VOGONS


486 motherboard won't post - DRAM config?

Topic actions

Reply 20 of 34, by evasive

User metadata
Rank Oldbie
Rank
Oldbie

No question is too much, no post in this thread was unnecessary. Fellow Vogons members will benefit from this. BTW, this uniflash is what we talk about:
https://www.rainbow-software.org/uniflash/

Reply 21 of 34, by dominik0801

User metadata
Rank Newbie
Rank
Newbie

Well I got DOS up and running with UniFlash. Weirdly, it can't detect my Realtek 8139 card, but it detects another D-Link card, which actually is better, since it has the 28dip socket rather than the 32pin one.
The problem is, that it can't recognize the BIOS chip, and I couldn't force it, since it's not in the chip list..

As I already mentioned, my TNT2 also uses the 27C512 chip, and uniflash also recognizes my GFX card, but won't recognize the card's BIOS either.

Should I search for a compatible chip that's also 64KBx8, then try to "alter" uniflash to think I use that chip? I mean, with flashrom I successfully flashed "mx25l12873f" chips forced as "mx25l12835f", because the first type is not in the compatibility list yet.

Reply 22 of 34, by evasive

User metadata
Rank Oldbie
Rank
Oldbie

hmmm good question. It is ofcourse still possible the chip is actually defective.

Reply 23 of 34, by dominik0801

User metadata
Rank Newbie
Rank
Newbie

I forced to read it as a M28F512 12V chip. I then saved the BIOS image to a file. I did the same with my TNT2 and saved it's BIOS aswell, also forced it to be read as the mentioned chip.
I then tried to write (more like... verify, as this is an OTP chip) the TNT2 BIOS with the motherboard's BIOS, and of course it fully failed, since (imo) nothing should be the same.

I then tried to verify a BIOS image I found on this page for my motherboard. The first half failed, a little part succeeded, then it failed again.

When trying to verify the same BIOS image I just saved from this chip, it fully succeeds.

I have a feeling that the chip could be working, and has some data on it, as it can "compare" the images with it's content. The fact that the little part succeeds when comparing the BIOS content with the BIOS image I found here, makes me believe that the chip is either fine, but has a slightly different BIOS version on it, or the chip's content is partly corrupted and party fine.

Of course this is only what I think, I might be completely wrong here, and everything I said could be BS. I will try to take a look in the saved Image file with a BIOS viewer, if possible..

Reply 24 of 34, by evasive

User metadata
Rank Oldbie
Rank
Oldbie

You can also attach it here for inspection. I think I have a pretty complete set of bios tools.

Reply 25 of 34, by dominik0801

User metadata
Rank Newbie
Rank
Newbie
evasive wrote on 2020-05-22, 05:52:

You can also attach it here for inspection. I think I have a pretty complete set of bios tools.

Okay. So, first of all, here is the BIOS I found on this forum and I would like to use:
download/file.php?id=78936
I also found another in this post, but I didn't try that one yet.. which is this:
download/file.php?id=79697

Here is the one I saved from the motherboard's BIOS chip:
http://dhosting.ddns.net/bios/saved.bin
Can you tell if it's corrupted, or in fact a working BIOS?

Also, there is a possibility that the way I tried to save the BIOS image is totally BS and in fact not working. So here is the image saved from my TNT2 M64 16MB, just if it helps something.. like, compare it to another BIOS file..
http://dhosting.ddns.net/bios/tnt2.bin

Also, I'm using a 21041-PB chip based D-Link DE-530CT network card to read chips.

Reply 26 of 34, by evasive

User metadata
Rank Oldbie
Rank
Oldbie

tnt2.bin is completely empty, so indeed, that didn't work.

saved.bin gives: 09/26/94-SIS-85C471B/E/G-2C4I9S23-00 and a checksum error in modbin. Structure looks quite different from the other two bios files that actually contain data. I also see blocks being duplicated as if the same section was written multiple times at different addresses. Question is, was this caused by forcing to read a different chip type or is it actually f-ed up. No matter how, it doesn't look healthy.

These two are virtually the same, the only difference maybe the saving time.
soyo 25k2.bin gives: 09/26/94-SIS-85C471B/E/G-2C4I9S23-00
bios.bin gives: 09/26/94-SIS-85C471B/E/G-2C4I9S23-00

About the binary checksum error, it is perfectly possible my version of modbin is a bit too new for these bioses. I'll have to dig up some even older tooling... You may want to get another eprom and a programmer.

Reply 27 of 34, by dominik0801

User metadata
Rank Newbie
Rank
Newbie

I saved the chip's content with 3 different chip force settings, and then compared the contents of them with total commander (is this a really dumb way? 😁 ), and all three got the same content. So it's either that every different forced chip messes the chips content up when saving it, OR the chips content actually bad.

I just asked my friend if they have an EPROM programmer at their job, and if yes, could they give me a pre-programmed chip with the contents I give them (as he works at a PCB assembler factory). I hope they can help me, so I don't have to spend that much of money.

