Need fan adjust (increase) since PC unstable/crashing

[nixda]

Since the PC crashed too often (it should never crash!) I checked for thermal problems.

To find out what happens the result of my experiences here:

The fan in general is fine/working but too low!

Sometimes it crashed during some minutes "doing nothing" with the PC. But then I started to figure it out and used a CPU und GPU stress test, too.

On the h56 I did:

(1) I cleaned and put on new thermal paste (didn't help).

(2) I removed the bottom of the device and put a big an very, very slow turning fan under it (5V DC powered, a standard PC case fan). Pushing cold air from the bottom...
With that I could increase the working time a lot (see attaced picture)!

(3) Then I adjusted the fan speed from 5V to 12V. Now no crash anymore!

With that I can rund a FurMark GPU plus in parallel a CPU stress test for over 40minutes (no crash, stopped then). I get stable max 61°C at the cpu termal diode. Without the additional (and strong fan) from the opened bottom the PC crashes after minutes.

Poorly the whole disaster with the PC started after ~5 weeks I got this PC. (AND DON'T TELL INSTALL THE SOFTWARE NEW! It crashes with everything... booting Linux from USB, starting memtest86 from USB...!!!).

I now would like to remove the additional fan at the bottom. I would like to connect my SSD again. I guess we could get it working with increasing the fan speed. I do not realy understand why the PC worked for some days/weeks (and then from one day to the other stopped). Maybe we have low quality parts in it ageing very fast?

Could I please get a tool to increase the fan speed/courve? New BIOS...?

Thank you and best regards

 

[nixda]

Hello Alliwava...?

 

[Alliwava]

We have debugged a new BIOS for you, try to download and update the BIOS.
https://1024terabox.com/s/1YGLJLvhCUNk3ep8Z67OEXg

 

[nixda]

@Alliwava Thank you for taking time and offering this bios.
I have tested this with a 40min run of FurMark 2.9.0.0 (in the default mode 1920x1080 window) plus running the so called "CPU Burner" from FurMark in parallel.
Two things I notices:

(1) after 40minuten it looked good (but wasn't completely) since the pc crashed during the shotdown.
(2) the overall benchmark from FurMark become much worse. Round half of the FPS!

Since I can compare with a second H56 I can not here a differen fan control. What did you change in this BIOS?

For me it looks like you tried to reduce some gpu power..?

Now I have added the PC tower cooling fan again. Maybe it will not crash if I only surf through the web. But to be save...

<del>But could you please so kind and tell where I can get the old bios?</del> found the bios in the other thread for the h56.

Maybe there is no chance with this h56 to get ir working stable(?).

 

[Alliwava]

Yes, we lowered the power.
We can understand that the performance of the H56 cannot meet your special scenarios.
A higher-end model may be required to meet your needs, such as the H90Pro or GH9.

 

[nixda]

Yes, we lowered the power.
We can understand that the performance of the H56 cannot meet your special scenarios.
A higher-end model may be required to meet your needs, such as the H90Pro or GH9.

Sorry I didn't explained clearly. I do not need more power.

Since my 1st H56 begun to crash after minutes (sometimes seconds!) doing noting (and before that it worked for round about 5 weeks) I bought an Intel N150 based mini PC from another reseller (or vendor - hard to find out). I just need a working "office" PC. This N150 is working so far.

Meanwhile I have a second H56!

This 2nd is not crashing so far. Keep fingers crossed(!)

But I want to understand "what is broken" with my 1st? Why did the 1st get that unstable? Is the thermal cooling the problem? "Broken" heatpipe? Fans too slowly? (I can get it working with the additional cooling fan below). Or are some circuit parts (voltage regulators) on the "bleeding edge" with thermal things or current draw?

Yes it would be very fine and helpfully if I had a thermal camera. Will by some one soon *g*
If you would like to recommend me an reasonable IR cam?

So in general:
I can throw this 1st h56 away and hope the second keeps working. Or/and I can try to find out what's wrong with the first. Maybe you know what is the most critical part of it?

Like said before: Next steps would be a thermal monitoring and maybe changing the complete cooling. My intetntion is not to lower GPU or CPU power. I don't see this as the clever step. I think we should try to get the components (the ryzen) in their general specs.

What do you think?

Best regards and thank you!

 

[Alliwava]

Targeted problem troubleshooting operations:
Start by running lightweight hardware monitoring software (HWiNFO64) on the first H56.
Key observations:
CPU (Tctl/Tdie) temperature
GPU Temperature
This is the most important thing if you can see the temperature sensor related to the VRM MOS or Motherboard (usually found in the "sensors" list in HWiNFO).
Observe the rate and peak of temperature changes. If you can catch any of them (especially VRM MOS) temperature spikes before the crash, you can lock the problem;

2. Check if the VRM is overheating:

1. Replace the grease: Remove the heatsink, thoroughly clean the old grease on the CPU/GPU core and heatsink base, and apply high-quality new grease (recommended: Shin-Etsu 7921, Honeywell PTM7950 phase changer, Limin TF7/TF8, Thermal Grizzly Kryonaut).

2. Clean the surface of the VRM components.
These heatsinks are attached to the MOSFETs and inductors to increase the heat dissipation area.

3. Check the fan: Clean the fan blades and fins for dust, listen for abnormal fan sounds, and try to manually toggle to confirm whether it is smooth.

If the above can troubleshoot the problem and need to repair the corresponding accessories, you can contact us to repair the corresponding accessories and solve the problem.

 

[nixda]

Excellent information and help!

Everything looks like the problems are caused by the original cooling compound. I did stress tests with sensor recordings (hwinfo) on both PCs. The older one with the crash problems. They become rare as I switched to a simple and old silikon based thermal paste. But then I bought a newer one (most seller on amazone) Artiv MX4 and did test with that, too.

I guess my 1st and older one PC got in trouble as your themal compound becomes dry. Look at the pic... This was the 1st and crashing

 

[nixda]

Now let's have a look at the second and new h56. If you can see at the pic it looks still a bit wet...

 

[nixda]

I bet if I would continue stress tests... if the paste get dry I get in trouble, too.

Now look at the csv with a 10 min stresstest, Look at the later thermal sensor values(!).

In the two results (i cut of a lof of not needed columns) you can see a dramtic thermal peak somehow. In general at the end, at the moment where I stop/start the cpu stresstest. I tried to reconstruct and yes there are very short peaks. I have a video of that but I know you do not need.

And then have a look at the second resolts after changing the paste and using the Artic MX-4. No longer very high peaks! looks much, much better. And in the middle the old/original cooling had about 67°C in average under 100% load. With the artic only about 62°C in average unter 100% load. And the peaks where about 64°C and not 91°C.

So my result is: The original cooling looks like to have a problem with the cooling paste. For me it looks like the fluid there at first is wet and the cooling is "nearby ok" but not realy good. As soon it is dry you get in trouble... Solution is a new and good prepared cooling compound. I already did test with a some years old silokone compound an it was not that bad(!). But with the newer compound it looks goog (looking foreward!).

 

[nixda]

Somethin completely different: In the bios the secure boot was not enabled. In the ending a win11 update stoppend somehow. It is a lot of work to get this vey old win11 22h2 to a reasonble 24h2 or so. ))

Thank you and best regards and go forward with your PCs. Thank you for a good help here. I just would recommend to look at some parts and think of if there would/could something changed a bit. The cooling (compund) looks like a potentional risk. Just my two censt...

Best regards