Homeroasters - Home Roasting Coffee Community » DATALOGGERS - CONTROLLERS - RATE OF RISE METERS » Dataloggers/Controllers/Rate of Rise Meters
Who is here? 1 guest(s)
Page 1 of 2: 12
 Print Thread
TC4/Artisan showing "-1"
jessep
Top  #21 Print Post
Posted on 01/19/2020 1:39 PM
Ok, I tried a few things and ultimately "fixed" the problem.

The LCD is causing the instability, when I unplugged it everything was fine and I did three roasts in a row with zero issues. I didn't recompile the code without the LCD or anything. In the end I guess I don't mind not using an LCD, but now I'm wondering why it's in an issue.

Spent the day working on Mark 2!

i.ibb.co/tXTGF5n/IMG-20200119-202845.jpg
i.ibb.co/xqgYCk5/IMG-20200118-110605.jpg
 
mg512
Top  #22 Print Post
Posted on 01/20/2020 11:32 AM
Ahhh, that actually makes sense. The LCD would be connected to the same I2C bus as the ADC and other chips, so if the LCD was somehow defective in a way that affected the I2C bus, all those chips wouldn't be able to communicate with the Arduino either. You could probably replace the LCD (and I2C board and cables!) and you should be fine, I would imagine it's a hardware defect somewhere in the LCD.
 
Web
greencardigan
Top  #23 Print Post
Posted on 01/20/2020 3:20 PM
Did you get the LCD from me? Let me know if you need a refund.
 
jessep
Top  #24 Print Post
Posted on 01/21/2020 1:18 PM
Well that could explain it!

Do you think I could have damaged the LCD when I plugged 5v into the ground?

Greencardigan, I did but don't worry about it. It was years ago and worked fine for a long time.

One thing that makes me wonder. I bought a cheap I2C OLED display that I was going to try and modify the program to use, but never got it to work. However, when it was plugged in, I'd get random drops in temperature values for one or two data points. Perhaps that's related?
 
jessep
Top  #25 Print Post
Posted on 03/22/2020 3:54 AM
I figured I'd update this, I replaced the Arduino and all the issues went away!

The roast is still offset from the background it's following, and could use some tuning. But the coffee tastes good.

i.ibb.co/hYLLzLj/fixed.png
 
renatoa
Top  #26 Print Post
Posted on 03/22/2020 7:30 AM
It's normal to be an offset, but under the curve, not above.
"Normal" PID settings led either to undershot following, either to oscillating following, not permanently above.
Check your lookahead value, the graph above suggest you have a lookahead of some 30 seconds, or even more.
 
ROSTARN
Top  #27 Print Post
Posted on 03/22/2020 8:07 AM
Hi
I have the exact same problem! The arduino would freeze up and showing -1 on all channels.
https://forum.hom...wstart=120
My setup is drum roaster with ot1 electric heat, ot2 ac fan and io3 dc motor. I had it up and running stable for two years and recently had to replace the tc4 and tried the tc4+ as I had that extra integrated circuit separate before. Maybe the lcd took some damage at the same time?

I also did try switching out the arduino but it didnt help.

I did have some problems with my chinese 12v psu and after replacing that the issues became less frequent but still occured. I do have cheap chinese stuff for lcd screen and I2C and maybe that is what is causing this. I will try to remove mine, maybe it will help.

Let us know if the issue persist

Thanks for posting everyone!
 
jessep
Top  #28 Print Post
Posted on 03/23/2020 3:17 PM

Quote

renatoa wrote:

It's normal to be an offset, but under the curve, not above.
"Normal" PID settings led either to undershot following, either to oscillating following, not permanently above.
Check your lookahead value, the graph above suggest you have a lookahead of some 30 seconds, or even more.


Sure did, 57s for some reason. Thanks!

So, the lookahead is mainly there to align the BT with the background? Makes sense.
 
renatoa
Top  #29 Print Post
Posted on 03/23/2020 3:37 PM
Yes, but not for the whole profile, just for sections with a slope between certain limits.
Is the PID setup method of those too lazy to hunt for the perfect PID tuning. Is simpler to set I and D to zero, and P to whatever value, say 10, then appropriate value for lookahead, and be done.
In such scenario, if we assume the average power level would be, say 60%, in the Maillard (yellow) section, where profile allure is more important than for drying, and slope more constant, then the offset will be 60% / 10(P) = 6 degrees, value known as error in the PID source code.
For an average RoR having value 10 C degrees per minute, in the yellow section, 6 degrees means 36 seconds. Set this value for lookahead and you have a good enough following, without many headaches, at least for final part of profile.
And also, very important for some guys, no oscillations, so a more pleasant and smooth RoR than a hard to tune true PID.
Not so accurate for drying phase, where following is not so critical anyway.
 
Page 1 of 2: 12
Jump to Forum: