tanis wrote on 2024-10-31, 22:14:

Still won't POST. No beeps. At this point I'm thinking (crazily) that the BIOS got erased? […]
Show full quote

Still won't POST.
No beeps.
At this point I'm thinking (crazily) that the BIOS got erased?

Suggestions?

Seems unlikely to me (but who knows) - BIOS shouldn't be easily writable by accident, and since you didn't
mention it, I'm assuming that when it did POST it wasn't complaining of any error in the BIOS?

I'd break it down to the barest minimum you need to see a POST screen - just RAM, Video card (try a different
known working one if you can) and power-supply.

I have seen some mainboards that wouldn't properly reset/POST without a hard drive
installed (IIRC these were older onboard IDS system).

I'd also connect a RESET button, just in case problem is power-on RESET (I had a system
that needed a manual reset at one time).

If you happen to have a device reader, you could see if the BIOS chip contains something readable.

If you have a scope - look at the BIOS chip select, and address/data lines immediately after RESET
and see if it's actually trying to read the chip.

Best of luck!

Dave ::: https://dunfield.themindfactory.com ::: "Daves Old Computers"->Personal

Bioses don't just erase for no reason.

If anything died, it's most likely the 25 year old capacitors.

Damage is rarely visible on electronics.

Yes probably caps. How did you verify PSU ? One sitting for 25 years is mostly not going to work proper as the caps dry out or internally leak..
FWIW: In my experience any ATI chipset board is just trouble down the road. Have never had one last more than a few years back in the day when I used them daily. Learned to avoid them.
Stuck with Intel and Nvidia chipsets for Intel cpu just because they seemed to last a lot longer before issues, but that is just my experience.

myne wrote on 2024-11-01, 02:38:

If anything died, it's most likely the 25 year old capacitors.

Agreed, most common reported failure - I've been very lucky over the years however - I've had A LOT of experience with very
old systems (see "Daves Old Computers") - and almost never saw caps fail (especially to the point where a system would appear
to be completely dead)- to be fair the 20+ years he mentions would certainly be a factor, although I've recovered systems out of
service longer than that without such problems. And the caps in a "modern" 486 board are likely denser and more affected by
age than some of the really old minicomputers I've played with!

Again, probably the first thing I would do on such a system is to scope the various power supplied and
capacitor banks to confirm "lack of noise" (You really can't invest much time in recovering old electronics
without a scope)

And with stuff this old, you really can't visually tell where failures are - CAPs are one of the first I'd
suspect - I've also had very old systems which used EPROMs for the boot-rom lose their content to do
age - I don't think this is a problem with "newer" PC's which tended to use flash - but you can never be
sure... you really need to be able to scope things to see what exactly is happening!

Dave ::: https://dunfield.themindfactory.com ::: "Daves Old Computers"->Personal

Try getting a POST card to find out if the POST actually starts and if it is then what step goes wrong - most of the POST routines halt the machine if some check is failed.

An EEPROM programmer might help if the contents of the flash chip has been corrupted - either due to aging (which is unlikely for a relatively new board such as yours) or due to a hardware or firmware bug - BIOS updates ESCD on boot by writing data into the flash memory, if it goes wrong then it could either overwrite something else, or the ESCD could be corrupted in a way that prevents the system to complete the POST. Reading out the flash chip would help matching the POST codes to the actual POST routines, writing a good known image into the flash could fix the system.

I'm also going to put a vote down to check the electrolytic capacitors, both on the motherboard and in the PSU.

Being that the motherboard is an ECS brand, crap cap brands are pretty much guaranteed. I've seen a few ECS s775/AM2 boards with Panasonic caps... but that's like 10% chance you get one. Most of the time, they use OST brand capacitors, which aren't very good on motherboards.

As for the PSU, what brand and model is it? How did you check it? Multimeter to "check" voltage and shorting PS-ON to ground is not really a useful test, other than "it powers On".

