Some temps drop down under load - maybe distance temps?

[hwwulli]

I'm using an Asus ROG Strix X570-F Gaming AMD X570. Starting temps of two sensors (Temp2 and Temp5) ar 95 and 96 degrees celsius. After some minutes on idle the temps go down to 80 and 83 degress. While gaming the temps are around 75 and 77 degrees. Under full load (CPU and GPU) the temps drop to around 61 degrees.

The PWM controller and VRMs are 12(4)+2(1) Phase ASP1106 (4+2) and 14x Vishay SIC639 DrMOS (50A). Max. operating junction temperature of the SIC639 is 150 degrees celsius. Junction temperature is 135 degrees.

Maybe this is a distance temp (150-95=55 (or 135-95=40) degrees after turning the pc on)? Didn't find the specs for ASP1106 but I think its 150 degrees, too. I guess you know the specs better cause you have implemented the ASP1106 in release 4.x

 

[Martin]

I think that those values are most likely some invalid readouts from sensors not connected and do not represent any real temperature.

 

[hwwulli]

Isn't it implausible if the temps go down under load and raise again without load. Even when I open the frontdoor and raise the fan speed the sensors react and go up a few degrees. They drop down as soon as I close the frontdoor. Shouldn't invalid readouts more random?

Couldn't it be a kind of Tjunction like the Intel core CPUs have implemented? The specs of the SIC639 mention following:

"Thermal Shutdown Warning (THWn)The THWn pin is an open drain signal that flags the presenceof excessive junction temperature. Connect with amaximum of 20 k-ohm, to VCIN. An internal temperature sensordetects the junction temperature. The temperaturethreshold is 160 °C. When this junction temperature isexceeded the THWn flag is set. When the junctiontemperature drops below 135 °C the device will clear theTHWn signal. The SiC639 and SiC639A do not stopoperation when the flag is set. The decision to shutdownmust be made by an external thermal control function. "

https://www.vishay.com/docs/76585/sic639.pdf

I have not seen any function for the SIC639 to read the temp of the internal temperature sensor. I think it's a bit late to set the THWn flag at 160°C if the max junction temperature is 150 degrees. Asus is not interested in RMA

Assuming a Bios programmer has 30 different boards with 10 different VRMs and 10 different max operating temperatures. He could use the max operating temp (150 for the SIC639) as a fixed value and substract the value of the temperature sensor. Regardless of the used VRM he can react at 0 degrees with an emergency shutdown. No need to check the tjunction flag or to change the code for different VRMs. Only one value in the header.

Just an idea. Your are much more expirienced in this things. I don't have a clue if there is any Bios working this way.

 

[Martin]

I agree that the behavior of those values seems to indicate some scheme, but I don't think this is real.
Those temperature sensors would have to be connected to true temperature diodes, which isn't the case I'm afraid.
Depending on SIO configuration those inputs might be just floating and/or mirroring something completely different.

 

[clearing68]

Good info, I'll ignore those 2 sensors. I'm going to post details here in case anyone else has a similar worry. It took me a bit of searching to find this.

New build, Asus X570-e Gaming, RTX 2080 Ti in the primarary PCIe (right over the PCH air intake for the chipset). 2x 140mm fans front and 1x 140mm fan rear blowing in, and a celcius s36 on top blowing out. All components reporting they were fine 35C-45C at idle. But Temp2 was running 80-82C, and Temp5 was 0C (not a typo - 0C for the better part of a day). I was concerned the video card was dumping heat onto the PCH, or I had jostled the fan loose or something. BIOS reports 2500-3500 RPM on that fan, so it's running and, even though BIOS doesn't report the PCH temp (come on, ASUS!). I opened the front and cranked the front fans. Temp2 started climbing, and then Temp5 started to report at 80C and started climbing (computer at idle). I lowered the front fans down to default (~900 RPM) when it got close to 90C, and Temp2 and Temp5 started to drop, returning to about 81C after 10 minutes. I added the thermistor sensor to the board and taped it over the PCH air intake (right below and next to the video card), and it read 35C at the intake.

So, nothing to worry about it sounds like, but it had me spooked for a day. Thanks for all the wonderful work you do, and the help.