Trouble

PMU running way too hot

5 1822
pi52  
I've got a Banana Pi M1 running 24/7 inside a ventilated enclosure in a vertical position (I read that this was better than horizontal), but am a little concerned if the PMU temperature readings are to be believed.

I've read through the various posts such as this one with regards to getting the temperature readings and their accuracy etc, but in every example I've come across, the CPU has always been the hottest, and temperatures have been fairly low in general.

I have not got any heatsinks or active cooling, but the case does have ventilation holes as I mentioned, and both the CPU and HDD temperature seem fine compared to the PMU.

The Pi is powered via 5v to the HDD power port, not the micro USB, and the connected HDD is also powered by this same 5v source, so not relying on the Pi. It is running OpenMediaVault acting purely as a networked storage device, and has small amounts of data written to it a few times an hour.

PMU Temp

PMU Temp


As you can see from the above image, the PMU temperature seems to be alarmingly high when you consider the lack of load on the Pi, and the small ~200mA current draw.

PMU Temp Graph

PMU Temp Graph


In the graph above you can see that the temperature decreases over night when it is cooler, and then increases again during the day, however even the lowest temperature reading of ~45 seems to be far too high.

Am I right in thinking that the Banana Pi will shutdown automatically if the PMU reaches a certain temperature? If so, at what temperature?

Does anyone have any thoughts on what could be the cause of this high temperature reading? Perhaps the sensor could be faulty and it isn't actually that hot? Or maybe the actual AXP209 is faulty?

Any advice would be greatly appreciated.
My BananaPI is running 24/7 on my basement within an ambient temperature of 23°C.
The PMU is about 30°C and the CPU about 40°C (uncalibrated).

Do you have any devices connected ? Could it be cause by an external sources (Sun, ...) ?

tkaiser  
There is clearly something wrong when you get these temperatures with this load/consumption.

I would do a simple test: Get a non-conductive glove and touch the AXP209 PMU while running in the early morning. If the temperature read-out increases then the thermal sensor inside the AXP209 is broken. If it decreases then the AXP209 seems to be broken (and you have an aching finger ).

If everything gets back to normal when you pull the Banana Pi out of its enclosure then you know what to blame for the high temperatures. But this shouldn't be possible given the low average consumption of just 1.2W/1.3W and the temperature values reported by the SoC.

BTW: I highly recommend using FPeter's "sunxi_tp_temp 1447" from here: http://www.lemaker.org/forum.php ... 47437&fromuid=33332

pi52  
Thanks for the replies.

The only connections on the Pi are the power input (through HDD connector), network, SD Card, and SATA. So nothing should be drawing any power except the Pi itself.

The enclosure is a dark room, so no sunlight on the enclosure. I'm fairly confident that the enclosure is suitable for the Pi given the locations of the chips, but I will take it out and run it in free air as suggested since this is a sensible test.

Could you explain a little why if the temp increases the sensor is broken, but decreases the actual chip is broken? So for both to be working the temp has to remain the same? What kind of tolerance are we talking about here.

Using the recommended "sunxi_tp_temp 1447", it reads 43.2, whilst the other RPi-Monitor page reads 50.6, so is a little better if that temperature reading is more accurate, however still worrying that it is above the 41.8.

tkaiser  
pi52 replied at Mon Jul 20, 2015 05:20
Could you explain a little why if the temp increases the sensor is broken, but decreases the actual chip is broken?

I would assume your body's temperature is around 37°C? I would also assume that the room temperature is below 30°C?
If you touch either SoC or PMU your finger will immediately lead to heat dissipation. Depending on the load/consumption the PMU's temperature should be just a few degrees above ambient temperature. Given the two assumptions above are correct, the PMU's temperature should rise by a few degree within seconds.

If it now reads 45°C - 55°C and the value still increases if you press a finger on the PMU's surface then the value reported is wrong (thermal probe inside the PMU broken). If it actually decreases then the PMU is really that hot and this is either a sign of a broken PMU or of a thermally broken enclosure.

BTW: sunxi_tp_temp measures the SoC's temperature and not the PMU's.




pi52  
I'm far too tired for my brain to function at the moment .

Okay that does make sense, thank you.

Also the 41.8 reading I mentioned was the CPU reading from RPi-Monitor, so if the 43.2 from "sunxi_tp_temp 1447" is the CPU too, I guess the readings are close enough.

I will do the various tests and report back, but it may be a little while as I don't have physical access to the Pi at the moment.

You have to log in before you can reply Login | Sign Up

Points Rules