nzoomed wrote on 2020-07-25, 05:22:

I have only cut off about 3 or 4 near the power connector. I think the short us somewhere else but blow if I know what would be shorting the 12V rail if the positive side of them all measure 1400 ohms to ground.

One of them should read as a short so those tantalums are probably fine.

Miphee wrote on 2020-07-25, 06:43:

nzoomed wrote on 2020-07-25, 05:22:

I have only cut off about 3 or 4 near the power connector. I think the short us somewhere else but blow if I know what would be shorting the 12V rail if the positive side of them all measure 1400 ohms to ground.

One of them should read as a short so those tantalums are probably fine.

Yeah that's what I thought too.
Have not gone through all the yellow ones yet. I'm pretty sure all those are rated at 5V.

I would suggest to first identify and mark out all tantalums that are connected to the +12V rail. Then desolder one leg of each tantalum and measure again for shot. Like that you can solder the good ones back later. If you cut the legs then you'll need to replace them later. There is another more tricky method of measuring the resistance precisely to identify the affected area. If the short is far away on the board, it will show slightly higher resistance (perhaps in the order of 0.1 ohm or less) as the meter is also measuring the resistance of the traces. For example in area A you get 1400.1, area B is 1400.2, area C is 1400, then it means the short is around area C. Then you can perform the "lift one leg" checks on all tantalums connected to +12V in area C.

If you have a bench power supply, then there is yet another method of supplying +12V to the 12V rail, and then increase the current limit step by step, and touch the tantalums to look for a hot one.

We used to hook a large battery to the bad supply lines and watch for what explodes then fix.

Of coarse I had an unlimited supply of free bad boards and would only fix the easy ones.

As long as the power isn’t for too long the worst thing is a blown trace to a bad component.

We used to blow traces on purpose to carbonized (overheated/shorted) portion Of pcbs so we could wire around the bad segment as well

Good Luck

kalohimal wrote on 2020-07-25, 10:25:

I would suggest to first identify and mark out all tantalums that are connected to the +12V rail. Then desolder one leg of each tantalum and measure again for shot. Like that you can solder the good ones back later. If you cut the legs then you'll need to replace them later. There is another more tricky method of measuring the resistance precisely to identify the affected area. If the short is far away on the board, it will show slightly higher resistance (perhaps in the order of 0.1 ohm or less) as the meter is also measuring the resistance of the traces. For example in area A you get 1400.1, area B is 1400.2, area C is 1400, then it means the short is around area C. Then you can perform the "lift one leg" checks on all tantalums connected to +12V in area C.

If you have a bench power supply, then there is yet another method of supplying +12V to the 12V rail, and then increase the current limit step by step, and touch the tantalums to look for a hot one.

Ok that sounds a good idea. I expect that there will be some that go hot or explode. If that is not going to damage the board, its probably the easiest way. My power supply also has a meter to display current draw.

Edit:

Just hooked up to a car battery after my power supply was unable to blow any capacitors.
Turns out it was a yellow cap that exploded, and these are rated at 25v.

The ones I thought were 5v must have had their writing rubbed off. Anyway, I think its probably ideal to replace them all eventually. Would you go for new tantalums or electrolytic next time?

Replace them with appropriately rated tantalum. I would not bother replacing the ones that still work. Tantalums don't dry out like electrolytics they just short out sometimes.

maxtherabbit wrote on 2020-07-25, 22:59:

Replace them with appropriately rated tantalum. I would not bother replacing the ones that still work. Tantalums don't dry out like electrolytics they just short out sometimes.

Ok that's good to know. Seems there is a 160 ohm short to ground on the 5v rails too. I need a beefier power supply to blow them i think. 3 amps is not doing it 🤣.
Heaps to test on this board. Have not gone through them all yet.

160 ohms is not a short 🤣

maxtherabbit wrote on 2020-07-26, 00:22:

160 ohms is not a short 🤣

Technically its not a "dead" short (zero ohms) 160 ohms is pretty low resistance and my supply is drawing its maximum 3Amp output when connecting it to the board.

Max is right in that 160ohm on the 5v is not a short, 160 ohm is a bit low but there is a lot of TTL circuitry on your board which draws more than say a 286 NEAT chipset where most IC's are LSI and draw very little. Your PSU has a max of 3A on 5V is way to small ! That is not even comparable to an old 120watt LASER XT PSU with 20A on 5v and 5.2A on 12v. Just sayin....

Horun wrote on 2020-07-26, 02:02:

Max is right in that 160ohm on the 5v is not a short, 160 ohm is a bit low but there is a lot of TTL circuitry on your board which draws more than say a 286 NEAT chipset where most IC's are LSI and draw very little. Your PSU has a max of 3A on 5V is way to small ! That is not even comparable to an old 120watt LASER XT PSU with 20A on 5v and 5.2A on 12v. Just sayin....

Yeah I wasn't too sure on what the current draw was supposed to be, but the cpu draws a good 1 amp alone.

Just decided to pull out other chips to test things and it powers on with tbis chip removed, but no image on screen. Probably needs the chip for something important.
The sticker says C0080B on it.
The socket is labelled 16L8A on it, I will see what's written on the chip under the sticker.

Perhaps its just a dead chip?

Good job ! Hope it is not a PLA, they generally cannot be replaced because there would be no documentation on how the array was originally programmed so you would not be able to program a replacement chip.

nzoomed wrote on 2020-07-25, 22:22:

Ok that sounds a good idea. I expect that there will be some that go hot or explode. If that is not going to damage the board, its probably the easiest way. My power supply also has a meter to display current draw.

You need to be very careful when using this method, as if the current is too large, the trace might blow before the cap explodes if it happens to be on a thin trace.

nzoomed wrote on 2020-07-26, 02:26:

Yeah I wasn't too sure on what the current draw was supposed to be, but the cpu draws a good 1 amp alone. […]
Show full quote

Yeah I wasn't too sure on what the current draw was supposed to be, but the cpu draws a good 1 amp alone.

Just decided to pull out other chips to test things and it powers on with tbis chip removed, but no image on screen. Probably needs the chip for something important.
The sticker says C0080B on it.
The socket is labelled 16L8A on it, I will see what's written on the chip under the sticker.

Perhaps its just a dead chip?

You could put this chip back to confirm it is indeed causing the short. The 16L8A is a PAL chip and if it's bad it'll be really troublesome. You can replace it with a newer GAL chip, but you'll need to find a programmer and also the fuse map (JED) file.

Horun wrote on 2020-07-26, 02:40:

Good job ! Hope it is not a PLA, they generally cannot be replaced because there would be no documentation on how the array was originally programmed so you would not be able to program a replacement chip.

Yup its a PLA unfortunately when I checked the part code.

Looks like some proprietary logic programmed to it I guess?

The sticker may have some info surrounding this if we can locate the board manufacturer and find another board to read another PLA?

kalohimal wrote on 2020-07-26, 02:54:

You need to be very careful when using this method, as if the current is too large, the trace might blow before the cap explodes […]
Show full quote

nzoomed wrote on 2020-07-25, 22:22:

Ok that sounds a good idea. I expect that there will be some that go hot or explode. If that is not going to damage the board, its probably the easiest way. My power supply also has a meter to display current draw.

You need to be very careful when using this method, as if the current is too large, the trace might blow before the cap explodes if it happens to be on a thin trace.

nzoomed wrote on 2020-07-26, 02:26:

Yeah I wasn't too sure on what the current draw was supposed to be, but the cpu draws a good 1 amp alone. […]
Show full quote

Yeah I wasn't too sure on what the current draw was supposed to be, but the cpu draws a good 1 amp alone.

Just decided to pull out other chips to test things and it powers on with tbis chip removed, but no image on screen. Probably needs the chip for something important.
The sticker says C0080B on it.
The socket is labelled 16L8A on it, I will see what's written on the chip under the sticker.

Perhaps its just a dead chip?

You could put this chip back to confirm it is indeed causing the short. The 16L8A is a PAL chip and if it's bad it'll be really troublesome. You can replace it with a newer GAL chip, but you'll need to find a programmer and also the fuse map (JED) file. To get the JED file you'll need to find another identical board and read it from the same chip with the programmer.

Yes I accept it was a risky method, but did it any way.

All traces are OK.
Have since tested all the other caps and no shorts.

I think the PLA is the issue. Looks like there is not much I can do unless I can find another board identical?

I was wondering if I dump the ROM chips that I might be able to find some info on the manufacturer if I opened them in a hex editor?

Looks like the BIOS is stored onn2 chips.

So if you insert the 16L8A back into the socket, the board no longer powers on? Because this is a 5V chip and shouldn't cause short on 12V. It's Vcc is pin 20 and Gnd is pin 10.

Yeah you can find the BIOS ID string if you could dump its content. But I won't put too high hope as the board is long discontinued and finding the PAL code would be very hard.

kalohimal wrote on 2020-07-26, 03:32:

So if you insert the 16L8A back into the socket, the board no longer powers on? Because this is a 5V chip and shouldn't cause short on 12V. It's Vcc is pin 20 and Gnd is pin 10.

Yeah you can find the BIOS ID string if you could dump its content. But I won't put too high hope as the board is long discontinued and finding the PAL code would be very hard.

Sorry, I thought I mentioned earlier that there is still enough of a drain on the 5V rail to cause the PSU to enter overload protection. I will take some measurements with the chip inserted and with it removed.
I tried another power supply and the board is draining 5A with that chip installed.

It's just to double confirm this chip is indeed the culprit. As mentioned if this chip failed, the hope of fixing the board is not high, unless you could find another identical board. Even so you might not be able to read the PAL if its security fuse is activated.

kalohimal wrote on 2020-07-26, 03:52:

It's just to double confirm this chip is indeed the culprit. As mentioned if this chip failed, the hope of fixing the board is not high, unless you could find another identical board. Even so you might not be able to read the PAL if its security fuse is activated.

Yeah not looking good.

I will check the pinouts and see what current draws when connected to the power supply.

Looks like I've got to try and track down another 286 board ...

5A is way too much even with all that logic. Yes Too bad it is a PLA. It may not be the whole issue as it can be used to turn on other chips based on the code it receives which could be the real culprits. I can see traces from it going down to ISA slots and also up to 74f245 and U39 (that has little or no label) plus some traces go up to J10 jumper pads from what little is visible in the picture.
If you can dump the BIOS please post it here. Double check the ISA slots for possible shorts or corrosion inside. https://en.wikipedia.org/wiki/Industry_Standard_Architecture
shows where the +5v and 12v is if you do not have another reference.

Lol put the chip back in and now it works!
It powered on but no image on boot up. I then went and moved the video card to another slot and hey presto it works!

Go figure!

Looks like i just need to replace the tants, that I've removed, a new cmos battery and good im to go.