I have seen past posts about how various temperature sensor from different manufacturers have different ranges of accuracy.
I have built a whole home thermostat system that takes inputs from throughout the house and control the upstairs and downstairs HVAC systems based on these results.
I have certainly found that different sensors (same and different brands) require an offset to calibrate them to get useful comparisons. That said, has anyone experimentally or by other means found any brand of sensor that stays very consistent? Alternatively has anyone found a brand that varies around the actual temperature, but within the advertised range of accuracy?
I am specifically looking for sensors that stay consistent even if they are consistently inaccurate.
I.E. If 3 sensors located together read s1 = 76.1 s2= 75 and s3 =75.2, and then while still located together in different circumstances they read s1 =61.1 s2= 60 and s3 =60.2, but not in different circumstances s1=60 sc=61 s3 = 60.5
Right now, I am getting these reading for various temp sensors in various rooms, but haven't calibrated them to each other:
One thing to consider is the time lag of your sensors. When making thermal rise measurements of electronic parts I would put the samples with the thermometers in a cardboard box (maybe 12" cube or so) and place it in a location with no nearby air currents. Let them sit for at least an hour until everything stopped changing. You could use a cooler instead of a cardboard box.
If you collect all the sensors (if possible) and test them all at once you can get a good idea of the variability within the group of sensors.
From there you will have to decide if they all agree enough or warrant using Ice and boiling water to "calibrate".
I was certainly thinking about this sort of method to calculate the differentials between sensors. My big fear is that if the sensor is +/- 1 degree and at time 1 sensor 1 is 2 degrees warmer that sensor 2 (i.e. one is actual temp +1 and one actual temp -1), and then after some time sensor 1 is 2 degrees colder than sensor 2 (i.e. they are both still within the marketed range, but have drifted to the opposite ends of the allowable (per specifications)range), then I can't really say always add 2 degrees to sensor 2 to match sensor 1 as sometimes this would cause them to match and sometimes would show them as 4 degrees off.
That is why I am concerned about which sensors have any given amount of drift within the specified range. I am fine correcting an offset as long as it is a consistent offset from a given point.
I've not experienced a significant amount of drift in my sensors. I will admit I've not performed any specific testing for drift.
It has been my experience with sensors in general that zero (or offset) drift is much more likely than span or gain shift.
I guess the only solution would be to test them today and again in one month or so.
Thank for the feedback. I have suspected you are right that drift is unlikely, but I was hoping someone had already done the experiment you suggested to verify this. I guess I either have to hope it is true or do it myself.
I have a bunch of Konke, 1 Aeotec Multi Sensor 6, 1 Advanced Honeywell T6 Pro Thermostat (Z-wave), and 1 annoying Lennox iComfort Thermostat.
Obviously for testing, the Thermostats are on the wall and would require extra efforts as when I set of the Konke on top of the iComfort, the heat coming off the thermostats electronics warmed the Konke by several degrees.
I'm not 100% sure but I think you can pull the Honeywell off the wall and the batteries will keep it going. Not sure you can get a temp from it though.
Yes in the box the units should be held off the bottom and spaced as best as you can.
A side note, a long time ago I didn't plug the wire hold adequately which caused havoc with my room temp.
I assume you meant wire hole, as in the air from behind the sheetrock reached your thermostat creating reading less about your living space and more about the non-climate controlled area behind the sheetrock.
My last concern is with the Aeotec Multi Sensor 6 generate enough heat from its electronics to give different readings depending on its orientation (I run in plugged in).
I doubt it, especially if you run it by battery. The electronics have to be so low powered to have an acceptable battery life I don't think they can afford to generate much heat.
