A very basic question about response time of an instrument

Regarding a temperature transducer, what is meant by response time in this context? It says it can be set to < 30 ms or < 300 ms. Is there any trade of between choosing low or high response time?

• Are you designing the control loop, or is someone else? The control loop designer should know. If the thermocouple is attached to a large thermal mass then you may want to go with the longer response time. – Mattman944 May 21 '19 at 13:34

Response time of a temperature sensor is the time it takes the measured value to get to a certain percentage (in general 95%) of the real temperature when you expose the sensor to a temperature step (for instance you have the sensor at ambient temperature and introduce it in an bucket full of ice).

Your temperature instrument behaves as a first order system. If you review a bit of the theory of these systems and look at the differential equation you'll find out the trade-off for a fast response comes in the form of static sensitivity. For more details you can take a look at the example at the end of this presentation: Sensors and Actuators TU Delft.

There are also quite a number of nuances that you should look at carefully. See, for instance, the thermocouple in a coffee cup example presented here: Dynamic Analysis and Time Response

Now, from a practical (industry) point of view, you should not worry about this particular problem that much, unless you want to measure a fast changing temperature (yeah, I know, how fast is fast?). This choice you have on your device (30ms vs. 300ms, which by the way are not realistic values, in the sense that you would not be able to measure temperature changing at 1/30ms=33Hz) are mostly to help you tune and or stabilize the response of a control system (in case your temperature sensor forms the feedback loop of a control system). I think the difference in between these two time responses is just the corner frequency of a filter at the end of the signal processing stage (this is just my guess, you would have to check with the manufacturer).

If you do need to measure temperature changes at fast rates, then you should start by looking at experimentally obtaining the real response of your whole measuring chain and maybe later looking at ways to compensate it (there is a whole lot of research literature on that topic).

In either case, you get min. 60 dB SNR with 18-bit resolution so it depends on how fast you sample and the temperature rate of change you expect for ΔT/Δt differences in sample delay.

• Is 300ms output is more averaged than the 30ms output? – cm64 May 21 '19 at 13:18
• @cm64 If the specs are the same for both output update rates, then you don't know if it is averaged or not. Usually, you get different specs because of what you say -- averaging performed for the longer option. If EE75 is right (I can't confirm because I didn't check) about the specs, and if you want to know for sure then I think you should write or call the manufacturer and just ask about the differences, if any. – jonk May 21 '19 at 16:37
• If you need more SNR than 60dB with 18bit = 108 dB dynamic range then I suspect it averages to increase SNR by up to 10dB or sqrt (10) – Tony Stewart EE75 May 21 '19 at 16:44