I'd reflash the bios.

That was my first guess... corrupted bios, but since sometimes it works fine, may I ask why you suggest so?

Nemo1985 wrote on 2021-08-05, 22:29:

That was my first guess... corrupted bios, but since sometimes it works fine, may I ask why you suggest so?

While you think that a bios is written to last forever, many times it can get corrupted by a random electrical discharge, a wee overvolt, undervolt...
Stuff I know from reading others' experiences and mine. Just recently I desoldered and reflashed a bios and voilà, the board is back on working 100%. It just got corrupted, no apparent reason.

It doesn't do bad and can solve some issues like this one.

The physics behind are a mystery to me, if that was your question.

I see, I refkashed the bios, but actually it didn't solve the problem, what I noticed is this: when I try to set the date time the bios hangs, while if I switch any other parameter it works fine.

I honestly do not know if a damaged eeprom could give those issues, but I doubt.

Nemo1985 wrote on 2021-08-06, 09:59:

I see, I refkashed the bios, but actually it didn't solve the problem, what I noticed is this: when I try to set the date time the bios hangs, while if I switch any other parameter it works fine.

I honestly do not know if a damaged eeprom could give those issues, but I doubt.

Finding the issue is like jumping in a meat grinder.
I'm trying to troubleshoot a board and it's a deadly deed.

I hope someone more knowledgeable hops in fast to help you.

My idea is maybe some voltage issue with r/w operation on the chip. Does it hang always when you modifiy date/time or it is more like random?
Could be bad caps too.

I see you have a dual bios board, you can solder a socket and add the second bios chip.
Totally works and you can do whatever you want. Not difficult to do.

My model is not dual bios, it is the ultra model that have dual bios and it is also socketed, mine has the pads for soldering on the second bios.
I checked on my stash and I haven't any similar bios chip (it is a W49F002U rated for 5 volts for reading and writing).
The most similar I have it's a SST39VF020, but it requires 2.7-3.6V

Unlucky soldering is something I'm really awful.

From what I could see it only stucks after I changed the date (day, month and year), I managed other options (in other menus) without issues.

Make a long term project.
If you solder the second socket it'll work despite not being the dual bios model from the start.
I remember reading it just detects two chips and works accordingly because the part of the flashing program is the same.

If you'll ever try you'll know.
Hope you get it going 100% soon enough.

Nexxen wrote on 2021-08-06, 15:22:

Make a long term project. If you solder the second socket it'll work despite not being the dual bios model from the start. I rem […]
Show full quote

Make a long term project.
If you solder the second socket it'll work despite not being the dual bios model from the start.
I remember reading it just detects two chips and works accordingly because the part of the flashing program is the same.

If you'll ever try you'll know.
Hope you get it going 100% soon enough.

I will have to ask anyone to do it for me, but I was thinking more to desolder the original bios and solder a socket, so any other try will be easier.
Now the board doesn't boot again so I can't do any try for bios settings

Ditto for possible BIOS problem.

Award BIOS on Gigabyte MB around early- to mid-2000's seems to have a high tendency to become unstable and/or corrupted. I've got not one but two GA-K8VM800M bricked because of corrupted BIOS.
The culprit *might* be an unbranded 512MB DDR but I have no way to confirm my hypothesis.

Nemo1985 wrote on 2021-08-06, 16:32:

Nexxen wrote on 2021-08-06, 15:22:

Make a long term project. If you solder the second socket it'll work despite not being the dual bios model from the start. I rem […]
Show full quote

Make a long term project.
If you solder the second socket it'll work despite not being the dual bios model from the start.
I remember reading it just detects two chips and works accordingly because the part of the flashing program is the same.

If you'll ever try you'll know.
Hope you get it going 100% soon enough.

I will have to ask anyone to do it for me, but I was thinking more to desolder the original bios and solder a socket, so any other try will be easier.
Now the board doesn't boot again so I can't do any try for bios settings

That's exactly my suggestion.
For a quick job you can just solder a socket on those empty pads and program the chip from bios or a programmer.
Sockets cost close to nothing, you can even go to a local retailer and buy 2.

dormcat wrote on 2021-08-06, 17:13:

Ditto for possible BIOS problem.

Award BIOS on Gigabyte MB around early- to mid-2000's seems to have a high tendency to become unstable and/or corrupted. I've got not one but two GA-K8VM800M bricked because of corrupted BIOS.
The culprit *might* be an unbranded 512MB DDR but I have no way to confirm my hypothesis.

When mine bricked I had the A-data cheapest + no brand almost blank El Cheapo PSU, lighter than aerogel (yes, feathers are heavier).
I recon I got lucky it didn't fry earlier. 😀

Nemo1985 wrote on 2021-08-05, 21:52:

What could it be? Bad solder joints? What can make a board works just turning it upside down and using some alcohol to clean so […]
Show full quote

What could it be? Bad solder joints? What can make a board works just turning it upside down and using some alcohol to clean some shades?

Edit:
And guess what?
It booted again! Slot 1 and 3 populated.

This time I also took pictures just to be sure i'm not crazy 😁

Exactly bad solder joints, IMO. It's typical especially for s478 mainboards.
Their CPU socket or north bridge may loose contact to the PCB. The boxing CPU cooler just tears them off.
Try to unfasten the CPU cooler and then press the CPU cooler to the CPU carefully but firmly before power on. The mainboard should be on the desk on some soft material.
The point is to restore contact temporarily for diagnostic purpose.

