Art of BIOS flashing

CHAPTER ONE. Dedicated for those lazy to read and those who just want to get it done A.S.A.P.

NOTE: if you're interested in this article, you're supposed to have some background in things like:

  • what is BIOS, what is it intended for and what types of BIOS-es exist?
  • what is DOS command-line interface and how to use it?
  • and maybe some other stuff...

At least, you should be able to find the required/missing information in the Net — this article won't help you with it. So, instead of starting from basics, let's do just the opposite and start with clearing up the most typical myths and confusions.

Myths

Myth #1: One has to (should) reprogram the FLASH memory chip (which contains the BIOS) using only the flashing software supplied by the BIOS manufacturer (for example, use AwdFlash with AwardBIOS, or AMIFlash with AMIBIOS).

That's wrong, but many people believe in this myth — particularly because there are so many misleading warnings on manufacturers' websites (especially for people for whom English is a foreign language!). You could ask: "Why is it wrong?". If you want to know the answer – don't be lazy to read this part of the article and the next one!

Myth #2: One has to (should) use the so-called Hot Swap method only with a motherboard of exactly the same type (same chipset, same manufacturer, etc.)

That's wrong, and it's the second most common mistake. It comes from the lack of knowledge about how different flash chips work, and this difference actually may cause some limitations on the type of motherboards that can be used to (re)program a specific flash memory chip. Anyway, if you need to reprogram the flash chip of a (dead) motherboard, and you have a working motherboard of exactly the same type — this would be an advantage.

Flashers and their usage

A flasher is a piece of software for flash memory chips (re)programming (process also commonly known as (re)flasing). The most popular "universal" flashers are AwdFlash, AMIFlash and UniFlash; other flashers are mostly special ones for specific brands, like aFlash for ASUS motherboards or special flashers for Intel boards (note that universal flashers may not work with these boards). In this article we'll concentrate only on the mentioned three "universal" flashers.

As most DOS programs, flashers can be controlled by command line switches. As there is plenty of information about them in the Net, so only the essential ones will be described here.

Image:45px-Stop.png ATTENTION! Running a flasher WITHOUT specifying any swithes (i.e. using the default ones) might damage the BIOS, even though you had no intention to reflash it and were just wondering how it works and/or just wanted to backup the current version of your BIOS. Although it's a very rare case and is related usually to some specific versions of AMIFlash, it has happened personally to the author, and not just once! cool

Executing some special utilities (like DMICFG), which are sometimes included on CD's that came with a motherboard (for "fine-tuning" BIOS), often causes the same problem — damages the BIOS, even when you had no intention to reflash the flash chip, but just wanted to have a look on what the utility does.

 

AwdFlash

The most popular flasher - but not the best one. Should be used with the following switches:

AWDFLASH bios_file /py/sn/f/r/cd/cp

These switches will cause AwdFlash to do the following:

  • /py — reprogram the flash memory with the contents of bios_file without asking for confirmations (here p means Program, y means Yes);
  • /sn — do not save the old BIOS (Save=No);
  • /f — do not check if the BIOS in the bios_file file is intended for this particular motherboard and if this is a BIOS file at all (Force reprogramming);
  • /r — reboot the computer after flashing is complete (Reboot). If you don't need this — just don't use this switch;
  • /cd and /cp — clear some setup data. It is highly recommended to use these switches (Clear DMI data, Clear PNP data). It is also recommended to add one more switch — /cc, to automatically Clear CMOS.

AMIFlash

Not so popular as AwdFlash, but actually the best one (despite even the problems with some versions when being executed without any switches — see the warning above). It should be used with the following switches:

AMIFLASH bios_file /b /n /-o /-c /-d /-r /v /-i /-k /-e /-g

To backup your current BIOS, run AMIFlash as follows (NOTE: there is no SPACE character after the /s switch):

AMIFLASH /b /n /-o /-c /-d /-r /v /-i /-k /-e /-g /sbackup_file

If you're going to reprogram a flash chip(s) more than twice — don't be lazy and prepare a .BAT file on your DOS system disk(ette), with a string like this:

A:\UTILS\AMIFLASH.EXE %1 /b /n /-o /-c /-d /-r /v /-i /-k /-e /-g

UniFlash

UniFlash is a much more advanced flasher (compared to AwdFlash and AMIFlash) and requires some major knowledge/experience from the user, so it will not be described in this article.

