Averaging temperature from two sensors



  • Hello,

    I would like to know if it's possible for example to connect two thermistors and average temperature between two? For example, I want to make a heated chamber - and have one thermistor on top of the chamber and one on the bottom, is that the way to go for precise temperature inside? Thank you!



  • In your example, I would think a small fan would be a better way to achieve temperature evenness in the chamber.

    But that ducks your original question: I know of no way to make the firmware average multiple thermistors (today), further, thermistors are very non-linear and therefore cannot be simply combined with Ohm/Kirchhoff calculations like ordinary resistors.

    So... the best way at the moment? Perhaps a small outboard board that combines two or more sensors and passes the results along, perhaps by emulating a single thermistor. Arduino, ATTiny, something like that.



  • @Danal, will 4 thermistors achieve some notion of 'average' ?

    E.g. Serial(Parallel(T1, T2), Parallel(T3, T4)) or Parallel(Serial(T1, T2), Serial(T3, T4)), with some physical distribution of the four thermistors in the chamber.



  • @zapta said in Averaging temperature from two sensors:

    @Danal, will 4 thermistors achieve some notion of 'average' ?

    E.g. Serial(Parallel(T1, T2), Parallel(T3, T4)) or Parallel(Serial(T1, T2), Serial(T3, T4)), with some physical distribution of the four thermistors in the chamber.

    Unfortuantely, not, they are WAY too non-linear. Even a few degrees difference in A would place it in a different ohms/deg area than B, and given that the signals are already summed by the parallel/serial wiring, nothing can decode reality.

    But why limit to just thermisistors?

    Another way, besides the "outboard" I mentioned above, would be to use multiple individual sensors that produce a linear output. I am not 100% sure... but... I believe that devices like this:

    https://www.heraeus.com/media/media/hne/datasheets/m_sensors/en_10/m_222_e.pdf

    could be successfully used 'parallel' and/or 'serial' in combinations that could then be read by RepRap firmware. With an M308 sensor type of "linear-analog" and the correct constants, this should work.


  • Moderator

    Small fan makes the most sense. Averaging the temperature doesn't suddenly make the actual temperature distribution even inside, but the fan would, and then a single thermistor would be able to give an accurate representation of the temperature of the air throughout.



  • @Danal said in Averaging temperature from two sensors:

    In your example, I would think a small fan would be a better way to achieve temperature evenness in the chamber.

    But that ducks your original question: I know of no way to make the firmware average multiple thermistors (today), further, thermistors are very non-linear and therefore cannot be simply combined with Ohm/Kirchhoff calculations like ordinary resistors.

    So... the best way at the moment? Perhaps a small outboard board that combines two or more sensors and passes the results along, perhaps by emulating a single thermistor. Arduino, ATTiny, something like that.

    Thanks for a detailed explanation Danal! I guess I will stick to fan solution then. 🙂



  • @Phaedrux

    Hello and thanks! Do you think it is better to leave the thermistor on top of the chamber and fan at the bottom? Also, do I need to control the speed of the fan or it is not necessary, as heater goes on the fan goes on automatically 100%? I am curious how its done in Stratasys machines... Thanks!


  • Moderator



  • @Phaedrux

    Thank you! That's very educative 🙂


  • administrators

    To get more-or-less average readings:

    • Two PT1000s in series or in parallel will give an average to within a few degrees
    • Two thermistors in parallel will give you a value that is weighted towards the higher temperature of the two
    • Two thermistors in series will give you an average that is weighted towards the lower of the two

    For thermistors you would need to change the thermistor R25 value and make a small change to the C coefficient; or you could just pretend that the series resistor is double or half its real value. For PT1000 sensors you would need to pretend that the series resistor is double or half its real value. Alternatively, if you have another spare thermistor input, connect two thermistor inputs in parallel, then that will work with parallel-connected thermistors or PT1000 sensors.


Log in to reply