13
$\begingroup$

I'm looking for advice on how to attach a thermocouple to a metal surface in a fashion that's relatively temperature resistant (more-so than) glue.

I've got some very fine gauge wire on hand to make the thermocouples and specialized spot welder that will join them to form a bead. My goal is to attach them to a 3-7 mm thick sheet some kind of metal. Could be steel, aluminum, copper etc. I don't have a strong preference on either the metal used or the exact thickness. I have access most commonplace metalworking equipment and can purchase basic tools and materials but can't justify the expense of a purpose-built welder for this kind of thing.

So far I've tried two things (unsuccessfully):

  • Soldering: I used a gas torch to heat copper plate enough to get a nice puddle of solder on it and then tried to dip the (already welded) TC. The wire simply would not wet, it would just push the solder out of the way and couldn't make a connection.

  • TIG welding: I tried to weld a both TC and bare TC wire to an aluminum plate by forming a small puddle on the plate and then attempting to dip the wire into the puddle while quickly extinguishing the arc. Mostly the wire just melted and "ran away" from the arc. Other times it just didn't fuse to the surface. I also tried putting the wire in contact with the surface in advance and starting the arc on it.

I know that I've heard mention of doing this, but haven't been able to find any good information about how to actually do it. I could use anything from a definitive answer to comments on how to improve the techniques that I've attempted already.

p.s. I weld semi-regularly but have very little experience soldering (esp. with a torch), so don't assume a lot of skill or knowledge in that department.

Improve this question
$\endgroup$
6
  • 1
    $\begingroup$ Could you include a picture of the thermocouples you're trying to use. From you question I have the impression that you are just using a pair of wires & I'm thinking screw them on with a bracket type system or sandwich the wires between the main metal plate & a smaller plate that is screwed to the main plate. $\endgroup$
    – Fred
    Feb 14 '15 at 8:35
  • $\begingroup$ Are you somehow electrically connecting the metal of the thermocouple to the metal surface? $\endgroup$
    – jjack
    Feb 15 '15 at 18:33
  • 1
    $\begingroup$ What is this thing's application, and what kind of temperatures do you expect to see? $\endgroup$
    – whatsisname
    Feb 15 '15 at 20:37
  • $\begingroup$ What kind of themromcouple (TC)? Copper-Constantin type T, solders just fine. How about putting the TC under a screw -clamp. $\endgroup$
    – George Herold
    Feb 17 '15 at 15:08
  • $\begingroup$ What temperatures are we talking about? Silicone for sealing chimneys can withstand up to 1200C. $\endgroup$
    – SF.
    Feb 24 '15 at 23:14

11 Answers 11

Active Oldest Votes
3
$\begingroup$

A little late to the game.

A method recommended in B&W's Steam, Its Generation and Use, is to drill small, closely set holes in the metal substrate, insert the individual wires into the holes, and peen them in place. This avoids the issues others have raised of welding affecting the properties. It provides intimate contact with the measured surface. It does not introduce any additional materials. It will serve for any temperatures that the substrate and thermocouple materials themselves can withstand. It should be moderately robust mechanically.

The net couple is the same as that of the standard wire to wire joint, but is made of each of the wires to the substrate, taken in series.

At my suggestion, a client used this method successfully in a fired boiler.

Improve this answer
$\endgroup$
2
$\begingroup$

I ended up finding a solution that works (haven't tried the cements suggested by others yet).

The trick was to use a spot welder rather than a tig welder.

With the correct power settings, that stuck the TC to the surface without damaging either component.

It's not exceptionally strong (I used very fine TC wires), but it should result in conditions that are as close as possible to what they would be if the TC were not there.

Improve this answer
$\endgroup$
3
  • 1
    $\begingroup$ Welding may change the electrical properties of the wires and affect the reliability of the readings. $\endgroup$
    – Wallace Park
    Jun 11 '15 at 15:36
  • 1
    $\begingroup$ @WallacePark Well, the thermocouples are normally welded together in the first place. The only difference here is that they'll be fused to the substrate as well. The law of intermediate metals indicates the involvement of a third component won't interfere with the TC reading, provided that temperature is locally uniform. $\endgroup$
    – Dan
    Jun 11 '15 at 16:05
  • $\begingroup$ @Dan - Will W answered your question suggesting a spot welder. Out of curiosity, why did you not accept his answer and then leave this as a comment? $\endgroup$
    – CBRF23
    Aug 12 '15 at 10:47
2
$\begingroup$

For an easily removable option you can use Kapton tape. It is heat resistant and works well for holding a thermocouple temporarily in place.

