Temperature Swings, Minimum Feed Rate and PID Controllers - An explanation

[Ruslan124]

I have seen this discussed many times, so creating a thread explaining the issue might be useful.

PID controllers are quite complex, and given it is unlikely Recteq makes their own, it is possible the support rep you speak to might not fully understand the theory of PID adjustments.

Auger Motors – An auger motor turns at a fixed speed. A 3RPM motor will dump twice as many pellets into the burn pot as a 1.5RPM (although they don’t seem to exist, it makes the math easy).

More pellets in the pot mean higher temperatures. If you set your controller to the max, the auger will run continuously and add lots of fuel to the pot, causing more extensive fire and higher temperatures.

The lowest temp is determined by how few pellets you can put in the pot without causing a flame out.

PID Controller – Cannot adjust the speed of the auger, it can simply turn it on and off. The minimum feed rate determines the lowest amount of time it can run the auger. The highest amount of time is continuous (Max on your dial). The controller can use any setting in between.

It is a sophisticated device that uses samples from the RTD to predict when it should turn on the auger and for how long. It controls the temperature using time, i.e., turn on the auger for x seconds. The time obviously determines how much pellet volume drops into the pot.

It uses an algorithm to predict how much fuel (pellets) is needed to reach the set temperature. Actually not the number of pellets but the time needed to generate the appropriate volume of pellet dump based on previous experience). The actual temperature is compared with the set temp. A significant difference will result in the auger running longer, resulting in fast temperature increases. As the gap reduces, smaller timings minimize the pellet feed into the pot. Once it gets to temp, it tries to maintain the temp using as few pellets as possible. The minimum feed rate determines the smallest fuel that can be fed into the burn pot.


How does that relate to temperature swings?

Imagine the PID controller has calculated that the set temp is close but needs several pellets to get there. It checks the minimum feed rate and dumps the minimum number of pellets.

However, you have your minimum feed rate set high. The number of pellets fed is multiples of what is needed. Once they are in the pot, they have to burn so the temperature surges causing an overshoot. Eventually, the fire burns down, and the temperature drops. The controller now knows that the minimum feed rate will cause a large spike, so it lets the temperature drop quite far below the set point. The temp will bounce wildly above and below the set point for a while before settling down (as the controller learns).I don't know if the controller has writable memory but I suspect not. If it doesn't that means it cannot use multiple cooks to learn.
Because the minimum feed rate causes a large spike, it cannot maintain a steady temp, so it will typically swing between a few degrees above-set temp and many degrees below.

So your goal is to allow it to minimize the number of pellets by setting the minimum feed rate as low as possible without flameout.

This setting will be influenced by how well your smoker holds temp. That is influenced by environmental conditions. It will also be influenced by pellet conditions and BTU. So although you should be able to set it and forget it, if you live in an area with a diverse climate, your pellets are damp or you change pellets for a different quality with different burn specifications, you will probably need to make adjustments. It is also possible the length of the pellets will have an effect. Shorter pellets are more concentrated in the auger than longer pellets.

I have a 3RPM Auger that quickly reaches over 700 degrees with the MFR set at 5. After 5 minutes of reaching the set point, the temperature rarely exceeds a +/- 3-degree range. I also live in FL so cold weather is not something I have to deal with.

Edit - Ensure your heat diffuser, drip tray, and grates are installed when tuning your grill. They all affect the air flow.

 

[Greg Jones]

I have a 3RPM Auger that quickly reaches over 700 degrees with the MFR set at 5. After 5 minutes of reaching the set point, the temperature rarely exceeds a +/- 3-degree range. I also live in FL so cold weather is not something I have to deal with.

What are you using to determine the +/- 3° variation?

 

[Ruslan124]

What are you using to determine the +/- 3° variation?

The temperature graph on my Recteq app and by monitoring the actual temperatures very closely..

 

[Greg Jones]

The temperature graph on my Recteq app and by monitoring the actual temperatures very closely..

The recteq controller and app only report what the projected temp is from the controller algorithm. You need a third-party pit probe to get accurate measurements of what the actual temp inside the grill is. The red line is from a ThermoWorks pit probe, and I’d be pretty confident is saying that the recteq numbers for this cook also did not vary more that +/-3°.

