PWR_GOOD signal not working properly perhaps? Even though there doesn't seem to be any significant battery damage. See that thin trace from power connector to nearby chip on the bottom of the PCB? That's the signal in question. Seems like there is another chip under the RTC, most likely a 4069, could be it died. The signal is probably getting inverted by one of the gates so check if you get that. You can do that with RTC out of socket, the mobo won't run without it but should at least show some codes.

Attach a reset button as well. Does it do anything? Usually the reset button is shorting a signal to GND, and the signal is held high with a pull-up resistors. If that resistor is open you might be in permanent manual reset mode. Check that too.

Deunan wrote on 2025-05-09, 17:05:

PWR_GOOD signal not working properly perhaps? Even though there doesn't seem to be any significant battery damage. See that thin trace from power connector to nearby chip on the bottom of the PCB? That's the signal in question. Seems like there is another chip under the RTC, most likely a 4069, could be it died. The signal is probably getting inverted by one of the gates so check if you get that. You can do that with RTC out of socket, the mobo won't run without it but should at least show some codes.

Attach a reset button as well. Does it do anything? Usually the reset button is shorting a signal to GND, and the signal is held high with a pull-up resistors. If that resistor is open you might be in permanent manual reset mode. Check that too.

thank you for answer!

I have tried to manually reset the motherboard shortening the pins before turning on, after but it does not make anything, any difference, or any indication that I have triggered the reset. The resistor here is not shortened, and the voltage on the reset pin is about 3,2V.

I have also checked the first pin, the trace is good, delivering 5V to the IC beneath the RTC. How to check if on of the the gates is inverted? --- sorry, just to make sure I will do the right thing.

thx, tomas

Well, first find out what that chip under RTC is excatly. Then find the datasheet. For example for 4069 type inverter the pin 1 is input to a gate, and pin 2 is the output of the same gate. So pin 2 should always have the opposite logic state to pin 1. The 4069 is a very commonly used as low power oscillator for the 818 family RTC (which needs 32768Hz clock) and the unused gates can do other things, like inverting reset signal for whatever chip that needs it.

A quick note, sometimes the mobo is actually working but the ISA RESET is stuck. It's a separate signal from the mobo and CPU reset. Another thing that can control the reset line is the keyboard controller (KBC). This mobo is old enough to use such solution. You can try running it witout the KBC, or try a different chip if you have one. It's socketed so might have bad contact or was replaced with incompatible variant.

Try reseating all reseatable chips (RTC, BIOS, KB ROM) to ensure the connections are good.

Also, JP4 and JP5 (between BIOS & KB ROM) could be on the wrong setting. Is JP5 BIOS clear?

MikeSG wrote on 2025-05-10, 08:36:

Try reseating all reseatable chips (RTC, BIOS, KB ROM) to ensure the connections are good.

Also, JP4 and JP5 (between BIOS & KB ROM) could be on the wrong setting. Is JP5 BIOS clear?

hi, I have tried to reseat them a few times, cleaned the contacts, etc... no difference. I have checked the leg on all ICs as well.

I do not have any documentation for the board but one of the jumpers switches between monochrome and color, the second maybe CPU speed, dont know for sure.

br, tomas

Deunan wrote on 2025-05-09, 23:00:

Well, first find out what that chip under RTC is excatly. Then find the datasheet. For example for 4069 type inverter the pin 1 is input to a gate, and pin 2 is the output of the same gate. So pin 2 should always have the opposite logic state to pin 1. The 4069 is a very commonly used as low power oscillator for the 818 family RTC (which needs 32768Hz clock) and the unused gates can do other things, like inverting reset signal for whatever chip that needs it.

A quick note, sometimes the mobo is actually working but the ISA RESET is stuck. It's a separate signal from the mobo and CPU reset. Another thing that can control the reset line is the keyboard controller (KBC). This mobo is old enough to use such solution. You can try running it witout the KBC, or try a different chip if you have one. It's socketed so might have bad contact or was replaced with incompatible variant.

thank you for the notes,

the IC beneath the RTC is a TC4069... I have enclosed the picture of it, and measured all voltages, and as well compared them to the alternative board - a picture of it enclosed as well. The readings are the same for both board expect that the alternative damaged board is passing the reset. I have used the old board to experiment a bit and disconnected the rail from the power connector that goes to the TC4069, and the damaged board went through the reset as before.

