Hi @caius, I definitely agree about minimizing wear on the fan. The way the code works stock, it It looks like it sleeps for 30 seconds between temperature tests so I’m trying with a 1 second sleep instead to see if that might tighten up the control. (It definitely does) That’s really their hysteresis - having a 30 second sleep in the code, at least that I see.
The other thing is I only muddle at python, so am not certain but I’m not finding any reasonable limit to the number of temperature/fan speed pairs you can have. That might be the easy way to input a sharper curve that isn’t so steppy with all the thermal cycling on the chip/board. And I have to say I don’t know if that is much ado about nothing. I know CPU and heat loads change all the time as a system runs.
But if it’s easy to tighten up how it runs with a shorter update cycle and more points, it might be easy to keep the system happy and minimize stress and fan hours.
Argononed is pretty lightweight. I shortened time.sleep() down to 1 second and that smoothed out the temperature cycles. I checked top and everything is dwarfed by dump-1090-fa and fa-mlat-client. 5 minute load average is 0.25-0.26 with python3 (where I think argononed resides but maybe other stuff is using it too) is less than 1% CPU.
I updated the config file to have 9 pairs - 40C through 48C, from 10% to 100% fan speed.
40=10
41=15
42=20
43=25
44=35
45=50
46=65
47=80
48=100
Temperatures are much steadier now. Next step is to give it better ventilation and some shade.
Here’s the edit I made to the /usr/bin/argononed.py file. I just changed time.sleep(30) to time.sleep(1). It’s near the end and there are two locations. Easier to post a screen grab…
