Who is here? 1 guest(s)
 Print Thread
Weird ROR peaks every 40s
I am having a weird artifact (or not?) that I am trying to figure out.
So in my apwollo roaster which is double wall insulated, no outside air besides the one from fan and burner can come in.
I am getting those weird peaks in ROR every 40s even when the burner and fan stay on the same setting.
I am trying to figure out where this could come from, any suggestions are highly appreciated.

I already checked the thermocouples for ground loops but could not find anything wrong.
Even did a long recording (2h) of beans spinning in case they cause ground loops when hitting the bean temp TC, but also nothing wrong in the log just a long almost constant line.
krampe attached the following image:

Edited by krampe on 11/12/2023 7:35 PM
How does your burner regulate it's output? Is it on-off? Hi-low? Or constant volume? Or?
Looks like lag in the control system to me: either the pid is set too sloppy on feedback or the burner cannot respond correctly to partial temp request.

Can you ask for different temps and then hit and maintain those? Say 200, 275, 350,
I have similar peaks in my electric non-drum roaster, but not so even distributed.
I don't worry too much about them, I attribute them to the imperfect agitation, which favors the exposure of the probe to the air more or less.
It was the burner on/off modulation.
Originally I had a single on/off cycle with different duty cycle.


That was not so good for the bean temp and ROR. My analog pressure gauge showed the swings.
With trial and error I found that my gas pressure stays stable if I do the on/off cycles 5 times per second (at 5hz).
The duty cycle overall stays the same but it's divided into smaller portions to reduce the off times and causing a fall of temp/ror.
For my specific setup this pattern (yes i literally translate it into true/false on/off and send it down every 10ms) causes precise pressure increments in 0.5 PSI steps with a regulated output of 6PSI.
Now this won't help anyone as it's specific to my setup (hose/tube length, diameter etc.) but I am very happy with it. At level 3 the system is stable, level 1+2 are fluctuating marginally around the target pressure.

L 0  "____________________________________________________________________________________________________",
L 1  "O_______________________________________O_______________________________________O___________________",
L 2  "O___________________O___________________O___________________O___________________O___________________",
L 3  "O___________________O_________O_________O___________________O_________O_________O___________________",
L 4  "O_________O_________O_________O_________O_________O_________O_________O_________O_________O_________",
L 5  "OO________O_________O_________O_________OO________O_________O_________O_________OO________O_________",
L 6  "OO________O_________OO________O_________OO________O_________OO________O_________OO________O_________",
L 7  "OO________OO________OO________OO________OO________OO________OO________OO________OO________OO________",
L 8  "OOO_______OOO_______OOO_______OOO_______OOO_______OOO_______OOO_______OOO_______OOO_______OOO_______",
L 9  "OOOO______OOOO______OOOO______OOOO______OOOO______OOOO______OOOO______OOOO______OOOO______OOOO______",
L 10 "OOOOO_____OOOO______OOOOO_____OOOO______OOOOO_____OOOO______OOOOO_____OOOO______OOOOO_____OOOO______",

Interestingly you can see the big jump in the on cycles between 10 and 11. All the settings in between resulted in max pressure.
Also note that the solenoid I use (Clippard ET) has a 5-10ms opening time.
So in the best case it's open for 10+6 ms in L1-L4, in the worst it's open 0+10.5 ms (assuming linear opening time, which is probably not the case).

This kept the temperatures way more stable.
I also replaced the thermocouple with an RTD (PT100) at the same diameter which brought another improvement (after finishing testing the burner with the thermocouple)
Now I can finally play around with PID, as you can see my I setting needs increasing and the P setting decreasing.
krampe attached the following image:
Jump to Forum: