Hi,

Is CMOS battery OK and a clear-cmos jumper set in normal position ?

there are things that shouldnt be repaired, this is one of them 😀

Roman555 wrote on 2022-12-20, 14:41:

Hi,

Is CMOS battery OK and a clear-cmos jumper set in normal position ?

Yes, they are.

Also I found pictures online of this board that show another clock chip. One I had much more luck finding a datasheet for (w83194R-39). So now having the datasheet for a compatible chip (my MB has cma8864bf-26) I can uld check all the pins.

The voltages are as they should be. The pci clock is being generated by the chip as well as the cpu clocks.

However, there is something "weird" on the sdram clock outputs. Instead of a clock they all have 61khz ringing. Same ringing is present on top of the cpu clock as well as on the CPU_STOP line. Now the CPU_STOP is active low and it should stop both the cpu1 clock and sdram when low. Somehow it is low (with this 60khz ringing on top), but cpu1 is active while all sdram clocks are off.

I measured resistance from the SDram clocks to GND and v+ it looks good at 100k. Do you think the clock chip might be damaged, or something else?

I see the same behaviour with and without ram/cpu.

Edit: Something is definitely pulling CPU_STOP low as I measured the resistance to ground and the moment I power it on it goes from 100k down to 10ohm. So the question is not why sdram clock if off, but why cpu clock is on with this... And why is this being pulled low.

Edit2: I'm now t sure I'm interpreting this right. I noticed my chip has different settings on the rj2 permanent jumper than the "compatible one". I really need its datasheet to be able to deduct anything.

Does anyone know a clock chip named cma8864bf-26 or has its datasheet? Google seems to fail me on this. This is the chip.

I doubt I can do much without a datasheet. What I thought to be CPU_STOP goes to pin 1 of rj2 (the unconnected pin on the left/top).

PCIRSTJ (reset in PCI bus) is active (low) , isn't it? A southbridge will deactivate it when power-good signal PWG gets active.
Or maybe it's some faulty chip pulls reset to low level. Or the southbridge is faulty itself.
The southbridge is ALi M1543/M1543C here. So you can find a datasheet and check the signals.
Also check a 32.768kHz crystal.

P.S. You can easily google datasheet for "CMA8864" - IMO it can be useful too.

rasz_pl wrote on 2022-12-20, 14:55:

there are things that shouldnt be repaired, this is one of them 😀

Why not - it's a good exercise for a brain and hands 😀 But sometimes it's really difficult , time consuming and expensive to revive a defective motherboard

flynth wrote on 2022-12-20, 15:15:

...Does anyone know a clock chip named cma8864bf-26 or has its datasheet? Google seems to fail me on this. This is the chip.

I doubt I can do much without a datasheet. What I thought to be CPU_STOP goes to pin 1 of rj2 (the unconnected pin on the left/top).

Maybe some help

Roman555 wrote on 2022-12-20, 17:38:

PCIRSTJ (reset in PCI bus) is active (low) , isn't it? A southbridge will deactivate it when power-good signal PWG gets active. […]
Show full quote

PCIRSTJ (reset in PCI bus) is active (low) , isn't it? A southbridge will deactivate it when power-good signal PWG gets active.
Or maybe it's some faulty chip pulls reset to low level. Or the southbridge is faulty itself.
The southbridge is ALi M1543/M1543C here. So you can find a datasheet and check the signals.
Also check a 32.768kHz crystal.

P.S. You can easily google datasheet for "CMA8864" - IMO it can be useful too.

Unfortunately despite searching for quite some time all I could find were some audio chips of the same name instead of the clock synthesizer, but the next poster found it 😀

I have however found the datasheet for the southbridge. So this will hopefully be very useful. Thank you. I'll make sure to check its crystal too.

I don't remember if the PCI reset is active low or active high (on a typical cpu it is active low), but my PC analyser card has a RESET LED that is lit all the time on the PCI bus. Regarding power good. I assume this is not an ATX PWG signal as this is an AT motherboard (It does support ATX PSUs too).

