My long way to the bliss of automated BBQ

[Juho A]

I finally have heatermeter in use, and did my first cook yesterday, after more than 2.5 months of trying to get a heatermeter! While it was not without challenges either, I am sure I will be able to work those out with some adjustments. The goal is very close now.


So, after spending something like 10 hours on finding and ordering stuff from various places (digikey, mouser, ebay, OSH Park), as I couldn't order directly from Heatermeter Store at the time I got started with this project as the store was closed due to COVID-19, I eventually had all the parts and boards in hand. I learned the hard way (spending another 10 hours trying to solder and test) that SMD soldering is not something I can do, nor is it something I want to do in the future either - it's a skill I have had very little need for until now, and I expect to be able to outsource any needs I might have in the future. I was right in my initial thoughts, that a kit with the SMD part done would have been the right option for me. Anyway, as I had tried and failed to build the SMD part of the heatermeter board, and having spent 2.5months on all this in total, I didn't want to wait until the whole BBQ season is over. So, Friday last week I ordered a fully assembled heatermeter 4.3 with thermocouple from the Heatermeter Store with expedited shipping. It was delivered on Thursday this week, so you can really call it expedited - 6 days from Florida to Finland.

So, I put it together, followed Bryan's very clear instructions and installed the Heatermeter on the Pi Zero W, started it up and plugged in the damper I had built. It worked right away! With the clear instructions I was able to quickly adjust the servo so that it opens and closes fully, as intended. And then yesterday on my day off I decided to make some meaty ribs. Got them out of the ribs and went to start up the grill with the heatermeter. Set the desired temp to 110C (the UI on the heatermeter unit is very clear and simple & intuitive to use!) and let it burn with the top vent fully open. That might have been my initial problem, but the temperature shot over, to around 135C and as by then the damper had been fully closed for 15min, the fire had died. Rinse and repeat a couple of times, I fired it up again, this time adjusted the top vent to open just a crack, set it to automatic, it shot over, died...


I found this post https://tvwbb.com/threads/hm-keeps-...fire-dies-and-needs-to-be-relit-at-225.67779/ and adjusted the PID values to those same values, as the Kamander I have is very similar to Akorn. After the next re-fire, when it shot over to 130C again, I then just adjusted the temp up to 135C and this way it didn't kill the fire but rather started maintaining it quite nicely. Honestly this was my last resort, as I had promised my kids to take them for a ride in the convertible to get some ice cream, and I was 2.5h in instead of the 1h I told them, so I was on borrowed time already. Luckily it worked, and the ribs were cooking all the while we were out, although in a bit higher temperature than I would have wanted.

These are my PID settings now:


Can anyone point out what would be the best way for me to adjust, as based on my earlier experience with my grill without heatermeter, once the temperature drops by ~5C, you should be blowing all the air in that you can, so that you can get the fire going again, and then once you see the temp increase by more than a degree, you should stop blowing and quite quickly adjust the damper valve in order not to have it overshoot. I would also like heatermeter to take the overshoot into account when heating up - it should stop blowing at least 20C before the desired temperature, and start to close the damper so that when the target temperature is reached, it should only be something like 15-20% open to maintain that temperature.

Oh yeah, someone might be wondering why there's no meat probe. That's because I wanted to test the thermistor probes that I had, if those could be used. I didn't get a sensible response from any of those when testing with different preset coefficients, so I bit the bullet and ordered some thermoworks probes, looking forward to getting them next week. I am not fully confident on my chinese thermocouple either, but when I get the thermoworks probes, I will be able to check whether they show similar temps for the pit. I "calibrated" the pit TC to 100C with boiling water, but I am uncertain if it's really trustworthy and linear above 100C.

 

[Gary V]

try these settings. They work for me with my Adapt-A-Damper. The only update to those setting are the lid open settings. I use the default settings which i believe are 15% and 240 seconds. It better to recover slowly to prevent overshoot.

 

[Bryan Mayland]

You got it hooked up, fantastic!

It is worth noting that the PID parameters are temperature units specific, so the defaults are roughly twice as high as they should be for Celsius users.

 

[Juho A]

Thanks Bryan and Gary. I finally had a moment to play around some more with this, now that I received a new thermocouple as well as the thermoworks thermistor probes. I decided to take the default values and divide those by two, and after firing up the Kamado I quite quickly learned that my primary problem wasn't the PID values...

I started off with my old thermocouple and one thermoworks probe. The old thermocouple was just painfully slow to react, and also seemed to eventually be showing less than the thermoworks probe. Then I swapped to the new thermocouple probe, which came with an insanely long cable, as you can see in the picture - there's probably 3-4 meters of cable on that roll. I was wondering if this affects the probe somehow?

The new thermocouple probe seemed to have a lot of "ripple", a constant fluctuation of the value, jumping back and forth some 0.3C all of the time, and this naturally was causing the same effect to the output, too. The new thermocouple was faster than the old, but still much slower than the thermoworks thermistor.

