286 Motherboard repair - ongoing w/ issues

[Shock]

Don't check resistance while the circuit is powered. It supplies it's own voltage in order to take a measurement and you will damage some multimeters by doing that. But 60 milliohms is about two test leads, so that is fine it seems like a good connection.

Swap the psu if you have another known good one to test if possible. If not you need to google how to load test a psu with a dummy load or electronic load.

I suspect if that checks out then there is either a partial voltage rail short on the board which could be caused by a component, a voltage rail short to somewhere it shouldn't. Or a missing ground or another shorted component connected via the cpu.

Lots of resistance testing or when you can... socket the cpu then touch test (being careful not to burn yourself) looking for other components getting warm.

[Nexxen]

Don't check resistance while the circuit is powered. It supplies it's own voltage in order to take a measurement and you will damage some multimeters by doing that. But 60 milliohms is about two test leads, so that is fine it seems like a good connection.

Swap the psu if you have another known good one to test if possible. If not you need to google how to load test a psu with a dummy load or electronic load.

I suspect if that checks out then there is either a partial voltage rail short on the board which could be caused by a component, a voltage rail short to somewhere it shouldn't. Or a missing ground or another shorted component connected via the cpu.

Lots of resistance testing or when you can... socket the cpu then touch test (being careful not to burn yourself) looking for other components getting warm.

How to load is something I'll look up. Learn once and forever done.
I ordered a load of sockets to test different components (maybe swapping from other boards). The seller connected a +5V to an input on the MC74, wrongfully thinking it was a dead trace but in fact it was the one next to it (even I firstly didn't notice). could have killed it.
While waiting I'll desolder other stuff and check for shorts and on a motherboard. Ruling out stuff is good + my TL-866 programmer can check those 74 logics and tell if dead.


I have a laser (or what tech it is) to test temperature. The days I was getting my finger burnt are over! 

[Shock]

Here is a simple test, but I recommend you learn the power or current ratings of each voltage on the psu. I'd not load more than 80% of the maximum power or current rating of each voltage. Reason why is you shouldn't be loading it more than that with the motherboard anyway. If there is a motherboard fault and it draws more than 80% then you got a motherboard fault or an underrated supply. Needlessly testing a vintage supply (if it is) at 100% may induce a failure so our goal isn't to create another repair unless you have some kind of intermittent psu problem.

Note that same color positive voltage wires may be the same rail connected inside the power supply. So if testing multiples at once careful to not needlessly overload the same rail.

So say you have a 60W 12V bulb. That is 5A or 60W load when connected to a 12V positive and common. You then check the voltage for DC and AC with a multimeter or an oscilloscope (don't blow your scope, remember never connect the ground lead of the probe to a grounded referenced, non floating voltage). The bulbs will get hot so put them on something appropriate and don't leave unattended. Voltage sag (loss of voltage) or excessive ripple can mean the psu is struggling to regulate.

You can buy or make fairly inexpensive electronic loads which can be adjusted but I still recommend reading how they are used. In your case you want to ensure it can handle the power/current/voltage ranges you require for psu testing.

[Shock]

Notice he is jumping the power good - PS_ON pin in the power supply so it starts up. You need to be aware that early power supplies used different wiring and standards. Don't just do random pin to pin without checking color coding. In the same way connecting some old psu to some old motherboard without checking can reveal that in fact they aren't pin compatible.

Automotive bulbs can be brought inexpensively, or acquired free if you know who to ask.

[Nexxen]

Ok, I think I got what you mean.

I also have other AT PSUs to try. I have a ATX-to-AT connector too.
Negative voltages aren't an issue for the sytem itself IIRC.

I'd like to have a bordview file of some 8086/286/386 now, I'm curious to see what it's all about with motherboard design.

[fzabkar]

Notice he is jumping the power good - PS_ON pin in the power supply so it starts up.

Power_Good (PWR_OK) and PS_ON are two different pins.

https://pinoutguide.com/Power/atx_v2_pinout.shtml

[fzabkar]

I also have other AT PSUs to try. I have a ATX-to-AT connector too.

Your 286 motherboard and AT PSU do not have a PS_ON pin.

https://pinoutguide.com/Power/MotherboardPower_pinout.shtml

[Nexxen]

I also have other AT PSUs to try. I have a ATX-to-AT connector too.

Your 286 motherboard and AT PSU do not have a PS_ON pin.

https://pinoutguide.com/Power/MotherboardPower_pinout.shtml

Ok, that I know too
It was to say that I have more and try more... just for good measure.


I desoldered all the electrolytics on the power rails. All 0.8-9% Vloss; ESR was between 0.8-1.2 ohm.

[amyk]

If the CPU is getting to 50C immediately, that doesn't seem normal for a 286.

I haven't touched one in a long time but I remember those early CPUs would only get warm to the touch, not burning hot, after prolonged use; late 486s and Pentiums is when heatsinks and fans started being necessary.

I'd like to have a bordview file of some 8086/286/386 now, I'm curious to see what it's all about with motherboard design.

Here's a 386: https://alexandrugroza.ro/microelectronics/system-design/isa-80386dx-sbmc/index.html

[Nexxen]

If the CPU is getting to 50C immediately, that doesn't seem normal for a 286.

I haven't touched one in a long time but I remember those early CPUs would only get warm to the touch, not burning hot, after prolonged use; late 486s and Pentiums is when heatsinks and fans started being necessary.

I'd like to have a bordview file of some 8086/286/386 now, I'm curious to see what it's all about with motherboard design.

