Time and Date lost with power off, New CMOS battery more details below.

[SixSpeedDeath]

I have run this Workstation for a few years now as my home lab, and 1 problem persists througjout every iteration: when I lose AC power, I lose time and date and certain BIOS settings, and ORIGINALLY RAID settings.

This will be a lot of typing, so to summarize: I'm looking for advice on IF A PCIe card can prevent time and date from saving, OR if I'm missing a CMOS clear jumper that has to bridge 2 pins to KEEP settings on the motherboard.

Diagnosis section:

I had replaced the CMOS battery as a slam dunk, and it didn't work.

I upgraded the whole motherboard to one that supported dual Opteron x6 processors, and used both X6 processors as well as 64GB EEC RAM.

Not that I THOUGHT there would be a change, but I still lose time, date, and some BIOS settings, as well as RAID info.

IMPORTANT NOTE:

I can load the default BIOS settings, and most of the configuration is not lost on AC power loss.

I upgraded to an LSI RAID card with a battery backup to run the drives, and as it will surprise no one, on power loss I no longer lose RAID settings! But I still lose time, date, and the Server doesn't spring back after a power loss.

I had a BIG whoopsie, where I made a mistake, and damaged my own RAID card, and the power supply.

I replaced the power supply (something I suspected could have caused the time and date errors in the CMOS?) as well as the LSi card... and I still lose time and date with AC power loss.

I tested the Litium 2032 battery, and it has a healthy 3.2v, but I replaced it as well, for good measure.

Nope, still lose time and date.

Suspected problem section:

So what I'm left with is the cooling package as original, as well as the Quadro card for video out.

I am wondering if there is a jumper for the CMOS that is supposed to be bridged in normal operating mode that I am missing. I know you can bridge the Clear CMOS pins to RESET the CMOS, but I've seen some motherboards that have a bridge for the CMOS in normal operating mode. 

Does the HP xw9400 have a jumper that I am missing?

Does anyone have a picture of their CMOS clear jumper on a board without a problem I can see for reference?

Thanks for helping.

[itsmyname]

@SixSpeedDeath --   I had replaced the CMOS battery as a slam dunk, and it didn't work.

That is very unusual. I would have expected that would have solved the issue, if the actual CMOS memory  chip (not the battery) was not the cause of the issue.

I upgraded the whole motherboard

It is really strange that a different motherboard, with a different CMOS chip and a different CMOS battery, would have the same issue.

Usually, the "clear CMOS" jumper connects pin #1 to pin #2, and to "clear", you move the jumper to connect pin #2 to pin #3, for a brief time, and then return the jumper to its original position.  Is that the arrangement on your motherboard?

Sorry to not be more helpful.

[DGroves]

the xw9400 like most workstation/consumer systems has power applied to the motherboard even when powered off

on some systems this trickle current also powers the realtime clock besides the sleep and remote start options on other systems such as the xw9400 the clock/bios cmos settings are kept by the cmos battery when the power is off

if you keep losing cmos, but the battery checks good then it's possible the chip part where the cmos settings are stored is defective, and getting a new replacement board that has the exact same issue is rather remote,

so what's left? as a prior poster "itsmyname" suggested leaving the cmos clear jumper on the wrong pins will always cause a bios reset on power on even if the battery/motherboard is good

download the service manual, and make sure the cmos and password clear jumpers are on the correct pins for normal operation (page 176/178)

https://www.compeve.com/Documents/HP%20xw9400/HP_xw9400service_compeve.pdf

https://www.mrmemory.co.uk/downloads/memory-configurations/49866.pdf

edit: there is one other possibility i can think of, a defective cmos reset button that is shorted causing it to always be in the on/clear state

[Doinkus]

So, the reset button resets setting as well.

Here's my thinking on this:

If the button is shorted internally, keeping AC plugged in and shutting down/rebooting would clear the CMOS as well, and a reboot, or shutting down and powering on still retains time and date.

AC power has to be removed to lose settings.

The CLEAR CMOS should just short the 3.2v to ground (probably disconnect battery, and short to ground) to clear the data, and a power supply would provide the ~5v power to retain settings, which is why I can reboot.

I *think* that either the 3.2v from the lithium 2032 is never reaching the CMOS (missing jumper?) Or that there is a mechanical short that is preventing that battery power from reaching the CMOS.

It's wacky-shack, but I'm thinking that this will be something like: HP installed an insulating nylon washer on one of the motherboard standoffs becUse the case grounded the CMOS, and somewhere between previous owners and me replacing components, something that LOOKS trivial, like a jumper that jas to be transferred from one MB to the next, or an insulating washer on the case went missing, and BOOM, wiped CMOS when AC power is lost.

I'm kiiinda hoping someone at HP is here in this forum that may have some insight into this.

For reference, I work on cars, so I see a lot of weird stuff like this that is caused by simple mistakes where someone along the line just thought "what the hell is this for?" And just threw it out, or didn't transfer it over, and you end up with something as weird as this.

A location and pinout of tye CMOS would be super helpful as well, as I could trace power from the battery to the CMOS that way.

[DGroves]

the xw9400 does not have a documented cmos reset jumper, only the cmos reset button

please take a pict of the installed cmos batt on your xw9400 (the cmos batt should have the + symbol pointed up towards you) 

[Doinkus]

I verified that the Battery is installed in the correct orientation (I was fairly certain, but I never want to be TOO over confident in the abilities of past-me

[DGroves]

it's quite rare, but sometimes the metal contacts that connect to the cmos battery "break" in a hairline crack that can't be seen

the top +  one can be checked with a continuity test from the batt side to the edge of the connector no the motherboard the neg one can be difficult if you can't reach underneath the board

again, the xw9400 DOES NOT HAVE A CMOS JUMPER, it only has a cmos reset button

[Doinkus]

I'll give it a shot and do some voltage drops across the connections.

Is there any resources that show where the CMOS battery + bus goes? I would assume to the CMOS itself, but you can never be too certain if it powers other circuits. 

I can bust out a DMM and trace ot down if we know everything it powers.

[DGroves]

cmos unlike in the pentium era, is now usually stored on the one of the chipsets chips