My first go-to check when a board doesn't post is to remove the RAM and see if it beeps a memory error at me. If it doesn't, I check all of the power rails on the board for any issues. If all power rails appear to be OK, then it's something deeper, usually - either a hardware fault (bad or busted component) or a software (BIOS) fault.
*EDIT* Beware of Intel's stock push-pin coolers too. These love to warp your board to hell (and back). Seen quite a few bust the BGA under the CPU, especially if you remove and re-install the CPU cooler a few times.
Whenever I get any of these for service, I delete the push pins and make my own backplate for the board to provide the HS tension (and not warp the board.)

Horun wrote on 2024-11-01, 03:20:

FWIW: In my experience any ATI chipset board is just trouble down the road. Have never had one last more than a few years back in the day when I used them daily. Learned to avoid them.
Stuck with Intel and Nvidia chipsets for Intel cpu just because they seemed to last a lot longer before issues, but that is just my experience.

For Intel chipsets, I completely agree with you - they are very reliable and stable.
nVidia chipset - helll NAWW! 😁 Run away and as fast as away as you can. ALL nVidia chipsets past nForce 3 are bumpgate issue -prone... with the nForce 6100/6150 being the absolute worst.
As for old AMD/ATI chipsets - these tended to be not so good in laptops, mostly because they were just rather hot chips... and laptop cooling really didn't do them any good. On desktop, they would usually do fine if the case had good cooling. RS480 is actually pretty rock-solid if not allowed to overheat. Mobo manufacturers often skimp on the chipset heatsink sizes, though, so not too hard to see even these chipsets die. They still do A TON better than nVidia.

Horun wrote on 2024-11-01, 03:20:

Yes probably caps. How did you verify PSU ? One sitting for 25 years is mostly not going to work proper as the caps dry out or internally leak..
FWIW: In my experience any ATI chipset board is just trouble down the road. Have never had one last more than a few years back in the day when I used them daily. Learned to avoid them.
Stuck with Intel and Nvidia chipsets for Intel cpu just because they seemed to last a lot longer before issues, but that is just my experience.

Used the PSU on a different mobo of the same era and it worked.

momaka wrote on 2024-11-01, 13:13:

I'm also going to put a vote down to check the electrolytic capacitors, both on the motherboard and in the PSU. […]
Show full quote

I'm also going to put a vote down to check the electrolytic capacitors, both on the motherboard and in the PSU.

Being that the motherboard is an ECS brand, crap cap brands are pretty much guaranteed. I've seen a few ECS s775/AM2 boards with Panasonic caps... but that's like 10% chance you get one. Most of the time, they use OST brand capacitors, which aren't very good on motherboards.

As for the PSU, what brand and model is it? How did you check it? Multimeter to "check" voltage and shorting PS-ON to ground is not really a useful test, other than "it powers On".

My first go-to check when a board doesn't post is to remove the RAM and see if it beeps a memory error at me. If it doesn't, I check all of the power rails on the board for any issues.

It's a Thermaltake TR2-430NP. I used it boot another mobo of the same era.

When I took out the memory it didn't beep either.

Your board warping stories are gonna make me check the area under and around the CPU though...

tanis wrote on 2024-11-03, 03:45:

Horun wrote on 2024-11-01, 03:20:

Yes probably caps. How did you verify PSU ?

Used the PSU on a different mobo of the same era and it worked.

Good to know. That usually points to the motherboard as the issue. Had had some boards that just suddenly die, some from bad capacitors, some from VRM parts, some just plain quit with no checkable reason.
Have you checked the voltages ? Need to check cpu and ram volts to make sure they are ok. That can be tricky as P4 core volts and RAM are not the 12v, 3.3v, 5v, etc from the PSU.
Do you have a good digital volt meter ?
Worst case: the Ball Grid array soldering under main chipset has gone bad, not easy fixes.

momaka wrote on 2024-11-01, 13:13:

I'm also going to put a vote down to check the electrolytic capacitors, both on the motherboard and in the PSU. […]
Show full quote

I'm also going to put a vote down to check the electrolytic capacitors, both on the motherboard and in the PSU.

