How does the "reset LED" on the POST card behave while you turn the system on and it fails to start?

sorrow wrote on 2023-03-25, 01:01:

It has some damage due to battery leak that I have cleared. I measured affected wire patches and they seems to be OK. The damage […]
Show full quote

It has some damage due to battery leak that I have cleared. I measured affected wire patches and they seems to be OK. The damaged part is shown on the attached picture.

The board often fails to POST when I turn the PSU on.

When I press reset button the board boots normally in 100% of cases.

My guess is that some component on the board needs too much time (after power on) to start functioning properly but I lack experience to identify it

It's not just that components might need time to start, they also might start up in an invalid state when operating voltage is applied, and need a RESET signal to get into a defined default state. This procedure is called "power-on reset" (POR) and an essential part of any complex electronic system. This usually works by simulating the push of the reset button for a fractional second as you apply power to a previously powered-off computer. The circuit responsible for generating the power-on reset signal is located right next to the CMOS battery on many mainboards, so damage to the POR circuit after battery leak is not uncommon. That's why sorrow asks for the behaviour of the RESET LED on a POST card: Just as pushing the actual reset button, also the "simulated push" of that button by the POR circuit is supposed to turn on that LED, so the RESET LED should be lit around 100 to 200ms after you applied power to the machine. If you don't get a RESET signal on power on, that's the cause for the malfunctioning initial POST, and troubleshooting can be limited to the POR circuit and the connection of that circuit to the chipset (which should handle sending RESET signals to the ISA bus and the processor).

I have the same situation with one 286 motherboard. I just left it as is.

It could be PSU fault. PWR_GOOD signal is supposed to have certain delay and pulse length, if it's not in spec mobo might not reset properly. Could also be "tired" chips or marginal filtering capacitors on the mobo, or a combination of both.
In general it's not easy to fix (unless PSU replacement cures it) unless you have some skills in modding electronics because in most cases it requires tinkering with this circuitry in the PSU.

Thank you very much, guys!

majestyk wrote on 2023-03-25, 06:11:

How does the "reset LED" on the POST card behave while you turn the system on and it fails to start?

Very good point. When I power up the system there is no visible light on reset LED. When I press reset button there the reset LED goes on for a short period of time.

mkarcher wrote on 2023-03-25, 09:23:

It's not just that components might need time to start, they also might start up in an invalid state when operating voltage is applied, and need a RESET signal to get into a defined default state. This procedure is called "power-on reset" (POR) and an essential part of any complex electronic system. This usually works by simulating the push of the reset button for a fractional second as you apply power to a previously powered-off computer. The circuit responsible for generating the power-on reset signal is located right next to the CMOS battery on many mainboards, so damage to the POR circuit after battery leak is not uncommon. That's why sorrow asks for the behaviour of the RESET LED on a POST card: Just as pushing the actual reset button, also the "simulated push" of that button by the POR circuit is supposed to turn on that LED, so the RESET LED should be lit around 100 to 200ms after you applied power to the machine. If you don't get a RESET signal on power on, that's the cause for the malfunctioning initial POST, and troubleshooting can be limited to the POR circuit and the connection of that circuit to the chipset (which should handle sending RESET signals to the ISA bus and the processor).

This might explain the behavior and is in line with the above mentioned fact that reset LED is not showing anything. I will investigate it further. Any further hint what particular circuit might be responsible for this is welcome - I did no find schematics for the board yet.

Deunan wrote on 2023-03-25, 12:55:

It could be PSU fault. PWR_GOOD signal is supposed to have certain delay and pulse length, if it's not in spec mobo might not reset properly. Could also be "tired" chips or marginal filtering capacitors on the mobo, or a combination of both.
In general it's not easy to fix (unless PSU replacement cures it) unless you have some skills in modding electronics because in most cases it requires tinkering with this circuitry in the PSU.

I have tested the board with another PSU and there seems to be no change. So the power-on reset issue is more probable. But I will try one more PSU just to be sure.

sorrow wrote on 2023-03-25, 16:54:

I have tested the board with another PSU and there seems to be no change. So the power-on reset issue is more probable. But I will try one more PSU just to be sure.

PWR_GOOD is the power-on reset signal. If your mobo had a battery spill near power connector then it's possible that trace got dissolved. If your POST card shows RESET LED activity when the button is pressed, but not on power-on, it's a big clue where to look.

Just had (nearly) the same issue today on a QDI SIS471 "Dark Green Mainboard":
https://theretroweb.com/motherboards/s/qdi-si … board#downloads

There was massive corrosion / damage in the battery-area. I had cleaned and repaired everything but forgot an interrupted trace on the flipside that goes to the KBC.
The board wouldn´t POST, only sometimes when the reset switch was used (but the keyboard never worked). After fixing this (hopefully final) trace all is well.

So my long-shot guess would be there´s still some trace / via unfixed...

Did you look under the ferrite beads, the keyboard connector and the PSU connector?

sorrow wrote on 2023-03-25, 16:54:

This might explain the behavior and is in line with the above mentioned fact that reset LED is not showing anything. I will investigate it further. Any further hint what particular circuit might be responsible for this is welcome - I did no find schematics for the board yet.

Typically, those circuits consists of a detector that checks whether all voltages are present (which can be substituted by just looking at POWER_GOOD), and an R/C delay circuit that delays the "all voltages good" signal. As long as the "all delayed voltages good" signal is not there, the RESET line needs to be activated. Some chipsets have a dedicated power-on reset signal, whereas on other chipsets, the reset button and the power-on reset signal are combined into one input pin. You might need to look for a transistor or open-collector output logic gate in parallel to the reset button.

If your board uses PWR_GOOD, tracing that signal is a good starting point.

majestyk wrote on 2023-03-25, 18:58:

So my long-shot guess would be there´s still some trace / via unfixed...

Did you look under the ferrite beads, the keyboard connector and the PSU connector?

You were correct. There was damaged trace hidden under another component transmitting POWER GOOD signal. I have put some solder on the spot and the board now posts after being turned on. Thanks a lot!

mkarcher wrote on 2023-03-25, 21:29:

Typically, those circuits consists of a detector that checks whether all voltages are present (which can be substituted by just looking at POWER_GOOD), and an R/C delay circuit that delays the "all voltages good" signal. As long as the "all delayed voltages good" signal is not there, the RESET line needs to be activated. Some chipsets have a dedicated power-on reset signal, whereas on other chipsets, the reset button and the power-on reset signal are combined into one input pin. You might need to look for a transistor or open-collector output logic gate in parallel to the reset button.

If your board uses PWR_GOOD, tracing that signal is a good starting point.

Thank you. Your explanation is much appreciated. And yes, you were also correct regarding the starting point.

There was a lot of (de-) soldering to do on my QDI board also: