Do you have a POST card and a phone with a high speed camera to record the sequence of the POST codes during the initialization?

Does the board act differently when you start it without RAM sticks? Are the slots clean?

If you have an oscilloscope or a logic analyzer, can you see any traffic on the SPI flash chip? If you don't, you can still try hooking an LED to one of the data pins to check for activity.

If you have a thermal camera, are there any overheating components on the board? If you don't, you can use thermal paper from old receipts.

Is the CMOS battery good? Some AM2/AM3 motherboards refure to start properly with a dead or missing CMOS battery.

kmeaw wrote on 2024-07-31, 07:37:

Do you have a POST card and a phone with a high speed camera to record the sequence of the POST codes during the initialization? […]
Show full quote

Do you have a POST card and a phone with a high speed camera to record the sequence of the POST codes during the initialization?

Does the board act differently when you start it without RAM sticks? Are the slots clean?

If you have an oscilloscope or a logic analyzer, can you see any traffic on the SPI flash chip? If you don't, you can still try hooking an LED to one of the data pins to check for activity.

If you have a thermal camera, are there any overheating components on the board? If you don't, you can use thermal paper from old receipts.

Is the CMOS battery good? Some AM2/AM3 motherboards refure to start properly with a dead or missing CMOS battery.

Negative on the post readout card, oscilloscope, logic analyzer and thermal camera, unfortunately.

With no RAM installed, I'm getting the RAM_LED. The slots should be clean enough since it detected the memory at least once. I tried a known good CMOS battery and it made no difference.

One possibly interesting finding that I made was that after flashing the BIOS chip with my CH341a programmer, I get one (consistent) CRC32 checksum. If I put the chip back into the board and power it on, next time I read the chip with the programmer I get a different checksum and I'm not sure that's normal. I also flashed the chip and put it in another working board (I have two 990FX Sabertooth boards) and the good board worked fine with the chip from the bad one, so the chip should be fine. I feel like the board is corrupting the BIOS somehow. Is that possible?

DragonClaw wrote on 2024-07-31, 12:24:

I also flashed the chip and put it in another working board (I have two 990FX Sabertooth boards) and the good board worked fine with the chip from the bad one, so the chip should be fine. I feel like the board is corrupting the BIOS somehow. Is that possible?

UEFI systems store usually their variables inside the same flash chip (in nv data volumes) so it is fine to have the motherboard update the contents of the flash by itself. I have seen only one instanse of it leading to self-corruption - when I installed the flash chip which had the wrong size and overflowing writes were wrapping around and overwriting the code in the "first half" of the supposed large flash.

Could you test it the other way around - flash the chip with good known BIOS, put it into the "bad" board, start it up, validate that the contents has been updated and try to boot the "good" board with the same chip.

kmeaw wrote on 2024-07-31, 19:03:

UEFI systems store usually their variables inside the same flash chip (in nv data volumes) so it is fine to have the motherboard update the contents of the flash by itself. I have seen only one instanse of it leading to self-corruption - when I installed the flash chip which had the wrong size and overflowing writes were wrapping around and overwriting the code in the "first half" of the supposed large flash.

That is my observation, too. Restoring the original chip in the good board brought back my saved settings. Though I'm not sure I would've expected the checksum to change despite not being able to post and edit any settings at all.

kmeaw wrote on 2024-07-31, 19:03:

Could you test it the other way around - flash the chip with good known BIOS, put it into the "bad" board, start it up, validate that the contents has been updated and try to boot the "good" board with the same chip.

That would be my next step. I'd also be curious to compare the checksum after attempting to boot the bad board against the checksum from booting up the good one following a CMOS reset and not changing any settings (not even going past the initial prompt to enter the BIOS after a CMOS reset).

No beeps may indicate failing north/south bridge.

DragonClaw wrote on 2024-07-31, 05:46:

and died

Try replasing the TIM under the north bridge.
Asus used a very cheap PCM (Phase change material).

The heat pipe from the VRM heats up the already hot bridge. And it tries to tear the heatsink off the bridge due to thermal expansion.
As the PCM ages, it loses the ability to soften when heated and simply cracks when displaced.

The victory of pathos over common sense at Asus

Thanks for all the suggestions, I should have a bit more time to try things out today and will report back!

Nice board, had one myself (R2.0), loved that build! Powered my FX-8350 and Crossfire dual Sapphire Tri-X R9 290X rig for many years. 😀

A few silly questions...

1. Have you tried pulling the battery?
2. Resetting via the jumper on the board and not just MEMOK?
3. Have you tried a different power supply?
4. Inspect the board for physical damage?

Okay, back with another update addressing the new suggestions (again, thanks a lot to you all!):

1. Re-programmed the chip from the bad board -> plugged it into the bad board -> powered on -> no post -> plugged the same chip into the good board (no re-flashing inbetween) -> posts perfectly fine, i.e. no, the board is not corrupting its own bios;
2. Replaced the TIM on both the NB and SB with some Arctic MX-4;
3. Have tried to boot with no battery in;
4. Have reset the CMOS via the jumper;
5. Have tried 2 known good power supplies;
6. Have inspected for physical damage, there's none that I could find... maybe - more on that below.

Today I decided to try plugging in my HD4870 reference card with debug LEDs and this configuration did the following:
1. VGA_LED on the motherboard solid red with no post (as I have been getting so far);
2. no debug LEDs on of any kind, but still no post;
3. CPU_LED on the motherboard solid red, D1601, D1602 and D1603 on the GPU solid red, indicating critical temperature fault, external power connector A removed and external power connector B removed.

The latter two seem like they were more of a fluke and maybe the GPU was getting confused as well. I was getting them an hour ago and now I can't reproduce - back to just VGA_LED. Throughout all my testing today I left the speaker plugged in and I haven't got any beeps whatsoever.

Regarding physical damage - I can't find any burnt/ripped/cracked components, but I did notice a few of the holes in the CPU socket didn't quite look like the rest. When the latch is about half-way closed, you usually catch a glimpse of the metal contacts inside the socket. In that case, some of the pins along the top edge of the socket in this picture don't look right (especially the first two from right to left).

I think this one is a no-go. I removed the socket cap to try and straighten the pins on the socket a bit, but that didn't yield any results. I think the sudden death after the initial POST and how I didn't even physically touch it between being it alive and then dead lends credence to the theory that it might be a chipset issue. Either that or some random component on the board somewhere gave up the ghost, but I don't have the tools and expertise to diagnose further and doing so at this point would be a poor investment.

Thanks for all your help!

DragonClaw wrote on 2024-08-01, 13:02:

it might be a chipset issue.

When the PCM “dries out,” it continues to conduct heat, but you just move the surfaces together and the contact disappears.

Many AM3(+) motherboards died due to the heatsink shifting even when assembling the config .
AM3(+) northbridges are very power hungry (20+ watts).
Alas.

If you have nothing else to try and the board is going to trash otherwise, a heat reflow option might be a last resort option. Try heating up individual chips with a heat gun (and lightly touching them with a toothpick to realign) and if nothing changes, reheat the whole board in an oven.