Some details about flashing BIOS-es of ASUS motherboards

"Regular" flashers shouldn't be used for reprogramming flash chips on ASUS motherboards, because vast majority of ASUS BIOSes (starting from ancient Pentium1) are blocking the regular methods of reprogramming. For these motherboards, ASUS' own branded utilities must be used. For the oldest boards it would be PFlash, and for the newer ones it would be AFlash (in many cases AFLASH will work with older boards as well). For the modern boards you will need to use Windows-based flashing utilities.

It should be mentioned, that the statement "you will need to" in the preceding paragraph doesn't mean that ASUS' own flashing utilities are troublesome. Quite the contrary: they are very comfortable to use, easy to understand, quite universal, don't ask hundreds of questions and successfully flash "everything on their way". In addition, AMIFlash is still usable, despite the error at the end of the flashing process (it is actually caused by DOS4GW rather then by AMIFLASH).

In any case, all happy owners of ASUS motherboards should be extra careful while flashing their boards, as the resurrection process in case of the BIOS corruption may be difficult. And even if you think you are an advanced user with a great experience - once more: be careful! ASUS engineers use very complex and sometimes weird BIOS protection techniques in their masterpieces, and this “other side of the medal” can make the process of repairing a motherboard with corrupted BIOS very complicated. Your motherboard may seem to be dead ("00" on a POST card), as well as the flash memory chip (e.g., you may not be able to reflash it on a board other then ASUS), but actually both of them are in good, absolutely operational condition - the "dead" chip will work well if put into another, exactly similar, but working board, and/or the "dead" board will return to live with the flash chip from that working board. If you're dealing with problems like these - you can read the chapter about the LPC interface (ASUS uses this interface long ago compared to other companies who have started using it later).

P.S. Similar problems (necessity to use branded flashers) may appear with some (usually - well known) other brands, for example - a couple of Abit motherboards have "fine points" in the flashing procedure. UniFlash will warn you that it "does not like" this brand.

Common issues and their solutions

  • BIOS image file's size does not match the flash memory chip capacity.

It's a rather common problem, most likely caused by the fact that manufacturers sometimes use flash memory chips of different capacity in different motherboard revisions. The MS-6117 motherboard (made by MSI) may serve as an example: look at its BIOS list and you'll see that all AMI BIOS releases are for a 1Mbit flash chip, as well as AWARD BIOS releases 1.x and 2.x; however, the AWARD BIOS release 3.2 is for a 2Mbit flash chip.

Rarely, you may run on a board with a flash chip of some specific capacity, while the manufacturer claims that the board is/was produсed only with flash chips of different capacity. That may just mean that your board is after repair, but it may also mean that you have a "fake" motherboard (i.e. some company has cloned (copied) a design of another manufacturer and has used a cheaper flash memory chip of different capacity).

  • A flasher refuses to accept the given BIOS image file, producing error messages like that the given file is not a BIOS image at all, or it's a BIOS for a motherboard of different model.

It's an even more common issue, but if you're 100% sure that you know what you're doing, just use the switches described previously to force the flasher to do the job.

  • A flasher cannot recognize the used flash memory chip and thus cannot (re)program it.

This issue is also a common one, and there could be a few reasons for that:

  1. the flash memory write protection is turned on (that's what most flashers would ask you to check in this case). Old boards hase a jumper that enables/disables flash chip protection, while newer ones have it as an option in BIOS Setup (something like BIOS FLASH PROTECTION = ENABLE/DISABLE);
  2. the particular version of the flasher that you're using doesn't know the particular type of a flash memory chip. It's a rare occasion, but it happens, usually when you are trying to flash some stone-age flash chip by a newer version flasher, or vise versa.
  3. the flash memory chip is not supported by the motherboard (this may happen e.g. when you try to hot-swap flash chips) - read the chapter dedicated to different interfaces like FWH/LPC;
  4. the flash memory chip is inserted in the wrong way. Double check that you have inserted it correctly before you turn the motherboard on, otherwise you may end up with the next option ;) ;
  5. the flash memory chip is dead;
  6. the flash chip is not actually a flash chip, but an OTP (one-time programmable) ROM or an Ultra Violet EPROM (UVEPROM, with small transparent window). May apply only to stone-age motherboards (the ones before Pentium 1); it's easy to recognize those chips by the marks on them (they would go like 27xxx).
Everything went fine, no visible errors, you turn the board on and - silence...

