Tzzantaru wrote on 2025-05-30, 18:41:

Hello again,

So, good news and bad news.

Good news is that I found someone that was kind enough to help me write the new chips. Bad news is that the motherboard is still giving me the cold shoulder.

Hi Tzzantaru,

The plot thickens.

I also asked the person to make a backup of the original chip and also the new chip #1 (that I've tried to blindly write the bios on the board itself by hot swap). I attached both backups, maybe they'll give any kind of idea, although I doubt it.

Well, these BIOS dumps actually revealed a lot!

To start with the backup of the original chip, at first glance the BIOS looked completely scrambled but the bootblock, including the decompression engine, was intact. At closer inspection I found 16KB blocks that held a correct section of a compressed module, but at the wrong place. Other 16KB blocks appeared to be cleared/empty.
Checking the backup from the hotflash, the scrambling was even worse.

The backup of the original flashchip represents the state of the BIOS after your last blind bootblock recovery attempt.
On a 128KB Award BIOS like yours, the start of the compressed main BIOS module (original.tmp) should be at offset 0000h in the BIOS file. However, I found the first 16KB of this module at offset 4000h!! So I’m beginning to think you have a stuck address-line on your board!

This would also explain the lack of video from your ISA card, because when the bootblock recovery code try’s to initialize the video card by calling its initialization routine, it first checks for the AA55h Video ROM signature at address C000:0000h. If it can’t find this signature at this address, it sends out a beep code and skips the video init.
With a stuck address line, this check will always fail because the video ROM presence check is send to the wrong address!
A stuck address line would also prevent a correctly programmed BIOS from functioning.

That the whole bootblock was still intact can also be explained by the fact that Awdflash.exe normally keeps this section of the flashchip write protected. Only when it detects that the bootblock needs to be updated as well, it will write to this flashchip section.

So, where to look for a suspect address line? Although the BIOS ROM and ISA bus are connected to the Southbridge, the RAM is connect to the Northbridge. My bet would be a faulty connection or shortcircuit near or in the RAM slots, or a damaged trace to the northbridge.

I hope this helps you further,
Jan

Hi again,

Maybe I’m double guessing myself, but after my previous message I’ve been looking further into this possible stuck address line issue.

Seeing that the Southbridge controls the BIOS ROM, ISA bus, and IDE ports from a common address and data bus (probably via buffers), I’ve been hypothesizing if the stuck or floating address line could be here.
The Southbridge is connected to the Northbridge via the PCI bus, so their address and data lines are physically separated. Although a stuck address line on the Northbridge’s memory bus could still be translated over the PCI bus to the Southbridge, I now think it is more likely that the problem is actually on the Southbridge’s address bus to the BIOS and ISA bus.

Because I found a 16KB shift in addressing a section of the BIOS, the most suspect address line is A14. This is pin 21 on the VT82C586B, pin 29 on the BIOS chip, and pin A17 on the ISA bus https://en.wikipedia.org/wiki/Industry_Standard_Architecture
Maybe you can find a broken connection or shortcircuit with another line here.

Note that address line A14 must be high when accessing the top 16KB of the BIOS. This is where the bootblock is so addressing this part of the BIOS is not affected if A14 is stuck high.

I’m curious what you will find.
Cheers, Jan

Hello Jan

Thank you for the reply.

I have read both posts and I'm now trying to understand everything and make some notes of the things I need to check. First of all I'm not really that competent in this department but I'm trying to research and read and educate myself on the matter.
Fortunately I have the vast internet at my disposal and lately Chat GPT does an awesome job of explaining things 😀)
Unfortunately I only own a multimeter and though I know my way around it to check stuff, apparently I also need an oscilloscope for what I need to verify going forward (or at least find someone with one and lend it 😁).

I did have a little time at work today to make a list of address lines between the bios chip and northbridge and/or southbridge and/or ISA slot but since I need an oscilloscope for the majority of tests, this is going to take a while.

One thing I'm trying to understand now. The first new chip that I've tried to hot swap and flash was empty but somehow the board was able to write the bootblock part perfectly and then mess the rest of the bios files. Why?
I'm still trying to figure out what handles the write process for the whole flashing. What manages the file transfer between the floppy and the chip?