 

[Ruslan124]

The recteq controller and app only report what the projected temp is from the controller algorithm. You need a third-party pit probe to get accurate measurements of what the actual temp inside the grill is. The red line is from a ThermoWorks pit probe, and I’d be pretty confident is saying that the recteq numbers for this cook also did not vary more that +/-3°.
View attachment 19104

To check variation you arguably don't need an accurate temp. If the reported temp is 5 degrees off it is 5 degrees off every time it is reported. The variation would be the same as it would if the reported temp is spot on.

Here is a recent cook.

 

[Greg Jones]

I’m not speaking to accuracy of temp, but variation of temp. You are relying on what the recteq technology is telling you to arrive at a conclusion that your grill only varies +/-3° during a cook. That’s just not what is happening, as is demonstrated in the graph I posted. It doesn’t matter if it’s a pellet grill or your home oven, they are just not capable of maintaining temps with +/-3° accuracy.

 

[Ruslan124]

Of course I am reporting what the technology is telling me. It is what everyone who is trying to set up their smoker will do. They run the grill and monitor the temps using the information provided.

If the controller reports a temperature and that temperature report is consistently measured by the same probe using the same algorithm it is good enough for what we are trying to do. You may not accept the variation is possible but that is what we have.

 

[Greg Jones]

Of course I am reporting what the technology is telling me. It is what everyone who is trying to set up their smoker will do. They run the grill and monitor the temps using the information provided.

If the controller reports a temperature and that temperature report is consistently measured by the same probe using the same algorithm it is good enough for what we are trying to do. You may not accept the variation is possible but that is what we have.

I said you are relying on what the recteq technology is telling you, and when it tells you that temps only vary by +/-3°, that’s just wrong. If you are satisfied that your grill is running with that degree of accuracy, that’s fine. However, there are people on this forum that do use aftermarket probes and I expect that none of them will report that the grill will operate with +/-3° variation. As a test, put one of your probes on the grate under the PID And see if the graph of that probe remotely matches the flatness of what recteq reports the pit temp is.

 

[Threadcutter]

The recteq controller and app only report what the projected temp is from the controller algorithm. You need a third-party pit probe to get accurate measurements of what the actual temp inside the grill is. The red line is from a ThermoWorks pit probe, and I’d be pretty confident is saying that the recteq numbers for this cook also did not vary more that +/-3°.
View attachment 19104

This mirrors my experience as well. It appears to me that Recteq (or their Contract Manufacturer of the PID Controller) have put in a very long time constant "smoothing filter" on the displayed (feedback) temperature.

When compared against my Thermoworks "Signals" probe (that I have installed in the barrel), I find that when the temperature is changing (setpoint vs. feedback), there can be as much as a 20 degree difference between the Recteq displayed temperature and the Thermoworks displayed temperature. This example is most evident when first starting the grill up and going up to 225 degrees. I think this is pretty well illustrated by Greg's graph above.

I do Also find that once set temperature is achieved and has settled out (2 - 5 minutes (?)), the Recteq displayed temperature and the Thermoworks displayed temperature match up quite closely. From what I've observed, it's even within +/- 1.5 degrees. Quite good, but I still rely primarily on temperature information from the Thermoworks unit.

So....moral of the story ? For a "semi bang bang" PID Controller, it doesn't do a half bad job (even if they are fudging the feedback numbers a little with a long time constant smoothing filter). I don't really pay any attention to the Recteq display when it's coming up to temp, but I will look at it after set temp has stabilized.

 

[Ruslan124]

I said you are relying on what the recteq technology is telling you, and when it tells you that temps only vary by +/-3°, that’s just wrong. If you are satisfied that your grill is running with that degree of accuracy, that’s fine. However, there are people on this forum that do use aftermarket probes and I expect that none of them will report that the grill will operate with +/-3° variation. As a test, put one of your probes on the grate under the PID.

I am not doing a good enough job making my point.

This post was intended to help with wild temperature swings. If your grill is swinging between 60 degrees above and 60 degrees below as reported by the recteq technology and you can adjust it so it now reports +/_ 3 degrees, you are demonstrating you have resolved the wild temperature swings. Given those new numbers you have also demonstrated that you are unlikely to do any better. The actual numbers are secondary.

I am not satisfied that I have a grill running with that degree of accuracy but I am satisfied I have the best consistency I can hope for. I know when I set a temperature I can trust it to be consistent within an acceptable range. This is preferable to the random swings you get from a poorly set up grill.

