Looking for some suggestions as the Artisan Software PID still feels like a bit of black magic to me (but awesome amazing magic).

This is my first PID roast (I'm very excited to say). I'm experimenting with controlling my Coffee Crafters Artisan 3-e with a servo. Post on that here:
Solved: Artisan-Scope and Yoctopuce YoctoServo - PID Heat Control

..and since threads are generally useless without pics, here's how I'm controlling the 3-e. I wanted to not have to modify the machine at all. This is my first pass and I will make something pretty for mounting later.




I'm mostly doing it for fun and the learning process of it, but watching the heat be controlled for the first time was super cool. Take a look at the roast. loads of things wrong (to be expected).

A couple of things to note on the roast below -
1. Im using PID background on a design I created. That design isn't super sexy or dialed in (ROR curve). Just experimenting at this point
2. The PID Control is a quite heavy handed with the heat. My heat dial was often between 0 (heat off) and 10 (max heat) and rarely in-between for very long.
3. I accidentally fat fingered drop at around 7/7.25 mins which made the final curve worse.





My PID Control looks like this (currently)




What should I change in the PID Settings to make the PID a bit more surgical and less heavy handed for next time?

Thanks!

Looks like there's a script that may help. Going to try this shortly!

https://github.com/marcov/pidtune

May not be understanding what I'm seeing but not seeing actionable data here from the script as it relates to PID for my machine.

I've been playing with the PID settings, thought I'd try the 33,0,0 but that resulted in the roast below.


I admit, at this point I'm a little lost. PID does control the heat but it seems to want to be at min or max and nothing in-between. (help)

Looks like I found a winning(ish) combination. The Artisan PID is now providing very fine control the way I expected it to.

PID Control settings for the roast below were: 1,0,0

So apparently, my roaster applies the heat so quickly I needed to back P WAY off. Probably still some tuning needed but its nice to see results.

I'm also noticing that the tail on the left side of my background profile is also likely causing some grief, because my beans dont rise like that.

Looking forward to roasting the next batch.

Latest Roast -



Current PID Settings (trying to go lower for the next roast)

Watching!

I'm getting closer but fairly trial and error at this point.

2 Roasts

Roast 1938, PID settings 0.66,0,0
Roast 2003, PID settings 0.79,0.01,24.06

Having a conversation with another gentlemen Dan Lawrence of the fluid bed roasters group on Facebook. He shared these with me and I thought I would add them to the thread for information. Dan has a custom fluid bed roaster with similar heat characteristics to my Coffee Crafters Artisan 3-e.


PID Settings
p=3.0, i=.08, d=65

Notes:

There's always a lot of instability at charge. You can minimize it by trying to match the starting temp on your background curve with room temp so that your machine isn't struggling to match the background. I have several background curves depending on current room temp. FYI I don't preheat the chamber.

After the first roast I run the blower only to get the machine cooled a bit with Artisan ON but not started. Then charge the chamber but don't start the roast till the temp stabilizes. Then hit Start and Charge. I found I needed a fairly large d value.

Here's my latest roast. You can see there's an initial peak as it corrects, then oscillates while correcting. The higher d keeps the oscillations down. The green trace is the voltage going to the ssr that controls the heat.

Following up. This is my last roast of a Brazil. I've been experimenting with charge heat so the first minute was a little wonky (but I'm not one to believe that early 1.5 min timeframe matters at all as it relates to ROR).

My current Artisan-Scope Software PID Settings are attached also. Higher P makes the temp rise on charge even faster. One thing to note on this roast there was no real noticeable flick/crash due to the bean temp being right on the line (and just ever so slightly below). Apparently when bean temp is right on target or just slightly below, the PID is still trying to raise the temp which puts the bean in the posture of being heat positive.n You'll notice more flick and crash when your bean temps are above your curve because the PID sees the crash as thermal relief. Something to consider. I do have a few roasts with even more stable ROR curves than this one, but overall, I'm very happy with this. This Brazil came out more even than anything I've ever roasted. Cant wait to try it in the morning.

Also of note, this is 8oz of beans. I perceive a larger charge would have even smoother curves. Will test 1lb tomorrow.

I'm super happy with the progress on my Artisan-Scope PID Controlled Coffee Crafters Artisan 3-e. Thank you again to Dan Lawrence who was a beast at putting me on the right track with the PID Control settings. I finally learned the trick not to have initial roast temp-spikes (on the 3-e). Charge at a higher temp than your SV Curve and the PID doesn't go nuts trying to get it up to temp. Video attached of the PID Controlled Servo. If you look real close you can see the Servo horn turning at incredibly-difficult-for-human-hands levels of rotation. Also notice there's a little slop in the velcro before the servo horn catches the knob. Once the servo is more permanently affixed I suspect the ROR adherence will get much tighter. My 3d printer arrives tomorrow. One of the first things I'm printing is a proper mount for the servo and perhaps something clever to affix it to the potentiometer. (perhaps a few new things on Thingiverse.com as a reult) One goal I had was to not modify the the core operation of the roaster. This pretty much concludes powerful roast control is possible with zero mods. I'm also SUPER impressed at how much heat this roaster has (huge nod to Ken Lathrop, his team and the Artisan Scope team), I now understand why many of the roast curves you see from Artisan-Scope users show a drop in the first minute or two. They just dont have the heat to deal with the thermal mass. Not the case on the 3-e. So much heat control I had to deal with it heavily in the PID. Side benefit is it takes some of the stress off of the roast, I can just relax and watch the loft. I probably don't touch it more than 2 or 3 times and that's just barely till I hit drop and Artisan returns the heat to zero like a boss...



Video link if the forum embed doesn't work for you.. http://www.youtube.com/watch?v=QG4nWAjHA3Q