Pin 1: 5V ... Pin 14: 5V
Pin 2: 0V ... Pin 13: 0V
Pin 3: 2,5V ... Pin 12: 5V
Pin 4: 2,6V ... Pin 11: 0V
Pin 5: 2,6V ... Pin 10: 5V
Pin 6: 2,6V ... Pin 9: 2,6V
Pin 7: 0V ... Pin 8: 2,6V

I have been tracing the RESET from the ISA slot to an HD74LSO4P, PIN 1. PIN 2 goes to the KBD bios on the RESET pin. On the board the PIN 1 is always high, on the alternative board PIN 1 goes high, and then low - this is without any KB, BIOS or RTC in the socket. Also, I have compared the voltage and resistance of the reset switch with the damaged board and they are the same. A bit worrying for me is the as well the fact that the CLK is no indicated by the post card and as well using the scope the ISA pin with the CLK has no signal. All OSC on the board are good.

I have tried to change the KBD bios for different ones, no difference. It looks to me like the chipset is not doing anything but I can not find a reason for it. Any suggestions what to try?

br, Tomas

Hm. So you checked both clock sources with a scope? The 66MHz one is simple, it's a complete oscillator so when it gets 5V power it should produce output on it's own. But the 14.31818MHz needs a phase inverting amplifier and I guess the 74LS04 is used for that. So first make sure this clock is actually working (if you haven't already). Usually the clock output is not taken directly from the crystal but goes through yet another gate, this acts as a separator so that the trace and other connected chips do not load the resonating circuit. So check which gate output is connected to the big mobo chip and if the clock signal is present there. If so, and the chipset still does nothing, it might be dead. But if there is no clock output (or it's not in spec - weak for example) then the chipset might not be releasing its own internal reset.

It could also be some power delivery issue to the chipset. Inspect all pins, resolder if questionable. IIRC this chip need to receive both the CPU clock (66MHz) and the mobo clock (14.3MHz) to work properly. This mobo looks to be made pretty decently, the clock traces (especially for the CPU) should be using more copper to lower the resistance. This might make it easier to figure out what pin it goes to.

There is also the 32768Hz clock for RTC but it doesn't affect the reset in any way even if missing.

Deunan wrote on 2025-05-10, 22:32:

Hm. So you checked both clock sources with a scope? The 66MHz one is simple, it's a complete oscillator so when it gets 5V power it should produce output on it's own. But the 14.31818MHz needs a phase inverting amplifier and I guess the 74LS04 is used for that. So first make sure this clock is actually working (if you haven't already). Usually the clock output is not taken directly from the crystal but goes through yet another gate, this acts as a separator so that the trace and other connected chips do not load the resonating circuit. So check which gate output is connected to the big mobo chip and if the clock signal is present there. If so, and the chipset still does nothing, it might be dead. But if there is no clock output (or it's not in spec - weak for example) then the chipset might not be releasing its own internal reset.

It could also be some power delivery issue to the chipset. Inspect all pins, resolder if questionable. IIRC this chip need to receive both the CPU clock (66MHz) and the mobo clock (14.3MHz) to work properly. This mobo looks to be made pretty decently, the clock traces (especially for the CPU) should be using more copper to lower the resistance. This might make it easier to figure out what pin it goes to.

There is also the 32768Hz clock for RTC but it doesn't affect the reset in any way even if missing.

hi,

I have checked everything once more, and I can confirm that the 66MHz and the 14,3MHz outputs as they should. The clock from both OSC is on the chipset as well, I have compared the readings from the 74LS04 with the damaged board and this one, and the readings are the same. To this moment, everything I have compared to the damaged board has the same readings and outputs. From this point the board looks dead but I will not give up so easily.

I have checked all pins, and did some re-flow there, just in case, no difference. I think I would try to compare the readings from the damaged board on all pins to this one, will take ages but seems to be the only approach I could think off for now. I will check for a possible trace damage once more, I have done this already a few time without any findings, but I could be missing something anyway.

