6
\$\begingroup\$

I want to use a thermocouple (K type) for a PID temperature controller in a heating system. The thermocouple should be pluggable to the system. I decided to use an industrial grade connector like below for this.

My question is: Is there any impact on correct measurements and/or accuracy of measurements when using thermocouple connections?

Aviation connector for connect thermocouple wire

\$\endgroup\$
5
  • 5
    \$\begingroup\$ That's a GX16 connector (mic connector) you will never be able to solder termocouple wires in that :D simply use the thermocouple connectors (which have screws and are made of the right metals) and try to estimate the mismatch. They also make special wire for extension, it depends on the distance and accuracy you need \$\endgroup\$ Dec 1, 2021 at 7:24
  • 2
    \$\begingroup\$ For soldering thermocouple wires, an easy trick is to crimp a ferrule on it first. \$\endgroup\$
    – jpa
    Dec 1, 2021 at 16:06
  • 2
    \$\begingroup\$ Thermocouples are pain to extend. Is it possible to use another type of sensor? \$\endgroup\$
    – fraxinus
    Dec 1, 2021 at 19:00
  • 1
    \$\begingroup\$ @jpa It's an easy trick if you don't have any specs to meet for temperature accuracy, sure, but it's not the correct way to connect thermocouple wires to anything. For a hobby project with no cares this "works", but it's wrong all the way through and should never be done in professional work of any kind. \$\endgroup\$
    – J...
    Dec 2, 2021 at 15:57
  • 2
    \$\begingroup\$ For temperatures from -55°C to +125°C and quantities below tens of thousands a DS18B20 is so much easier to use. \$\endgroup\$
    – Vorac
    Dec 4, 2021 at 8:04

5 Answers 5

18
\$\begingroup\$

The "correct" way to do this is to use thermocouple-specific connectors which use the appropriate metals as the pins. Here is an example from Omega:

conn

As Neil_UK said, you can also find these in bulkhead styles, too.


Another solution could be to use a "thermocouple transmitter" which would allow you to use your preferred connector.

They are installed closer to the thermocouple, using the appropriate thermocouple wire. They perform the temperature conversion and then provide a standard industrial output (e.g. 0-5 VDC or 4-20 mA).

This output can then use standard copper wiring and connectors back to your PID controller.

Here's an example of what they look like. I've never used this specific one:

tc_transmitter

\$\endgroup\$
3
  • \$\begingroup\$ You may note that the terminal strip in the transmitter is not composed of thermocouple materials. ;-) \$\endgroup\$ Dec 2, 2021 at 20:00
  • 1
    \$\begingroup\$ Usually not. For one thing, they're often 'universal' and can be set to to different thermocouples. The key thing is to keep the connector isothermal with "something". In the case of OP's arrangement the in and out wires should be the same temperature. In the case of the transmitter there is a temperature sensor internally and it should read the temperature of the junction from tb to wire accurately. Thick wires make that more difficult! \$\endgroup\$ Dec 2, 2021 at 20:06
  • 2
    \$\begingroup\$ @SpehroPefhany Thanks :) I had made a typo in my comment and deleted it, but now nobody knows what I said that you were responding to! Regardless, your info is good and helpful :) \$\endgroup\$
    – bitsmack
    Dec 2, 2021 at 20:08
14
\$\begingroup\$

It depends what wire you use to connect the circular connector back to your PID controller.

If you use K-type thermocouple wire, then the temperature error will be limited to the difference in temperature between the connector contacts, which, given the continuous metal shell around the connector, ought to be fairly small.

If you use copper wire, then you're effectively setting up the cold junction at the connector, and you could experience a temperature error of the order of the difference in temperature between the PID controller (where it thinks the cold junction is) and the connector. Depending on where the heat sources are in the equipment, it could be possible to have tens of degrees difference between them.

You can buy panel mounting female sockets specifically for thermocouple extension wires. They are made from the thermocouple materials, so their temperature becomes irrelevant when used to extend thermocouple wire. Most of them are not as robust as the connector you've shown though. IIRC, if you buy an expensive enough circular connector, you can order that specific pins are made of themocouple material for this very purpose.

An important consideration is maintenance and fault finding. If there's a standard type-K thermocouple socket on your front or back panel, then any maintenance technician will know what it is, and will get accurate and repeatable results when they plug in their type-K equipment. If you've used some non-standard contact materials in a non-standard connector, are you going to be around forever to fault-find?

\$\endgroup\$
1
  • 1
    \$\begingroup\$ You can indeed get pins made in certain thermocouple materials, though you may have trouble if you want to use a type R thermocouple or something (the ones made of platinum-rhodium alloys). Most people only care about types K, J, and maybe T, though. \$\endgroup\$
    – Hearth
    Dec 1, 2021 at 5:45
6
\$\begingroup\$

Neil_UK explained most of the issues. But I noticed something else, also.

That looks like a solder connector. I don't think you will find it easy to solder type K thermocouple wires. While I have never tried to solder them, I have read that it is difficult.

Also, I have stuck type K thermocouples (with exposed junction) into molten solder before (to measure the solder temperature) and when I did I found that the solder did not stick to the thermocouple wires at all.

Therefore, if you use this connector, you may end up with inaccurate results just due to the difficulty of soldering the wires.

One way around this might be to use crimp ferules on the thermocouple wires, and then solder the ferule to the connector. Not sure if all that will fit inside the housing, but maybe it will. Ferules for thermocouple wire are pretty small.

\$\endgroup\$
1
  • 2
    \$\begingroup\$ It's possible to solder K-type wires, though they need to be pre-tinned with something more aggressive than ordinary multicore solder. I bought a length of aluminium-soldering multicore decades ago, that works for K-type. \$\endgroup\$
    – Neil_UK
    Dec 2, 2021 at 12:13
4
\$\begingroup\$

If there is a temperature difference between the two pairs of contacts in the connector, it may also generate a thermal EMF and affect your accuracy -- perhaps by the equivalent of 1 °C or so.

Use a good quality connector (will make a better metallic connection), and if possible use multiple contact pairs for each thermocouple wire.

\$\endgroup\$
3
\$\begingroup\$

Yes, you can use these connectors for thermocouples. It's not ideal but they'll work.

The error would be worse if there is a temperature gradient across the connector. In benign conditions it won't affect the inherent inaccuracy much- usually the cold junction sensor is the biggest source of error.

Soldering may or may not be an issue. You can extend type K with type T without too much error provided the extra junctions are not crazy hot or cold. Type T (copper-Constantan) is quite solderable.

Sometimes the type K wires are copper plated to make them solderable.

Failing that, one trick is to tin the ends with silver solder, which itself is quite solderable (you can then soft-solder to the connector- unlikely the connector would survive direct silver soldering temperatures). Solder (and the requisite corrosive flux) used for stainless steel would also probably work since type K metals are similar to SS. Be sure to clean the flux thoroughly.

\$\endgroup\$

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.