Inkbird PID AT not working?

[blkspade99]

Hey all,

Finished my oven build. Works great, but I'm having a 20-30 degree fluctuation. I tried to set it to AT and it runs for like a few min but its still having this 30 degree swing. Am I missing something? The manual is just so vague on what to do. Is there a way to set the rate of which it heats? Maybe that's why its overshooting so much.

Heres a video

This is the thermocouple i bought.. Unless its this that's the problem. I also bought one of the cheap chinese ones everyone uses.. I could try that one as well.
https://www.amazon.com/gp/product/B0142S9J4S/ref=ppx_yo_dt_b_asin_title_o09_s01?ie=UTF8&psc=1

 

[J. Keeton]

I'm am extremely far from a pro. On my tempting oven I went though weeks of tuning tribulations until the good patrons on this forum helped me figure it out. What worked for me was to set the integral and derivative to 0 then adjust the proportional band by itself (starting low around P200... then working up to around P600 to reduce undulations).


This was my auto tune:



Started with just the proportional and started working up:




Final results VS the auto-tune:





I stopped there, but I'm guessing you could start adjusting the I and D to get it a little more dialed in after getting your P right? Just food for thought.

 

[tkroenlein]

Let's see the inside of your oven. I have that same Inkbird and auto tune will keep me within 10F at 1700F.

 

[blkspade99]

Let's see the inside of your oven. I have that same Inkbird and auto tune will keep me within 10F at 1700F.

-Its 2800 watts
-Tried changing the thermocouple and running the AT again. Still the same. Has a 30c degree swing

 

[Gab-e]

Hi Blkspade99. I have used one of these for a tempering oven. Being an instrumentation and control technologist, you would think it would have took me minutes to tune this up but it turned into a very unplesant experience. First, the P, I and D parameter units listed in the manual are all incoherent with the implemented algorithm. Secondly, their auto tune logic is a pure POS, you'll only waste time and energy trying to make it work this way. As J.Keeton mentionned above, try to disable I (9999) and D (0) first and adjust P until the process value oscillates just under your setpoint. (A good trick while tuning is to display the controlled output instead of the setpoint, this way you'll actually see how the controller is reacting vs the temperature in your oven.) Since the units are all jacked up, I suggest starting with a P value of 5000, let run for a couple of cycles(oscillations) then divide/multiply P value by 2. You'll see the process reacton improve in one case and worsen in the other. Keep doing this until you have reached a satisfatory result. Then you'll be ready to lower P just a bit and start adding some Integral. Begin with a value (in seconds) close to the dead time of your oven ( time between a change of the controller output to the beginnign of a temp. change in the oven) then, multiply or divide value by 2 until you get a satisfactory result. Once you get there, you can start fine tuning by adjusting parameters individually, remember to let the process run for a couple cycles between changes and monitor closely. I hope this helps, this controller is definitely able to reach a temperature and not overshoot, it just takes more work and figuring out than an 'industry standard' type controller.

Gab

 

[blkspade99]

Hi Blkspade99. I have used one of these for a tempering oven. Being an instrumentation and control technologist, you would think it would have took me minutes to tune this up but it turned into a very unplesant experience. First, the P, I and D parameter units listed in the manual are all incoherent with the implemented algorithm. Secondly, their auto tune logic is a pure POS, you'll only waste time and energy trying to make it work this way. As J.Keeton mentionned above, try to disable I (9999) and D (0) first and adjust P until the process value oscillates just under your setpoint. (A good trick while tuning is to display the controlled output instead of the setpoint, this way you'll actually see how the controller is reacting vs the temperature in your oven.) Since the units are all jacked up, I suggest starting with a P value of 5000, let run for a couple of cycles(oscillations) then divide/multiply P value by 2. You'll see the process reacton improve in one case and worsen in the other. Keep doing this until you have reached a satisfatory result. Then you'll be ready to lower P just a bit and start adding some Integral. Begin with a value (in seconds) close to the dead time of your oven ( time between a change of the controller output to the beginnign of a temp. change in the oven) then, multiply or divide value by 2 until you get a satisfactory result. Once you get there, you can start fine tuning by adjusting parameters individually, remember to let the process run for a couple cycles between changes and monitor closely. I hope this helps, this controller is definitely able to reach a temperature and not overshoot, it just takes more work and figuring out than an 'industry standard' type controller.

Gab

Thank you. I will give this a shot. I just hope its the tuning of the PID.

- Jon

 

[blkspade99]

I'm am extremely far from a pro. On my tempting oven I went though weeks of tuning tribulations until the good patrons on this forum helped me figure it out. What worked for me was to set the integral and derivative to 0 then adjust the proportional band by itself (starting low around P200... then working up to around P600 to reduce undulations).