For recovering in this case, you will need to read the article further...

CHAPTER TWO. Dedicated for those who hasn’t read the first chapter on time and now needs a recovery.

Considering that a person would read this chapter not to study about updating the BIOS of a working board but for recovering it after a failure, let’s go ahead and see what options do we have.

Option 1. Reprogramming the flash chip using a hardware programmer unit

Plain and simple - you will have to break your piggy-box, pull out your flash chip, take a floppy disk, bottle of beer (keg sounds better :) and go to one of your tech-buddies. If the flash chip is soldered in, then you will have to break another piggy-box and take the motherboard with you. In this case, consider asking the technitian to solder in a chipbed for the flash chip - this may lighten your future expenses.

Option 2. "Hot swap"

You will need another WORKING motherboard. As was mentioned before, in most cases it neither has to be a one with the same chipset, nor of the same brand. What is most important, indeed, is the type of flash chips that the working board is compatible with - there are two major types:

  • the "regular" ones - all flash chips in DIP32 ("rectangular") package are of this type. They are usually marked as xx28xxxx, xx29xxxx, sometimes xx39xxxx, xx49xxxx;
  • the ones with LPC and/or FWH interface (sometimes called hub chips) - vast majority of these chips are in "square" PLCC32 package. Most of those are marked as xx49xxxx and sometimes xx39xxxx, while Intel's chips of this type - like 82802xx.

So, if our recovery board uses the same (or compatible) type of BIOS flash chip as the dead one (and of same or larger capacity) - everything should work fine.

How-to

Use your imagination. The most important thing - during this process do not drop any ironware on the motherboard or spill any liquid. For chips in DIP32 – everything is easy. Usually it is recommended to tie the “native” chip with a fishing line or similar non-conducting thread for easier pulling it out of the chipbed before reprogramming. As for myself, I use a hoop ring belt (an old COM-port wiring, for example) and set it up under the chip so it would be comfortable and safe to pull the belt without cutting it and without damaging the chip and the chipbed.

.....................PICTURE SPOT......DIP32 PLIERS..................

As for the square chips aka PLCC32 it's a bit harder. One can try using fishing line as well or any non-conducting type of threads or if you have a good access point and your hands are straight - just use a needle, slowly lifting the chip from access corners in chipbed. If you, however, reprogram chips everyday or just want to be 100% sure - glue something like a piece of plastic or even melt some glue from the glue-gun to the top of the chip. This would allow you to pull it out. You can also just buy PLCC32 extractor. Anyway, it is your choice.

....................picture spot ....extractor pliers for PLCC32......

Option 3. Using a floppy disk (for modern motherboards – HDD or CD-ROM)

If the BIOS is not completely dead (during the boot-up, the board tries to access the floppy disk drive and/or displays a "BIOS CHECKSUM ERROR" message) - it may be possible to recover the BIOS without additional hardware.

Note: on most boards up to PentiumIII-class ones, the error message won't be displayed if you have an AGP/PCI video card. If the board has ISA slot(s), try using some ancient ISA video card; for boards not having such slot - try using a PCI video card. As of many modern boards, the bootblock (which actually produces and displays this error message) supports AGP video cards.
In addition: if you're using a POST card (or the motherboard has a built-in one), the following POST codes may help you to determine that the bootblock tries to recover the BIOS from a floppy:
  • for Award BIOS: POST code 41h shown in the beginning of the POST procedure (something about the 5th-6th POST code of those you could see on the POST card). In Award BIOS this code corresponds to FDD controller and FDD(s) initialization.
  • for AMI BIOS: POST codes F0h-FEh will mean that AMI BIOS attempts to boot from a floppy drive.

Mostly all you need is:

  • a "minimal" bootable diskette (i.e.: only IO.SYS, MSDOS.SYS, COMMAND.COM);
  • a flasher (e.g. AWDFLASH.EXE)
  • of course, the BIOS image file (in many cases, must be (re)named to bios.bin) that you'll have to get (from the board's manufacturer's website or from somewhere else)
  • AUTOEXEC.BAT with appropriate string inside:
  • for AwdFlash:
awdflash bios.bin /cd/cp/py/sn/f/r
  • for AMIFlash:
amiflash amibios.rom /b /n /-o /-c /-d /-r /v /-i /-k /-e /-g

It is not recommended to use other names for the BIOS image file and/or flashers (use only the "standard" names mentioned above) - some BIOS-es (basically, their bootblocks) may just ignore them, producing an error message like NON SYSTEM DISK or DISK DRIVE NOT READY.

