0
\$\begingroup\$

I'm using 4 MAX31855 ICs to measure 4 K-type thermocouples and realized that 2 of my thermocouples have grounded junctions; I simultaneously also realized the MAX31855 is stated to not support grounded thermocouples (returns a "nan" error). The 2 grounded TCs in my setup (of the 4 total) continuously give me errors, no matter which channel they're connected to - so clearly the issue is the grounded junction. Using non-grounded TCs for my specific application isn't a feasible solution since I need my device to be compatible with existing systems which use welded/grounded TCs. What options are there to read the grounded TCs in this kind of scenario? So far I've come up with the following:

- Modifying the circuit: I looked into what I could do to change my IC circuit and/or the TC setup, but haven't been able to find anything yet that works as a solution. The grounded TCs are welded to earth-grounded components which are grounded to prevent operators from getting shocks when servicing the units. Un-grounding these components is not a good option, and so far tricks to make the IC work with the grounded TCs haven't provided any success. Isolating my TC-measuring device isn't very feasible since I need to permanently run off the target system's internal power supply and also connect via USB to a laptop - it may be possible to isolate but likely not practical.

- Replacing the grounded TCs with Un-grounded TCs: This doesn't work as a solution because my TC-reading device needs to be able to work with existing system setups that already use welded grounded TCs which can't be swapped out without incurring large expenses.

- Alternative ICs: While looking for ICs that can read grounded TCs my search results are much more meager than I hoped. I'm still combing through, but any specifically recommended ICs would be great. Writing new code for a different IC wouldn't be ideal but it's a potential solution.

My MAX31855 circuit is the following: enter image description here

Below are examples of TC grounding configurations. I believe the thermocouple weld in my grounded TCs is like the middle configuration in this figure, though it's possible it may also be like the top option. My multimeter beeps during continuity check between ground and each side of the TC, implying a grounded junction weld, but I don't trust this as an "official" test since the resistance of each TC wire is rather low to begin with, so I suspect it may be possible it could be the top configuration in the figure:

enter image description here

\$\endgroup\$
  • 1
    \$\begingroup\$ I Suspect that you're stuck. The only suggestions I would have (aside from finding a chip that directly supports your setup), is to isolate power to the MAX chip, and add isolation to the SPI bus (not a fun task, which still may not work). I'm looking forward to suggestions from other folks... \$\endgroup\$ – Chris Knudsen Oct 5 '18 at 17:54
  • 1
    \$\begingroup\$ Rule of thumb : If sensors are grounded, use galvanically isolated DAQ. You need to add a DC/DC converter, LDO and SPI isolator. \$\endgroup\$ – Marko Buršič Oct 5 '18 at 18:11
  • \$\begingroup\$ May I recommend MAX14850 SPI isolator; there is MAX14850PMB1 assembled board for prototyping. \$\endgroup\$ – MarkU Oct 5 '18 at 22:13
  • \$\begingroup\$ related question: Grounded thermocouple \$\endgroup\$ – Nick Alexeev Oct 7 '18 at 0:53
2
\$\begingroup\$

The only tried and tested method I used for monitoring a number of thermcouples that were an assortment of grounded and ungrounded devices of correctly and inadvertently-reverse wired types was to use a +/- 5 volt supply and a series of multiplexers (DG409) with cold junction measurement (RTD) and a voltage reference measurement also fed through the multiplexers. This then fed an InAmp with decent gain and mid-rail offset on the reference pin then to a decent ADC with serial interface.

Using +/-5 volts meant I could still read wrongly polarized/connected thermocouples and have a decent breadth of ground voltage. All the compensation was done in software with the voltage reference feed being used to compensate for any error introduced by the multiplexing.

If you are using several thermocouples then I can recommend this method but it was some years ago and maybe there is a reduced chipset version available.

\$\endgroup\$
  • 1
    \$\begingroup\$ +1 for correct implentation. I think ADS1248 has a input multiplexer and it can be powered with bipolar supply +/-2.5V. \$\endgroup\$ – Marko Buršič Oct 5 '18 at 18:33

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.