HP Omen 875-1000 Thermal error shutdown alert then Cmos reset error on every user initiated reboot

[h1csvntdracones]

I have been having this issue for a month now, it started occasionally, but now happens every time I reboot or shutdown then restart the system. I get a 90D thermal shutdown error, then a cmos reset error, then it boot loops. I have tried all the cmos reset pins and restoring the bios, the only thing that fixes this is to pull the battery from the system, which requires removing the rather large and expensive water cooled graphics card, every time I reboot..

No thermal issues exist, i've checked everything. One thing, is that when this happens, the bios shows all of my fans at 65,000 rpm, which they aren't. After battery pull, the rpm's are back to standard. I've updated the bios also, The part that sucks is this started the day after my warranty ended and has just been getting worse, I am not sure if the system is reading a reboot incorrectly as a thermal error or if it is getting corrupt. I have tested it with Ubuntu and had the same issue, so it is not a Windows 11 issue.

I can't really afford a new board right now, had to invest heavily into liquid cooling the included RTX 2800 Super that was overheating constantly and heating up my HDMI cables so much they fell out of the port constantly.. So looking to see if anyone has seen this before and might have a solution? Also, any options for doing the same as a battery pull without physically pulling the battery? None of the jumper options do it, but I am really worried with removing the graphics card every time. Not just with the connectors going bad, but the liquid cooling tubes are bound to wear out also, not too mention it is very hard to maneuver it.

This eventually went away after a windows 11 reset following a constant explorer crash at sign in. No idea why that reset got rid of the problem, but verifies that this was a windows issue causing a bios issue, in case anyone encounters in it.

HP Omen Obelisk 875-1000

Intel i9 9900k liquid cooled

RTX 2800 Super liquid cooled

Operating system

Windows 11 Pro Insider Preview 64-bit Version: 22494.1000

Microprocessor

Intel(R) Core(TM) i9-9900K CPU @ 3.60GHz

System memory

32 GB

Memory slot 1

16GB Kingston 2666MHz

Memory slot 2

16GB Kingston 2666MHz

System board

8509 00

System BIOS

F.34

Video

Graphic device 1

NVIDIA GeForce RTX 2080 SUPER

Current resolution

3840 x 2160

Refresh rate

60 Hz

Version

30.0.14.7239 (10/13/2021)

Audio

Device 1

Realtek High Definition Audio

Status

Enabled

Driver

RTKVHD64.sys

Version

6.0.8924.1

Device 2

NVIDIA Broadcast

Status

Enabled

Driver

nvrtxvad64v.sys

ersion

1.0.0.5

Device 3

NVIDIA Virtual Audio Device (Wave Extensible) (WDM)

Status

Enabled

Driver

nvvad64v.sys

Version

4.13.0.0

Device 4

NVIDIA High Definition Audio

Status

Enabled

Driver

nvhda64v.sys

ersion

1.3.38.60

[itsmyname]

@h1csvntdracones --  the bios shows all of my fans at 65,000 rpm

A binary-value of 0000 0001 is a decimal-value of 1.

A binary-value of 0000 0011 is a decimal-value of 3.

A binary-value of 0000 0111 is a decimal-value of 7.

A binary-value of 0000 1111 is a decimal-value of  15.

A binary-value of 0001 1111 is a decimal-value of 15+32 ==> 47

A binary-value of 0011 1111 is a decimal-value of 15+32+64 ==> 111

A binary-value of 0111 1111 is a decimal-value of 15+32+64+128 ==> 239

A binary-value of 1111 1111 is a decimal-value of 15+32+64+128+256 ==> 495

A binary-value of 0000 0001 1111 1111 is a decimal-value of 256+495 ==> 751

and so on, as you add more "1" bits.

By the time you get to 1111 1111 1111 1111, you get a decimal value of 65535 -- slightly higher than the 65000 that is reported.  I think that "1111 1111 1111 1111" is a "fake" value, when the fan is not returning a valid number. 