This was my auto tune:



Started with just the proportional and started working up:




Final results VS the auto-tune:





I stopped there, but I'm guessing you could start adjusting the I and D to get it a little more dialed in after getting your P right? Just food for thought.

Gave this a shot without much luck. What PID Controller are you using? Mine is having 30c swings AFTER the AT. Even playing with it. I can't get it closer then that.

 

[Cushing H.]

Hey blkspade99. I am one of the people that helped out J Keeton. Also, i very much agree with Gab-e that both the units for the PID parameters are confusing/mixed up, AND that the autotune just really does not work. You need to do tjis manually from the start. Disable both the integral AND the derivative functions (how will depend on your unit - get this from the manual). Now go about setting the D parameter. The P parameter is actually a band width around the setpoint: when the temp is outside of that band, the oven heaters will be going full blast, and when inside of the band, the heaters will decrease linearly until they are at zero (off) at the setpoint. Think about it: if that band is too small, the heaters will blast, but then not shut off fast enough to avoid overshoot - and you will get the oscillations you see. So start with a BIG band on the P setting (again, read the manual to figure out whether for this unit whether that corresponds to a big or small parameter setting). You should see a SLOW rise towards the setpoint, settling out at some level below the setpoint, but with small oscillations. From there decrease the band size (again, see the manual for what the parameter really means) in increments. The rise to temperature should be faster, and the final temp should be closer to the setpoint. Continue like that until you start getting overshoot. At that point back off to the last setting that did NOT give you overshoot. STOP THERE. We can talk later about whether to bring in integral control, but you might not need it.

stay totally away from derivative control: that is WAY too sensitive (to noise) for the equipment we are using, and will just lead to massive confusion.

folks in general: this is the second time in a few months this has come up. Is it worth my putting together a tutorial on tuning a PID, and see if it is understandable enough to put into the stickies?

 

[blkspade99]

Hey blkspade99. I am one of the people that helped out J Keeton. Also, i very much agree with Gab-e that both the units for the PID parameters are confusing/mixed up, AND that the autotune just really does not work. You need to do tjis manually from the start. Disable both the integral AND the derivative functions (how will depend on your unit - get this from the manual). Now go about setting the D parameter. The P parameter is actually a band width around the setpoint: when the temp is outside of that band, the oven heaters will be going full blast, and when inside of the band, the heaters will decrease linearly until they are at zero (off) at the setpoint. Think about it: if that band is too small, the heaters will blast, but then not shut off fast enough to avoid overshoot - and you will get the oscillations you see. So start with a BIG band on the P setting (again, read the manual to figure out whether for this unit whether that corresponds to a big or small parameter setting). You should see a SLOW rise towards the setpoint, settling out at some level below the setpoint, but with small oscillations. From there decrease the band size (again, see the manual for what the parameter really means) in increments. The rise to temperature should be faster, and the final temp should be closer to the setpoint. Continue like that until you start getting overshoot. At that point back off to the last setting that did NOT give you overshoot. STOP THERE. We can talk later about whether to bring in integral control, but you might not need it.

stay totally away from derivative control: that is WAY too sensitive (to noise) for the equipment we are using, and will just lead to massive confusion.

folks in general: this is the second time in a few months this has come up. Is it worth my putting together a tutorial on tuning a PID, and see if it is understandable enough to put into the stickies?

Thanks!
Ok after work today I'll give it another go. Setting I, and D to zero should be the same as disabling them correct? I got the P up to like 600 and noticed it was the same. But we'll see.
I also ordered a MYPIN controller that arrives today. Just to try as a comparison. Seems MOST people use this one, run the AT and go from there as a basepoint. The inkbird just doesn't have alot of info on it.

I will work on it today and let you know
Thanks

- Jon

 

[Cushing H.]

Ok after work today I'll give it another go. Setting I, and D to zero should be the same as disabling them correct? I got the P up to like 600 and noticed it was the same. But we'll see.

The settings really depend on the controller. Zero may or may not disable the P and D fnction. You need to check the manual. If not clear, send me the manual (or a link) and i can help decipher it...

 

[Cushing H.]

Seems MOST people use this one, run the AT and go from there as a basepoint.

I guess I do not trust the autotune functions on these things in general. Whether they work depends on the behavior of the system they are controlling ... and not all systems are the same. Especially, I would like to emphasize, do not use (or let the AT use) the differential function - even a little bit of noise in the system can cause that function to go haywire, and really confuse things for you. Most of the time (for the kind of stuff we are doing) just the proportional function, if appropriately tuned, will work just fine - you just need to understand what each function is doing, and take it one step at a time to see where you can get ... and how good a control you can get (again - this is system dependent).

