286 Motherboard repair - ongoing w/ issues

[Nexxen]

Hello!!

I bought a 286 motherboard, advertised as "damaged". Pics from the seller were pretty straightforward.
He added that he couldn't make it POST, the cpu was scalding hot and that no BIOS reflash could help.

After a thorough inspection I found many undetected issues that might have prevented the board from POSTing at all.

I attached the seller's pics.

Most prominent is the jump wire.

[Nexxen]

Also, many solder joints were cracked - and that alone took quite some time to inspect and reflow.

It was a little dirty and had some mild corrosion on some other DIP-14 ICs. Nothing too serious but needed the vinegar+soap wash combo.
Everything came out spic and span.

The jump wire was soldered to the wrong leads:

the seller made a R5 to pin11 of a MC74HC14AN (https://pdf1.alldatasheet.com/datasheet-pdf/view/164836/MOTOROLA/MC74HC14AN.html),
that in turn connects to pin 13 of a SN7406N Datasheet (https://pdf1.alldatasheet.com/datasheet-pdf/view/27354/TI/SN7406N.html);
when in fact it was connected to R6.
-- my guess is that he followed the wrong trace and on the wrong assumption that is wasn't making contact he wired it.

I hope nothing happened there.

Traces:
some were actually good on continuity test, but when I scratched away the old solder mask a couple came off in bits;
another one was gone;
two others had indents of corrosion.

I tinned everything and restored the dead traces. Nothing fancy, it was easy.


Socketed chips:
- BIOS chips were so oxidised that my TL-866 struggled to read them indicating bad contact. Cleaned and perfect reads and writes.
- Keyboard BIOS, same, heavily oxidised. Cleaned and I hope good to go.

Sockets deoxidised thoroughly, let's hope it's good too.

I had to replace a cap that had a broken leg (it came straight out) (C63 pic), removed a broken 10K resistor (replaced with a smaller size - see R4).


Yesterday I was too tired to fire it up and watch it fail - LOL
Later today I'll test it.

[Nexxen]

Questions:

My last concern is the size of the R4 10K resistor, smaller size could mean that it will go smoking away.
I have no match for this one. I don't know the wattage of the bigger and of this one.

Also, in order to replace the two ICs involved in the jump wire, SN7406N is avilable, but MC74HC14AN looks way lessa vailable.
Is there a newer perfect replacement for this one?

[Nexxen]

I have issues with the crystals. I'll try to post pics from the oscilloscope.
BTW, board powers on but no POST.
Cpu goes up to 50+ °C (normal to me).
Edit: cpu is good

32KHz:
normal, 32 on both leads;
it displays a sine wave and on the other lead a square wave.

14MHz:
- on one lead sine wave, other oddly "dented" wave
Edit, the wave is squared by an ic and is correct (my mistake interpreting)

32MHz has only one out of the four leads issuing a signal, 32MHz, the other 3: 2 dead, one just Vcc in
edit: Datasheet is 2 nc 1xVcc and 1xOutput

8MHz (keyb):
one 94 KHz (but sometimes 9-8 MHz) and other lead dead.
Edit: without turbo, with turbo on it's correct

[Nexxen]

I closed the turbo switch and the 8MHz is now 8.

On the cpu, CLK is 16MHz; no change with turbo ON or OFF.

ISA Pin B2 is 94KHz, is this frequency ok for some communication?

Also, it won't RESET. Shorting the leads does nothing.
It does light up on pwron, and at poweroff (displaying FF 00, or FF --  for a split second).
Edit, even with bios chips removed; irrelevant then.

Next step: try some other BIOS. But I don't think that's the problem.

If BIOS still fails posting I'll desolder the cpu and socket it.


Edit: One Vcc on cpu is 5.17V, the other one, straight from the 5V rail as well, reads 3.17, 3.74V.
Maybe the cpu is dead after all.

[Nexxen]

I tested the BIOS on another 286 and it POSTS up to a certain point.
I get codes and that is enough to know it can work.

I checked the CPU Vcc and the one that is getting 3.74V on this board, gets the full 5.15V required.

The cpu is actually dead as per seller's prognosis or this is a motherboard issue.
I'm pretty sure it's the cpu at this point.

When I hit reset nothing happens (it has to be drawn HIGH), and it just loses "Frame" on POST card, no reset happening.
I'll desolder and socket the cpu. Maybe I'll try it on another board, but I also have plenty of PLC68 286...

[HwAoRrDk]

I wonder what the justification of the "fastest keyboard BIOS" tagline on that Jet-Key keyboard controller is. Not a brand I've ever seen before.

[Zero999]

Questions:

My last concern is the size of the R4 10K resistor, smaller size could mean that it will go smoking away.
I have no match for this one. I don't know the wattage of the bigger and of this one.

Don't worry. The voltage would have to be much higher to overheat, even a small 10k resistor, which would just dissipate 2.5mW at 5V.

Also, in order to replace the two ICs involved in the jump wire, SN7406N is avilable, but MC74HC14AN looks way lessa vailable.
Is there a newer perfect replacement for this one?

No, don't use the 74HC series. It has different logic levels than old TTL. The 74HC14 also has a push-pull output, where as the 7406 has an open collector. A modern equivalent, which is still available, is the 74LS06.

I tested the BIOS on another 286 and it POSTS up to a certain point.
I get codes and that is enough to know it can work.

I checked the CPU Vcc and the one that is getting 3.74V on this board, gets the full 5.15V required.

The cpu is actually dead as per seller's prognosis or this is a motherboard issue.
I'm pretty sure it's the cpu at this point.

When I hit reset nothing happens (it has to be drawn HIGH), and it just loses "Frame" on POST card, no reset happening.
I'll desolder and socket the cpu. Maybe I'll try it on another board, but I also have plenty of PLC68 286...

It does sound like the CPU. Have you tried removing it to see what voltages you get? The low voltage on VCC could be due to it drawing too much current and pulling the supply voltage down.

[wasedadoc]

No, don't use the 74HC series. It has different logic levels than old TTL. The 74HC14 also has a push-pull output, where as the 7406 has an open collector. A modern equivalent, which is still available, is the 74LS06.

The OP is not proposing to use a 74HC14 to replace a 7406. Both types of chip are already on the board and both are to be replaced.

[Nexxen]

1) Thanks! No, I'm looking for an equivalent replacement. I knew there was better and "more modern".

2) I desoldered the cpu, put it in another socketed board and it works 100%.

However I discvoered that the Vcc on pin 30 wasn't connected, or just a small piece of the trace was.

Later I'll press down the cpu on it to see if it was that.

[Nexxen]

No, don't use the 74HC series. It has different logic levels than old TTL. The 74HC14 also has a push-pull output, where as the 7406 has an open collector. A modern equivalent, which is still available, is the 74LS06.

The OP is not proposing to use a 74HC14 to replace a 7406. Both types of chip are already on the board and both are to be replaced.

Yes, exactly.
If nothing works I'll buy new chips.

[Nexxen]

I'm out of PLCC68 smd sockets. All used.
I'll have to order and wait 15 days. Well, that's it so far I believe.

Any other advice is welcome and don't feel cold posting any

[Haenk]

Obviously the battery leaked, and is missing completely now. A via has been drilled out. That really could be an issue.
Also several traces show damage from leakage, you certainly need to take care of that.
No battery at all might be an issue as well.

[Nexxen]

Obviously the battery leaked, and is missing completely now. A via has been drilled out. That really could be an issue.
Also several traces show damage from leakage, you certainly need to take care of that.
No battery at all might be an issue as well.

I'll desoler the sockets and chips and check for more broken traces, probably right to check more in the area.
As I don't have a desolering gun it's gonna take some time... Maybe next week, to give me some new view on the whole matter.

[Nexxen]

The cpu was proven to be functional. But the 2nd Vcc input was below minimum required (+5V).

I tested the smal 104 cermaic capacitor the +5V was coming from (a small trace from hole to pin), and it was a 130nF /0.1% /2.4ohms ESR, the V was 5.xxV (as intended).

Question:

if it wasn't being siphoned by this cap, why Vcc is lower on the cpu? What else could be affecting it?
Vss were = 0V.

https://pdf1.alldatasheet.com/datasheet-pdf/view/1284116/AMD/80286.html

[Haenk]

Maybe check/replace the electrolytic caps as well, 30 years is quite some time, and Samsung (SEMCO) was not exactly top-tier back then.

[Zero999]

No, don't use the 74HC series. It has different logic levels than old TTL. The 74HC14 also has a push-pull output, where as the 7406 has an open collector. A modern equivalent, which is still available, is the 74LS06.

The OP is not proposing to use a 74HC14 to replace a 7406. Both types of chip are already on the board and both are to be replaced.

Yes, exactly.
If nothing works I'll buy new chips.

Sorry. I misread. Replace the 7406 with 74LS06, but replace the 74HC14 with similar. It should be fairly easy to get hold of. There are better, faster, alternatives such as the 74AC14, but more speed isn't always better. The faster rise and fall times can cause instability on a board which is not designed for it. Don't worry about the MC bit, any 74HC14, in a DIP package will do.

The cpu was proven to be functional. But the 2nd Vcc input was below minimum required (+5V).

I tested the smal 104 cermaic capacitor the +5V was coming from (a small trace from hole to pin), and it was a 130nF /0.1% /2.4ohms ESR, the V was 5.xxV (as intended).

Question:

if it wasn't being siphoned by this cap, why Vcc is lower on the cpu? What else could be affecting it?
Vss were = 0V.

https://pdf1.alldatasheet.com/datasheet-pdf/view/1284116/AMD/80286.html

Have you measured the resistance of the VCC pin to the power supply rail? Perhaps it's a bad trace/solder joint somewhere.

[evac]

The cpu was proven to be functional. But the 2nd Vcc input was below minimum required (+5V).

I tested the smal 104 cermaic capacitor the +5V was coming from (a small trace from hole to pin), and it was a 130nF /0.1% /2.4ohms ESR, the V was 5.xxV (as intended).