Also, if you add "1" to "1111 1111 1111 1111" and keep only the last 16 bits, you get "zero" (plus an "overflow" signal). So, "1111 1111 1111 1111" could represent "minus-1". Adding "1" to "minus-1" gives "zero".

Are you replacing the CMOS battery, or always using the same one?

[h1csvntdracones]

The fans actually all report 65535 during the issue, I was going off memory and thought it was 65000, but was wrong. I am using the same battery, I just remove it and put it back in, that fixes it. I did test the voltage with a multimeter and the voltage is still good

[h1csvntdracones]

The odd thing is that pulling the battery fixes it, even switching the jumper to BRR and doing a full bios replacement doesn't fix the issue. Think it would be possible the   1111 1111 1111 1111 is getting stuck in the cmos somewhere and not releasing the value? that would account for the battery pull working and would explain a thermal error along with a cmos crc mismatch error. If the system thinks all fans are running at absolute max rpm, then I wonder if the system treats that as a thermal issue? 

Also, occasionally it will get past the boot loop long enough to try starting windows, but ends with a blue screen, I think the error is required application failed to start or something similar,. Moist of the time though I just get the 2 errors, then it turns on turns off and repeats, but does it within seconds, too fast for the GPU to have time to initialize.

The first time it did this, I was rebooting for a Win 11 insider update, and as it went to reboot, I started getting an application failed to start error repeatedly, like filled up the screen. I think it was adb for Android. It did eventually reboot, but that is when I first saw this. Then no issues for a week or so, and it started happening more often. Now it is literally at every reboot.

I think the reason windows errors on start during this issue, is because the cmos resets , and might not have TPM 2 working or setup at that point. I saw a very similar windows issue when someone replaced a motherboard and the new board didn't have tpm 2 available until after a bios update. Win 11 had the exact same error and reaction, so I am hoping that is able to account for the windows issue as a result of the corrective actions taken by the system rather than a windows issue itself, making me think it is something with the reboot process. 

I did test just pulling the plug. and it started up again without issue, so it happens during the shutdown or reboot process the system goes through.

[Prométhée]

hi

Hello
there are different Hp document
in general it is a problem of overheating, bios problem, motherboard
Check temperatures if possible, in different situations.
The correct functioning or not of the cooling system
Even if it is not documents indicated for this model, it remains plausible:

https://support.hp.com/au-en/document/c06158628

https://support.hp.com/lt-en/document/c05606404

[itsmyname]

@h1csvntdracones -- I did test the voltage with a multimeter and the voltage is still good.

The chemical reactions inside the battery always produce the same voltage.

You also need to test the amperage -- how many simultaneous chemical reactions are occurring.

Compare to one person pushing an automobile out of a snowbank to having 5 people pushing.

Each person has about the same strength (voltage), but combined, their "amperage" is much higher.

A brand-new "AA" battery can produce up to 7 or 8 amperes. A very-used one will produce less than 1 ampere. Both batteries still produce 1.5 Volts.

[itsmyname]

@h1csvntdracones --  a cmos crc mismatch error

When your computer is told to write a block of data to the disk-drive, a CRC (Cyclic Redundancy Check) is calculated, based on the number of "0" and "1" bits in the block of data. Then, both the data and the CRC are written to the disk-drive. When the computer tries to read that block from the disk-drive, both the data and that CRC are read. then, the CRC is recalculated, based on the actual value read from the disk-drive. If the "new" CRC does not match the "old" CRC, you have a "mismatch" error, indicating that the disk-drive did not PERFECTLY write the data, or did not PERFECTLY read the data.

The same CRC algorithm is used when writing data to the CMOS. When the motherboard reads from the CMOS, it does the same verification. Any errors in reading from the CMOS can cause that "mismatch" report.

At less than $4 for a new CMOS battery, you really should try a brand-new CMOS battery.