Being that the motherboard is an ECS brand, crap cap brands are pretty much guaranteed. I've seen a few ECS s775/AM2 boards with Panasonic caps... but that's like 10% chance you get one. Most of the time, they use OST brand capacitors, which aren't very good on motherboards.

As for the PSU, what brand and model is it? How did you check it? Multimeter to "check" voltage and shorting PS-ON to ground is not really a useful test, other than "it powers On".

My first go-to check when a board doesn't post is to remove the RAM and see if it beeps a memory error at me. If it doesn't, I check all of the power rails on the board for any issues. If all power rails appear to be OK, then it's something deeper, usually - either a hardware fault (bad or busted component) or a software (BIOS) fault.
*EDIT* Beware of Intel's stock push-pin coolers too. These love to warp your board to hell (and back). Seen quite a few bust the BGA under the CPU, especially if you remove and re-install the CPU cooler a few times.
Whenever I get any of these for service, I delete the push pins and make my own backplate for the board to provide the HS tension (and not warp the board.)

Horun wrote on 2024-11-01, 03:20:

FWIW: In my experience any ATI chipset board is just trouble down the road. Have never had one last more than a few years back in the day when I used them daily. Learned to avoid them.
Stuck with Intel and Nvidia chipsets for Intel cpu just because they seemed to last a lot longer before issues, but that is just my experience.

For Intel chipsets, I completely agree with you - they are very reliable and stable.
nVidia chipset - helll NAWW! 😁 Run away and as fast as away as you can. ALL nVidia chipsets past nForce 3 are bumpgate issue -prone... with the nForce 6100/6150 being the absolute worst.
As for old AMD/ATI chipsets - these tended to be not so good in laptops, mostly because they were just rather hot chips... and laptop cooling really didn't do them any good. On desktop, they would usually do fine if the case had good cooling. RS480 is actually pretty rock-solid if not allowed to overheat. Mobo manufacturers often skimp on the chipset heatsink sizes, though, so not too hard to see even these chipsets die. They still do A TON better than nVidia.

ECS from that era also loved to forge Panasonic caps. You'd be surprised how much fake Panasonics I could find during ECS' P4/Athlon 64 phase.
The easiest way for me to spot them is the bottom bung - authentic Panasonics of the series used on ECS boards never used the bullseye-type bottom bungs.

Seems extremely unlikely a BGA failure would happen between reboots.

Had some time today to tear everything back apart and try a few things.

Removed CPU and inspected the BGA and LGA, everything looks fine there. Clean, pristine, undamaged.

Installed the reset switch connector to the front panel; no affect.

Gonna order a POST analyzer, but I suspect I'll just see -- -- 🤣

tanis wrote on 2024-11-03, 04:01:

It's a Thermaltake TR2-430NP. I used it boot another mobo of the same era.

Hey, I have one of those too! 😀
Bought it new back in late 2006 or mid 2007 somewhere.
It's a pretty decent PSU, though not really a 430 Watt. More like 300-350 Watts tops. But still decent, nevertheless. Only issue with these (and the HEC Orion HP5xx units, which are the same thing more or less... and I have one of these too) is that they are also prone to getting bad caps. Mine (the TR2-430NP) blew a few of its caps a few years back. Was powering an OC-ed Athlon 64 3200+ Venice fine for many years. Then took it out of storage one day to test some video cards and the system was unstable + slight screeching from PSU. Opened it up and surely saw the reason why - 3 bad caps (only two visibly bulging/leaking, one was found bad during testing.) Replaced all the affected caps + any others from the same brand and all was well again.

So in short, open your TR2-430NP and give it a visual inspection, just to be sure that's not also an issue (the mobo may have a separate issue, but dirty power from bad caps won't help any with the troubleshooting.) And until you get that PSU recapped, perhaps get in the habit to check it from time to time. Yearly would be OK, though YMMV depending on how much and how often you use it. Not a bad idea to check the 5VSB voltage, either, especially after some time with the PSU sitting in storage. Unfortunately, these PSUs have a relatively dangerous 2-transistor 5VSB circuit that can do crazy over-voltage when either the output caps or the "critical" 5VSB cap on the primary side fails.