I have seen boards that looked like after a zombie apocalypse with shortened legs on the chipset, missing ICs, missing legs, etc... still delivering at least the CLK and releasing the RESET.
I have still hope for this board, at least now, heh.

apart of comparing the readings with the damaged board on the chipset, anything else I could try?

br , tomas

tsalat wrote on 2025-05-10, 17:41:

MikeSG wrote on 2025-05-10, 08:36:

Is JP5 BIOS clear?

[...]maybe CPU speed, dont know for sure.

br, tomas

Try switching this jumper... or finding the Bios Clear jumper. 386sx motherboards that are stuck in this mode exhibit the same symptoms. Black screen, appear to be stuck in reset. At least that would be ruled out.

MikeSG wrote on 2025-05-11, 11:53:

tsalat wrote on 2025-05-10, 17:41:

MikeSG wrote on 2025-05-10, 08:36:

Is JP5 BIOS clear?

[...]maybe CPU speed, dont know for sure.

br, tomas

Try switching this jumper... or finding the Bios Clear jumper. 386sx motherboards that are stuck in this mode exhibit the same symptoms. Black screen, appear to be stuck in reset. At least that would be ruled out.

hi, I have added all the ICs back (KBD Bios, Bios, RTC), tried to switch the jumper but with no difference or effect. Also, I have tried to switcht he jumper for the mono/color output, just in case, no difference what so ever, still no CLK and reset high.

br, tomas

tomas

Since you have a scope, check few more things. On ISA slot test B20 (CLK) and B30 (OSC) pins. I would expect the 14.31818MHz B30 to be present at all times, even with chipset in reset state. But the B20 (~8MHz) might be missing, that's why it doesn't show on POST card. That clock is usuall derived from CPU clock via software set divider, so if the main chip is not working or held in reset it would be missing.

I suppose the main question is if it chipset is held in reset by some external factor we missed, or is it broken and not exiting internal reset state even though clocks are present. After some thinking the KBC reset control should not affect the rest of the mobo, this is just to reset the CPU (the only way to exit protected mode for 286). The mobo/chipset reset is pretty much only 3 things:
1) The power-on reset from PSU via POWER GOOD signal
2) Manual reset via the button
3) Any possible reset delay via internal timer in the chipset itself

ISA reset is sort of bi-directional (cards can force it low and reset the system) but this is held high at 5V so not the issue here. And it's not the same as the reset via button on POWER GOOD but the actual implementation is somewhat mobo specific.

I have a working mobo based on M1217, if I have some free time I will try to figure out how the button and POWER GOOD are routed to the chipset. Or you can try to do that yourself. See if that arrives to the chip properly.

Deunan wrote on 2025-05-11, 15:49:

Since you have a scope, check few more things. On ISA slot test B20 (CLK) and B30 (OSC) pins. I would expect the 14.31818MHz B30 […]
Show full quote

Since you have a scope, check few more things. On ISA slot test B20 (CLK) and B30 (OSC) pins. I would expect the 14.31818MHz B30 to be present at all times, even with chipset in reset state. But the B20 (~8MHz) might be missing, that's why it doesn't show on POST card. That clock is usuall derived from CPU clock via software set divider, so if the main chip is not working or held in reset it would be missing.

I suppose the main question is if it chipset is held in reset by some external factor we missed, or is it broken and not exiting internal reset state even though clocks are present. After some thinking the KBC reset control should not affect the rest of the mobo, this is just to reset the CPU (the only way to exit protected mode for 286). The mobo/chipset reset is pretty much only 3 things:
1) The power-on reset from PSU via POWER GOOD signal
2) Manual reset via the button
3) Any possible reset delay via internal timer in the chipset itself

ISA reset is sort of bi-directional (cards can force it low and reset the system) but this is held high at 5V so not the issue here. And it's not the same as the reset via button on POWER GOOD but the actual implementation is somewhat mobo specific.

I have a working mobo based on M1217, if I have some free time I will try to figure out how the button and POWER GOOD are routed to the chipset. Or you can try to do that yourself. See if that arrives to the chip properly.

yes, the B30 (OSC) on the ISA slot is showing 14,31818MHz, the B20 (CLK) is high at 3,6V but without any signal, just flat. On the damaged board the B20 is at 3,6V as well but with a 8MHz CLK.
documentation for this would be great, at least I would know what pin to look at but I hope using the damaged board I can find some difference.

any help determining if the mobo is dead or there is just something wrong would be appreciated. I will try to compare the chipset readings with the damaged board that is still outputting something to see if I will find any difference. The week is coming and this will take me some time I guess.

br, tomas

It took me a while, but I have some info on the mobo. There are more chips hidden inside the sockets.