PC Hoarder Patrol wrote on 2022-12-20, 19:59:

Maybe some help […]
Show full quote

flynth wrote on 2022-12-20, 15:15:

...Does anyone know a clock chip named cma8864bf-26 or has its datasheet? Google seems to fail me on this. This is the chip.

I doubt I can do much without a datasheet. What I thought to be CPU_STOP goes to pin 1 of rj2 (the unconnected pin on the left/top).

Maybe some help

cma8864bf-26b.pdf
cma8864bf-26d.pdf

Great stuff, thank you. Where there was CPU_STOP on the previous chip, we have CPU_CLK3 on this one. This makes much more sense.

However, this leaves the issue of the missing SDRAM clocks.

There is pin 15 on the clock chip named BUFFER_IN described as "Input from fan-out buffers for SDRAM input.". The only reason for there being a BUFFER_IN on a clock generator is if the chip is not supposed to generate SDRAM clocks, but it expects to get SDRAM clock from "somewhere" and it just buffers them. This would make sense.

So now the question is, what is supposed to generate that SDRAM clock? The northbridge probably. Unfortunately again it is next to impossible to find a datasheet for it.

It is Ali Alladin-Pro II M1621 A1. If anyone is aware of a datasheet for it please let me know.
Edit: Actually I found it here: http://bitsavers.org/components/ali/M1621_Ala … orth_Bridge.pdf

I have this motherboard in a somewhat working state last time I checked. It had some instability and would often crash, so I just put it away, but it should still boot if you need me to measure anything on it. I don't have an oscilloscope though.

Another idea - the RESET button circuit may be faulty and cause the system to be in a reset state.
A PWG signal for the southbridge may be formed by a PWM controller of CPU VRM (as example).

Make sure you are running the board in FSB66 mode. 100 Will only run stable if the stars align, as the board has piss-poor power delivery.

Also, measuring resistance on powered-on equipment is a big NO-NO!

I haven't had a chance to look at this MB again yet. There is lots of holiday related stuff to do, but I'll definitely have some updates in few days.

devius wrote on 2022-12-21, 18:16:

I have this motherboard in a somewhat working state last time I checked. It had some instability and would often crash, so I just put it away, but it should still boot if you need me to measure anything on it. I don't have an oscilloscope though.

Thank you, I'll get back to you on this 😀

Roman555 wrote on 2022-12-21, 18:21:

Another idea - the RESET button circuit may be faulty and cause the system to be in a reset state.
A PWG signal for the southbridge may be formed by a PWM controller of CPU VRM (as example).

I'll make sure to check it. Thanks.

Oerg866 wrote on 2022-12-21, 18:25:

Make sure you are running the board in FSB66 mode. 100 Will only run stable if the stars align, as the board has piss-poor power delivery.

Also, measuring resistance on powered-on equipment is a big NO-NO!

I've been trying both as my P2 (known good) CPU is nominally 100MHZ FSB, but I'll just keep it on 66MHz for now.

As for measuring the resistance while powered on, yes it is a big no-no. However 😀 I have a "special" battery operated multi meter that probes the resistance with only 0.4V so it is very unlikely to damage the circuit being probed. Also, I knew there is no voltage where I'm measuring to harm the meter. So I wanted to check if this pin is close to GND due to a pull down resistor, or is there some logic actively pulling it down (As I thought it was CPU_STOP , then it turned out this pin is irrelevant).

Honestly I have no idea how this multimeter can measure resistance in mega-ohm range with well under 1V. That requires measuring currents in the range of 40~400nA, but I'm pretty sure it uses only 0.4V as I connected its measuring output to various voltmeters and they always showed the same. Specially that this is a pretty cheap no-name brand device. While my desktop expensive HP multimeter uses 7V for its resistance measurement so I can't use it on any delicate logic.

I wish more multimeters came with an ability to select its resistance/continuity probing voltage as well as max current.