Also, I'm 99% sure that there is no physical damage to the board/traces as I've always took care of the components. Never scratched the board, never hit it with anything, never dropped anything on it. I was always careful when removing it from the case and putting it in storage. There are no caps that leaked so there's no damage from that.
I never damaged any board that I handled so my intuition would say that the problem is a component that's just given up due to age. Problem is I need to find out which one and hope that it's not the northbridge/southbridge as I don't think I could find a replacement for them and it's going to cost a truckload of money to replace.

I appreciate your answers and advices

Thank you

Hi,
Apologizes for resurrecting/hijacking this thread, but I have the same motherboard, it has been sitting dead in my closet for a good 23 years. Ironically it died while I was trying to install Windows ME. I've been meaning to try fixing the configuration, already have a working identical replacement motherboard. But reading this thread made me hopeful that maybe the original board could be fixed. I'm partial to it since it was the first PC we owned back in 1998. Also it supports a ridiculous amounts of stuff, even has onboard USB (not keyed, and ~25 years ago it managed to fry my USB MP3 player, an expensive problem at the time. much to my dismay, the USB standard only became worse since then).

The bad board has the usual no screen on boot problem. Already tried with another identical CPU and got the same result. I've dumped the BIOS and attempted to run it in x86box by picking a similar config (FIC VA-503+) and replacing the BIOS with ones belonging to the P5MVP3. With the BIOS files downloaded from Retro Web, x86box runs fine. With the BIOS I dumped, it just shows no screen. Additionally I moved the bad BIOS to the working board, and it also shows no screen. So problem #1 is that the BIOS is probably corrupted. I've attached it here, in hopes that some guru can take a look at it: we owned the board since 1998 December, so it may be an undumped version predating the REV A on Retro Web. It does not match any of the files I found online, I suspect because it's corrupt. The chip is a MX28F1000P.

Anyhow, I flashed the Rev D (the latest) BIOS to the chip, verified it, plugged it to the good board. It works! Counts the memory etc, then before the no boot disc error, it says "Unknown Flash Type". Other than that it works fine. Both boards are set to +12V Flash ROM Voltage (JP20). Maybe this BIOS just doesn't know that chip, I'll try using REV A on it later, but this is a minor issue because the board works. Odd that a newer version of the board would not know this chip type.
The good board has a Winbond chip, I suspect a W29EE011P, I did not pull the cover sticker off but I can see the top of the logo. My programmer could partially dump it and it matched the REV A chip, but I could not make a full dump because the programmer keeps locking up (first time ever it locked up like this).

Okay, so now I have a known good BIOS. Plugging it back into the bad board... no screen still.
Upon visual inspection, the only fault I see is a slightly discolored part on the bottom of the board, right under the KV1083 chip. Sure enough it's scorching hot to the touch when the board is powered on. On the good board, it's completely cool. So possibly the KV1083 is the problem. However, I measured the input/output voltages and they are an almost complete match on both boards: 2V, 3V, 5V on the good board, 2V, 3.3V, 5V on the scorching hot bad board. So maybe it's not the fault. It should be noted that bad board is running an AMD K6-2 450 and the good one a Pentium 166 (A80502166 SY016), so possibly the KV1083 is cool as a cucumber because under that chip it doesn't have as much to do, just a guess.

I'll try swapping over the KV1083 to see if it works with that, then write up which caps the board uses and see about sourcing replacements. The two ones near the KV1083 (1000uf) are already ones I don't have in stock at home.

The PC Speaker (beeper) is entirely silent on the bad board.

So, questions:
- about the P5MVP3 bios I uploaded, is there anything salvageable or interesting in it?
- is the "Unknown flash type" error something to be concerned about?
- what's a modern drop-in replacement for the KV1083 in case it's the broken part?
- if swapping the KV1083 won't fix the board, how should I continue debugging? I'm only really armed with a multimeter and soldering iron.
- can the Pentium 166 run without a heatsink? It gets scorching hot on touch but works for the less than minute I have it running.

Also what's with Super Socket 7 board prices, I've been looking online and they go ~75€ for untested boards and from 100€ up to 400€ for known working ones. WTF?

> "Unknown Flash Type"

just annoying text and slightly slower boot, related to "Updating ESCD" "Unknown Flash Type" message right before DOS boots (solved)

>swapping the KV1083 won't fix the board, 2V, 3.3V, 5V on the scorching hot bad board

yes it wont, its hot because AMD K6-2 450

>Pentium 166 run without a heatsink

for maybe a minute before it crashes, leave it for longer in that state and it will eventually fry

>caps

might actually make some sense replacing if they lived their lives near that hot KV1083. Wont fix the board, but will make supply more stable when its actually fixed and running

POST card would be helpful in diagnosing, or a scope/logic analyzer

Thank you kindly. I ordered one of those cheap chinese PCI testers, may take 1-2 weeks to arrive. Until then I guess I'll just set up the good board with the K6-2 and see if I can figure out how to install Win98 (it's been a while and my spindle of blank CDs may have chemically decomposed since I last touched them). I might even have to pre-install Win98 on x86box and then copy the VHD over to my IDE drive, since I have an adapter for that.

Hi rorirub,

> about the P5MVP3 bios I uploaded, is there anything salvageable or interesting in it?

I have been analyzing this LT-P5MVP3 BIOS you uploaded, and it is corrupted for sure! 🙁

The upper 20KB of this 128KB BIOS is correct. This is where the Bootblock and decompression engine are located.
The remaining 108KB is the space for the compressed BIOS modules. All three required modules are present, however the main BIOS module (original.tmp) should always decompress to 128KB but in this BIOS it is only 120KB. The top 8 KB are missing, and as this is the area where the BIOS runtime code is, with all its interrupt service routines, this BIOS will never work…
In addition, the checksum byte of the compressed main BIOS module is incorrect.

This missing 8KB is also the area where the BIOS sign-on message and BIOS-ID are located. So I have no idea what BIOS version this is and if it is even meant for your board.

How this corruption inside a compressed module could have happened is anybody’s guess.
I doubt that even the 1998 CIH virus could have done that.
https://en.wikipedia.org/wiki/CIH_(computer_virus)
But who knows, you may have had a virus infection in your Windows Me installation files.

I hope you can revive this nice SS7 board.
Cheers, Jan

Thank you for the analysis. Since it is corrupt there's no telling if it was some other revision or not. Given the dates, I was hoping it could be some earlier, undumped BIOS but I guess it does not matter now.
I've tinkered with the jumpers today, found out the bad board was set to run at 66MHz AGP clock. Set it back to 100MHz SDRAM clock, but sadly it made no difference, board still does not run. I suspect something just broke on the board and when it did, it took the BIOS with it. It's not the first time I see a flash chip get corrupted on its own. Mental note: still need to repair my console save carts by soldering a new flash chip on them.

Today I put the K6-2 in the good board, and it booted immediately, so that's a good sign. The cooling fan is dead, so I have to run it passively. Set it down to 2.0V and 2.5x 100MHz for now (it's a 2.2V 450MHz chip), it seems to boot fine, heatsink runs hot but not third-degree-burns-hot, and it doesn't seem to boot the CD I tested it with. Tried every boot order in the BIOS, two different CD drives, three different boot discs (freshly burned 98SE disc, known good XP installer, known good Memtest CD that I used last month). Nothing will budge. The system doesn't even say boot disc failure, it just sits at the PCI device listing.

I remember this machine having issues with CD boot, but mostly because I didn't have a self booting 98SE disc at the time and had to use a boot floppy with MSCDEX and then launch any discs. I recall it used to boot properly bootable CDs but I could be wrong... nevertheless, it doesn't even seem to touch the CD drive right now. I can hear the discs spinning up on boot, but they are not accessed at all.
It's going to suck if this board can't boot from CDs since I have zero floppy disks any more, nor any means to copy anything on them (well, I do have a working 486DX but no idea how I'd make a bootable floppy from there, and also, no floppy disk to use).

Maybe I could install Win98SE in x86box and then write out the VHD or whatever to an IDE drive (if I can find anything that can do that).
Also the floppy disk led is continuously lit up. I thought it should only light up when accessing.

EDIT: on the up side, I changed the FLASH ROM Voltage Select jumper from 5V to 12V and the board is no longer complaining about "Unknown Flash Type", and can update the ESCD as well. But it still can't boot from CDROM with the Rev D BIOS either (the only difference I see is the system config screen says the CDROM is ATA 33, the REV A BIOS said UDMA2 instead).

Well, damn. The "good" board requires onboard cache to be disabled to be able to boot. If I leave cache on, it hangs after the system info screen. If I set the motherboard jumpers to Interleaved Cache, it hangs after POST instead.

I guess in the end only one of the boards will run, since I'll have to use the other as a parts donor.

bad L2 might be just one of the cache chips U15 U16 legs loose from age and proximity to CPU - clamping cpu cooler can bend the board slightly when done in situ and that can crack joints on cache