In the end, using the thermoworks thermistor everything seemed to work fine, I would get the grill to settle to a temperature and maintain it nicely. That got me wondering why I insisted upon having a thermocouple pit sensor in the first place... That causing the SMD-soldering-PITA actually made me spend hundreds of euros extra on this project, and I can't even remember why I didn't just build a thermistor-based system to start with.

Anyway, here we are, so I guess I should just try and eventually get it to work with a thermocouple, but in the meanwhile I can already use the system with a thermistor. I am just wondering if all thermocouples are slower, or are the thermocouples I have bought just substandard? The first one was an anonymous chinese unit, that is already in the trashcan now, but this second one I bought had at least gotten some positive reviews on Amazon, it's a brand called "Keenso": https://www.amazon.de/dp/B07R9K4GJ5/

So, wonder if the long cable on a roll could cause the ripple and/or slowness on a thermocouple, or is it just such a low quality sensor? And if it's the latter, which thermocouple (available in Europe) would be a good one for a reasonable price?

 

[JKalchik]

Cable length on a copper/constantan type thermocouple (Type K?) shouldn't much of an issue. Back home, when my brother was working on a grain dryer efficiency project for his M.S. (and the DoE, Michigan State, etc.,) the grain bins were instrumented back to a remote recorder. Some of the runs were over 200'. What is important is as few joints as possible, and no crossing of materials. Each leg of the thermocouple needs to back to the electronics in it's native material. Splicing in some twisted pair, bell wire, etc., creates new thermocouples at every joint where dissimilar metals come together.

How fast thermocouples will react depends on encapsulation. A bare thermocouple reacts very quickly, but is also very sensitive to air currents. A potted thermocouple in a stainless capsule will react far slower and is fairly resistant to minor variations due to air current.

 

[Juho A]

Thanks JKalchik. Yes, it's a K-type thermocouple. I wonder if it affects anything that the thermocouple came with abiko (crimped) terminals, and I cut those off to install the TC plug that goes into heatermeter's port. The plug is the kind where you put the leads under screw-on terminals.

Do you have any thoughts what might cause the fluctuations / ripple, as to me it seems heavily encapsulated, which would go against sensitivity and thus ripple too?

 

[Bryan Mayland]

You should check the noise graph to see how the signal looks and maybe it will help find a solution. Press N on the keyboard when logged in and on the home screen. You can see if it looks like a solid signal or if it is all over the place and the average is just moving around. Like my thermocouple here. The temp is bouncing around by around 0.8F but when you click the noise indicator you see it is a real mess. My kitchen outlets are all like this and if I want a clean signal I have to run an extension cord to an outlet on a different circuit. This one you may note the period of oscillation is ~60Hz (the numbers on the bottom are milliseconds) which would be cleaned up by the 60Hz noise filter option, but still isn't ideal.



It should look like this, here in the other outlet:


You can also try powering the HeaterMeter from a 12V battery to see if it magically fixes the problem, which would indicate a power supply problem or a ground loop or something.

 

[Gary V]

I run on battery power so that i do not introduce any 60hz waveforms. I have even enclosed heatermeter in a grounded metal box with a spot for the thermocouple to enter box and connect. This test still showed the same amount of noise. I have even done common mode grounding practices with the thermocouple to see if that reduces the noise. It`s still there. I have reduced noise best by cleaning my boards. I do see some benefit with filter, but not enough to use full time. Does the noise cause issue with control. No, if you have a healthy thermocouple connected to the heatermeter. I have explored this noise issue ever since I got my heatermeter up and running and the source could be servo noise, internal clock harmonics, and lastly noisy power sources. The most important point I am making is, make sure your thermocouple is performing properly. In times when you see a mysterious spike, as I have found, has been traced to either a faulty intermittent thermocouple, or faulty servos. Last item to cause weird charts is sticky dampers. This looks like a bad thermocouple, but it is because the damper is screwed too tight to the servo. I always detect noise, but with a healthy thermocouple, control issues do not happen on my heatermeter. We need to make a bumper sticker that says "NOISE HAPPENS". One other item to bring up while on thermocouples. Exposed weld couples are extremely fast reacting too temp change. From my experience, it so fast that it can cause instability and make it hard to tune the PID loop. The thermocouple circuit has RC time constant components which create a usable band for the circuit. Any signal below the cut-off point is used and anything above is rejected. All my exposed tips have a piece of Kapton tape on the exposed tip to slow down the reaction speed of the thermocouple. The current thermocouples I am making have a response time that matches well with the RC time constant of the amp. I still have noise, but my thermocouples work better than anyone I have ever purchased. I also still see noise on the graph.

 

[RalphTrimble]

I like the new noise feature Bryan! I see you have polished the UI quite a bite recently, I'm seeing all sorts of user friendly stuff popping up... NICE WORK!