, DQ2, or
RY/BY#. Refer to the Write Operation Status section
for information on these status bits.
Any commands written during the chip erase operation
are ignored. However, note that a hardware reset im-
mediately termina tes the erase operation. If tha t occurs, the chip erase command sequence should be
reinitiated once the device has returned to reading
array data, to ensure data integrity.
Figure 6 illustrates the algorithm for the erase operation. Refer to the Erase and Program Operation s tables in the AC Characteristics sectio n for parameters,
and Figure 20 section for timing diagrams.
Sector Erase Command Sequence
Sector erase is a six bus cycle operation. The sector
erase command sequence is initiated by writing two
unlock cycles, followed by a set-up command. Two additional unlock cycles are written, and are then followed by the address of the sector to be erased, and
the sector erase command. Table 10 shows the address and data requirements for the sector erase command sequence.
The device does not require the system to preprogram
prior to erase. The Embedded Erase algorithm automatically programs and v erifies the entire m emory for
an all zero data pattern prior to electrica l erase. The
system is not required to provide any controls or timings during these operations.
After the command sequence is written, a sector erase
time-out of 50 µs occurs. During the time-out period,
additional sector addresses and sector erase com mands may be written. Loading the sector e rase buf fer
may be done in any sequence, and the number of sectors may be from one sector to all sectors. The time
between these additional cycles must be less than 50
µs, otherwise erasure may begin. Any sector erase
address and command following the exceeded
time-out may or may not be accepted. It is recommended that processor interrupts be disabled during
this time to ensure all comm ands are accepted . The
interrupts can be re-enabled after the last Sector
Erase command is written. Any command other than
Sector Erase or Erase Suspend during the
time-out period resets the device to the read
mode. The system must rewrite the command se-
quence and any additional addresses and commands.
The system can monitor DQ3 to determine if the sec-
tor erase timer has timed out (See the section on DQ3:
Sector Erase Timer.). The time-out begins from the rising edge of the final WE# pulse in the command
sequence.
When the Embedded Erase algorithm is complete, the
device returns to reading array data and addresses
are no longer latched. Note that while the Embedded
Erase operation is in progress, the system can read
data from the non-erasing sector. The system can determine the status of the erase o peration by readin g
DQ7, DQ6, DQ2, or RY/BY# in the erasing sector.
Refer to the Write Operation Status section for information on these status bits.
Once the sector erase operation has begun, only the
Erase Suspend command is valid. All other commands are ignored. However, note that a hardwarereset immediately terminates the er ase operation. If
that occurs, the sector erase command seq uence
should be reinitiated once the device has returned to
reading array data, to ensure data integrity.
Figure 6 illustrates the algorithm for the erase operation. Refer to the Erase and Program Operatio ns tables in the AC Characteristics sectio n for parameters,
and Figure 20 section for timing diagrams.
Erase Suspend/Erase Resume
Commands
The Erase Suspend command, B0h, allows the system to interrupt a sector erase operation and then read
data from, or program data to, any sector not selected
for erasure. This command is valid only during the
sector erase operation, including the 50 µs time-out
period during the sector erase command sequence.
The Erase Suspend command is ignored if written during the chip erase operation or Embedded Program
algorithm.
When the Erase Suspend command is written during
the sector erase operation, the device requires a maximum of 20 µs to suspend the erase operation. However, when the Erase Suspend command is written
during the sector erase time-out, the device immediately terminates the time-out period and suspends the
erase operation.
After the erase operation has been suspended, the
device enters the erase-suspend-read mode. The system can read data from or program data to any sector
not selected for erasure. (The device “erase suspends” all sectors selected for erasure.) Note that unlock bypass programming is not allowed when the
device is erase-suspended.
Reading at any add ress within e rase-suspe nded sectors produces status information on DQ7–DQ0. The
system can use DQ7, or DQ6 and DQ2 together, to
determine if a sector is actively erasing or is
erase-suspended. Refer to the Write Operation Status
section for information on these status bits.
