I have a sensor board placed in box with an ESP8266 and 220V to 5V adaptor board which both are considerable heat sources. But sensor is very far away from them at least 5cm.

When I do not close the box it measure air temperature correctly. But when I close the box, it increases by 1 Celcius or so. Also box has air nozzles.

What I wonder is, there a billions of room temperature devices in a box sold on stores. How do they make the right measurement? Is there something like a constant temperature offset can be substracted and real temperature inside the room can be found?

  • \$\begingroup\$ So when you reduce ventilation, the temperature rises? Seems like your sensor is working fine \$\endgroup\$
    – PlasmaHH
    Jan 31, 2016 at 17:17

1 Answer 1


If you keep the power dissipation in your box as low and as constant as possible then a small delta-T can be subtracted. To a first approximation the difference in temperature is proportional to the power dissipated. This is essentially Ohm's law for thermal resistance.

For example, if you have a 360mW relay you could change it for a latching relay (low dissipation) or add a resistor and transistor to waste 360mW when the relay is off.

Devices like thermostats usually have vents and encourage air flow where it maximizes the accuracy. Sometimes forced flow is induced. I have one product that sucks air past the sensor.

Keep in mind that the thermal resistance is a separate question from the thermal capacitance (heat capacity). It might take 30 minutes for the temperature difference to stabilize- and it will be similarly slow in following changes in external temperature.

  • \$\begingroup\$ Well for my case it is not possible to replace an smps module and wifi chip which are the core items. But considering your answer the smps module & wifi chip has constant heat dissipation. In which case I can approximate the delta-T method you mentioned. I will make measurements at 10, 25, 45 Celcius degrees and share the results here to how well a delta-T can approximate (I'm really wishing it to be linear). Thanks for the answer. \$\endgroup\$
    – cheour
    Jan 31, 2016 at 21:48
  • \$\begingroup\$ The offset should be almost constant, not just linear. Be sure to allow enough time for things to settle down. An hour per point might be reasonable, depending on your system. \$\endgroup\$ Jan 31, 2016 at 22:54
  • \$\begingroup\$ I have made different measurements and yes it seems heat difference is constant. Following graph i64.tinypic.com/2s92mft.png is inside a room and nobody is there. It changes around 1.20 Celcius -+10%. So it seems right to assume change is constant. I have also made a measurement at outside. But there the difference extend to 2 sometimes 3 Celcius. I assume that it is because unboxed sensor effected by wind and outside conditions. Also as far as I see boxing gives a little bit filtering so changes become smooth. \$\endgroup\$
    – cheour
    Feb 3, 2016 at 12:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.