I also think you mistakenly wrote PID (easily done) when you suggested putting a probe under it. The PID is the controller outside of the grill. Did you mean RTD? I have measured the Temps at multiple locations in the grill with a 4 probe system including near the RTD and adjusted the offset so that they report the same temperature. I am happy my RTD is reporting temperatures that are within 2 or 3 degrees of actual as measured by a couple of different systems.

 

[Greg Jones]

Yes, RTD.

 

[Ruslan124]

This mirrors my experience as well. It appears to me that Recteq (or their Contract Manufacturer of the PID Controller) have put in a very long time constant "smoothing filter" on the displayed (feedback) temperature.

When compared against my Thermoworks "Signals" probe (that I have installed in the barrel), I find that when the temperature is changing (setpoint vs. feedback), there can be as much as a 20 degree difference between the Recteq displayed temperature and the Thermoworks displayed temperature. This example is most evident when first starting the grill up and going up to 225 degrees. I think this is pretty well illustrated by Greg's graph above.

I do Also find that once set temperature is achieved and has settled out (2 - 5 minutes (?)), the Recteq displayed temperature and the Thermoworks displayed temperature match up quite closely. From what I've observed, it's even within +/- 1.5 degrees. Quite good, but I still rely primarily on temperature information from the Thermoworks unit.

So....moral of the story ? For a "semi bang bang" PID Controller, it doesn't do a half bad job (even if they are fudging the feedback numbers a little with a long time constant smoothing filter). I don't really pay any attention to the Recteq display when it's coming up to temp, but I will look at it after set temp has stabilized.

Unless the thermoworks is installed very close to the RTD probe you will always get differences. The air circulation produces different zones in the barrel.

I agree any measures until 5 minutes after the set point is reached is meaningless.

However for the topic (reducing temp swings)of the post the accuracy of the probe is secondary. As I have done a poor job of explaining, if I have a probe that reports 200 degrees but reads 20 degrees high when calibrated, and it the reports 230, the variation is 30 degrees (assuming the probe is not damaged).

If I calibrate that probe and measure at the exact same two points I will get 180ndegrees and 210 degrees. The variation is still 30 degrees.

In tuning my grill, my goal is to reduce that variation and make it consistent. I can do that with a consistent but innacurate probe.

 

[Threadcutter]

Yes, my Thermoworks (barrel) probe is right next to the Recteq barrel probe.

Couple of other random thoughts.......

OP/Ruslan has done a very good job of explaining PID. It's a lot of work and greatly appreciated !

I don't have any expectation of the Thermoworks probe displayed temperature and the Recteq probe displayed temperature matching each other. What I do expect is that they accurately track each other. If there is a 5 degree difference between the two when the set temperature is static, that's fine, that's just an offset and can easily be adjusted out. However, when the temperature is changing up or down and the difference between the two is 10, 20, 30 degrees, then there's something else going on. In the Recteq case, I believe that is due to what I referred to as the "smoothing filter".

When I said "semi bang bang PID Controller" above, what I mean is this;
A "Full" PID control system would include a variable speed motor controller for the auger motor. As it is now, it's just "on/off" control, hence the "bang bang" terminology. A variable speed motor controller would allow the PID control loop to infinitely (within the bandwidth capabilities of the PID controller) adjust/control the speed/feed rate of the auger. This would allow for better/tighter/more consistent heat/temperature regulation.

I spent over thirty years in the variable speed motor controller/power electronics industry and am well versed in PID applications. I "could" rip out all the guts and install a "full" PID control system in my 700. Am I/will I ? Not a chance. Why not ? From my experience (YMMV), the stock unit does what I need it to do. The recognized "shortcomings" do not interfere with my ability to achieve a very good end product.

So, why doesn't Recteq utilize variable speed motor controls and go "full PID" ? Simple; $$$.
Plus, what they have works. However, VS motor controllers (particularly for fractional HP motors) have become cheap, cheap, cheap. I think we'll see more and more grills with them in the future because of them getting down to certain price points.

Ruslan; you have done a very good job and I'm not trying to diminish anything that you've said. I would however like to comment that the PID control and the auger regulate "volume" of pellets, not the "number" of pellets. Having said that, I think your choice of using "number" of pellets was probably better for the sake of illustration. Thank You again !

 

[Ruslan124]

Ruslan; you have done a very good job and I'm not trying to diminish anything that you've said. I would however like to comment that the PID control and the auger regulate "volume" of pellets, not the "number" of pellets.

