Over the past couple days I've run some tests. I'm using a small chest freezer and an arduino based controller. Note, my controller does not use a differential. If it measures a temp over the target temp and it is in cooling mode, it turns on. Once the temp falls below the target, it goes off. No fuzz. The opposite is true when it is set to heating.
Day 1:
- Filled ale pail with 5gal tap water (59f) added ice to drop to 53.5f since the ambient temp in my basement is ~61f I wanted a little separation.
-Left controlling probe in freezer air, placed water/wort probe in a thermowell close to the center of the pail, set controller temp to 54f.
-The water gradually came to a rest at about 53.1f. With the controlling probe in the air, the equilibrium temp will be below the max point on the curve of the freezer air.
- Once the water and air equalized, the freezer was allowed to cycle for about 12hrs. The freezer cycled every 45min - 1hr and stayed on for 30 seconds each time. The freezer air temp swung about 2-3 degrees each compressor cycle while the water temp fluctuated no more than .1f.
Day 2:
- Swapped probes so that the probe in the water was now the controlling probe. The controller code stayed exactly the same. I literally just swapped the probes.
- Set target temp to 53.1 to match where it had settled on the previous day.
- The freezer was allowed to cycle for about 12hrs again. This time it cycled every 1hr - 90min, and stayed on for about 2min each time. The freezer air swung about 4-5 degrees each cycle while the water swung no more than .1f.
In conclusion, with an outside temp just under 10 degrees higher than the target wort temp, and assuming that fermentation has mostly concluded, the difference in accuracy of the two methods is nearly identical. I'd like to try this with a higher difference in temperatures but it will likely result in the same conclusion.
My concern now is this. Once I had cycled the freezer using both methods, I tried dropping the target temp to 52.8 with the controlling probe in the pail. Only a 0.3f drop. Naturally the freezer turned on right away. It took about 3 minutes to cool the freezer to freezing (30f+ swing) and my probe bugged out once it got there so I set the target temp back to 53.1. It took another 2-3 minutes before the water temp fully dropped to 52.8 then continued to drop to 52.1 over the next 40min with the freezing air around it. It has remained there for another 20 minutes and is due to meet the freezer air temp in a few minutes as it raises. It will likely take at least an hour and a half to make it back up to 52.8 again so for more than three hours it will have hovered about a full degree off of the target. It would have been even farther off had I let the freezer continue to run once it hit freezing. If a controller has a half a degree differential, it seems like this would happen every time the freezer cycles. What if the air was cold and you needed to heat the chamber? Would you then have a fire hazard if the heat was left on too long? If the controlling probe was in the air of the freezer I would have set the target a few degrees lower until the water dropped to my real target, then set the target back to about a half a degree above that. A gentle change involving no freezing temps that could freeze my air lock or cause a target to be overshot.
In final conclusion, until that last test I was ready to come on here and say I was wrong (which I am still mostly doing) and put my controlling probe in the wort. But, the final test blew it for me. I don't like the freezer actually reaching freezing temps, nor do I like the way the wort temp continues to drop once the compressor has stopped. The only way I would switch to controlling with the probe in the wort is with some sort of dynamic min/max air temps. Like if you want to go to 54 from 55, not allow the freezer air to drop past 44 or 10 degrees below the target. At that point you'd be using data from both probes to control the system so your venturing towards another method entirely.
So yeah. Not sure where that leaves the debate but I'm tapping out, rescinding my argument for lack of conclusive evidence, and continuing on my merry way. I'll report back in a new thread once I've tested out this new hybrid system. Until then, your minds will remain unblown. Sorry about that.