Overheating odroid c2

[dutchmaster18]

Can someone tell me if maybe i have a faulty board. My cpu tems reach 150 Fahrenheit while watching video then just cuts off. I believe my board is faulty but was wondering if anyone else is having like issues

[mad_ady]

65C is not a high temperature, theboard can do 80-90 without crashing.
What player/os/power supply are you using?

[dutchmaster18]

Using a 2 amp samsung phone charger. Using latest android release

[rooted]

Using a 2 amp samsung phone charger. Using latest android release

Do you have the jumper removed?

[mad_ady]

You should try with a different power supply just to rule out power as a problem
I'm guessing power comes over the otg cable. You should try something with a barrel connector so you can remove the jumper. That would buy you ~8C in temperature. It's also worth to try with a short usb cable (the one supplied with power banks) to reduce resistance.

[dutchmaster18]

I think it was the charger giving me my problems... Im hovering around 45C to 55C now with a barrel connector from amazon for the c2... It is just weird because before with the samsung charger i never had a proble with it on my other c2. I will run it overnight as see what happens.... gotta do some more testing with videos and the such... thanks for all the imput

[graysky]

You should try with a different power supply just to rule out power as a problem
I'm guessing power comes over the otg cable. You should try something with a barrel connector so you can remove the jumper. That would buy you ~8C in temperature. It's also worth to try with a short usb cable (the one supplied with power banks) to reduce resistance.

Can you cite a source on that delta 8°C/OTG vs DC Input? I have both power supplies and in testing the CPU temp under both, I notice only ~2°C cooler load and ~3°C cooler at idle. All tests were carried out using the ondemand CPU governor, and of course, this is under Linux, not Android but I can't see that making a difference.

Load
CPU load generated using stress -c5 which ran for 1100 seconds totally. I waited until the temps leveled out (800 seconds) and averaged the CPU temp reading for the next 300 seconds. Temp data were collected once every second for the load test.

Load Results
Mean temp with (jumper removed) DC power (800s-1200s) = 64.0°C ± 0.2 (n=3 runs)
Mean temp with (jumper installed) OTG (800s-1200s) = 66.2°C ± 0.2 (n=3 runs)

Idle
For idle, I just sampled the CPU temp while the machine was "idle" once every 4 hours. I use quotes around idle because this box runs pihole and openvpn server. I didn't stop either of those for the sampling but given size of the data sets and then fact they I sampled them over multiple days, I would think we could discount them as present for both tests and averaged out. For the OTG power, I had about 5 days-worth of data (27 points); for the DC power, I had about 10 days-worth of data (61 points).

Idle Results
Mean temp with (jumper removed) DC power = 42.1°C ± 0.6 (27 data points)
Mean temp with (jumper installed) OTG = 39.5°C ± 1.0 (61 data points)

EDIT: based on some comments, I edited the text in this post (discussing cpu governor and correcting a mistake I initially made).

[rooted]

Last I checked the idle temperature was at least 5° C lower.

[mad_ady]

The difference was what I noticed under linux in idle conditions with ondemand governor. If you're using performance it probably won't make much difference.

[rooted]

Definitely, using ondemand makes the difference. Performance wouldn't see much if any change.

[graysky]

I edited the post a few above to address the question about ondemand (which I was using) and to correct a mistake I made in pulling the data into gnumeric. TL;DR summary is DC power for my test case was about ~2 degrees cooler under load and ~3 degrees cooler under idle.