ASRock Z490 Taichi VRM sensors missing/MB sensors duplicated

Lukas0610

Member
Hey, I just got my new mainboard and wanted to report some sensor issues.

The sensor values for the Intersil ISL69269 VRM are missing completely and the values for the NCT6796D are duplicated, with the second version showing strange values like 13.536 V for the +12V sensor. I tested this with CPUID's HWMonitor which only shows the first version of the sensor (although with fewer values).

I already read about the PCH problem in another thread, if you have any solution you want to test I'd be happy to help out.

Debugging output and report are attached.

Thanks in advance
 

Attachments

  • Report+Debug.zip
    111.6 KB · Views: 5
I will try to fix the duplicated NCT6796D with correct data in the next build. But I don't know yet how support the ISL69269 on ASRock boards.
 
Thanks for that, and the thing with the VRM is unfortunate, but it doesn't seem like ASRock's own A-Tuning cool can't read them either, so it seems like they aren't accessible at all.

While looking at A-Tuning I found out a few more more things:
The Vcore sensor values is always twice as high as it is supposed to be and the fan sensor named AUXFANIN4 is MOS1.
Further the temperature of MOS2 doesn't seem to be correct as A-Tuning gives 25°C while HWiNFO shows 32°C (Goes on with AT: 26 and HWi: 33, etc.; seems to be off about +7°C).

Everything else seems to be fine, thanks for the great work Martin
 
Thanks for that, and the thing with the VRM is unfortunate, but it doesn't seem like ASRock's own A-Tuning cool can't read them either, so it seems like they aren't accessible at all.

While looking at A-Tuning I found out a few more more things:
The Vcore sensor values is always twice as high as it is supposed to be and the fan sensor named AUXFANIN4 is MOS1.
Further the temperature of MOS2 doesn't seem to be correct as A-Tuning gives 25°C while HWiNFO shows 32°C (Goes on with AT: 26 and HWi: 33, etc.; seems to be off about +7°C).

Everything else seems to be fine, thanks for the great work Martin

Thanks for the feedback. Please try this build: www.hwinfo.com/beta/hwi64_627_4197.zip
And let me know if it's reporting correct now, or you see any issues.
 
The Vcore voltage is now correct and the fan is named correcly, but the MOS2 temperature is now off A-Tuning's value by 8 to to 9°C.

The second NCT6796D disappeared and shows an NCT5585D instead with some values A-Tuning also reports like CPU Standby 1 voltage and VCC PLL voltage, but those temperatures are still strange and I don't see any equivalent to them in A-Tuning. Further there seem to be a few negative readings of some temperatures, but I couldn't reproduce that yet.
 

Attachments

  • 1592838262808.png
    1592838262808.png
    42.1 KB · Views: 1
Thanks for the feedback. I will remove the other invalid temperatures, but still don't have a clue about the difference in MOS2 temperature.
 
It seems like ASRock multiplicates a few temperatures by 0.8, which seems to be correctly as when I apply that to the values shown by HWiNFO they always are equal to the values shown by A-Tuning.

This applys to three sensors as far as I know (internally called M2_1_TEMP, M2_2_TEMP and M2_3_TEMP) and those are used on a few other mainboards too, but mostly just X399 and Z490 ones.

If you want to I could send you the configurations used by A-Tuning so you would exactly know which sensors are affected by that "workaround".
 
Thanks. I have the XML files for ASRock boards, but can't see there any mention of such multiplier for those temperatures.

<Item Name="M2_1_TEMP" Bank="6" Reg="0x70" PatchMB="0">M2_1_TEMP</Item>

Have you seen that somewhere else ?
 
I found it when I decompiled A-Tuning's main executable and looked for the references to M2_1_TEMP. At one of those references it multiplied the sensor value by 0.8, which I then tried out with various MOS temperatures and compared with A-Tuning's displayed values.
 
That's interesting! I'm wondering why does ASRock reduce the temperature value as the raw readout should be straight in degrees of Celsius.
Is this perhaps some intentional attempt to hide higher VRM temperatures? The fact that this adjustment is not done in the XML, but hidden in the EXE looks suspicious as well.
I don't think I'm going to apply the same adjustment in HWiNFO unless ASRock proves that this is really required to get accurate temperature.
 
I don't really think that they want to hide higher MOS temperatures, but as the name M2_1_TEMP hints it seems to be have been intended to be used for M.2 temperature measuring, but I agree, it looks quite suspicious as any other sensors reports the values straight without any further manipulation.

Plus I just looked at the BIOS's temperature reading and those match the readings provided by HWiNFO, so it seems ASRock was just to lazy to implement new temperature sensors into A-Tuning and just used those already available...
 
Yes, M2_x_TEMP values were used for M.2 sensors in the past (X399) boards. The fact that they re-used this internal name for the VRM on Z490 might be just a coincidence and the additional scaling was not noticed by the developers. Difficult to say if this was intentional or not, I have asked ASRock about this..
 
Back
Top