So anyway, the 4069 is a false lead. Yes, it's connected to the POWER_GOOD but all it does is shape (invert the signal twice) for the RTC chip. So PG goes to pin 1, out of pin 2 to pin 11, and out of pin 10 to RTC. That's it. Even if broken the 4069 would not prevent the chipset from running.

POWER_GOOD is also connected directly to chipset, pin 129. Also, the RESET button goes there but not directly. From the pin on the header it goes through R51 (between slots). There is R50 resistor that is a pull-up from 5V, capacitor C21 to filter/debounce the signal, and diode D5 nearby that is connected to 5V to discharge C21. From R51 the signal goes to pin 6 of 7407 chip that is hidden under the keyboard controller. And the output of that is pin 7, which also goes to chipset pin 129.

So in effect both the POWER_GOOD and the reset signal from button meet at pin 7 of the 7407 and the resulting wire AND goes to pin 129 of the chipset. So check if that connection is OK, and then test if the voltage here changes when you press the RESET button. Testing on the chipset is difficult due to dense pinout so once you confirm the connection is good it's better to check voltage as the 7407. You can even solder temporary wire to the pin from underside so you don't have to remove the KBC from socket for this.

If the voltage changes in response to RESET button and the chipset doesn't start the clocks then I'm out of ideas. Could be a bad solder on the chipset, or actually fried chip.

Deunan wrote on 2025-05-22, 23:11:

It took me a while, but I have some info on the mobo. There are more chips hidden inside the sockets. […]
Show full quote

It took me a while, but I have some info on the mobo. There are more chips hidden inside the sockets.

So anyway, the 4069 is a false lead. Yes, it's connected to the POWER_GOOD but all it does is shape (invert the signal twice) for the RTC chip. So PG goes to pin 1, out of pin 2 to pin 11, and out of pin 10 to RTC. That's it. Even if broken the 4069 would not prevent the chipset from running.

POWER_GOOD is also connected directly to chipset, pin 129. Also, the RESET button goes there but not directly. From the pin on the header it goes through R51 (between slots). There is R50 resistor that is a pull-up from 5V, capacitor C21 to filter/debounce the signal, and diode D5 nearby that is connected to 5V to discharge C21. From R51 the signal goes to pin 6 of 7407 chip that is hidden under the keyboard controller. And the output of that is pin 7, which also goes to chipset pin 129.

So in effect both the POWER_GOOD and the reset signal from button meet at pin 7 of the 7407 and the resulting wire AND goes to pin 129 of the chipset. So check if that connection is OK, and then test if the voltage here changes when you press the RESET button. Testing on the chipset is difficult due to dense pinout so once you confirm the connection is good it's better to check voltage as the 7407. You can even solder temporary wire to the pin from underside so you don't have to remove the KBC from socket for this.

If the voltage changes in response to RESET button and the chipset doesn't start the clocks then I'm out of ideas. Could be a bad solder on the chipset, or actually fried chip.

Hi @Deunan,

thank you for your answer, and I am sorry for my belated feedback. I could not find time to reply or test the board until now.

The analysis you have made is highly appreciated.
As for the 7407, just to make sure, you have mentioned pin 7 which is according to the datasheet GND, hence I guess you meant pin 6 instead. Correct?

I took the old damaged board that is resetting at first to have something to compare, and tried this at first. The pin 6 on the 7407 beneath the KBC goes low while triggering the reset. The voltage on the 7407 pin 6 was something below 3V (if I recall correctly) and upon triggering the reset switch, the voltage dropped down, and the board indicated a reset.

After verifying it does what you have outlined above, I took the second board that is stuck on reset. The voltage on the 7407 pin 6 was around 5V. The boards are not exactly the same, and hence it could be different right? Triggering the reset switch got the pin 6 to low but the board is still stuck. The continuity to pin 129 of the chipset is there - again, just to make sure, it is the pin next to the 125 which is printed on the PCB, right? 😁

From the above I would assume the chipset could be fried or I am still missing something. Unfortunately I wasn't able to do anything expect the above on the board due to insufficient time but I will try this weekend once more.
Thank you a lot for your help.

best regards, Tomas

tsalat wrote on 2025-06-09, 04:26:

As for the 7407, just to make sure, you have mentioned pin 7 which is according to the datasheet GND, hence I guess you meant pin 6 instead. Correct?