After an eras e-sus pende d prog ram o pera tion is complete, the device returns to the erase-suspend-read
24Am29PDS322DAugust 7, 2002
Page 26
ADVANCE INFORMATION
mode. The system can determine the status of the
program oper ation u sing the DQ7 or DQ6 s tatus bits,
just as in the standard word program operation.
Refer to the Write Operation Status section for more
information.
START
In the erase-suspend-read mode, the system can also
issue the autoselect command sequence. Refer to the
Autoselect Mode and Autoselect Command Sequence
sections for details.
To resume the sect or erase operation, the system
must write the Erase Resume command. The address
of the erase-suspended sector is required when writing this command. Further writes of the Resume command are ignored. Another Erase Suspend com mand
can be written after the chip has resumed erasing.
Write Erase
Command Sequence
(Notes 1, 2)
Data Poll to Erasing
Bank from System
No
Notes:
1. See Table 10 for erase command sequence.
2. See the section on DQ3 for information on the sector
erase timer.
Data = FFh?
Yes
Erasure Completed
Embedded
Erase
algorithm
in progress
Figure 6. Erase Operation
August 7, 2002Am29PDS322D25
Page 27
ADVANCE INFORMATION
Table 10. Am29PDS322D Command Definitions
Command
Sequence
(Note 1)
Read (Note 6)1RARD
Reset (Note 7)1XXXF0
Manufacturer ID4555AA2AA5555590X000001
Device ID (Note 9)6555AA2AA5555590X01227EX0E2206X0F
SecSi Sector Fa ctory
X = Don’t care
RA = Address of the memory location to be read.
RD = Data re ad from location RA duri ng read operation.
PA = Addre ss of the me mo ry lo catio n to b e pr ogramme d. Addre sses
latch on the falling edge of the WE# or CE# pulse, whichever happens
later.
FirstSecond Third Fourth Fifth Sixth
Cycles
AddrDataAddrDataAddrDataAddrDataAddr DataAddr Data
4555AA2AA5555590X0380/00
4555AA2AA5555590
555AA2AA5555520
3
XXXA0PAPD
2
XXX90XXX00
2
Bus Cycles (Notes 2–5)
2201/
2200
(SGA)
X02
PD = Data to be programmed at location PA. Data latches on the risi ng
edge of WE# or CE# pulse, whichever happens first.
SGA = Address of the sector group to be verified (in autoselect mode)
or erased. A ddress bits A20–A12 uniquely select any sector.
XX00/
XX01
Notes:
1. See Table 1 for description of bus operations.
2. All values are in hexadecimal.
3. Except for the read cycle and the fourth and fifth cycle of the
autoselect command sequence, all bus cycles are write cycles.
4. Data bits DQ15–DQ8 are don’t care in command sequences,
except for RD and PD.
5. Unless otherwise noted, address bits A20–A12 are don’t cares in
unlock sequence.
6. No unlock or command cycles required when device is in read
mode.
7. The Reset command is required to return to the read mode (or to
the erase-suspend-read mode if previously in Erase Suspend)
when the device is in the autoselect mode, or if DQ5 goes high
(while the device is providing status information).
8. The fourth cycle of the autoselect command sequence is a read
cycle. The system must provide the bank address to obtain the
manufacturer ID, device ID, or SecSi Sector factory protect
information. Data bits DQ15–DQ8 are don’t care. See the
Autoselect Command Sequence section for more information.
9. The device ID must be read across the fourth, fifth and sixth
cycles. The sixth cycle specifies 2201h for top boot or 2200h for
bottom boot.
10. The data is 80h for factory locked and 00h for not factory locked.
11. The data is 00h for an unprotected sector group and 01h for a
protected sector group.
12. The Unlock Bypass command is required prior to the Unlock
Bypass Program command.
13. The Unlock Bypass Reset command is required to return to the
read mode when the device is in the unlock bypass mode.
14. The system may read and program in non-erasing sectors, or
enter the autoselect mode, when in the Erase Suspend mode.
The Erase Suspend command is valid only during a sector erase
15. The Erase Resume command is valid only during the Erase
Suspend mode, and requires the bank address.
