I have an instrument I want to maintain at as constant an average temperature as I can conveniently manage. There is enough thermal mass for a fairly long thermal time constant, like 30 minutes, and the ambient only changes a few degrees anyway. I have high-resolution temperature sensors. I was thinking about running a PWM-controlled heater to make my enclosure a stabilized oven. I'm not sure yet, but probably a few watts of power is enough. Now it seems to me the difference in heat dissipation of a Pi 5 at max CPU load, and at minimum load, is at least several watts*, so if I need some computer to record and transmit data anyway, I can also use a Pi to double as the temperature stabilizing element itself, by turning on and off some dummy computations under control of a PID loop, as needed to maintain the temperature of the whole package near a setpoint.
Is there already some stress-test code, that not only generates a peak thermal load, but takes a parameter and dissipates some amount of heat more or less linearly proportional to that parameter? Ideally would like to be able to adjust the parameter by very fine steps, and with low latency so I do not need to wait for some long computation cycle to run before changing the power consumption rate.
* Adjustable from 3.2W idle to 9.8W max, according to https://www.jeffgeerling.com/blog/2024/ ... er-savings
Is there already some stress-test code, that not only generates a peak thermal load, but takes a parameter and dissipates some amount of heat more or less linearly proportional to that parameter? Ideally would like to be able to adjust the parameter by very fine steps, and with low latency so I do not need to wait for some long computation cycle to run before changing the power consumption rate.
* Adjustable from 3.2W idle to 9.8W max, according to https://www.jeffgeerling.com/blog/2024/ ... er-savings
Statistics: Posted by jbeale — Fri Feb 07, 2025 4:38 pm