PDA

View Full Version : Thermostat Probe Location



clearchris
30-04-2013, 12:26 PM
So I got tired of the cheap analog thermostats breaking in my cooler. I build a solid state one with an arduino (arduino.cc for the curious).

Why are the thermostat probes located in the evaporator coil? In my (short) experience, the coil temp doesn't have much to do with the cooler temperature. What I'm seeing is that the air leaving the evaporator is at 32F (0C) and the cooler air can be anywhere from 50F (out of spec) to 32F. So a reading in the coil isn't really helping me much.

Do the analog thermostats basically have a time delay built in? I.e. it takes maybe 15 mins for the gas in the probe to cool down thus ensuring that the circulating air has time to cool the contents?

Do you think I should
1) Build in a time delay (seems like a hack)
2) Move the thermocouple to a place that I want to actually measure the temperature.

I'm leaning towards #2 but I want to know why the thermocouple is in the coil. I imagine there must be a good reason for doing it this way.

Thanks!

Tesla
30-04-2013, 02:16 PM
Hi clearchris
I really like Arduino. Sometimes the coil sensor is used to terminate the defrost. Some times there can be an air on, coil and space sensor for control. The analogue (technically speaking all temp sensors are) gas sensors are much slower in reacting and increase the delay like a second order. If you google carel and skope you should find something on what the different places for sensors are for or some one who knows will answer (this would be for forced draft). You have not stated what type of fridge you have so I will assume a domestic from this point on using natural convection. There will be an amount of time for the frost line to travel down the length of the evap coil and the delay of the gas in the sensor element to reduce pressure then activate a spring switch. So go for your option of 1. This has a disadvantage when the fridge gets older an springs a gas leak as the frost line never reaches the sensor so the comp keeps running. 15 mins would be too long in my mind depending on the load so you might be better to down load a PID loop control algorithm from one of the libraries and use P and I only. Let us know how you go please.

clearchris
01-05-2013, 06:15 PM
Thanks! Looks like I have some reading to do.

Brian_UK
01-05-2013, 10:47 PM
I know that this is a ready made controller but the operation details may be of interest/use to you...

http://www.parker.com/parker/jsp/documentdisplay.jsp?mgmtid=20be166a991b9210VgnVCM10000048021dacRCRD