Bug in Z240-SFF Fan-Control

[marcmeyer]

Hello Community,

I believe I have found an issue with the fan control of the Z240-SFF workstation and am seeking assistance. After weeks of research, I would like to share the following experience with you:

Z240-SFF, E3-1270V6, 64GB-2133, 2TB NVMe, 4TB SATA SSD, NVIDIA Quadro K1200 (BIOS 01.92 Rev.A, ME v11.8.94.4494)

The CPU and case fans do not accelerate under CPU stress. For example, when running Prime95, the CPU reaches about 95 degrees Celsius in approximately 5 minutes. During this time, neither fan speeds up significantly, only minimally from about 700rpm to 800rpm. After several weeks of validation, I have found that the baseline for fan control appears to be the chassis temperature rather than the CPU temperature. (I consistently noted the RPM of both fans and compared it to the chassis temperature, and I can confidently say that the chassis temperature is controlling the fans. -> This is an error -> The CPU temperature must be the baseline! (Adjusting a higher minimum fan speed in the BIOS does not resolve the issue.)

(Tested with Standard Heat Sink (810285-002) and Performance Heat Sink (810281-001) in Win10 x64, Win11 x64, Ubuntu, Debian, VMware ESXi.)

The problem is that the CPU reaches a critical temperature (95 degrees Celsius) after 5-6 minutes, causing all fans to suddenly accelerate to 6000rpm, making it sound like an airplane is taking off. The CPU fan can then immediately cool the CPU down to 65-70 degrees Celsius, so within a few seconds, the 6000rpm drops back to 700rpm. This means that both coolers (Standard and Performance) are capable of cooling the CPU without any issues, but the fans need to accelerate with increasing CPU temperature (which they do not do until 95 degrees Celsius)!

I urgently request assistance.

[NonSequitur777]

@marcmeyer,

Welcome to our HP Community forum!

Thank you for the PM to make me aware of your issue, let's take a look at this.

And as you mentioned in your PM, an HP Z240 SFF has been an upgrade project of mine a while back.

I didn't have that non-responsive cooling fan issue, whether I used an i7-7700K or my favorite, the Xeon E3-1280 v6.

Perhaps you want to try HP's Performance Advisor: https://www.hp.com/us-en/workstations/performance-advisor.html, as this tool offers real-time monitoring, diagnostics, and performance tuning capabilities to enhance overall user experience.  The reason for your issue could be a bad motherboard processor board sensor circuit or an identical processor issue -meaning, you may have to replace either your motherboard or your CPU.  In order to troubleshoot this, see if you can swap your processor with another one, and see if the same problem reoccurs.

Also, I want to make sure that you use 4-pin PWM cooling fans for both the CPU (CPUFAN) and the System (CHFAN2) fans:

Kind Regards,

NonSequitur777

[marcmeyer]

Thank you for your response! 🙂

May I ask which Intel Management Engine Firmware version you are using? (I have read about issues when an incorrect version is installed.)

Below, I am sending screenshots from the Performance Advisor. One showing a CPU temperature of 51°C and the other showing a CPU temperature of 91°C. It’s clear to see that the fans are not speeding up.

Regarding the motherboard headers: Yes, the original HP CPU fan is connected to the CPU port (CPUFAN), and the original HP chassis fan is connected to the CHFAN2 port (so both are PWM fans).
CPU @51CCPU @91C

I am away on a business trip for the next three days and will take care of the issue once I return. Thank you in advance for any further information. Thanks!

[SDH]

Problems with PWM fan speed control are usually ascribed to originating from the motherboard end of things. Don't forget that there is a fan rotor PMW response circuit board that is at the other end... that is another thing that can go bad. The fan label is usually on the side that faces into the CPU cooler so to get a new used replacement off eBay you likely need to take the fan off to get its part number(s). It is an easier and cheaper fix if that is where the issue is rather than on the motherboard.

[NonSequitur777]

@SDH,

You mentioned the "fan rotor PMW response circuit board".  If you don't mind, I would like to know where that board is situated.

Kind Regards,

NonSequitur777

[SDH]

Any new major projects?

The fan's spinning hub side almost never has a label on it. The side the air is blowing towards is the label side, not spinning. The wires come in under that sticky label which can be lifted up and looked under. That is right where the fan's tiny rotor controller transistor is. The motherboard gets its RPM data input from the "tach" wire #3 and sends back its control signals via the "PWM" #4 wire. Then the controller on the PCB under the label raises or lowers the RPMs. If you snip the PWM wire the fan defaults instantly to full 12VDC RPMs, the same as if only wires #1 (ground) and #2 (+12VDC) were connected. HP uses quite high RPM versions of the PWM fans it buys I think because those would be engineered to have longest life. It then applies quite high PWM braking so those fans last even longer.

That is why if you replace a HP original PWM fan with a nice slow Noctua that Noctua PWM fan will end up running way too slow. A trick I'll use is to buy lower amperage HP PWM fans (shown on the label) and stick them where a higher amperage HP fan previously was (if I know the workstation is to be optimized for quiet and does not do prolonged rendering, etc.). In our work we're supposed to slow down and actually look at the pictures...

92mm 0.40A HP PWM fan

[NonSequitur777]

@SDH,

The only new (upgrade) project I am working on at the moment is this: Re: Upgrading an OMEN by HP Obelisk Desktop 875-1xxx - Page 3 - HP Support Community - 9221092.  Not without its challenges either.

Anyway, you may very well be on to something: @marcmeyer, experiment with different 4-pin PWM cooling fans if you can.  As @SDH alluded to, HP computers can be finnicky that way.

Kind Regards,

NonSequitur777

[marcmeyer]

Thank you for your responses.

I rule out the electronics in a single fan as the cause of the problem. However, I have tested four different PWM fans, and they all behave the same. A notable aspect of my issue is that none of the fans accelerate: neither the CPU fan, nor the chassis fan, nor the PSU fan.

Unfortunately, I currently don't have another CPU available.

I suspect there might be an issue with the Management Engine firmware, or perhaps an incorrectly installed version. Do you know your version? (I rule out the BIOS, as I have already tested several downgrades.)

Also, does anyone know where exactly the fan control is located? In the CPU or on the motherboard? -> For a visual check.

I'm running out of ideas. 😞

[NonSequitur777]

@marcmeyer,

Well, for all intents and purposes, the primary temp control is situated in your processor: it is supposed to relay temps to your motherboard which then is supposed to ramp up your cooling fans.

Personally, I have yet to experience a faulty CPU, so that seems to be the least obvious culprit.  If I were to troubleshoot this issue, I would try a different motherboard.  HP Z240 SFF motherboards are inexpensive to come by -at least in the US.

Kind Regards,

NonSequitur777

[SDH]

To the OP... I agree with NS777 and have found new used motherboard transplants to be my usual solution for these strange things that crop up. I do dink around for a while but have learned to shift more quickly over to a transplant because our time has some value.

But first, I'd stop doing speed testing for a while, enjoy what you have, and recover from this distraction. Your processor has a single sSpec code and is easy to find via that on eBay from China for about $27.00 USD including shipping, plus a couple of weeks to get here. That is a reasonable first step, if you wish, or you could go straight to a MB transplant if needed. You clearly have the skills to do that.