I will update this thread, but until then, I happily read every new ideas or anything.

Reply 28 of 34, by evasive

User metadata
Rank Oldbie
Rank
Oldbie

comparing content with total commander (if done with setting binary) will compare the files bit for bit so that is a valid way to do so. All three methods giving the same result sounds like indeed the chip was reprogrammed wrongly. We will have the answer once the pre-programmed eprom from your friend comes back.

Reply 29 of 34, by dominik0801

User metadata
Rank Newbie
Rank
Newbie

Hello there, again. Don't worry, I didn't forget about this thread.

Today I got a new chip, pre-programmed with the BIOS that I found on this forum earlier. They verified the chip, and was told it was ready to go. After putting it in the motherboard however, still nothing happens - everything remained as before. At this point I feel like i have tried everything and there's nothing left to do, other than let the motherboard rest.

The only thing I could think of as a "fixable" problem, that the motherboard currently has no CMOS battery at all, but I think, without a battery, it should still POST or BEEP, but it still won't do anything.

If you would have some ideas, I would still like to hear them, but if you have ran out of them too.. that's it for the board, and thanks to everyone who tried to help.

Reply 30 of 34, by quicknick

User metadata
Rank Oldbie
Rank
Oldbie

It is said that there are boards that won't boot without a CMOS battery. You should try fitting one.

Also get a POST diagnostics card, the ISA/PCI one is good enough and very cheap.

In most cases, one single broken trace or via will make the board appear completely dead, so you should double-check all that you have fixed and all that look suspicious. Thin traces can have barely visible breaks due to the battery corrosion. Maybe post some close-up photos with the area, more pairs of eyes have a better chance of discovering what's wrong.

Reply 31 of 34, by mkarcher

User metadata
Rank l33t
Rank
l33t
mpe wrote on 2020-05-14, 18:44:
dominik0801 wrote on 2020-05-14, 18:12:

Then I started measuring voltages, and noticed, that the -5V line has -6.80V coming out of the PSU.. so I switched to another, which had the right output voltage, but that just didn't help.

-5V level is unlikely to cause any problems as it is only used by less common expansion cards. But might be a sign that the PSU isn't healthy.

"high" -5V level likely is nothing to be concerned about. In a lot of AT supplies, -5V is generated from the -12V line using a 7905 linear regulator. The 79xx series (depending on manufacturer) needs a couple of milliamps load, or the output will be too high. So it is very likely that -5V will be spot on as soon as something is actually trying to use it.

Reply 32 of 34, by mkarcher

User metadata
Rank l33t
Rank
l33t
dominik0801 wrote on 2020-05-14, 18:12:

The PSU is on, it's fan is spinning, but the motherboard won't post, won't beep or anything. The CPU gets slightly warm over time, but not hot, and it's not getting warm fast either.

[...]

How should I jumper the board, and which RAM slot should I use, if I only want to use a single 16MB 72pin memory?

While I understand that this post is old, the question is a common one, so this thread deserves an answer: If you start troubleshooting a non-working board (and you don't have a port 80 diagnostic card), you should install a known-good speaker, *no* cards, *no* RAM, *nothing* connected to onboard I/O if any. If processor, mainboard and BIOS are working, you will get some kind of beep signal to tell you that the RAM is bad. If you don't get a beep in this configuration, the CPU, BIOS or board is bad.

Reply 33 of 34, by dominik0801

User metadata
Rank Newbie
Rank
Newbie

This thread is not that old, but actually I have already tried what you mentioned. Also, other CPUs and a new, pre-programmed BIOS chip. The reason I wanted a DRAM configuration to get a beep, is because I have read a post here, that a guy's motherboard wouldn't POST or beep as long he hadn't the right DRAM configuration with RAM in the slots.

Attaching a battery didn't help, neither starting without a KBC chip (which I have read about that if it boots without one, the KBC went bad).

Also, I took out the KBC chip to check under, and I can see some kinda blackened traces, which I can't figure out where they go/went, because unfortunately the RAM slots and the KBC socket is over them.. so if I would like to check those traces, I either have to desolder half of the components on the board, or find someone with the same motherboard and ask them to measure those traces for me, so I can check/bridge my board according to theirs, which sounds really unlikely that I will find someone.

However, I don't like to give up, so I will most likely start to fiddle around with it again, as soon as I get the motivation..

Reply 34 of 34, by dominik0801

User metadata
Rank Newbie
Rank
Newbie

Since the last post, I desoldered a lots of chips off the board, the RAM slots as well, and the KBC socket too.. I found some more traces that were broken, and I fixed them too. No progress.
Yesterday I got new SIS85C407 and SIS85C471 chips, so I replaced the old ones, still the same.

At this point I really can't see what else could be wrong here. I am thinking about replacing the crystals, but that idea really is just some "last try" stuff..