A poor choice of a word on my part. I agree volume would be more accurate. Thanks for the added contribution.

 

[Threadcutter]

A poor choice of a word on my part. I agree volume would be more accurate. Thanks for the added contribution.

Like I said, I think it worked better for the sake of illustration

 

[Round Hill]

Like I said, I think it worked better for the sake of illustration

Rusian124, great thread. Thanks for the effort put into it. I instrumented my 410 with a Thermoworks Chef Master. Not as sophisticated as their Signals device but accurate and trusted just the same. I found a big variance between it and the set temperature (and the "actual" temp shown on the display which is really not the actual temperature). The variance followed a curve. As I increased the set temperature, the variance got bigger. Fortunately, the compensation tool in the app allowed me to dial in a 10% offset which brought the control panel readouts in line with the Thermoworks reading +/- a few degrees. Despite the complexities of the PID/RTD system, I think Recteq should have had a better handle on quality control related to this.

 

[Ruslan124]

Hi Round Hill

Don't take this as an automatic defense of Recteq, I am just spitballing.

The temperature on the display comes from the RTD. Recteq does not manufacture this component but buys it from a third-party supplier. If I were the owner of Recteq, I could insist on better accuracy and quality control from the supplier, but that would almost certainly add cost. I could do the quality checking in-house, which would also add cost. Or I could add an offset setting in the software to the controller (a third-party component) with no additional cost. It is also possible that better accuracy is not possible even when switching suppliers, and all sensors are equally inaccurate.

Given that the RTD works by detecting the change in a current passed through it and influenced by the metal of the RTD expanding and contracting, some factors can influence the expansion and contraction that differ between users (dirty probes, poorly positioned in the grill, pitting on the probe).

The manufacturer will almost certainly have variations in the quality of their manufacturing materials that produce different expansion and contraction qualities between batches. Recteq could be chasing a losing battle.

Additionally, it is possible/probable that the variation is different for not only the different models but for other factors like different locations with different climate conditions.

The RTD technology is not that sophisticated, and as such, I think highly accurate results would be difficult, if not impossible, to achieve.

It seems inevitable that you will never deliver a probe that is accurate for all users, so why bother?

Given all of these factors, I would consider supplying a reasonable RTD quality and ensure that the user can, as you did, make corrections.

 

[Round Hill]

Hi Round Hill

Don't take this as an automatic defense of Recteq, I am just spitballing.

The temperature on the display comes from the RTD. Recteq does not manufacture this component but buys it from a third-party supplier. If I were the owner of Recteq, I could insist on better accuracy and quality control from the supplier, but that would almost certainly add cost. I could do the quality checking in-house, which would also add cost. Or I could add an offset setting in the software to the controller (a third-party component) with no additional cost. It is also possible that better accuracy is not possible even when switching suppliers, and all sensors are equally inaccurate.

Given that the RTD works by detecting the change in a current passed through it and influenced by the metal of the RTD expanding and contracting, some factors can influence the expansion and contraction that differ between users (dirty probes, poorly positioned in the grill, pitting on the probe).

The manufacturer will almost certainly have variations in the quality of their manufacturing materials that produce different expansion and contraction qualities between batches. Recteq could be chasing a losing battle.

Additionally, it is possible/probable that the variation is different for not only the different models but for other factors like different locations with different climate conditions.

The RTD technology is not that sophisticated, and as such, I think highly accurate results would be difficult, if not impossible, to achieve.

It seems inevitable that you will never deliver a probe that is accurate for all users, so why bother?

Given all of these factors, I would consider supplying a reasonable RTD quality and ensure that the user can, as you did, make corrections.

Rusian124, I agree with all of your points. Thanks for the feedback. I recognize that I used PID and RTD somewhat interchangeably and that they are individual elements of the temp control system - my bad.

 

[Motodad1776]

NERDS!!!!!!!!! I'll bet ya that a stick burner has more variance than our grills, also much less work/hassle with these grills, do whatever makes ya happy or stimulates your brain.

 

[Threadcutter]

The thing to remember about RTD “Accuracy” is that a "poor quality" RTD is in the range of 5% error against the measured temperature.

For example, if a setpoint is 200 degrees and the displayed/feedback temp is 210/190 degrees, thats technically within specification.

Well within the desired cooking range, but “mathematically undesirable”.

Me ? I dont care that much. I just watch my Thermoworks “Signals” feedback and make my decisions from there.