I also wrap the wires in Kapton for electrical isolation.

Improve this answer
$\endgroup$
1
$\begingroup$

If you aren't too worried about response speed, you might be able to embed the thermocouple in a thermal grease, then weld a piece of shim steel over the top to provide mechanical support.

I have only ever seen thermocouples welded directly to a substrate before with spot welding equipment.

Improve this answer
$\endgroup$
1
  • $\begingroup$ Thanks. Do you know if an ordinary spot welder might do the trick? I'm sure that I can get access to one if it might yield better results. $\endgroup$
    – Dan
    Feb 15 '15 at 18:51
1
$\begingroup$

I would use a thermal epoxy (potting agent) for this. I have done this successfully with a thermistor before. Also I'm not sure, but you probably want to electrically isolate the thermocouple so that you don't lose voltage to an earthed component and thus get an incorrect reading. So a potting agent is great for this (high thermal conductivity, but not electrical conductivity).

You can buy split-bag packs from electronics suppliers, take off the clip so the liquids can mix and then pour/blob it on as needed. It then cures in about 24h. However if this is just a one-off application and you only need a small amount, this is probably a very expensive way to do this task, so find a local electronics hobbyist (try your local hackspace) and ask them for a small blob when they're doing a larger job with the stuff.

Improve this answer
$\endgroup$
2
  • $\begingroup$ Does it have to be an epoxy? I've been using thermal silicone glues (see, e.g., a product called HC910) that are electrically insulating by thermally conductive. About \$5/ounce, but you can just squeeze the amount you need out of a single tube, vs. epoxy potting compounds that are \$3/ounce but not easy to get or mix in tiny quantities. $\endgroup$
    – feetwet
    May 27 '16 at 1:13
  • $\begingroup$ @feetwet the silicone glue sounds better then for one-off applications. $\endgroup$
    – jhabbott
    May 27 '16 at 12:50
1
$\begingroup$

Drill small hole in the aluminum, big enuf to put in the TC junction head, then pound the edge of the hole with a center punch, that'll deform the hole to hold the TC in the hole mechanically. Works great in high temperature applications.

Improve this answer
$\endgroup$
1
$\begingroup$

I read all the comments regarding affixing a thermocouple to a surface. Every one has nice contribution.

I have done testing myself to attach temporarily j type thermocouples wire to surfaces and found out that adding epoxy to attach the thermocouple to surfaces decreases the measuring accuracy.

The second solution I used was adding a Kapton tape on its tip and using the same tape to attach it to the surface and I found that two layers of Kapton tape decrease thermocouple accuracy.

Using one layer of Kapton tape at the thermocouple tip increases the accuracy but still the difference is around 1.5 degrees with the reference.

Improve this answer
$\endgroup$
1
  • 1
    $\begingroup$ I have removed the question that was at the end of the text. If you have a question raise it as a separate question, do not ask it in the answer section of an existing question. When posted as a separate question you can link you question to this question if you want to. $\endgroup$
    – Fred
    Dec 16 '21 at 15:58
0
$\begingroup$

How about Ceramic Cement? When properly mixed and used OB-400 is good to 1425C (2600F). Soldering a TC to metal surfaces takes skill and patience. Even more of the above when TIG welding. Have you considered using a self-adhesive thermocouple?

Improve this answer
$\endgroup$
0
$\begingroup$

Wire T/C's require a bead or stress relief if separate at the surface to be measured, both methods work well on metal. Peening a place for your bead is best if surface is not delicate. Attach with either epoxy (without metal in it), or attach with aluminum tape covered with rtv for stress relief. If attaching bare wires, spot weld the wires with stress relief loops in the wires (this is so the wires don't break upon thermal expansion) then cover with epoxy.

The spot welded wires are more accurate than the beads.

Improve this answer
$\endgroup$
-1
$\begingroup$

Try using a t a u (thermocouple attachment unit) they normally work

Improve this answer
$\endgroup$
1
  • 3
    $\begingroup$ You could improve you answer by adding more detail about how this technology works and adding references to illustrate it. $\endgroup$
    – Chris Johns
    Apr 18 '16 at 19:55
-2
$\begingroup$

Al tape is also good option.

Some epoxy mixed with metal powder (Iron powder) is another method.

Improve this answer
$\endgroup$
1
  • 3
    $\begingroup$ Hello user6012, welcome to engineering.SE. We have a focus on quality question and answer here, and quality answers need more than a few sentences. Why is Al tape a good option? What is the point of mixing the metal powder into the epoxy? $\endgroup$
    – Chris Mueller
    Apr 15 '16 at 11:39

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.