26Am29PDS322DAugust 7, 2002
Page 28
ADVANCE INFORMATION
WRITE OPERATION STATUS
The device provides several bits to determine the status of
a program or erase operation: DQ2, DQ3, DQ5, DQ6, and
DQ7. Table 1 1 and the following s ubsections descr i be the
function of these bits. DQ7 and DQ6 each offer a method
for determining whether a program or erase operation is
complete or in progress. The device also provides a hardware-based output signal, RY/BY#, to determine whether
an Embedded Program or Eras e operation is in progress or
has been completed.
invalid. Valid data on DQ0–DQ7 will appear on successive read cycles.
Table 11 shows the outputs for Data# Polling on DQ7.
Figure 7 shows the Data# Polling algorithm. Figure 22
in the AC Characteristics section shows the Data#
Polling timing diagram.
DQ7: Data# Polling
The Data# Polling bit, DQ7, indicates to the host system
whether an Embedded Program or Erase algorithm is in
progress or completed, or whether the device is in Erase
Suspend. Data# Polling is valid after the rising edge of the
final WE# pulse in the comma nd sequence .
During the Embedded Prog ram algorithm, the devi ce outputs on DQ7 the complement of t he datum pr ogr ammed to
DQ7. This DQ7 status also applies to programming during
Erase Suspend. When t he Embedded Pr ogram algorit hm is
complete, the device outputs the datum programmed to
DQ7. The system must provide the program address to
read valid status in formation on DQ7. If a program address
falls within a protected sector, Data# Polling on DQ7 is active for approximately 1 µs, then the device returns to the
read mode.
During the Embedd ed Erase algorith m, Data# Pollin g
produces a “0” on DQ7. When the Embedde d Erase
algorithm is complete, or if the device enters the Erase
Suspend mode, Data# Polling produces a “1” on DQ7.
The system must provide an address within any of the
sectors selected for erasure to r ead valid statu s information on DQ7.
No
START
Read DQ7–DQ0
Addr = VA
DQ7 = Data?
No
DQ5 = 1?
Yes
Read DQ7–DQ0
Addr = VA
Yes
After an erase com mand sequen ce is written, if all
sectors selected for erasing are protected, Data# Polling on DQ7 is active for approximately 100 µs, then
DQ7 = Data?
Yes
the device returns to the read mode. If not all selected
sectors are protected, the Embedded Erase algorithm
erases the unpr otected secto rs, and i gnores the se-
No
lected sectors that are protected. However, if the system reads DQ7 at an address withi n a protected
sector, the status may not be valid.
Just prior to the completion of an Embedded Program
or Erase operation, DQ7 may change asynchronously
with DQ0–DQ6 while Output Enable (OE#) is asserted
low. That is, the device may change from providing
status information to valid data on DQ7. Depending on
when the system samples the DQ7 output, it may read
the status or valid data. Even if the device has com-
FAIL
Notes:
1. VA = Valid address for programming. During a sector
erase operation, a valid address is any sector address
within the sector being erased. During chip erase, a
valid address is any non-protected sector address.
2. DQ7 should be rechecked even if DQ5 = “1” because
DQ7 may change simultaneously with DQ5.
PASS
pleted the program or er ase operation and D Q7 has
valid data, the data outputs on DQ0–DQ6 may be still
Figure 7. Data# Polling Algorithm
August 7, 2002Am29PDS322D27
Page 29
ADVANCE INFORMATION
RY/BY#: Ready/Busy#
The RY/BY# is a dedicated, open-drain output pin
which indicates whether an Embedded Algorithm is in
progress or complete. The RY/BY# status is valid after
the rising edge of the final WE# pulse in the command
sequence. Since RY/BY# is an open-drain output, several RY/BY# pins can be tied together in parallel with a
pull-up resistor to V
CC
.
If the output is low (Busy), the device is active ly erasing or programming. (This includes programming in
the Erase Suspend mode.) If the output is high
(Ready), the device is in the read mode, the standby
mode, or the device is in the erase-suspend-read
mode.
Table 11 shows the outputs for RY/BY#.
