Update on the progress with PID auto-tune shows how a measurement of the maximum

Update on the progress with PID auto-tune shows how a measurement of the maximum rise time, the dead time and the heat transfer coefficient can be used to better the relay tuning skew. The tuning is completed on the -16 min mark here, and the testing shows that the P and I-values are perhaps too low. But the biggest problem with this tuning method is that it does not work reliably in the face of forced convection, IE a fan blowing on it. Even though the pre-calibration and tuning setup only works without a fan blowing, it should serve as a starting point for determining the right values for the PID and then move on to making it more robust.

663e8c118109851f170ea079c8f4a6f1.png

I would argue that having a fan blowing on the hot end is bad hardware design and should be addressed in hardware, not software… But it’s pretty common for people to run auto-tune with the hot end airflow at max. Does that work here?

I’ve not tried having a fan blowing on the hot end, but I’ve tried having active cooling on the cold part of the hot end and that also makes the calculations wrong. From what I can tell, the calculations for both max heat rate and cooling constant seems right, but the set point power value is not right. Overshoots by about 25 degrees every time.

What kind of hot end are you using for these tests?

I’ve got a hexagon hot end that I’m testing with now, but I’ve got an E3D that can be hooked up as well. Actually, let me do that right away…