PID settings on the Tiny Fluid Bed Roaster

by Tije de Jong

Trial and error
A few days ago I made a first successful attempt to improve the PID settings on the roaster.
I’m certainly not an expert in this: I have only a faint idea of how a PID works, and what the
effect is of each of the P, I, D-vaues.
Therefore, a bit of trial and error seemed the best way to go for me.

The settings I had been using since we fist got the roaster to work  (P=1.0  =I=0.5   D=0.0), didn’t
lead to the desired controllable roasting process.



In particular during the drying phase, where the inclination of the profile is rather steep, I had to
slide the background back and forth in order to get the machine on the right track.
Once passed dry end, the system gained some stability.

D for derivative
I was told that the P-value could be regarded as the gas throttle of the machine.
The function of the I and D were still a mystery to me.
At first I tried increasing the I, which gave a slight better control, but still not satisfactory.

It seemed that the machine was responding too late and too much to any deviation from the desired curve.

After that I raised the D-value a little which worked a little better.
When raising the D substantially I noticed a large improvement in the functionality of the machine.
It appeared to me that this D-factor gave the system a better ability to look ahead and anticipate
what’s coming.

The settings at this moment:  P=20   I=5   D=20.



Now, the machine reacts way more subtle and accurate.

I will use these settings for a while and in the future I will probably try a few more adjustments.

Reacties

Corneel zei…
Hallo Frans/Tije,

Ik ben zelf ook aan het spelen met mijn net gebouwde minibrander en was benieuwd naar je curve-settings. Bij mij schieten ze alle kanten op. Pas als ik de smoothing heel hoog zet begint het ergens op te lijken.
En klopt het dat je PID settings P: 20, I: 5 en D:20 zijn? Of moet de laatste D 2.0 zijn?

Groeten,
Corneel
Frans zei…
Hoi Corneel, indien je je mailadres erbij zet kunnen we je beter bereiken. Heb je ergens een verslag staan van wat je precies hebt gebouwd, welke hardware gebruikt, je profielen etc? Het vinden van die settings kan per apparaat verschillen en het is een zaak van 'trial & error'.
Anoniem zei…
I had the same problem (and solution) with a fluidbed roaster similar to your Tiny. I tried many PID configurations I found on the internet with little success. They all shared low D values and even the "general knowledge" was to keep D low or even zero.
Then I found a basic PID-tuning method - increase P to get a slight overshoot of the first oscilation -> then increase I to get the temp oscilating around the SV -> increase D to smooth out the oscilations.
This worked great and I ended up with settings of P=10.0 I=5.0 D=85.0. That is a configuration I would have never tried on my own but it gives me rock solid performance (except the dip during the FC). I guess that "our" fluidbed design has big tendencies for temp overshoot and oscilations and high D value (and in my case lookahead of 10s) is needed to tame it down.
In words of the young folk - "She wants the D!" :D
Frans zei…
Thank you for your contribution!
♠vsc♠ zei…
Hello Frans! I am also building a fluid bed roaster (from a popcorn popper) and I am struggling with the PID configuration. Are you still using this conf (P=20 I=5 D=20) in your latest videos?
I sense my roaster is way too powerful (in a bad way for the PID) so I limited its power by software... but still need a good PID conf.
Thanks in advance!
Frans zei…
Tije has played with PID settings, not sure which he has currently but I think that is different for every design, a fascinating 'trial and error' learning curve -- a PID like the Fuji can auto-tune very well.

Populaire posts van deze blog

Tiny Cheap Fluid Bed Roaster by Tije and Jan

Finished Pavoni & Fuji PXG4 pid: Probe Inside the Ring

TC4+ boards and Arduino controlled roaster by Matthias Gerstgrasser