Asus Embedded EC chip get bugged T_Sensor values

ssateneth

Well-Known Member
On my Asus Rampage V, I use the ASUS EC chip to report temperature value on my liquid coolant (its a wired temperature sensor that you can place anywhere. I simply taped it to the outside of the tubing and insulated it, works well).

I'm not concerned about lag or CPU usage as I put it on a conservative 20 second interval on sensor refreshes. Periodically, sometimes the sensor bugs and returns a -115C or 116C value but returns to normal on the next read. But eventually, within a couple days, it gets stuck at -115C. There are 3 T_Sensor values. Hot-plugging in the wired sensor to a different header doesn't fix the issue, and neither does using a different wired sensor.

AI Suite, Asus's program for doing some motherboard specific stuff (fans, sensors, volt setting, clock setting, etc) also reports the T_Sensor values, but when its bugged in hwinfo, its bugged in AI Suite as well, simply reporting the last known valid value, or nothing at all.

I wanted to know if this could be a hwinfo-related issue that triggers the bug, and if it could be fixed.

I'll update to the newest 5.xx version, currently on 4.64-2530. I'm on 5.02 now.

https://dl.dropboxusercontent.com/u/9768004/temp/HWiNFO64.DBG
https://dl.dropboxusercontent.com/u/9768004/temp/SSATENETH-PC.HTM
 
Reading this sensor is a bit tricky (I think it's not well designed by ASUS) and that's why you get the warning ;)
I'm not sure whether you run HWiNFO together with AI Suite, but I strongly recommend not to run them both, as this might cause multiple issues because of the nature of these sensors.
But if the problem you describe happens when you run HWiNFO only and when the value gets stuck after a while, which you then confirm by AI Suite (close HWiNFO, run AI Suite), or if it happens when running AI Suite only, then I think the problem is somewhere in the BIOS/firmware.
 
I closed AI Suite as soon as I restarted and just had hwinfo64 probing the sensor. After 2 days, it returned to a -115C value. I'll have to see if AI Suite by itself causes this problem too.
 
Bumping this thread from the dead. I have reason to believe that the AMDA00 Interface Driver is responsible for pulling information from the ASUS EC sensor. I found a minor revision update for this driver (1.0.0.0 -> 1.0.1.0). I can now poll the sensor very quickly with minimal CPU usage. It's possible the sensor hanging and returning invalid values for the user-placed thermal diodes is fixed too, unless you've done something recently. Running HWiNFO64 v5.05-2633
 
Yes, I have optimized readout from the EC chip in recent versions, so you should see a significant reduction in system load/CPU usage while polling it ;)
 
Thanks for the followup post. I see it in the 5.04 changelog, well done :)

Fingers crossed to see if the polling doesn't return invalid values after a long period of time again. Hopefully not, due to optimized EC reading or later AMDA00 driver revision.
 
Back
Top