Here's a 386: https://alexandrugroza.ro/microelectronics/system-design/isa-80386dx-sbmc/index.html

286 can get quite hot, not abnormal. On my working 286 board it's 50°C.

It gets to 50 on a curve, it's not 0 to 50. Plus I desoldered it and it works 100%.

As soon as the socket comes in I'll be able to do more tests.

[Nexxen]

I tried the keyboard microcontroller (Jet Key), in other boards it would not work (no POST) or give Gate 20 error.

Could this be the sign of a malfunctioning of the keyb controller?

[Nexxen]

I tried the keyboard microcontroller (Jet Key), in other boards it would not work (no POST) or give Gate 20 error.

Could this be the sign of a malfunctioning of the keyb controller?

In any board that has a keyboard micrcontroller, the Jet Key displays Gate A20, nothing or 0605 0807 post codes - and hangs there.

[Nexxen]

I tested the keyboard controller, Jet Key, and it won't work in any other motherboard.
Maybe it is this?

I can't wait to get the PLCC socket and test other keyb controllers in it.

[Nexxen]

DISCLAIMER: this is a pure "I have no idea but I'll try to push my solitary brain cell to do some work".

Let's reason on this:

my assumptions:

1) the 286 is divided into 2 zones, powered by 2 different Vcc inputs
2) each zone has a Vdd (gnd) - those are pointless here as they are GND

my thoughts

1) if zone2 is powered by Vcc2, and zone 2 is associated with Addresses (A0 to A23)
2) there is something shorted to ground in the ICs/other connected to Addresses
3) this could be avoiding zone2 from working properly and never reach the POST stage

My possible culprits:

i) Keyboard microcontroller (8042)

ii) 2 chips associated to it, located underneath

iii) some other transceiver/multiplexer to ISA bus

Nothing is shorted to ground, but internally something went wrong and an input is now associated to an output to ground, hence my transceiver/multiplex/8042 failed components.



Yes, this is baseless mainly. But it's a starting point to learn from mistakes 

[Nexxen]

PLCC68 socket came in today. From now on I can try cpus.

I put in the BIOS of an old Compaq 286 and I get a few codes from BIOS. 01/08/09, 09 0E 07....
With one Keyb controller it issues 00 01.
These Compaq BIOS images are 128 bit, not 256. Modifying the size via jumper changes nothing.
Reset LED lights up only at Power on and off. Is this good?

The bios images it came with work with another board but give checksum error. They should issue some code, probably they aren't this board's images.
I have to try new bios images.

There is a problem that I can't figure out, RESET does nothing when triggered. Only with one keyb controller it works, but just cycles IRDY and starts over.


Bottom line:

I can safely assume that the CPU is decoding something.

[Nexxen]

Chip HD14818P ha issues:

RESET should go LOW during power on, but it immediately goes and stays HIGH.

I'll desolder and socket it.

[Nexxen]

Tested the HD14... works in another board.

Issue is starting to smell like chipset.

[Nexxen]

I desoldered sn7406N and MC74HC14N (the two icd under the keyboard controller).
I tested them with XgPro, and they both passed.
IDK if this in conclusive or if I should buy replacements.

Whatever keybaord controller I stick in it won't have the reset line working anyway.

I'm out of ideas.

[Nexxen]

I desldered the keyboard socket and cleaned under it. Replaced with a new one.
Some traces on the back of the board had microcracks that became cracks and that was probably preventing some porcesses.

Now RESET works when triggered.
But IRDY on POST card isn't lighting up. Something isn't being read.

I socketed the cpu, but it broke. Bad quality stuff.

[Nexxen]

8MHz Xstal, connected to Keyboard wasn't working anymore.
When "alive" it had an erratic signal (not a nice wave but a very irregular shape, didn't take a pic so go with a sine that had many creases).

Ordered one, waiting to arrive and solder it back in place.

[Nexxen]

I replaced the xstal and still nothing out of it, no 8 MHz.

Next thing I'll check is a dead cap between xstal and keyboard controller.

I have RESET working, something is alive but without that 8 MHz nothing is happening.

[Shock]

If the clock is derived from the keyboard itself you will need it present. Some boards need the keyboard to boot. See if the keyboard leds flash or light up, the old quick test of toggling the caps and numlock is good to see if it's in a hung, off or normal state.

Don't assume some operation previous hasn't hung either. Keep updating.

[Nexxen]

To see if it works I simply use a POST card.

I needed some time to wash it off my brain. I'm planning on going back to it soon.
I'll test it witha keyboard connected. When I say it takes time to solve mysteries... it's my turn to hunt down one

Thanks for the input

[Nexxen]

Keyboard lights up but no signs of life after all LEDs lighting up at power on.
Numlock doesn't work.


My next step is to remove all the sockets and check for broken traces.

[Nexxen]

Latest news:

- RESET signal is asserteed both at power on and off.

- removing keyboard controller and RTC and BIOS chips (or any combination) makes no difference with the POST card:

reset lights on at power on and at power off;
FRAME is lit;
CLK is lit;
IRDY is always off

The keyboard controller that came with the board is dead, Fujitsu, Holtek and VIA make the 8MHz oscillator work, albeit @4.45V


My next step:

desolder all sockets and check for broken traces

Before I desoldered the keyboard controller and the RTC, Reset was stuck and POST card was stuck at 02 03 05 06 07 08 09 codes.
Cpu is good and working.


Any suggestions for a good desoldering gun? I have a hand pump but it doesn't work that well.