I hear that thermocouples, especially type K, got really low output voltage ( 41µV/°C. )

Someone told me that if I had 16 bit DAQ device I would need a range from - 1 V to 1V to measure 1 °C.

So the problem is if my DAQ usb device would record the temperature changes without any signal amplifier ?

I got in mind two devices :

I. USB 2416 24 bit Voltage Mode: ±20 V, ±10 V, ±5 V, ±2.5 V, ±1.25 V, ±0.625 V, ±0.3125 V, ±0.15625 V, ±0.078125 V

II. NI USB-6218 16 bit Input range : ±10 V, ±5 V, ±1 V, ±0.2 V

According to the mentioned statement, I should be fine with both but I need to know for sure because it's not a cheap ride.

BTW Does differential measurement have anything to do with the "sensitivity" level apart from noise elimination ?


Ok so it seems that I'm forced to use signal amplifier so I'd better start looking for wires, connecting amps with my DAQ device. Thanks for answers.

  • \$\begingroup\$ Well, why not use a k-type thermocouple amp? Surely you realize you're not the first person to have the problem. Check out the AD8495. You can also just read it digitally with this. Keep an eye on those XY problems you're having. \$\endgroup\$
    – Samuel
    Nov 17 '15 at 17:21
  • \$\begingroup\$ You've heard about 41µV/°C. Have you heard about cold junction compensation (CJC)? \$\endgroup\$ Nov 17 '15 at 17:36
  • \$\begingroup\$ Yes, I heard about it. I know it's better to use DI method than SE ( single ended ). \$\endgroup\$
    – Galjuntur
    Nov 17 '15 at 18:22
  • \$\begingroup\$ The low output volts of thermocouples means that they are best at really high temps where other sensors would die. Do you really need to use a thermocouple? \$\endgroup\$
    – Autistic
    Nov 17 '15 at 19:28
  • \$\begingroup\$ @Autistic Thermocouples have another important advantage that they can be made very small. But you raise an important question: does this have to be a termocouple? \$\endgroup\$ Nov 17 '15 at 19:36

Things you need to decide- whether your junction will be grounded or not. If they are grounded you will need differential measurement with good common-mode rejection. If they are ungrounded you don't need that (assuming low leakage). Advantages are that grounded-junction is usually much faster response and can be cheaper.

Also you need to provide cold-junction compensation. That means that you have to measure the temperature (with something other than a thermocouple!) of where the wires from the thermocouple are connected to copper (usually a screw terminal block). A bit of math is required to do the compensation, depending on the accuracy you need. The error in your cold-junction sensor's reading is about 1:1 with error in your thermocouple reading- meaning if you want accuracy of 1°C you might need to have the terminals isothermal and measure to within a total error of +/-0.5°C. This is non-trivial if the environment isn't rather benign.

Resolution as you calculated is only part of it- you need to make sure that the zero drift of your amplifier/ADC is low enough. The output of a type K thermocouple is not insanely low, but you need to be fairly careful. For example, a cheap op-amp could easily drift that much with a few degrees C change and could have 50 or 100 degrees initial error due to the offset.

Remember, you must use thermocouple or thermocouple extension wire and appropriate connectors all the way to your measuring instrument, where the cold-junction compensation takes place. A thermocouple is not just like a battery supplying a fixed voltage for temperature. The temperature of the connection points matters about as much as the temperature of the measuring junction!

  • \$\begingroup\$ USB 2416 is provided with 4 cjc sensors. Manual says that absolute accuracy for 0,078125 V is 34 micro volt. It's strange for me that I buy expensive ADC with dedictated diferrential inputs for thermocouples and i need to use signal amps. \$\endgroup\$
    – Galjuntur
    Nov 18 '15 at 10:33
  • \$\begingroup\$ If it has CJC then it probably has dedicated linearized thermocouple ranges. Strongly suggest reading the manual. \$\endgroup\$ Nov 18 '15 at 16:53
  • \$\begingroup\$ It should work without any amps or compensation wires. I learnt that manual almost by heart but I am not so familiar with such devices. I will investigate this "domain" in the future without a doubt. \$\endgroup\$
    – Galjuntur
    Nov 18 '15 at 17:26
  • \$\begingroup\$ The wires to the device must be compensating type (thermocouple material) from the thermocouple junction right to the terminal strip, and they must be connected correctly (no double reversal) or you will have large errors. \$\endgroup\$ Nov 18 '15 at 18:07
  • \$\begingroup\$ Ah, I see. I'll look into it then. \$\endgroup\$
    – Galjuntur
    Nov 18 '15 at 20:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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