There is no "standard" procedure for BIOS recovery from a floppy drive. A plenty of brand-name computers made by Intel, HP, Compaq, Dell, IBM, Fujitsu and some others could be restored using a special floppy disk - its image is usually availible for download from their website along with detailed descriptions of how to use it.

It is also possible that you will need just a flasher program and the BIOS image file (the diskette should not even contain any system files), or even only the BIOS image file itself, because both the flasher and the BIOS would be built into that file. Usually, this may be applicable to newer boards.

To be short - there may be some unexpected variations. Taking into account a possibility of hardware and/or bootblock corruption, this method (using a floppy drive) is rather not reliable and promising (and statistics prove it), but it's the best method to start BIOS recovery with - who knows, maybe you'll be the lucky one!

CHAPTER 3. Dedicated for those who want to know not only "how to do", but also "how and why it works"

PP-type Flash ROM chips interface (DIP32 and PLCC32 packages).
PP-type Flash ROM chips interface (DIP32 and PLCC32 packages).

A little about operating principles

The "regular" flash chip operates with the following signals:

  • DQ0-DQ7 - 8 data lines;
  • A00-A16/A17 - 17 or 18 address lines (depends on the chip's capacity: 1Mbit chips have 17 lines, 2Mbit ones have 18 lines);
  • CE# - "chip enable" (or "chip select"); this signal enables the chip to be read/written (think of it as a "turn on" button);
  • OE# - "output enable" (or RE - "read enable"); this signal allows to read a data byte.
  • WE# - "write enable"; allows to write a data byte.

It is quite simple to understand even when you don't have a degree in electronics.

Now we can see that a 8-bit data bus seriously limits the read speed from the chip (you can think of it as a road with 8 lanes, while e.g. the nForce chips access the RAM through a "highway" of 128 lanes!), so the startup process of even the fastest computer could become a turtle run. To avoid that, the chip is connected to the south bridge in a special way, allowing the latter to shadow the contents of the flash to RAM, thus improving the speed dramatically.

However, if the flash chip you use is (partially or totally) damaged, or is inserted incorrectly - the consequences may be bad not only for the chip itself (which could be expected), but for the whole chipset - that's due to the direct connection between the chip and the south bridge.

The flash chip occupies upper addresses E0000-FFFFF of the address space (True for 1Mbit) and because of similarity to other chips let's take 1Mbit/AwardBIOS for example. After you turn computer on and RESET# signal disappears, the CPU executes the first command located by F000:FFF0 address. After initialization process beginning, CPU recalculates different checksums (there are quite a bit of those in BIOS) but the main is at E000:0-FFFF+F000:0-BFFC. If this one is wrong, then bootblock located in *most* upper addresses (F000:C000/E000-FFF -the last 8/16Kb) is executed trying to initialize floppy drive and to read the system and/or flasher and/or BIOS file from floppy disk in attempt to recover/boot normally.

And yet, another conclusion: if you want to force boot block execution (for reflashing purposes, as an example) you can do it purposely corrupting CRC (checksum) by shortening address lines above bootblock's. As for myself, I usually shorten bylocated A15 and A16 lines (pin 2 and 3). Bootblock size=16Kb=2^14 and so this would not lead to bootblock corruption, but checksum of the *rest* will be corrupted. If you can not understand why you need that, you will when you meet soldered in flash chip fashed with *similar* bios and, by the way, such shortening is harmless for address lines and chip itself.

After successful CRC check, bios is shadowed i.e. the flash contents is copied to the ram and flash chip is turned off by south bridge. To force all programs working with shadowed bios as with ROM contents it is marked with READ ONLY atribute by means of programming south bridge SB (and/or Model Specific Registers MSR's for AMD K7/K8 CPUs). All foregoing bios calls are redirected to the shadowed copy of the bios.