DQ6: Toggle Bit I
Toggle Bit I on DQ6 indicates whether an Embedded
Program or Erase algorithm is in pro gress or complete, or whether the device has entered the Erase
Suspend mo de. Toggle Bit I m ay be re ad at any address, and is valid after the rising edge of the final
WE# pulse in the command sequence (prior to the
program or erase operation), and d uring the sector
erase time-out.
During an Embedded Program or Erase algorithm operation, successive read cycles to any address cause
DQ6 to toggle. The system may use either OE# or
CE# to control the read cycles. When the operation is
complete, DQ6 stops toggling.
After an erase command sequence is written, if all sectors
selected for erasing are pr otect ed, DQ6 t oggles f or appr oximately 100 µs, then returns to reading array data. If not all
selected sectors are protected, the Embedded Erase algorithm erases the unprotected se ctors, and ignores the selected sectors that are prot ected.
The system can use DQ6 and DQ2 together to determine
whether a sector is actively erasing or is erase-suspended.
When the device is actively erasing (that is, the Embedded
Erase algorithm is in progre ss), DQ6 t oggles. When the device enters the Erase Suspend mode, DQ6 stops toggling.
However, the system must also use DQ2 to determine
which sectors are erasing or erase-suspended. Alternatively, the system can use DQ7 (see the subsection on
DQ7: Data# Polling).
If a program address falls within a protected sector,
DQ6 toggles for approximately 1 µs after the program
command sequence is written, then returns to reading
array data.
Tabl e 11 shows the outputs for Toggle Bit I on DQ6.
Figure 8 shows the toggle bit algorithm. Figure 23 in
the “AC Characteristics” section shows the toggle bit
timing diagrams. F igure 24 shows th e differences b etween DQ2 and DQ6 in graphical form. See also the
subsection on DQ2: Toggle Bit II.
START
Read DQ7–DQ0
Read DQ7–DQ0
Toggle Bit
= Toggle?
Yes
No
Note: The system should recheck the toggle bit even if
DQ5 = “1” because the toggle bit may stop toggling as DQ5
changes to “1.” See the subsections on DQ6 a nd DQ2 for
more information.
DQ5 = 1?
Yes
Read DQ7–DQ0
Twice
Toggle Bit
= Toggle?
Yes
Program/Erase
Operation Not
Complete, Write
Reset Command
No
No
Program/Erase
Operation Complete
DQ6 also toggles during the erase-suspend-program
mode, and stops toggling once the Embedded Pro-
Figure 8. Toggle Bit Algorithm
gram algorithm is complete.
28Am29PDS322DAugust 7, 2002
Page 30
ADVANCE INFORMATION
DQ2: Toggle Bit II
The “Toggle Bit II” on DQ2, when used with DQ6, indicates whether a particular sector is actively erasing
(that is, the Embedded Erase algorithm is in progress),
or whether that sector is erase-suspe nded. Toggle Bit
II is valid after the rising edge of the final WE# pulse in
the command sequence.
DQ2 toggles when the system reads at addresses
within those sectors that have been selected for erasure. (The system may use either OE# or CE# to control the read cycles.) But DQ2 cannot distinguish
whether the sector is actively era sing or is e rase-suspended. DQ6, by comparis on, indicates whether th e
device is actively erasing, or is in Erase Suspend, but
cannot distinguish wh ich sectors ar e selected for erasure. Thus, both status bits are required for sector and
mode information. Refer to Table 11 to compare outputs for DQ2 and DQ6.
Figure 8 shows the toggle bit algorithm in flowchart
form, and the section “DQ2: Toggle Bit II” explains the
algorithm. See also the DQ6: Toggle Bit I subsection.
Figure 23 shows the toggle bit timing diagr am. Figure
24 shows the differences between DQ2 and DQ6 in
graphical form.
Reading Toggle Bits DQ6/DQ2
Refer to Figure 8 for the following discussion. Whenever the system initially begins reading toggle bit status, it must read DQ7–DQ0 at least twice in a row to
determine whether a toggle bit is toggling. Typically,
the system would note and store the va lue of the toggle bit after the first read. After the second read, the
system would compare the new value of the toggle bit
with the first. If the toggle bit is not toggling, the device
has completed the program or erase operation. The
system can read array data on DQ7–DQ0 on the following read cycle.
