Markus
Good question. There’s a hell a lot of thermal inertia. To tune it, I did steps from ambient to 26C (step up for 20-30min) - it took ages to do. Followed these three standard(ish) steps
1. Set P-value (proportional gain): start with PID = 0,0,0. Increased P until it could at least closely approach the step set point without much oscillation when it got there. NB: there will be some error over time when it reaches the new stable value.
P=100 was about right.
2. Set I-value (integral gain): with P=100, D=0, increase I to reduce the long-term error. This will be at the expense of some oscillation at the new set point.
I=150 was about right.
3. Set D-value (differential gain): with P=100, I=150, increased D to reduce oscillations (without too much initial overshoot).
D=10 was about right
After getting it set up on a one heater strap with my 80mm refractor, I then tried the same settings with 2 other things – 50mm guidescope & a 30mm guidescope. They were all pretty similar. And you can pick than from my graphs – have to change PID values a lot to get any change in responsiveness.
You don’t even really need PID but I used it just cause there’s a nice library for it. I originally just used difference between set-point and actual heater temperature but I found it to be very sluggish and always below the set point - that’s the nature of P-only control.
This has a great graph stating the above process
https://forum.arduino.cc/t/pid-library/333329
Hope this helps