Conclusion: So you can extract the flash chip just after the report "Starting Windows/DOS/linux" on the screen. And now if you wish to work with the flash itself (for example write ESCD/DMI), you have to fulfil the special procedures (chipset specific !). You have to connect the flash to the bus before the operation and disconnect after.
Conclusion: because of different methods of turning flash chips ON/OFF (for different chipsets), flasher should know how to do it i.e. flasher has to *know* chipset of the system where flashing is performing.

PARTICULARITIES OF FLASHING INTEL 28FXXX CHIPS

If your motherboard has an INTEL flash chip you have to be triple careful while flashing it, because intel chips use hardware botblock protection. What does it mean ? This means that intel chips have one more signal RP# (pin 30), and this signal allows chip bootblock programming. For 2Mbit it is A17, 1Mbit -NC (not connected, not used) i.e. for correct flashing of _all_memory_space_, there should be +12Volt present on RP# pin. The special jumpers were made on many Socket7 motherboards and named: INTEL/NON-INTEL for this purpose.

Conclusion: If the particular motheroard has Intel 28FXXX flash chip and PCB (Printed Circuit Board) does not support correct commutation for programming those chips (a common matter) - BIOS will die after the first attempt to flash the chip without any chances for revival (on this board).

For flashing Intel 28FXXX on such boards (without RP# pin) you have to connect pin30 to +12Volt source manually, but so it would not get to the motherboard itself (A17). To avoid this you may just bend this pin so it will not have a connection with chipbed.

SOFTWARE PARTICULARITIES OF FLASHING PROCESS

The main aspect you have to know is that you can not reprogram this flash chip *at once*. You can not also write one separate byte. This type of chips supports blocks (sectorization) reprograming methods only. Different type of flash chips support different type of sector organization. For example, a very popular flash Intel 28F001BX has the following blocks:

  1. First block, in the very beginning and the biggest one -*main* block, occupies 112Kb of space, it usually contains main body of the bios (that is why it is the biggest).
  2. Next are two identical blocks, each occupies 4Kb space – the data related toESCD/DMI/CMOS/PASSWORDS etc. located usually in these blocks.
  3. Last block - 8Kb -bootblock

Speaking technicaly - i28F001BX has 112+4*2+8 organization. Simple and logical: the biggest block, 112Kb, anyways being rewritten only during bios renewal process, the next follows changing data area, thats why there are two small blocks used. They can be rewritten faster and easier. And the non-changeable bootblock purposely was separated from other blocks for bios protection in case of emergency recovery from floppy drive. Bootblock is always separated from oher blocks so it could peacefully reside during normal updates of the other blocks.

Some flash chips are made with very high sectorization, for example: SST29EE020 (2Mbit) has 2048 similar sectors each occupying 128 bytes, but most part of chips has organization similar to Intel one, because smaller blocks have drawbacks as for example quite complicated write protection realization.

CHAPTER 4

HUB TYPE FLASH CHIPS

Old school...How simple it was...address bus, data bus CE/WE/OE...But the progress is not standing on the same spot (especially In certaIn. companies) and with appearing of the i8XX chipsets, there came so called hub flash chips.

LPC Interface

The end of year 1997. Intel decided that powerful processor must have BIG bios (4-8Mbit and more) eventhough, many that time companies were OK with what they had and present companies still use 2Mbit.

  • Regular* (PLCC32), widely used flash type chips, would not have enough pins for address lines, so Intel invented LPC standart (Low Pin Count) but decided to keep PLCC body. Besides possibility to have huge flash chip capacieties - up to 4Gb, one of the other main reasons to have LPC, was soon-to-be-useless ISA bus, absense of which would remove address/data lines and allowed to have less traces in between Super I/O chip and the south bridge (obvious possibility to hide SIO in to the South Bridge were kept in secret).

LPC chips are using only 5 lines: LAD0-LAD3 + LFRAME# (thats why it is called LOW PIN COUNT) See typical LPC flash chip pinout at picture below (this standart may be used not only in flash chips, thats why iam being specific)

LPC/FWH interface
LPC/FWH interface

As you can see this type of flash can work in 2 different modes: LPC and Parallel Programming (PP). Mode is chosen by logic level on IC pin (Interface Configuration Pin) during computer startup process as follows: logic "0" -LPC mode and logic "1" -PP mode.

PP mode is the closest to *regular* flash chips and somewhat compatible with them, but not completely by all pins, so be aware! PP mode uses 11 (0-10) address lines, but those are multiplexed i.e. first, addresses A00-A10 are read and then, through the same lines, addresses A11-A21 are read as well. This multiplexing thing and dissappearance of CE# signal is the only difference between PP and *regular* flash chips.

