Deunan wrote on 2021-04-24, 20:23:

OL? You mean close to zero ohms I guess?

I have yet to see a shorted tantalum cap that would not burst into flames when power is applied. You really can't miss that. PSU might shut down instead if the short is on -12V line but the +12V line has more than enough amp rating to just feed the fire and not care. And obviously that should be indicated by the POST card LEDs, that one rail is missing due to short. On the other hand, if the cap went open, it would not prevent the mobo from booting. It's just a 10uF and there's plenty of other local bypass caps to make up for that.

OL comes from OverLoad, which means that the capacitance is either too high or the capacitor has shorted. Contrary to popular beliefs, the tantalum capacitors burst into flame only rarely. They often fail without any signs of damage (http://minuszerodegrees.net/failure/failure.htm).

Oh I get it now, you meant a capacitance test. Well, not all meters (and especially not cheap ones) can measure capacitance, plus measuring it in-circuit is not always possible, but pretty much all meters can measure resistance. Finding a shorted cap is way easier in resistance mode, and usually you can even use the buzzer to speed it up.

As for popular beliefs, I had quite a few tantalums go bang on me. One did just a few days ago in fact. And all the photos on the page you linked show clearly damaged capacitors so I'm not sure why you'd call that "without any signs ". Sure, some tantalum can fail more subtly but not the ones connected to power lines. In the end though, damage or not, this is not magic - I've already explained there would be other obvious signs of shorted caps, like POST card LEDs being off. Frankly the mobo should boot with just +5V and clearly that is working as some codes do appear. If there was to be a fault here I would suspect the PSU and excessive ripple first.

Deunan wrote on 2021-04-26, 19:47:

Oh I get it now, you meant a capacitance test. Well, not all meters (and especially not cheap ones) can measure capacitance, plus measuring it in-circuit is not always possible, but pretty much all meters can measure resistance. Finding a shorted cap is way easier in resistance mode, and usually you can even use the buzzer to speed it up.

As for popular beliefs, I had quite a few tantalums go bang on me. One did just a few days ago in fact. And all the photos on the page you linked show clearly damaged capacitors so I'm not sure why you'd call that "without any signs ". Sure, some tantalum can fail more subtly but not the ones connected to power lines. In the end though, damage or not, this is not magic - I've already explained there would be other obvious signs of shorted caps, like POST card LEDs being off. Frankly the mobo should boot with just +5V and clearly that is working as some codes do appear. If there was to be a fault here I would suspect the PSU and excessive ripple first.

The linked page contains the following statement: " Do not assume that because a tantalum capacitor shows no damage, that it is good. Faulty tantalum capacitors often show no visible indication of failure." This is also my experience. I have a collection of more than a hundred mobo-s, and as time goes by some of them fail to POST. However, of the 7 I repaired recently, only one tantalum capacitor has burst into flames. It was from an old IBM 5160. The remaining ones showed no visual signs of capacitor damage.

techgeek wrote on 2021-04-27, 15:17:

The linked page contains the following statement: " Do not assume that because a tantalum capacitor shows no damage, that it is good. Faulty tantalum capacitors often show no visible indication of failure." This is also my experience.

I don't want to argue, my point is that statement can be true for any and all electronic parts on the motherboard. And for example chips, especially old DRAMs, do fail on their own without any signs of damage. Are we supposed to replace all the parts because they could, potentially, be somehow faulty? I say there's always other signs one should look for.

Not to derail this thread further but your experience is so different from mine that I have to ask - what were your faulty mobos doing, exactly? Starting but not fully not booting - like in this case? Or was the PSU tripping, thus there were obvious signs of shorted power line? Did you check all the voltages being present before removing the capacitors and did you verify the cap was really bad afterwards, because maybe the problem was somehow unrelated.

Interesting case, I had created an exact problem (POST always halted at 0D) two months ago, when I was modding Award BIOS for one 486 motherboard. The mod was about correcting a CPU clock calculation and model string display for AMD DX4 and 5x86 CPUs. The problem, that caused 0D after modifyng CPU clock calculation and identification areas was that BIOS could no longer load default BIOS settings from a special table. That table is selected by a subroutine (subroutine is called every time when the CMOS checksum is bad), that uses CPU type and clock as a reference and after the mod, of course, it didn't know what to do with the never seen before CPU identification data and just crashed every time it had to load default BIOS values.

It might be that you have a problem with the integrated RTC chip there (it stores CMOS data). Check 82C471 datasheet for the RTC control pins.

figbash wrote on 2021-04-17, 22:33:

11. Went outside and screamed at the clouds

😁