Yes, sorry, I was counting the pins from the end and somehow thought the chip is DIP-16, not DIP-14. So the pin numbers on the 7407 in my post are +1 off the mark.

tsalat wrote on 2025-06-09, 04:26:

After verifying it does what you have outlined above, I took the second board that is stuck on reset. The voltage on the 7407 pin 6 was around 5V. The boards are not exactly the same, and hence it could be different right? Triggering the reset switch got the pin 6 to low but the board is still stuck. The continuity to pin 129 of the chipset is there - again, just to make sure, it is the pin next to the 125 which is printed on the PCB, right? 😁

The voltage here will depend on the pull-up resistor and any possible leakage in the PSU. But 3V or 5V is a valid H level for TTL. So all good, and it also changes. The pin 129 is, on my mobo, close to ISA slots. Close to 125 marking, correct.

As I've said, the main clock and reset being off should be enough for the chipset to work. But keep in mind that the other ISA clock is programmed, so the CPU must be running for that, and the BIOS code must be properly executing.
So make sure the 66MHz clock (the CPU clock) is also connected to the chipset, and running. I've had problems with these socketed clock generators before, the socket contact can be poor and the clock is not running or the signal is weak. Re-seat just in case. Check the CPU reset signal (use 386SX datasheet to locate it). If the chipset holds the CPU in permanent reset, even with CPU clock running, then I would suspect the chipset is dead. But if the CPU reset goes away (it depends on the mobo reset but is generated by the chipset, not connected directly) then you need to start probing the BIOS socket. If there are any signs that CPU is fetching code from ROM. Could be a bad data bus (a broken trace or something).

Deunan wrote on 2025-06-10, 23:53:

Yes, sorry, I was counting the pins from the end and somehow thought the chip is DIP-16, not DIP-14. So the pin numbers on the 7 […]
Show full quote

tsalat wrote on 2025-06-09, 04:26:

As for the 7407, just to make sure, you have mentioned pin 7 which is according to the datasheet GND, hence I guess you meant pin 6 instead. Correct?

Yes, sorry, I was counting the pins from the end and somehow thought the chip is DIP-16, not DIP-14. So the pin numbers on the 7407 in my post are +1 off the mark.

tsalat wrote on 2025-06-09, 04:26:

After verifying it does what you have outlined above, I took the second board that is stuck on reset. The voltage on the 7407 pin 6 was around 5V. The boards are not exactly the same, and hence it could be different right? Triggering the reset switch got the pin 6 to low but the board is still stuck. The continuity to pin 129 of the chipset is there - again, just to make sure, it is the pin next to the 125 which is printed on the PCB, right? 😁

The voltage here will depend on the pull-up resistor and any possible leakage in the PSU. But 3V or 5V is a valid H level for TTL. So all good, and it also changes. The pin 129 is, on my mobo, close to ISA slots. Close to 125 marking, correct.

As I've said, the main clock and reset being off should be enough for the chipset to work. But keep in mind that the other ISA clock is programmed, so the CPU must be running for that, and the BIOS code must be properly executing.
So make sure the 66MHz clock (the CPU clock) is also connected to the chipset, and running. I've had problems with these socketed clock generators before, the socket contact can be poor and the clock is not running or the signal is weak. Re-seat just in case. Check the CPU reset signal (use 386SX datasheet to locate it). If the chipset holds the CPU in permanent reset, even with CPU clock running, then I would suspect the chipset is dead. But if the CPU reset goes away (it depends on the mobo reset but is generated by the chipset, not connected directly) then you need to start probing the BIOS socket. If there are any signs that CPU is fetching code from ROM. Could be a bad data bus (a broken trace or something).

Hi, I had some time to measure the 66MHz OSC once more, and to confirm if the OSC reach the chipset. I have also de-soldered the socket and soldered the OSC directly to the board, just in case.
The 66MHz is present on the OSV, and it goes to the chipset as well, the voltage level is something above 2,7V. The chipset however does not provide the 33MHz not the CPU gets anything on the CLK2 pin. The CPU RESET pin is always high. The pin on which the chipset should provide 33MHz is high but without any signal.

