Using my BrewPi for the first time and wondering what the typical expected overshoot should be, or if I have a calibration off. My variance isn’t that much, but I do want to get it dialed in as best as possible. Also, I’m wondering if there is a ‘learning’ period where the estimator learns my setup and the cooling accuracy improves over time. See the attached screenshot. After the initial start at 65F, it bypassed the setup of 65F and went all the way down to 63F. Then over night it climbed up to the new set point of 68F, and now its been bouncing around a little between 67.2F and 68.3F. Other than the drop to 63F (not sure why that happened), is the bouncing around a set point (e.g. 68F) normal?
Why is your beer temperature dropping so fast? There are only a few very short spikes where the cooler is on.
Is your beer sensor even in a volume of beer?
Also please also show your fridge setpoint and temp when asking for advice.
This is in a conical setup with a coil. The coil/wort is cooled by a glycol mix pumped from a separate chiller/hold tank. That take is currently set to 15F, so I can cold crash down to 35F. I might be able to come up a little more, but that is likely the reason for the quick chill spikes that you are seeing. I’d rather the BrewPi offset that for me though so I never have to change the chiller temp (which will be necessary eventually when multiple fermenters are supported), but I guess it has no idea how cold the chiller/glycol is.
I currently have the beer temp probe in a thermowell inside the conical. The Fridge temp probe is just fixed to the outside of the conical, but basically room temp. I got this idea from @sunadmn who has a similar setup with success.
Screenshot including fridge temp/settings is below.
Hey @Elco Denny has a setup almost identical to mine. He is using a Glycol setup with the SSBrewtech 1/2 barrel and FTSs coils. He and I talked about how I have things setup but since I am still running mine with the older Arduino I am not sure we are talking apples to apples. I still use two sensors one in thermowell and one on side of conical. As you and I had discussed I cranked my chiller way up so my overshoot wasn’t as bad. My understanding is with the newer code and the PWM functionality this type of setup would be much better since only one temp sensor would be needed. If I am off and understood things wrong I apologize.
I hope that helps elaborate on what’s going on and you might have some advise for @Denny_Deaton (I too will be very interested).
Cheers,
-Stephen
Thanks @sunadmn for adding the additional context! I’m intrigued by the single sensor! That’s what I’d ideally like to get to.
It is better to use fridge constant mode if you have just one sensor.
The cascaded control assumes that the beer temperature will slowly respond to the fridge temperature. If you have such an uncommon setup, why not mention this if you ask me why there is overshoot? You ask me what is happening but let me guess all the physics involved.
Switch to fridge mode, reduce the minimum on time and set the period to 5 or 10 minutes. Put the fridge sensor in the beer.
Profile mode is not available now in fridge mode, but this will change. Single sensor beer mode will come.
If you want brewpi to offset the glycol temp, you will have to put the fridge sensor inside the glycol and let BrewPi control the compressor. But that will leave what controls the pump. It will have to be turned on periodically. Good idea to support in the future.
I also changed the thread title, because your situation is far from typical so asking for typical expected overshoot is misleading. With proper configuration in fridge constant mode you should be able to get a flat line.
Thanks @Elco. Sorry for not including those details. I have in previous posts, but I shouldn’t expect you to remember everyone’s setup. 
To confirm, if I switch to fridge mode, is it the ‘Cooler minimum ON time’ constant that I want to change to 5 or 10 minutes? It’s currently set to 120 (which I assume is seconds). So I change that value to 300 or 600?
Ideally, for the future with my setup, I would like to be able to do the following with the BrewPi (for two ferms).
- Independently control power to my glycol cooler compressor (window-mount A/C unit), based on a single temp probe in the glycol.
- Control power to a pump which circulates coolant from the glycol cooler through a coil in my fermenter, based on a single temp probe in the fermenter thermowell (for multiple fermenters).
No reduce the MINIMUM ON TIME , and set the PERIOD to 5 or 10 minutes.
How you want to control it will be possible in the future.
I am running the latest version (0.4) and do not see a Minimum On Time in the Constants. I do however see Cooler PWM period (seconds) and Cooler Minimum Time On, so I assume that is what you are talking about. I set the Cooler PWM period (seconds) to 300 (5 minutes) and reduced the Cooler Minimum Time On from 120 to 30. I changed to Fridge Constant mode and its been going like this for about an hour. It was below my set temp of 68F initially, but eventually has risen to 68.6F. I expected the pump to initiate by now. Thoughts?
A quick update, I realized that the Idle time kept getting reset, so I watched it closely and noticed that it will come on for a few seconds and then turn off. It doesn’t run long enough to do much, for example it only dropped from 68.7 to 68.7. Then it started running more frequently and for a little longer each time (up to 30 seconds one time). The temp is now down to 68.3F.
That’s the integrator at work. If it did not turn on long enough based on the temperature difference, you need to increase Kp. Take a look at the control algorithm tab to see how the PID value of the cooler is formed by P, I, and D.
Your pump should never be ON shorter than the minimum cool on time.
Just wanted to post a quick update that I think the integrator has figured things out now. It has really smoothed out over the last 12 hours! Now if we could just get an Fridge Profile feature and multiple ferm support, I’d be set! 
’
Thanks for the help, @Elco!
