# ACI’s latest Tech Tips Video: Saving Controller Inputs by Averaging Room or Zone Temperatures There are two methods used to provide an average output from your sensors. The method you use will depend on the type of sensors, resistive or analog, in your application. Watch as ACI’s Tech Support Manager, Matt Buchholz, demonstrates the wiring of both methods.

### 3 Responses

1. Geoff Potts says:

Thanks Matt, reasonable tips for a small system with limited inputs but with the resistive solution, to me this assumes the sensors are linear. Most resistive sensors I use are non-linear, meaning differing resistance changes can affect readings. So how to calibrate?
I used to do this quite a bit but as DDC hardware costs are reducing and systems are becoming so much smarter I now prefer to just allow more IO capacity (remember most specifications now require spare capacity anyway).

Also (for both types) sensor failure is not allowed for. Most BMS systems now are able to read a separate input and determine if the sensor has failed. Then a bit of smart programming “removes” the failed sensor from calculations and hence control is unaffected.

2. Matt Buchholz says:

Very welcome! Great question.
You are correct that the thermistors are non-linear. But, the averaging calculation still works correctly even though they are non-linear. You must use the same thermistor Type for this to work correctly.
Let’s look at a example of 4 rooms.
Room 1= 72.5F (11167.79 ohms)
Room 2= 73.1F (11003.49 ohms)
Room 3= 72.8F (11085.30 ohms)
Room 4= 73.5F (10895.48 ohms)
R1 + R2 = 22171.28 ohms
R3 + R4 = 21980.78 ohms
Take the parallel calculation and the output will be 11037.81 ohms. If you look at the resistance vs temp chart, this equates to 73F.

To double check that, you can take the average of just the actual temperature number. If you average by actual temp numbers, this comes out to 72.975F.
72.5 + 73.1 + 72.8 + 73.5= 291.9
291.9 / 4 = 72.975

For calibration, it gets a little harder since you have 4 total sensors. A thermistor output cannot be adjusted at the sensor. Normally, if you have one thermistor reading off, you would need to do a single point offset in the controller. So the only option would be to add an offset at the controller, but that would affect total average of all sensors. My best recommendation would be to the replace the 1 sensor that would be reading off.
That’s one big advantage of using an active Temp Transmitter. You can adjust the output at just one sensor.

Very good point about the extra IO capacity. That’s always my preferred method, but we still see applications where the system is maxed, and they do not want to add an extra IO module. We run into these with retrofit jobs.

For the failed sensor application, unfortunately there would not be a good way to “remove” it from programming. If one would fail, it would skew the entire reading. Now there should be a very significant drop in temp reading if one would fail. So that should be red flag. If one would fail, both sensors on that branch would be affected. An example if R2 failed. This would also affect R1. The average output would be around 36F. That should be red flagged if a room temp was reading that.
Any other questions, please let me know! 