I once had a system that was TOO well insulated, and the PID controller just could not be made to settle the thing at its setpoint. I actually had to degrade the insulation in the system a little, and then the PID worked just fine (I am changing what I actually did to make the equivalency to thermal insulation - but the idea is the same....)

 

[blkspade99]

I guess I do not trust the autotune functions on these things in general. Whether they work depends on the behavior of the system they are controlling ... and not all systems are the same. Especially, I would like to emphasize, do not use (or let the AT use) the differential function - even a little bit of noise in the system can cause that function to go haywire, and really confuse things for you. Most of the time (for the kind of stuff we are doing) just the proportional function, if appropriately tuned, will work just fine - you just need to understand what each function is doing, and take it one step at a time to see where you can get ... and how good a control you can get (again - this is system dependent).

I once had a system that was TOO well insulated, and the PID controller just could not be made to settle the thing at its setpoint. I actually had to degrade the insulation in the system a little, and then the PID worked just fine (I am changing what I actually did to make the equivalency to thermal insulation - but the idea is the same....)

If you ever make a tutorial this might help.

 

[Natlek]

Hey all,

Finished my oven build. Works great, but I'm having a 20-30 degree fluctuation. I tried to set it to AT and it runs for like a few min but its still having this 30 degree swing. Am I missing something? The manual is just so vague on what to do. Is there a way to set the rate of which it heats? Maybe that's why its overshooting so much.

Heres a video

This is the thermocouple i bought.. Unless its this that's the problem. I also bought one of the cheap chinese ones everyone uses.. I could try that one as well.
https://www.amazon.com/gp/product/B0142S9J4S/ref=ppx_yo_dt_b_asin_title_o09_s01?ie=UTF8&psc=1

That thermocouple has temperature range from 0 to 800 celsius ? Could that be problem ? It works on max.limit maybe accuracy is not good in that case ? Why you don t buy sensor with higher range ?

 

[blkspade99]

That thermocouple has temperature range from 0 to 800 celsius ? Could that be problem ? It works on max.limit maybe accuracy is not good in that case ? Why you don t buy sensor with higher range ?

I swapped it out for one of those chinese ones everyone uses off aliexpress. That one says 1200c. Who knows though, the $15 dollar one on amazon says 800c, and the $1.50 one on aliexpress says 1200? I wonder if they're the same thing haha.

 

[J. Keeton]

folks in general: this is the second time in a few months this has come up. Is it worth my putting together a tutorial on tuning a PID, and see if it is understandable enough to put into the stickies?

I would think a detailed tutorial on tuning a PID controller would be pretty valuable. With terminology, component selection, tuning method, and considerations for HT Ovens, Tempering Ovens, and PID controlled Forges.

 

[Cushing H.]

ok ... will give it a try. I might be able to also include some plots from simulations with various settings- will need to re-teach myself Python though :-(

 

[Gab-e]

While not being a step-by step tutorial, the Wikipedia page about PID controllers (https://en.m.wikipedia.org/wiki/PID_controller) actually has a good amount or pertinent info along with animations showing the typical behaviour of each parameter. It'll help you better understand how it works and consequently, what to look for when tuning your oven.

Gab

 

[blkspade99]

ok ... will give it a try. I might be able to also include some plots from simulations with various settings- will need to re-teach myself Python though :-(

So last night I rewired my oven with the new MYPIN pid controller. It was late so I didn't have a ton of time, but I ran the AT really quick and the oven is only having swings of 8-9 vs the 30 that the other controller was doing. Its also a lot nicer. The buttons are easier to press. I feel like the longer the oven was running the more accurate it was getting. I only had it running for like 10-15 min.

At least its close now. Tonight I'll try to tune it myself, and return the other controller.

Edit: Did it at lunch. Just dropped alittle D off and its good to 1c degree now! Can't ask for better then that
P-3.58 I-15.50 D- 2.98
- jon

 

[Cushing H.]

Good to hear! One thing to consider: if the 1 degree is oscillation, then just live with it. If it is an offset ... then you can change your setpoint so that actual temperature reached is the one you want.

On the other hand - with something like a one degree difference - you will need ask yourself just how accurate your temperature measurement system is. that one degree could well be in the noise of your ability to measure temperature precisely.

 

[blkspade99]

Good to hear! One thing to consider: if the 1 degree is oscillation, then just live with it. If it is an offset ... then you can change your setpoint so that actual temperature reached is the one you want.

On the other hand - with something like a one degree difference - you will need ask yourself just how accurate your temperature measurement system is. that one degree could well be in the noise of your ability to measure temperature precisely.

Oh I'm happy with it! I was going to be happy with it being 5 off lol. Thanks again. You really should do a tuning sticky. Would help alot!

- Jon