I'm operating a board at cryogenic temperatures (65 degrees Kelvin and higher) and need to monitor the PCB temperature. All of the standard surface-mount thermistors I've seen have operating ranges down to -80 degrees C for standard thermistors. There have been several thermistors with leads that are intended to be immersed in liquid that I've managed to find. Is there some physical limitation to designing thermistors for this temperature range? Our mechanical setup makes thermocouples and temperature-sensitive diodes a very difficult option to use. Are there some other temperature measurement methods that I am not aware of? Price isn't a problem.

  • \$\begingroup\$ Google: Cryogenic Temperature Sensors. First listed is Omega. My company used silicon diodes when I worked with cryogenic coolers many, many years ago. \$\endgroup\$
    – Mattman944
    Jul 24, 2019 at 20:54
  • \$\begingroup\$ I'm surprised that you can put a thermistor somewhere that you can't put a diode. @Mattman944 said pretty much exactly what I was going to -- a diode or a forward-biased BE junction of a transistor (short the collector to the base) works a charm. \$\endgroup\$
    – TimWescott
    Jul 24, 2019 at 21:05
  • \$\begingroup\$ How about a coil of wire? If you know the resistance at room temperature, then calculation should be possible for other temperatures. \$\endgroup\$
    – rdtsc
    Jul 24, 2019 at 21:16

1 Answer 1


Lakeshore Cryotronics is one of the main worldwide sources of cryogenic temperature sensors. They have relatively inexpensive silicon diode sensors DT-670 that work fine for rough accuracy. I've used those and the Germanium temperature sensors (which are much more expensive and accurate but require specialized circuitry (AC energization at very low current and 4-wire measurements). The latter also come with individual calibration curve constants fit to the individual response of each sensor serial number. You might possibly re-think "price is not a problem" when you see the price list of instruments and sensors.

In both cases I've designed circuitry to provide signal conditioning. In the latter case we were able to do much better than 2\$\mu\text K/\sqrt{\text {Hz}}\$ using DSP techniques and some very precise and stable circuitry, in an airborne instrument. The Ge types are more suited for lower temperatures, they are very nonlinear and give best resolution below 10K.

65K might be a bit low for regular silicon diodes (carrier freeze out effects might show up at sub 100K) but you could probably calibrate thin film Pt1000 elements using a calibrated reference sensor.


Your Answer

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

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