PcBytes wrote on 2024-11-03, 05:26:

ECS from that era also loved to forge Panasonic caps. You'd be surprised how much fake Panasonics I could find during ECS' P4/Athlon 64 phase.
The easiest way for me to spot them is the bottom bung - authentic Panasonics of the series used on ECS boards never used the bullseye-type bottom bungs.

Those aren't counterfeit/forged Panasonic caps. Just like the 6.3V 820 uF Rubycon MFZ caps found on Xbox 360 boards (those also have bulls-eye bungs), these Panasonic caps are simply "large OEM" orders... and sometimes even the Japanese brands will use "cheaper" bungs for these.
One easy way to tell a genuine Panasonic from the fakes is the date code. The ones I saw on ECS and ASUS boards never appear to use the same "generic" date code throughout. And the fakes usually have a hard time copying that stylized "T" vent (save for Hermei, which are pretty good at copying it.)
Now, if you want to see some decent-ish looking fake Nichicon HE caps I got recently (a box of 45x 50V 2700 uF for $1.75 total 🤣), lemme know. 😉

tanis wrote on 2024-11-04, 03:18:

Had some time today to tear everything back apart and try a few things. […]
Show full quote

Had some time today to tear everything back apart and try a few things.

Removed CPU and inspected the BGA and LGA, everything looks fine there. Clean, pristine, undamaged.

Installed the reset switch connector to the front panel; no affect.

Gonna order a POST analyzer, but I suspect I'll just see -- -- 🤣

In regards to BGA failure / intermittency under the CPU socket, you won't be able to see anything damaged with your eyes.
Best hope in detecting it (and I use the word "hope" very loosely here) is to try powering the board on with and without the CPU heatsink clamped tight on the board and also attempting to flex the board a little in the CPU socket area (by applying and removing pressure) between reboots.

That said, do get a POST analyzer card first, just to see if it spits out any codes. If it does, then doing the above (testing with the CPU cooler clamped and unclamped) to see if the POST codes changed may give you a better idea of what's happening.

I'll pull apart the PSU and inspect the caps inside. I've got a POST analyzer otw.

Regarding PSUs, I would assume a modern one can be used if it has the required connectors. What would be an appropriate wattage to get? 500ish? Heck I thought the Thermaltake 430 was overkill.

P4 3GHz
2 GB DDR2 RAM
eGeforce 7600 GT PCI-e
SoundBlaster Xi-fi PCI
Couple of IDE and Sata HDDs
IDE cdrom

tanis wrote on 2024-11-06, 04:51:

Regarding PSUs, I would assume a modern one can be used if it has the required connectors. What would be an appropriate wattage to get? 500ish? Heck I thought the Thermaltake 430 was overkill.

Yes, a modern PSU will work, since late P4 motherboards and onward started using the 4-pin 12V CPU connector for power... and thus rely more on a stronger 12V rail, which all modern PSUs are designed around.
Even a 200-250 Watt PSU will do the job if it's built well and has a strong-ish 12V rail (16-18 Amps minimum for your build, I'd say).
One reason I don't like to go too overkill, especially with modern retail* PSUs, is that they have a very strong single 12V rail rather than having the 12V rail separated into several weaker ones. In the event of a "weaker" short-circuit, especially with old faulty hardware, the PSU may not trip right away or at all... and that could do further damage or burn something severely. I prefer my 12V rail to be separated into several ones with a 18-20 Amp limit each. AFAIK, OEM PSUs like the ones in Dell, HP, and Lenovo PCs still do this. So I often grab these as my go-to PSUs for older builds. If you do go that route though, just make sure you'd don't get one with some proprietary connector.

For a P4 3 GHz, you need about 100-110W from the 12V rail. 7600 GT is another 40W or so from the 12V rail. And two HDDs spooling up can use up about 2 Amps each... or about 50W spinning up (and much less once they are up and running.) So that equals about 200W absolute worst case on the 12V rail... or roughly 17 Amps. Simple as that. 😀

Welp, got my POST card today.

straight -- -- from both PCI slots

LEDS show that the required power rails are working, unfortunately the RESET led is constant on.

tanis wrote on 2024-11-12, 00:40:

LEDS show that the required power rails are working, unfortunately the RESET led is constant on.

