0
\$\begingroup\$

The system in question has just one type K thermocouple.

Is it a bad practice to Y a thermocouple into 2+ thermocouple inputs?

For example, Y a thermocouple into:

  • DAC thermocouple input card
  • Limit controller (for over temperature safe shutdown)

The temperatures I work at are between 0°C and 42°C, with a tolerance of +/- 0.1°C.


Y'ing Method

  • Using type K thermocouple extension wire
  • Soldering (63% Sn, 37% Pb) the two Y's together
  • Individually heat shrinking the connections afterwards, so they won't touch

ying thermocouple

I think this works because:

  • Solder beads are at the same atmospheric temperature as each other
  • Solder beads are at the same atmospheric temperature as the electronics
  • The solder beads aren't touching --> no thermocouple junction is formed there
  • The same dissimilar metal is used in each solder bead

From what I have read on thermocouples, this seems to indicate that it should be fine.


Assumptions

  • Cannot add more thermocouples or replace with dual element thermocouples
  • Do not want to replace control devices to work with temperature transmitter
  • If the thermocouple were a probe, it would be ungrounded

Possible Solutions

Just to address these two solutions (outside of the scope of this question) up front.

From How do you split a thermocouple signal?

  1. Dual element thermocouple (only relevant if thermocouple is a probe)
    • Pros:
      • Doesn't rely on Y'ing small mV signal to several inputs
    • Cons:
      • Higher cost
      • Probably not readily available in your local factory
      • I have observed them to be more fragile in bending
  2. Thermocouple transmitters
    • Pros:
      • Doesn't rely on Y'ing small mV signal to several inputs
    • Cons:
      • Added system cost/complexity
      • Usually requires swapping out pre-existing hardware as analogkid2digitalman did in this thread

**Edit**: Other Relevant Information

NTC to multiple inputs

One or both controllers might not even have one end of the thermister connected to ground (or any other shareable reference node), further complicating matters.

In other words, you need to know a lot about how both controllers work internally before you can consider how to get them to share a sensor.

As @KevinWhite brings up in the comments, knowing the internals of the input hardware is relevant, as grounding is an important parameter.

\$\endgroup\$
2
  • 1
    \$\begingroup\$ Does the controller/ADC ground either lead of the thermocouple? If so then there may be a cross-connection that could induce noise/error into the readings. \$\endgroup\$ Mar 4, 2020 at 18:43
  • \$\begingroup\$ @KevinWhite good thinking. Currently I am using a Beckhoff EL3318 thermocouple input terminal (beckhoff.com/english.asp?ethercat/el3318.htm), which should hopefully not do that. \$\endgroup\$ Mar 4, 2020 at 21:30

1 Answer 1

1
\$\begingroup\$

Generally that's not a problem provided there is no ground issue.

The thermocouple connection per se is perfectly fine, and I've personally done it in many different applications (such as operating a data acquisition system, recorder or indicator from the same sensor as the controller) .

The individual Chromel / Alumel (type K) wires are not solderable with the usual flux unless they happen to be plated. They may be solderable with very active (corrosive) stainless-steel flux. You could also use a terminal block, but the block should be kept out of the way of air currents etc. The temperature is not particularly important provided the gradients are kept to a minimum (obviously proper extension leadwire must be used, and with the correct polarity).

It's also possible to simply daisy-chain the thermocouple extension wire from one terminal block to the next. Since the limit controller is less critical as to accuracy you could put it last. The temperature at the terminal block of the first instrument will be a few degrees higher than ambient, but that does not cause an error.

Note: there is a small chance that one device could fail in such a way as to affect the other. So this may not be a kosher approach in safety-critical system as the redundancy is not present. There is also a possible issue with break protection, which typically puts a bit of current through the thermocouple to detect a broken sensor. If one instrument is set for high and the other configured for low they may fight. The current adds when properly configured so you might need much less total resistance of connections to keep a bit of extra error from creeping in.

One advantage to using two completely different sensors (rather than a dual thermocouple) is that the controller can do a really good job of controlling its own sensor temperature, but not so great a job of controlling the item you're really interested in, and a separate sensor can help pinpoint that kind of issue.

\$\endgroup\$
1
  • \$\begingroup\$ Thanks for mentioning why having two process values from two locations can help in debugging! Interesting point on using the corrosive solder. This agrees with my understanding of the Seeback effect, and potential consequences when Y'ing a thermcouple input. \$\endgroup\$ Mar 5, 2020 at 1:52

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.