As of LPC mode, communication between flash chip and south bridge is made through 4 bit bus and fifth LFRAME# needed as a start point for data to travel. LRESET# and LCLK# which are identical to their cousin signals used on motherboards all other lines are optional. This is how interface got simlpyfied, but, its still not serial interface as I2C and still allowing it to use DMA and BUS MASTERING (Which are used in devices like DiskOnChip/LPC)

PP mode is used during manufacturer flashing process of the chip (because it is faster), LPC is used for communication when already set in motherboard, so chipset must know how to communicate this way, thats why this standard got so popular after appearance of Nforce2 (ISA bus was never imlemented for use in this chipset)

FWH INTERFACE

Intel was very pleased with own work and decided to carry the flag forward and made another standart - FWH (Firm Ware Hub, thats how *HUB* type of flash chips appeared). It was fully electrically and by pin number compatible, but _absolutely_ incompatible by communication protocol. As you can read in many papers it was made in purpose, so those chips can be set up in one system, on the same bus (electrically compatible) but during communication each chip would take own data packets, because of the difference in communication protocols. by the way it is possible to hook up to 16 *HUB* type flash chips in one system, because of additional ID0-ID3 signals.

(by Author: Please think of all my stones towards the Father of x86 as a joke, however every joke has a piece of joke ;0)

As you can see FWH is similar to LPC, only pins LAD0-3 + LFRAME# renamed to FWH0-3 + FWH4. PP mode is absolutely same.

FWH differs from LPC by protocol, i.e. all commands for operation have different code values in the field of data exchange, plus, the ability to use up to 16 FWH devices, thanks to newly implemented identification signals ID0-3. As of system default ID configuration for flash chip it is assigned ID=0000, you may easily check it (intel boards) with multimeter (ohms), all those pins are connected to the ground.

Common way to check what kind of (FWH/LPC) flash chip we have is: using Ohm meter, check continuity between PINs 16/28 (flash chip) and GND (anywhere on the board) in *HUB* type flash chip resistance will be 0 Ohms (or close to it) while *regular* will have about 1KOhms.

 

If you have the need of flashing chips every-so-often, at work, or for any other reasons, i would have a few suggestions for you.

Someone may recommend you, in this case, to use a flash chip programmer device, but personally myself i would not advice that and not even recommend to use one. There are two major reasons. Reason one is : many programmer device will not recognize all flash chip types and those which do, cost well over three digit numbers. Second reason would be: just any programmer device cost would be, as low as, a three digit number, that being said here is my solution - all you may need to flash 99% of flash chips (talking about the chips that are used in motherboards) are next three type of motherboards:

  1. Something out of stoneage, like P1 (pentium) - for flashing well spreaded 1Mbit Intel chips
  2. Something out of PII (Pentium 2) age with 2Mbit BIOS chip - for flashing *regular* 1Mbit and 2Mbit flash chips.
  3. As you already guessed, something out of P!!! (Pentium 3) for example, any motherboard with i8xx chipset for flashing *HUB* type chips.
  4. (added for last version of edited article) Something, that is built on nForce1/2/3 chipset.

As of myself, towards the first point, I use ASUS motherboard built on i430HX chipset - P55T2P4, for cases like this one i recommend ASUS, because of very usable asus made flasher program, an old PFLASH or newer version - AFLASH, very simple to use, stable and may show you memory address where error occured, in case you will have an error of course.

Talking about second point, I use motherboard built on i440BX, PC Partner-928.It has 2Mbit flash chip, which allows us to flash *regular* chips with 1 and 2Mbit capacities.It has a DIP32 bed, for flashing PLCC32, I use PLCC32<->DIP32 adapter, which was made out of old non working board with PLCC32 bed (see picture)

Hand-made PLCC32->DIP32 adapter (made by a figured groove by a fret saw on faulty motherboards;) plus manually soldered pins)
Hand-made PLCC32->DIP32
adapter (made by a figured groove by a fret saw on faulty motherboards;) plus manually soldered pins)
Hand-made PLCC32->DIP32 adapter - bottom view
Hand-made PLCC32->DIP32 adapter - bottom view

As of point 3, I use motherboard built on i820 - Chaintech 6CTA2, it has i82802AB (4Mbit), that allows us to flash intel *hub* chips including HUBs themselves.

