TPfancontrol on t420s

The software can only detect one active sensor, the one for CPU. Other sensors, such as battery, pci and power, are "n/a" in the list of sensors.

Is that normal?

 

The embedded controller on the 420 and 520 series is different than previous models. You'll want to be sure you are using the latest version of tpfancontrol.

Find it here:
forum.thinkpads.com • View topic - W520 w/ tpfancontrol issue

But yes, a reduced sensor list is normal.

 

I am using v064. That's the latest I can find on their website. I saw in the page you listed that there seems to be a version for W520. Is there a version for the new 420 series?

 

Should be the same.

 

I tried the W520 version of TPFC on my T420s (Intel only graphics) and it didn't seem to run as well, so I reverted to the normal version. After a little tweaking of the temperature settings I have a good balance between heat and fan activity. Failure to read the other sensors doesn't worry me. If you want to check their values then try the Sensors tab of HWiNFO32 (but be forewarned that it does clock up a significant amount of CPU use if left running in the background)

John

 

Here are the problems with the older versions: From what I can gather, reading the sensors in the new embedded controller leads to some errors. Periodically you'll get a cpu readout that's just totally wrong. So say you are running something very cpu intensive. Suddenly you get a cpu readout of 5 degree C (erroneous reading). Tpfancontrol kills the fans because the cpu is cold. After a couple read cycles the value returns to normal (say 85C). Tpfancontrol dials in a fan speed of 7. You just spent a period of time with your fans not spinning while running. More so, this embedded controller doesn't always react immediately to fan speed changes. Sometimes, changing from 7->0->7 takes 10-15 seconds to bring the fans back up to speed.

Another issue is that I think there's now more than one location that needs to be written to. With the old version, I've had cases where I'm running something with the fans set manually to 7, and yet somehow, something else briefly takes control of the fan for 10-20 seconds and they change speed. This issue is further compounded with the quirky behavior of the new EC. If I go from fan speed 7 -> say..... 5. The fans on mine don't always simply go from 3800 rpm down to 3000 rpm. They'll go from 3800rpm -> 1000 rpm -> 3000 rpm.

All this odd behavior led to a VERY bad combination of events happening when I was running intensive things with the old version. If you aren't utilizing 100% of the cpu and you have integrated graphics this may not be as big a deal, but on my W520, running all cores 100% and the gpu 100%, 15 seconds of "confusion" where the fans stopped spinning sent my temps up to around 96-97 degrees C. Not acceptable.

I never see this behavior with the new version. It plays some games to eliminate the sensor readout errors and writes the fan speed to multiple locations. One thing I found helped with the new version, by default it allows the temp to change by 10 degree increments. I reduced that to 2 degree increments. So basically, if the cpu temp climbs quickly, the program will ever only register 2 degrees of change every cycle. That's how it prevents the erroneous readouts from killing the fan. So as an example, if the cpu was 50 degrees and spontaneously became 70 degrees, it would take 10 readouts for tpfancontrol to show 70 degrees (52, 54, 56, 58, 60, 62, 64, 66, 68, 70). This kind of behavior works well. It takes a touch longer to spin the fans up to speed but also spins them a bit longer when the temps dropping. A bad readout for a couple cycles won't affect the fan speed.

 

The software doesn't have any problem monitoring and responding to the CPU temp sensor... It is just that I am not sure whether that sensor alone is enough to monitor the whole system and control the fan...

 

The cpu has almost always been the lone trigger for fan activity.

 

for me, most ot the time /if not always/ is the GPU