The LEDs might not tell you much other that "something is there", you really want the check voltage (multimeter) and "lack or ripple"
- I usually use an oscilloscope, but you can do a pretty good job with a high capacitance capacitor.
The cap blocks DC but AC (ripple) passes, if your multimeter has a good low AC range (couple volts) that's all you need, otherwise use a diode to rectify

On older mainboards, the RESET LED output often directly reflected the state of the RESET input (Either by button or internal power-on reset circuit)
I've run into some newer boards where it wasn't quite so simple... and a at least one which needed the BIOS to turn it off...

But RESET LED ON suggests the system/CPU is held in RESET.

Confirm that there's nothing on the "RESET button" input, also check that PSU is stable and has all signals the system needs.

I've run into a couple systems which wouldn't RESET properly without a hard-drive installed (I think I worked it out to reliance on a reset single
on IDE interface produced by the drive - manual RESET button would always work)

It could be a power-on RESET circuit has failed (often these use a CAP which is a likely point of failure on old hardware)

Worth checking the CPU reset line - for most CPUs this is LOW to reset, and should go high shortly after power-on.
Also worth looking to see if it changes state when front-panel RESET signal is activated.
Depending on the busses the board has, RESET may also be available there.

Dave ::: https://dunfield.themindfactory.com ::: "Daves Old Computers"->Personal

The fact that the board doesn't spit out beep codes even without RAM and the POST card showing "--" on the display confirms it's a pre-POST issue. Still cannot pinpoint with that what exactly the issue is, though. It could be a BIOS issue (though would have to be a completely dead chip or very corrupt BIOS for that to happen, which I think is unlikely), a power rail issue (rare - usually that would be caused by a bad MOSFET... which if it was the case, there would be a short-circuit or overload situation and the mobo+PSU will just show a "blip" of power before powering off), or some other hardware issue (mechanical issue with CPU socket BGA, burned out trace on an inner layer, etc.)

Seeing the state of the RESET line may indeed shed some light onto the issue... though tracing that could turn into a rabbit hole. As such, my preference is first to check the power rails on the motherboard to see if they are delivering the correct voltages. So that would be the CPU VRM low-side voltage (CPU V_core), memory V_DIMM voltage (2.5-2.6V for DDR or 1.8-2.1V for DDR2), memory V_TT termination voltage (about half of V_DIMM... so around 1.25V or 1.4-1.5V), CPU termination voltage, if any (don't remember if CPUs had it), Northbridge Vcc, Southbridge Vcc, AGP comm voltage, and Standby 3.3V.

Using the picture from TRW as a reference (this one)...
CPU VRM /V_core can be measured directly any any of the 4 toroidal inductors around the CPU socket.
RAM V_DIMM should be present on that toroidal inductor in the upper-right corner by the RAM slots
RAM V_TT should be present on that MOSFET-like IC with the 5 pins on the bottom (not sure which pin exactly, but one of them... be careful when you poke around there with your multimeter.)
Northbridge Vcc (or possibly CPU V_TT) can be measured on the toroid directly to the left of the Northbridge
The AGP comm/termination voltage is likely present on the tab or right-most pin on that large linear regulator at the upper-right corner of the PCI-E 16x slot
Southbridge Vcc should be on one of the pins on one of the SOT-223 regulators near (to the left) of the Southbridge IC
The standby 3.3 Vcc should be on one of the pins on the other SOT-223 regulator near of the Southbridge IC

I think that's about it with all of the voltage rails on this motherboard... or at least what I can tell from the pictures on TRW.

BTW, have you tried to see what happens when you connect your monitor to the onboard VGA? I know the chance of this is really small, but perhaps the POST card is not able to pickup POST cods for whatever reason *and* perhaps also the boards does not beep for successful POST by default (i.e. typically called something like "silent POST" in BIOS)... Yeah, I know, extremely small chance for this to be the case, but still worth trying perhaps, just in case.