CMOS battery mystery

[Junk Parts]

Re: CMOS battery mystery

When it comes to soldering in new wires and such I have no idea. Now another thing came to mind. My bios has a setting for "Reset Configuration data automaticly". Why don't you see if your bios also has this setting and disable it if it is enabled. Reboot and see if the bios now remembers.

[pentium]

Re: CMOS battery mystery

When it comes to soldering in new wires and such I have no idea. Now another thing came to mind. My bios has a setting for "Reset Configuration data automaticly". Why don't you see if your bios also has this setting and disable it if it is enabled. Reboot and see if the bios now remembers.

Whenever I do a major hardware upgrade or a BIOS flash in any of systems, I select that to clear wahtever exists.
Yes, I already tried that and still no dice.

[Junk Parts]

Re: CMOS battery mystery

Then Junk Parts has run out of "Knowledge" again.... This happens to me everyday now! Sorry but I don't know what to tell you next.

[LesCoke]

Re: CMOS battery mystery

"Reset Configuration data" setting clears out stored PCI card / slot settings. I have not seen this setting cause a lose of clock or other CMOS settings.

I see you have tied other batteries in the board. I did not see any mention of that when I made my prior post. I find it unlikely they are all dead, but you did not say known good. Some systems are more picky than others about what state of discharge starts causing problems.

At this point I would suspect the circuit path to the CMOS may be open. If there is an external diode in that path, I would check to see if it is open (or shorted). If open, nothing will be getting to the CMOS. If it is shorted, the other circuits on the 3p3vstby could be loading down the battery. If you have a DMM, I would try checking the voltage at the battery with the system unplugged, and trace the circuit to see where the battery voltage is lost. Failing that it, I would point to the chip implementing the CMOS...

Les

[gg1978]

Re: CMOS battery mystery

I did forget to mention that the intel ref board schematic had a resistor inline with the southbridge Vbat to the center pin, but if that were bad, you would lose the settings every time, not just when the ATX PSU is all the way off..

If you can find that trace, that's well and good, but the thing is that when the computer is on, i would assume that the +3.3VSB would tend to backcharge the battery, if you have it directly connected, therefore it would probably kill the battery quite prematurely, as you'd be bypassing the series diode..

Also, since the battery would be running it all the time, instead of just when the ATX PSU is all the way off, it would have a much shorter lifetime..

If you wanted to connect an external battery somehow, i would think that you'd have to put a series schottky diode in with it, like the board schematic, to prevent the +3.3VSB from damaging the battery.

I'm assuming that the ICH5 uses +3.3VSB for the RTC functionality.. I don't know exactly when +3.3VSB started to be implemented, obviously after BX sometime. I was going to grab the ICH5 datasheet, but it's a almost 8M PDF file, and i'm on dialup, and i lack that amount of patience, just to look up one voltage spec. Pity Intel doesn't provide a reference schematic for anything newer than the BX..

[gg1978]

Re: CMOS battery mystery

Well, it turns out i do have the ICH5 datasheet already downloaded. For VccRTC, it specs 3.46V max, and for normal operation it is assumed that Vbat will be betwen 3.0V and 1.0V.. This probably doesn't take into account the voltage dropped across the diode, but i would think that as long as Vbat is at least 1.8V, a board with the ICH5 should still be keeping time and date.. Specced current draw at 3.0Vbat is 6uA.. I don't think most DMMs will be able to measure currents that low..

[gg1978]

Re: CMOS battery mystery

Heh, i didn't look at the BX schematic hard enough. Even BX had 5VSB dropped to 3VSB for powering the RTC.

[pentium]

Re: CMOS battery mystery

"Reset Configuration data" setting clears out stored PCI card / slot settings. I have not seen this setting cause a lose of clock or other CMOS settings.

I see you have tied other batteries in the board. I did not see any mention of that when I made my prior post. I find it unlikely they are all dead, but you did not say known good. Some systems are more picky than others about what state of discharge starts causing problems.

At this point I would suspect the circuit path to the CMOS may be open. If there is an external diode in that path, I would check to see if it is open (or shorted). If open, nothing will be getting to the CMOS. If it is shorted, the other circuits on the 3p3vstby could be loading down the battery. If you have a DMM, I would try checking the voltage at the battery with the system unplugged, and trace the circuit to see where the battery voltage is lost. Failing that it, I would point to the chip implementing the CMOS...

Les

I'll check the loaded and unloaded voltages on the cells again but as for checking the trace, I don't think I can do that right now until I have the time to pull the system out from under the desk, completely dis asemble it and then remove the motherboard and inspect it.
Also, just getting to the battery is fun. It sits right under the AGP and top PCI slot at the very back of the board so every time I ahve to access the battery I have to power the system off, pull the video and two PCI cards and then move a few wires before I can reach the battery. What terrible placement. x_x

[LesCoke]

Re: CMOS battery mystery

You will also find that with very few exceptions, the trace will most likely be buried on one of the layers you cannot visually inspect. Best you could do is make educated guesses about the components connected to that trace and use them to verify continuity.

Les

[pentium]

Re: CMOS battery mystery

You will also find that with very few exceptions, the trace will most likely be buried on one of the layers you cannot visually inspect. Best you could do is make educated guesses about the components connected to that trace and use them to verify continuity.

Les

But if it has to pass through diodes and the likes, somewhere along the line it has to come to the surface of the PCB and that's where I can effect repairs, assuming that the damaged trace is not between the southbridge and the surface mount components.