Question:

if it wasn't being siphoned by this cap, why Vcc is lower on the cpu? What else could be affecting it?
Vss were = 0V.

https://pdf1.alldatasheet.com/datasheet-pdf/view/1284116/AMD/80286.html

Are you able to load test that supply rail to check if it can deliver the required current?

[Nexxen]

No, because I don't know how. I have to ask for your help on this (or link a guide).
I'd look it up by myself but no clue what I'd be looking at.

PSU is working 100%, AFAIK. I could check voltages once again to rule that out.

[Nexxen]

Thanks and will do as suggested.

It's a nice little neat board, I'm pretty sure it's a thing so obvious I'm not even thinking about...

[evac]

No, because I don't know how. I have to ask for your help on this (or link a guide).
I'd look it up by myself but no clue what I'd be looking at.

PSU is working 100%, AFAIK. I could check voltages once again to rule that out.

the point is just to get a realistic load on the rail and see if the voltage stays stable and doesn't drop
I don't know in this case if the rail is directly provided by the PSU or regulated on the board itself

ideally you would use an electronic load connected to the rail and increase the current until you either reach the expected load or the voltage starts to drop too much (indicating failure)
a barebones electronic load can be had for relatively low money from the usual places
alternatively, you can also use an adequately sized resistor (in both value and power dissipation) connected to the rail to draw some current

[Nexxen]

EDIT1: after checking voltages on another board:
1. -3.25 as recorded on CAP pin, a negative voltage is correct.
2. +5 on Vcc pin 30 is correct but not present on this board.
3. the 473 cap is there too, I guess that even 60nF should be good, as no short and no res is present (at least up to 2M)
4. without cpu CAP is = to 0.00V as no cpu is installed

Last issue is no Vcc = +5V on pin 30

What is down here is outdated.



I checked my notes taken a few days ago.

I have one cap (C43 - ceramic) with a reading of -2.4V (minus 2.4V), connected to CAP pin (see attached pic), that might be preventing the internal circuit to reach reset before POST operations.
Couldn't find its source, nothing beeps. I'll need a wider probe (i.e. a cable with smaller threads) to find where it is generated.

I'll post more in a while.

Edit: no cpu, cap43 readings are 0.00V on +.
Check Intel's CAP definition. Can't get what it means, looks like CAP is connected to Gnd via a ?? capacitor? Makes no sense to me.

Cap43 is a 473 value, 0.047µF or 47nf, reads 62nF. Maybe it was leaking to G?? Other 104 (100nF) read +20% or 30% higher. Vloss on all is 0.1%, ESR 2.1 to 2.4 ohms. Could be bad too??
I can buy new ones.

[Nexxen]

No, because I don't know how. I have to ask for your help on this (or link a guide).
I'd look it up by myself but no clue what I'd be looking at.

PSU is working 100%, AFAIK. I could check voltages once again to rule that out.

the point is just to get a realistic load on the rail and see if the voltage stays stable and doesn't drop
I don't know in this case if the rail is directly provided by the PSU or regulated on the board itself

ideally you would use an electronic load connected to the rail and increase the current until you either reach the expected load or the voltage starts to drop too much (indicating failure)
a barebones electronic load can be had for relatively low money from the usual places
alternatively, you can also use an adequately sized resistor (in both value and power dissipation) connected to the rail to draw some current

This is beyond my expertise.
I'll have to understand how in order to get there. Learning curve...

Thanks for the input!

[Shock]

So the only fault you are aware of is a missing voltage rail? So one VCC is good, ones missing? If you added a socket for the cpu then check your soldering work and that the problematic VCC is a low ohms connection and not shorted anywhere (to ground or a neighboring pin for instance). Then trace the VCC back to the power connector. You can also see if that rail is used elsewhere or other power rails are missing.

Perhaps a good time to take some better images of the board.

[Nexxen]

So the only fault you are aware of is a missing voltage rail? So one VCC is good, ones missing? If you added a socket for the cpu then check your soldering work and that the problematic VCC is a low ohms connection and not shorted anywhere (to ground or a neighboring pin for instance). Then trace the VCC back to the power connector. You can also see if that rail is used elsewhere or other power rails are missing.

Perhaps a good time to take some better images of the board.

+5V is present everywhere.
The Vcc from the capacitor (as on other 286 boards) is measured +5V on the leg; the pin of the cpu is 3.74V. 1.2 is missing.
PSU +5V to the capacitor is 0.06 ohm, powered off; powered on 0.00.

I probably measured Amps making a mistake, probes became hot, 0.25mA from G to +5V (I don't know how to use a multimeter to measure A, but I think I killed it some time ago when I got it sparking...). Have to buy a new one.

What, inside the cpu, could be drawing out 1.2V? Some short on a BUS line? Some 74 chips? LOL when they say learn by getting kicked...

Chipset isn't getting warm, probably because of a missing cpu.


I didn't solder a socket yet, I bought 5 and used them all elsewhere, I have THT sockets that can't be used. I ordered some and have to wait 15 days.
I also oredered new caps, in the meantime I'll use tantalums like on any other board.