I have again compared everything to the damaged board, right now already without the CPU, the board still release the reset, and provide the CLK to ISA slot, even without the CPU. The CLK2 is present on this board, and the reset is LOW on the CPU pin.

not sure how to simulate this state on the damaged board to reproduce the same issue. the chipset on the damaged board is missing, I think, 4-6 legs, and was certainly shorted in the past. Still, the board release the reset and provide all the OSC that the nice one does not.... what a shame. not sure what to explore now. seems to me I am missing something or the chipset is completely fried.

br, Tomas

best regards, Tomas

tsalat wrote on 2025-06-15, 09:03:

not sure what to explore now. seems to me I am missing something or the chipset is completely fried.

Well as last attempt I would try to add a pull-up resistor to the pin 129 (as explained above best to actually add it at 7407 side, and not mess with the chipset itself). See if you can bring it up closer to 5V. A 1k resistor or even something lower, down to 330 ohms or so. Maybe it really needs higher voltage for the H level to work for some reason. If you cant get it up above 4V even with additional pull-up to 5V, consider removing the 7407, might be faulty. Keep in mind this pin is also driven by the POWER_GOOD from PSU, so make sure you use the same good PSU that you tested with the broken mobo, that does have 5V there and good reset sequence.

Deunan wrote on 2025-06-16, 13:55:

tsalat wrote on 2025-06-15, 09:03:

not sure what to explore now. seems to me I am missing something or the chipset is completely fried.

Well as last attempt I would try to add a pull-up resistor to the pin 129 (as explained above best to actually add it at 7407 side, and not mess with the chipset itself). See if you can bring it up closer to 5V. A 1k resistor or even something lower, down to 330 ohms or so. Maybe it really needs higher voltage for the H level to work for some reason. If you cant get it up above 4V even with additional pull-up to 5V, consider removing the 7407, might be faulty. Keep in mind this pin is also driven by the POWER_GOOD from PSU, so make sure you use the same good PSU that you tested with the broken mobo, that does have 5V there and good reset sequence.

hi Deunan,

thank you for coming back to this. I have tried to add the pull up resistor from 5V on the 7407 to pin 6. The voltage went up to 5V but the board is still in the reset mode. I have used the same PSU I have used for the other board.
Now, knowing that this is not working - I am thinking about three ways of how to approach this. a) assume that the chipset is fried, and thus I will try to find a board with the same chipset with all legs - hence I will just put this board on the side until then, b) I will try to replace the 7407 and other components but I was not able to find anything wrong with them and thus it would be just a blind shot, c) I will try to remove parts from the damaged boards and wait until it will get stuck as this one - could be a dead end of course since there could be more scenarios when the board will be in this state.

Speaking about the damaged board, I have extracted the CPU, the 32kHz and 14MHz OSC from the board, the board still release the reset and provide CLK to the ISA port. I would assume that this will prevent the chipset to be running but it seems that the board does not care... heh.

br, tomas

I'm out of ideas to try. I assume you've already resoldered all the chipset pins with some flux and a touch of soldering iron tip.

What to do next - well, depends on how much free time you have. Desoldering these big chips is always a bit tricky, the PCB is old (and was probably made cheaply) so there is always some risk of lifting the pads. But if you want to try something, how about this approach:
- Remove the chipset with missing pins from the broken mobo, you want to preserve the PCB - a chip with missing pins is dead anyway.
- Remove the chipset that doesn't work, again trying not to damage the mobo.
- Assuming the mobos survived this, transplant the chip. But not the one with missing pins, that is broken for sure. But we know it at least still leaves reset state even on mobo with removed parts. So put the suspect chipset on the mobo with removed parts. Now test again to see if it resets properly or not.

So there are two possible outcomes - the chipset still doesn't reset properly. So it's confirmed dead. If it does reset now then either the original mobo is faulty somehow, or it had to be resoldered completly and just touching the pins with soldering iron didn't fix it. Also you can now inspect the "good" mobo under the chipset for any obvious damage. And then you can decide - transplant it back, to see if the chipset works, or strip this mobo for parts as well, as suspect.