Is the memory controller of this board on the NB?
Because I had an ASUS P2B with a scratched northbridge on the side and that board would only post with the ram stick only on the first slot.

First of all I wish to thank you all for the educated guess.

So we have 2 main suspects: the bios and solder joints.

Eeprom damaged: When I tried to flash the bios with the utility provided with the bios I had this error, then I used uniflash and everything went fine, it may help to troubleshoot the eeprom issue. I also noticed that when the board can't boot it reset itself after some seconds, over and over. That's not the first time I see this behaviour in a non working motherboard, it's like it tries to initialize itself over and over.
If the culprit is the Winbond W49F002U, do you have any advice of what other 256K × 8 32 pin plcc I should use?

Bad solder joints. Pressing the cpu, or the nb heatsink or the sb didn't help. I tried to do so while having the motherboard on my cardboard test bench. The only way to make it work for a brief time it's to the motherboard wit hisit over for some second, but it back on his place and boot it.
The only sure way to make it works is to turn the motherboard upside down then put it on his original position and make it boot.
What can I do if it is bad solder joints? I have no hot airgun the only thing I could do is to put it in the oven, but it would be quite desperate I suppose.

This is what happened some moments ago, it hanged before get into the bios and dead again, notice please the 3 white dots around the screen.

aaronkatrini wrote on 2021-08-06, 19:55:

Is the memory controller of this board on the NB?
Because I had an ASUS P2B with a scratched northbridge on the side and that board would only post with the ram stick only on the first slot.

Yes the memory controller is still on the nb, the first cpu to have it was the athlon 64, still in this motherboard there is the heatsink protecting the northbridge.

If you have a magnifying glass inspect the joints and use a tool to see if by moving the pins cracks get visible.

When it turns on black screen, probe the power pins to see if they do connect.

Also could be a bad cap.

Does it now boot with all ram sticks? or just 1 3 combo?

I gave a look with a magnifier loupe, everything seems in order, do you have any suggestion what I should watch precisely?

The only thing I could notice is a capacitor which is lifted from one side (but still soldered), other than that the board seems in pretty new conditions.

The capacitor lifted is the MBZ:

The board stopped booting at all, it gave just a sign of life with c100 on the debug card once, then nothing only ----, turning it upside down didn't help.

Edit: I now soldered it properly, obviously no results... still dead.
I also noticed that sometimes the debug card doesn't update itself between a reset and the next. So I suppose that something is wrong at the very beginning, but I have no idea about what chips come in place while starting other than the bios.

Nemo1985 wrote on 2021-08-06, 20:35:

First of all I wish to thank you all for the educated guess. […]
Show full quote

First of all I wish to thank you all for the educated guess.

So we have 2 main suspects: the bios and solder joints.

Eeprom damaged: When I tried to flash the bios with the utility provided with the bios I had this error, then I used uniflash and everything went fine, it may help to troubleshoot the eeprom issue. I also noticed that when the board can't boot it reset itself after some seconds, over and over. That's not the first time I see this behaviour in a non working motherboard, it's like it tries to initialize itself over and over.
If the culprit is the Winbond W49F002U, do you have any advice of what other 256K × 8 32 pin plcc I should use?

Bad solder joints. Pressing the cpu, or the nb heatsink or the sb didn't help. I tried to do so while having the motherboard on my cardboard test bench. The only way to make it work for a brief time it's to the motherboard wit hisit over for some second, but it back on his place and boot it.
The only sure way to make it works is to turn the motherboard upside down then put it on his original position and make it boot.
What can I do if it is bad solder joints? I have no hot airgun the only thing I could do is to put it in the oven, but it would be quite desperate I suppose.

photo_2021-08-06_22-30-13.jpg

This is what happened some moments ago, it hanged before get into the bios and dead again, notice please the 3 white dots around the screen.

aaronkatrini wrote on 2021-08-06, 19:55:

Is the memory controller of this board on the NB?
Because I had an ASUS P2B with a scratched northbridge on the side and that board would only post with the ram stick only on the first slot.

Yes the memory controller is still on the nb, the first cpu to have it was the athlon 64, still in this motherboard there is the heatsink protecting the northbridge.

1) I think BIOS is OK as the mainboard sometimes starts.
2) So solder joints under s478 and the NB maybe OK too. The airgun and oven doesn't need in your case.
The airgun doesn't help for massive BGA chips by the way The oven is very unpredictable so I wouldn't use it.

My second wild guess is a MOSFETs APM3055. Maybe you can see it between the SB and DDR slots. It might make voltage for the NB an the SB.
Measure voltage on PIN3 of the MOSFET (red pin of voltmeter here and black pin to GND). Do it carefully to not short something. It should be 1.8v.
MOSFETs APM3055 are very unreliable if they works in linear voltage regulators under big load.

P.S. The model of the mainboard is GA-8S655FX (there is a typo in the topic name).

Thank you for the tips.

Typo fixed, I checked the voltage it is: v1.883, so I guess it is ok...

The board boots less and less, during last try it booted just once and it did after some time it was turned on, I was busy pushing chips and it suddenly started.

Anything gets hot, cpu, nb and southbridge if it matters.

That's a picture of the board, if you need any detailed picture or anything I should measure, that's something i'm able to do 😀