My desire for precision and accuracy occasionally leads to trouble. It wasn’t enough to have constructed my own immersion circulator, I had to test it to make sure it would be accurate through the range of temperatures (60°C – 85°C) I would need for sous vide cooking.
My plan was simple: I’d set the circulator at various temperatures, starting at 50°C, stepping through 5°C increments, and ending at 90°C. I would plot the value reported by the circulator thermocouple (green LEDs in this photo) against the actual temperature recorded by my calibrated digital probe thermometer (Thermapen). If the numbers lay along a straight line, I would be able to program an offset into the controller.
If only it had been that simple. The first thing I had to do was change the pre-set upper temperature limit in the controller from 40°C to 100°C, which would give me the testing range required. I immediately noticed a huge discrepancy between the set temperature (orange LEDs), reported value, and actual temperature. After a bit of fiddling with sub-menus (a process not improved by an interface that uses only four buttons) I programmed an offset of -28°C – a negative temperature correction that seemed to resolve the discrepancy at 60°C.
As I began the stepwise testing, I realized that the offset would have to change, and dusted off my algebra knowledge to calculate a slope (y = mx + b, remember?). Then it occurred to me to check all of the settings for the thermocouple, which is when I discovered that the reported value was being stated in °F instead of °C. Once I fixed that, I was ready to begin in earnest.
I stared at 55°C and stepped up, but when I got to around 75°C my readings seemed to level off. As I stared at the circulator, watching vapor escape from the tub of water, it hit me: I had reached the equilibrium point where evaporative cooling was perfectly balanced with the heat input. I solved that problem by floating about 100 ping pong balls ($12 for a gross at Amazon) on the surface of the water bath. They prevented further heat loss, and could easily be moved aside to add or remove cooking bags.
I got the water up to 75°C, then 80, and 85°C, when I noticed that the pump was making a lot of noise. I reached 90°C, shut off the pump, and let the whole rig cool down overnight. The next day, when I turned on the circulator, I didn’t hear the pump, and couldn’t see the water circulating. That’s when I noticed it had melted, as seen above. The cover was completely ruined and the case had begun to warp. Much to my surprise, the pump still worked with the cover removed, but I didn’t want to risk having it crack open and dump raw current into the ungrounded water bath.
After a bit of online research I learned that most of these pumps were rated only to 85°C. I bought a replacement, and instaled it this past weekend:
It’s larger and sturdier, but I decided to seal the snap-on cover to the rest of the unit with silicone cement. It runs quietly and survived another test run, but as added insurance I reprogrammed the upper limit to not exceed 86°C. I’m satisfied enough with the performance of the circulator that I have removed the electrical tape securing the front lid and permanently sealed it with more silicone.
What about the calibration curve that caused all of this trouble in the first place? It’s accurate to a tenth of a degree, with no offset required in the controller. Science: it may get you in trouble at first, but it always works out in the end.