Motherboard short Ohms "draining" up?

[AndrewB]

Hi All,

I'm working on a motherboard that has a short on the 5V rail.  So at the beginning the short was constant and low Ohms.  I eventually found a few ceramic  capacitors that were getting hot and removed them.  But now I've encountered a new situation which I'm hoping someone might be able to better identify and point me in the proper direction.

So this is the observation now.  If the board is left for a while and then with a multi meter tested across the 5V and ground - the multi meter will let out a short beep and register a reading of between 85-90 Ohms.  BUT if I just apply power to the motherboard for a few seconds only, remove the power, and test again.  Now the multi meter will emit a constant beep and shows a reading 14-16 Ohms.  Here's the kicker, if I leave the board about a minute or so and then test again the multi meter will again give off the short beep and show a reading of 85-90 Ohms.  Now this behavior happens whether the multi meter is hooked up or not during that minute, so I don't believe it's the multi meter playing a part in this behavior.   If I leave the multi meter connected, I will suddenly see the Ohms start to rise around that one minute mark.

So does anyone have any idea what this might be?  Why is it right after power is applied the short is constant but then a minute later seems to "Drain" away?  What should I be looking at?  Could it be electrolytic caps or ceramic caps causing this?  or could it be something completely different?

Thanks.

[Tomorokoshi]

Ceramic MLCC capacitors can crack such that opposite layers can short. However, it might only be a thin strand of the conductive layer. Therefore it could have a cold resistance of 15 ohms, but when it heats up there is enough thermal expansion to reduce the connection to perhaps 100 ohms.

Expect a current of 5V/100ohms = 50 mA, and that it dissipates 250 mW.

That can be detectable on a thermal camera, but it will be difficult to feel directly.

However, because it seems to be consistent, there are some diagnostic methods you can use:
1. Gently twist the board. Does the reading change?

2. Press on the parts when it's powered. Does the reading change?

3. Spray on the parts when it's powered with that compressed air spray. It cools down the part, so the reading might change.

[AndrewB]

Hi,

Thanks for the feedback.  Following the advice that what I'm seeing is probably correct, I went ahead and re-installed all the caps I pulled.  The power supply appears to be running fine when connected and doesn't go into any sort of protection mode.  The chips all appear to be getting their voltage.

Unfortunately it's still booting to a black screen.  The monitor gets a signal just black.  No beeps.  The only two chips that get warm are the processor and FDC chip.   I'm suspecting maybe a bad BIOS.  I'm waiting on the programmer to see if I can write some new images and hope that was it.

[perieanuo]

bad bios is rare, I reprogrammed only 2-3 bios chips coming from 'tuning guys'
what I suspect is bad caps, but they usually are with the top ready to explode
if you have the guts and it worth the time/cost, replace the eletrolytics
of course, reprogramm the bios if you can and you've cleared the subject

[amyk]

MOSFET gates can hold a charge for a surprisingly long time. That might be what's happening here.

[AndrewB]

Just an update on this. I had my hunch about a bad BIOS sort of confirmed today. I remembered that while digging through a bad of old ICs I remembered there was one with a Compaq sticker on it. Turns out it was a BIOS from one of their XT clones. So I figured "Compaq was an XT clone - Commodore is an XT clone. What's the worse that could happen". Well popped it in and it posted. Gibberish but it posted. So pretty sure the board is fixed just need to program a new BIOS.

[amyk]

It might just be a BIOS that doesn't touch the defective device and attempt to initialise it... because it doesn't know about its presence.

[james_s]

Well it's about at that age where EPROMs can start to fade. It wouldn't hurt to burn a new chip and pop it in there. There are even some more modern BIOS's out there that can give you updated features.

[perieanuo]

eeprom same capacity/reading algo will also do the job

[Rasz]

I like the part where you mention its a PC10/PC20
btw 5V × 5V / 15 ohm = 1.6 W, there is no short