Heatermeter PID Loop Tuning Help

[MJPeisher]

Hello all,
I have recently had some serious difficulties maintaining temperatures and stuff while using my Primo Oval XL with a heatermeter connected. A buddy of mine also has the same grill with a heater meter, and we tried his out, and then mine out using the exact same parameters on his grill. His worked just fine, and mine let the fire go out multiple times. My graphs are attached, and he will post his in a reply. Can anyone give me some guidance on why this is happening?

I am going to adjust my parameters on the heatermeter, and do a test run on my grill, and see how it goes (with presumably updated parameters).
Any help would be appreciated.
Thanks.

 

[Chris Peisher]

Hello all,
I have recently had some serious difficulties maintaining temperatures and stuff while using my Primo Oval XL with a heatermeter connected. A buddy of mine also has the same grill with a heater meter, and we tried his out, and then mine out using the exact same parameters on his grill. His worked just fine, and mine let the fire go out multiple times. My graphs are attached, and he will post his in a reply. Can anyone give me some guidance on why this is happening?

I am going to adjust my parameters on the heatermeter, and do a test run on my grill, and see how it goes (with presumably updated parameters).
Any help would be appreciated.
Thanks.

I am his “buddy”. Here is my graph from a recent cook with identical settings except for the set point. My HM and grill have always had rock solid temp control since my initial PID calibrations years ago. I can also take the same thermocouple probe, HM and blower over to a 3’x5’ concrete fire pit big enough for a whole pig and get identically accurate temp control. MJP has had a lot of trouble lately so I asked him to post here to get some help with his tuning.

 

[MartinB]

Why do you have your fan not coming on till 75% output. Are you running this with a damper?

 

[MJPeisher]

Yes there is a damper. The thought being that if opening and closing the damper will maintain, no sense bringing the fan on. So the fan only comes on when the damper can't cover it.

 

[Gary V]

Here is a configuration I have used with my PrimoJr. One thing i want to make clear is that I use the d-plates when cooking. I have found these help smooth out the control when cooking. Also, I use 2 different thermocouples in different point when using grill. When cooking ribs, I drop thermocouple down through the damper and position it by the thermometer. I also use the newer thermometer. When cooking something large like a turkey or beer can chicken, I use a probe mounted to the grill grate about 2-3" in from edge of grill body. I also use a Adapt-t-Damper.

 

[MartinB]

Yes there is a damper. The thought being that if opening and closing the damper will maintain, no sense bringing the fan on. So the fan only comes on when the damper can't cover it.

Obviously opening and closing the damper isn't doing it so why don't you change that......

Set your fan to come on at about 20-30%, see if it improves right now your controller is cycling the output mostly through the range that isn't doing anything. If it overshoots then try limiting the maximum output of your blower ( except for startup) to something lower as well

 

[IvanTrail]

I agree with the fan statements. The fan has proportional control, so it's best to use that to it's full capability.

As for pid tuning, here are my general thoughts on such:

It's easiest to think of pid in terms of geometric calculus.

P is proportional, which is equivalent to the zeroeth derivative. That means it calculates error based on the position relative to the set point. In general, you'll set P to get full on or full off just outside the allowable range of error you can handle. So, if you're ok with 5° variance, you'll want full open or closed at ±7° or so from the set point. This is with I and D off. A well tuned P value will nearly settle at the set point and not oscillate too far away. If it is too high, it will overshoot wildly. If it is too low, it will never get to the set point.


The next one to consider is D, which, as the name suggests is the first derivative. When looking at the graph, it controls the slope of the line. As the line slope increases, the output tries to offset that. A higher D value tries to more aggressively flatten the slope. As P drops off, D helps control the approach to the set point. So, when Pnis set well, a good D value will show up as it gets closer to the set point, limiting how fast the actual is allowed to approach the set point. This helps limit overshoot on P. If D is too low, the approach will be too steep and P will cause an overshoot. If D is too high, it will over compensate and not let the temperature get close to the set point.

I is integral, which is the -1 derivative. It looks at "area under the curve". If the error stays away from the setpoint, I compensation increases with time. This helps keep minimal errors not covered by P and D from just staying there. If your actual comes within a couple degrees of set but never really crosses it, increase I a bit. If I is too large, you'll get slow fugoid oscillations.

It's all a balancing act. On my phone, if I tap the left side of the fan status bar, it brings up the PID values, and you can easily tell who is the dominant compensator at the time. When the error is large, P should dominate. When the slope is large, D should dominate. When a small error has been there for a long time I should dominate.

So, keep your eye on the PID values. If you're thinking "This should have more compensation because the error is huge", set P higher. If you're thinking "Man, that is just really jumping all over the place and doesn't settle down", increase D. If you're thinking "It just never quite gets to the correct temperature", increase I.


Hope that helps you a bit.