However, if after the initial two read cycles, the system
determines that the tog gle bit is still togg ling, the sys tem also should note whether the value of DQ5 is high
(see the section on DQ5). If it is, the system should
then determine ag ain whether the toggle bit is to ggling, since the toggle bit may have stopped toggling
just as DQ5 went high. If the toggle bit is no longer
toggling, the device has successfully completed the
program or erase operation. If it is still toggling, the device did not completed the operation successfully, and
the system must write the reset command to return to
reading array data.
The remaining s cenario is that th e system initia lly determines that the toggle bit is toggling and DQ5 has
not gone high. The system may continue to monitor
the toggle bit and DQ5 through successive read cycles, determining the status as described in the previous paragraph. Alternatively, it may choose to perform
other system tasks. In this case, the system must start
at the beginning of the algorithm when it returns to determine the status of the operation (top of Figure 8).
DQ5: Exceeded Timing Limits
DQ5 indicates whether the program or eras e time has
exceeded a specified int ernal pulse cou nt limit. Under t hese
conditions DQ5 produces a “1,” indicating that the program
or erase cycle was not successfully completed.
The device may output a “1” on DQ5 if the system tries
to program a “1” to a location that was previously programmed to “0.” Only an erase operation canchange a “0” back to a “1.” Under this condition, the
device halts the operation, and when the timi ng limit
has been exceeded, DQ5 produces a “1.”
Under both these conditions, the system must write
the reset command to return to the read mode (or to
the erase-suspend-read mode if the device was previously in the erase-suspend-program mode).
DQ3: Sector Erase Timer
After writing a sector erase command sequence, the
system may read DQ3 to determ ine whether or not
erasure has begun. (The sector erase timer does not
apply to the chip erase com mand.) If additional
sectors are selected for erasure, the entire time-out
also applies afte r each a dditional se ctor eras e command. When the time-out period is complete, DQ3
switches from a “0” to a “1.” If the tim e between additional sector erase commands from the system can be
assumed to be less than 50 µs, the system need not
monitor DQ3. See also the Sector Erase Command
Sequence section.
After the sector erase command is written, the system
should read the status of DQ7 (Data# Polling) or DQ6
(Toggle Bit I) to ensure that the device has accepted
the command sequence, and then read DQ3. If DQ3 is
“1,” the Embedded Erase algorithm has begun; all further commands (except Erase Suspend) are ignored
until the erase operation is complete. If DQ3 is “0,” the
device will acce pt additional sector eras e com mands.
To ensure the com mand ha s been ac cepted, the sys tem software should check the status of DQ3 prior to
and following each subsequent sector erase command. If DQ3 is high on the sec ond status ch eck, the
last command might not have been accepted.
Table 11 shows the status of DQ3 relative to the other
status bits.
August 7, 2002Am29PDS322D29
Page 31
ADVANCE INFORMATION
Table 11. Write Operation Status
DQ7
Status
Standard
Mode
Erase
Suspend
Mode
Notes:
1. DQ5 switches to ‘1’ when an Embedded Program or Embedded Erase operation has exceeded the maximum timing limits.
Refer to the section on DQ5 for more information.
2. DQ7 and DQ2 require a valid address when reading status information. Refer to t he appro priate subsect ion f or fur ther detail s.
Embedded Program Algo rith mDQ7#Toggle0N/ANo toggle0
Embedded Erase Algorith m0Toggle01Toggle0
with Power Applied . . . . . . . . . . . . . –65°C to +125°C
Voltage with Respect to Ground
(Note 1) . . . . . . . . . . . . . . . . .–0.5 V to +2.5 V
V
CC
A9, OE#, and RESET#
(Note 2). . . . . . . . . . . . . . . . . . . . . –0.5 V to +11 V
WP#/ACC . . . . . . . . . . . . . . . . . .–0.5 V to +12.6 V
All other pins (Note 1). . . . . . –0.5 V to V
Output Short Circuit Current (Note 3) . . . . . . 100 mA
Notes:
1. Minimum DC voltage on input or I/O pins is –0.5 V.
During voltage transitions, input or I/O pins may
overshoot V
Maximum DC voltage on input or I/O pins is V
See Figure 9. During voltage transitions, input or I/O pins
may overshoot to V
Figure 10.
2. Minimum DC input voltage on pins A9, OE#, RESET#,
and WP#/ACC is –0.5 V. During voltage transitions, A9,
OE#, WP#/ACC, and RESET# may overshoot V
–2.0 V for periods of up to 20 ns. See Figure 9. Maximum
DC input voltage on p in A9 is +12.5 V which m ay
overshoot to +14.0 V for periods up to 20 ns. Maximum
DC input voltage on WP#/ACC is +12.6 V which may
overshoot to +12.0 V for periods up to 20 ns.
3. No more than on e output ma y be shorted to ground at a
time. Duration of the short c ircuit should n ot be greater
than one second.
Stresses above those listed under “Absolute Maximum
Ratings” may cause permanent damage to the device. This
is a stress rating only; functional operation of the device at
these or any other co nditions above those i ndicated in the
operational sections of this data sheet is not implied.
Exposure of the device to absolute maximum rating
conditions for extended per iod s may affe ct dev ice relia bili ty.
to –2.0 V for pe riods of up to 20 ns.
SS
+2.0 V for periods up to 20 ns. See
CC
+0.5 V
CC
+0.5 V.
CC
SS
to
+0.8 V
–0.5 V
–2.0 V
V
+2.0 V
V
+0.5 V
2.0 V
20 ns
20 ns
Figure 9. Maximum Negative
Overshoot Waveform
20 ns
CC
CC
20 ns
Figure 10. Maximum Positive
Overshoot Waveform
20 ns
20 ns
OPERATING RANGES
Industrial (I) Devices
Ambient Temperature (T
V
Supply Voltages
CC
for standard voltage range . . . . . . .1.8 V to 2.2 V
V
CC
Operating ranges define those limits between which the functionality of the device is guaranteed.
Note: VA = Valid address; not required for DQ6. Illustration shows first two status cycle after command sequence, last status
read cycle, and array data read cycle
Figure 23. Toggle Bit Timings (During Embedded Algorithms)
Enter
Embedded
Erasing
WE#
Erase
Erase
Suspend
Erase Suspend
Suspend Program
Read
Enter Erase
Erase
Suspend
Program
Erase Suspend
Read
Erase
Resume
Erase
Erase
Complete
DQ6
DQ2
Note: DQ2 toggles only when read at an address within an erase-suspended sector. The system may use OE# or CE# to toggle
DQ2 and DQ6.
Figure 24. DQ2 vs. DQ6
42Am29PDS322DAugust 7, 2002
Page 44
ADVANCE INFORMATION
AC CHARACTERISTICS
Temporary Sector Unprotect
Parameter
DescriptionAll Speed OptionsUnitJEDECStd
t
VID Rise and Fall Time (See Note)Min500ns
VIDR
t
VHHVHH
t
RSP
Rise and Fall Time (See Note)Min500ns
RESET# Setup Time for Temporary
Sector/Sector Block Unprotect
Min4µs
t
RRB
Note: Not 100% tested.
V
ID
RESET#
VSS, VIL,
or V
CE#
WE#
RY/BY#
RESET# Hold Time from RY/BY# High for
Temporary Sector/Sector Block Unprotect
IH
t
VIDR
Min4µs
Program or Erase Command Sequence
t
RSP
t
RRB
Figure 25. Temporary Sector Group Unprotect Timing Diagram
t
VIDR
V
VSS, VIL,
or V
ID
IH
August 7, 2002Am29PDS322D43
Page 45
AC CHARACTERISTICS
V
ID
V
RESET#
IH
ADVANCE INFORMATION
SA, A6,
A1, A0
Valid*Valid*Valid*
Sector/Sector Block Protect or UnprotectVerify
Data
60h60h40h
Sector/Sector Block Protect: 150 µs,
Sector/Sector Block Unprotect: 15 ms
1 µs
CE#
WE#
OE#
For sector group protect, A6 = 0, A1 = 1, A0 = 0. For sector group unprotect, A6 = 1, A1 = 1, A0 = 0.
Figure 26. Sector Group Protect and Unprotect Timing Diagram
Status
44Am29PDS322DAugust 7, 2002
Page 46
ADVANCE INFORMATION
AC CHARACTERISTICS
Alternate CE# Controlled Erase and Program Operations
ParameterSpeed Option
JEDECStdDescription1012Unit
t
AVAV
t
AVWL
t
ELAX
t
DVEH
t
EHDX
t
GHEL
t
WLEL
t
EHWH
t
ELEH
t
EHEL
t
WHWH1
t
WHWH1
t
WHWH2
t
t
t
t
t
t
GHEL
t
t
t
t
CPH
t
WHWH1
t
WHWH1
t
WHWH2
Write Cycle Time (Note 1)Min100120ns
WC
Address Setup TimeMin0ns
AS
Address Hold TimeMin60ns
AH
Data Setup TimeMin60ns
DS
Data Hold TimeMin0ns
DH
Read Recovery Time Before Write
(OE# High to WE# Low)
overhead (Note 5)
Accelerated Word Program Time5µs
Chip Program Time (Note 3)20100sec
Notes:
1. Typical program and erase times assume the following conditions: 25°C, 2.0 V V
, 1,000,000 cycles. Additionally,
CC
programming typicals assume checkerboard pattern.
2. Under worst case conditions of 90°C, V
= 1.8 V , 1,000,000 cycle s.
CC
3. The typical chip programming time is considerably less than t he maxi mum chip programmi ng time lis ted, s ince most words
program faster than the maximum program t imes l isted .
4. In the pre-programming step of the Embedded Erase algorithm, all bits are pr ogrammed to 00h bef ore er asure.
5. System-level overhead is the time required to execute the two- or four-bus-cycle sequence for the program command. See Table
10 for further information on command defin ition s.
6. The device has a minimum erase and program cycle endurance of 1,000,000 cycles.
LATCHUP CHARACTERISTICS
DescriptionMinMax
Input voltage with respect to V
(including A9, OE#, and RESE T#)
on all pins except I/O pins
SS
–1.0 V12.5 V
Input voltage with respect to V
Current–100 mA+100 mA
V
CC
on all I/O pins–1.0 VVCC + 1.0 V
SS
Note: Includes all pins except VCC. Test conditions: VCC = 1.8 V, one pin at a time.
DATA RETENTION
Parameter DescriptionTest ConditionsMinUnit
Minimum Pattern Data Retention Time
150°C10Years
125°C20Years
August 7, 2002Am29PDS322D47
Page 49
ADVANCE INFORMATION
PHYSICAL DIMENSIONS
FBD048—48-Ball Fine-Pitch Ball Grid Array (FBGA) 6 x 12 mm package
xFBD 048
6.00 mm x 12.00 mm
PACKAGE
0.20
0.84
0.94
12.00 BSC
6.00 BSC
5.60 BSC
4.00 BSC
8
6
48
0.250.30
0.35
0.80 BSC
Dwg rev AF; 1/2000
1.20
0.40 BSC
* For reference only. BSC is an ANSI standard for Basic Space Centering.
48Am29PDS322DAugust 7, 2002
Page 50
ADVANCE INFORMATION
REVISION SUMMARY
Revision A (December 4, 2000)
Initial release.
Revision A+1 (February 16, 2001)
Erase Suspend/Erase Resume Commands
Noted in the third paragraph that unlock bypass programming is not allowed when the device is erase suspended.
Ordering Information
Added “U” des ignator to package markin g. Deleted
burn-in option.
Revision A+2 (August 31, 2001)
Autoselect Command Sequence
Modified sectio n to point to approp riate tables for autoselect functions.
Revision A+3 (February 18, 2002)
Global
Removed “Advance Information” designation from
data sheet.
Revision A+4 (August 7, 2002)
Distinctive Characteristics
Removed “Supports Common Flash Memory Interface
(CFI))
Table 10. Am29PDS322D Command Definitions
Changed the Command Cycle Device ID cycle from 6
to 4.