# Measuring Cryogenic Temperatures

I need an inexpensive instrument to measure cryogenic temperatures (down to -200C).

I can build a thermistor-based thermometer using an Arduino that is accurate to under 1 degree for 0 to 100C.

First of all, can ordinary NTC thermistors be used at cryogenic temperatures? Second, if I were to try to use this approach, I would need ways to calibrate the device.

I have found a chipset from Analog Devices that would let me use a thermocouple down to that range.

I would take other ideas on how to build one, but then I need some equipment to calibrate whatever I build. Is there anything I can buy for < $100, I don't need huge accuracy, but I need something. • I expect you don't want to use liquid nitrogen at 1atm as a calibration? What accuracy do you need? – Nicolas Oct 21 '13 at 15:09 • Since liquifying air is the goal, I need some data points along the way. Besides, How would I know if we're liquifying Nitrogen as opposed to oxygen? – Dov Oct 21 '13 at 15:51 ## 2 Answers You ether can take usual K-type thermocouple (like in multimeters), and calibrate it - it supposed to work fine in LN2. I personally tried that with DMM and it worked (though showed large error). Multimeters use simplified conversion formulas and could give huge errors at cryogenic temperatures - so remember that you ether need table-based conversion, or non-linear formula. For example, you can use this data: http://www.omega.com/temperature/z/pdf/z204-206.pdf or in Google "K-type table" Another possibility is to take PT100/PT1000 platinum resistance probe/sensor. They are not expensive, some even under 5$. You might get decent (I guess <5C) accuracy with it even without calibration, but measuring resistance might be slightly trickier, than voltage.

• I have type K thermocouple and multimeters, but since the range does not extend to cold temperatures, what do I do with it? It's not like the thermocouple just has a small voltage, it's not that simple. There's the compensation circuit, noise reduction. What should I try, exactly? – Dov Oct 21 '13 at 21:38
• I just looked up pt1000 spec. omega.com/pptst/F1500_F2000_F4000.html It says 0.003 ohms/degree C. So This would require amplification at the very least because the current through this had better be uA in order not to disturb the temperature being measured? In general I think I am more comfortable with thermocouple solutions as they don't resistively heat the environment, but they must cool it. Am I being irrational? – Dov Oct 21 '13 at 21:43
• when you say you "tried that" do you mean you put the K-type thermocouple across a known temperature and a cryogenic temperature and measured the voltage with a DMM? Or that you just used a stock DMM and the number came out??? – Dov Oct 21 '13 at 21:47
• Regarding K-type thermocouple: yes, to get correct readings you will need to amplify & shift signal using operational amplifier. Regarding pt1000 - it's ~0.003 Ohm/Ohm/C, not 0.003 Ohm/C - i.e. resistance change will be much larger, you have to multiply it by base resistance value. Your self-heating point is valid - so you will defenitely need as low as possible current flowing. – BarsMonster Oct 21 '13 at 22:15
• @Dov Regarding "Working K-type thermocouple" - yes, I've checked results using DMM and got wrong temperature reading in LN2 (it was off by ~50C). But there is nothing stopping form using it with more precise formula/table. – BarsMonster Oct 21 '13 at 22:17

If this case is still open, here are a couple of options. Both PRT-100 and thermocouples are feasible solutions. If using a thermocouple, E type has the highest sensitivity at cryogenic temperatures (even down till 40 K, LN2 is 77 K as you know). However as pointed out above, thermocouples generate very low voltage and this output signal is prone to EM interference. Hence it needs proper signal conditioning. The advantage is thermocouples are cheap.

On the other hand, there are PRT-100 which work perfectly accurate down till about 50 K. They in fact follow a standard IEC-751 calibration. There are PRTs specifically developed for cryogenic temperature measurements and so calibrations may not be necessary. These PRTs are a little more expensive than thermocouples, but would still fit in your budget of \$100. The output can be measured on any DMM (2/4 wire).

The cryogenic thermocouples and PRTs can be bought from Lakeshore Cryotronics, Inc. which is an industry standard for cryogenic sensors and controllers (I buy all my thermometers from there).