Thermistor in parallel

Question

I want to measure a very low temperature (<-50C) with an NTC thermistor, but the error in my system is dominated by my ADC frontend characteristics due to the high resistance of the thermistor (>500k). Are there any downsides to putting two NTC thermistors in parallel next to each other? Assuming that they are very close, and thermally bonded, it seems like this would decrease the resistance measurement error of my system, and not really increase the resistance error of the NTC readings at that temperature?

It seems like you could do the same thing for high temperatures, with NTCs in series?

Obviously both of these approaches have downsides in the other end of the scale (parallel is bad for high temp, series is bad for low temp). Its also possible to get thermistors with lower resistances for the low end range, although they seem to be in short supply.

Answer 1

Are there any downsides to putting two NTC thermistors in parallel next to each other?

In terms of actually measuring the temperature, there should be no problem with putting multiple identical parts in parallel.

• There is extra cost for more parts. If you can it might be cheaper to just find a part that has a lower resistance to begin with (like 1K or 10K).

• There is extra space taken up.

• More parts to put down means more assembly labor.

There are also other options.

• Buffering the thermistor with a precision op-amp would probably get the best results.
• If the sample rate is low, and the bias current on the ADC is low, then you could improve the measurement by putting a 1uF ceramic capacitor in parallel with the thermistor but near the ADC. The capacitor would serve as a low impedance charge reservoir to feed the ADC each time it takes a sample.