Point 4. I use motherboard on nForce1 - ABIT NV7-133R (it is not a big deal if it be NF2 /NF3).This board is used for heavyest cases, for flashing *PURE LPC* chips.Many mobo manufacturers, if not all of them, that built boards on nForce1/2/3/etc chipsets are using chips that are compatible with both LPC and FWH protocols, which allows us to flash those on equipment mentioned in point#3

Talking overall, about facts towards motherboard choices, there are no such thing, so it is solely depends on your personal needs, but i will mention the case with an ASUS for Intel flash chips combination.

It is not necessarily has to be some kind of advanced board, it is better to act counterwise - trying to find cheapest and simple boards, even so called No-Name boards are good, Acorp or PC-Partner are just a couple of such. Other important fact is easy access to flash chip bed.

As of the software I use for major flashing purpose - AMIFLASH (v.8.37 with keys, of course) Flasher knows many flash chips/chipsets and very friendly to use (I prefer command prompt).

For rare cases, when AMIFLASH cannot handle the task (some kind of exotic chips/bioses/chipsets)- I recommend to use AWDFLASH with /f key.

The heavyest case is when we need to reprogram very exotic/rare chips/bioses/chipsets, either very old or newest boards) I recommend to use UNIFLASH, it can be used for *visual* check of an error spot in memory space. Happy owners of gigabyte/epox or other mobo that has nForce2 Chipset with LPC flash chips like PMC.

PMC's are compatible with all three protocols LPC/FWH/PP and that is the probable cause of an error in case of *regular* flasher usage (those may even just hang/lock the PC), however UNIFLASH works just fine.

p.s. Latest versions of Uniflash are so knowledgeable and stable that i use it in most cases now.

Some extreme flash/reprogram methods.

There are cases when you may need to use two flashers to flash one chip, for example you reprogram chip with amiflash, at the end it may give you an error, then, you will reprogram it rightaway, again with uniflash, but you should not use it counterwise, because you will fail, its more likely will not work other way.

Probable cause - are programming algorithms of particular chips, when one flasher will be able to accomplish first part and other one will finish the job (by the way, this is not insanes speach, but my personal experience with amd chips, for example, that do have some non-standart read/write protection or damaged or non stable memory area.)

If you have a board that has only 1Mbit chip, but you need to reprogram 2Mbit chip - you may do it in two sets, first megabit last and second megabit first, as separate processes, when your first part is done, you will need to short PIN#30 (A17 for 2Mbit and PIN#NC in 1Mbit) to the ground and then flash second part.You will need to split file in two pieces, 1Mbit each (1Mbit=128kb), you may do it with any HEX editor, the only limitation for this process will be no support from flash chip for sector writing.

This situation may evolve in to case when we need only a bootblock. the second part (end part) of the splitted file. After this will be done, you need to insert this chip in to motherboard and start the pc, because of unfinished flashing, chip will have bad crc and bootblock will be loaded for recovery purposes, where you can use a floppy disk with bios file.

FLASHING CHIPS WITH DIFFERENT FLASHING VOLTAGES.

Different chips require different programming voltages. Most popular ones are 12V/5V for older boards and 3,3V/3V/2,7V and lower for modern. According to the theory, you should set onboard jumpers into appropriate position (if there are such junpers) to accomplish reprogramming. Practically we see that the chips designed for lower voltages can be reprogrammed with higher voltages as well.

For example, if the motherboard has 5V chip (usually 29xxxx series), you can flash the chips designed for lower voltages on this board and you will succeed.

But in the case when you try to flash 12V chip in 5V motherboard (or with jumpers set to 5V), you may succeed and may not, thence when you reprogram 28xxxx series chips (most of them are 12V ones) you have to pay attention to the jumpers and don't try to reprogram those using HOT-SWAP method on newer boards (12V chips were mostly used in P1/P2 period). So, the older motherboard you have, the bigger chance to succeed.

If the motherboard jumpers have no marks, you can use multimeter to see the pin32 voltage (the first one by the right side from the key, when looking to the chip's top). You also can bend or isolate this pin on the board/socket and connect the chip's pin right to 12V manually (from the power supply).

 

p.s. Original version of article in russian. (11.26.2003)

Ленты новостей