Welding Thermocouple to Test Sample

Question

I have a 3mm copper test sample that is going to get heated up, and I would like to know how hot it is actually getting. Previously, we have just held the thermocouple close to the sample, but we are considering mounting it to the sample to get a more accurate reading. I was told that it is possible to weld an exposed thermocouple to the sample, but I am concerned about making a ground loop.

I guess there are two questions here:

1. Will a ground loop be created and do I need to be concerned about metal-on-metal contact with the thermocouple and copper for any other reasons?
2. Does anyone have advice on the best way to actually weld the thermocouple to the test sample?

Answer 1

I normally do this by twisting the thermocouple wires together, then crimping into a standard crimp-type terminal lug. That is then either bolted or soldered to the test item.

So long as the two thermocouple wires are in direct contact with each other, you won't have any appreciatiable accuracy problems.

Note that your temperature measuring device must be able to accept a grounded thermocouple. This isn't a problem with a stand-alone / hand-held meter but does have to be taken into account if connecting to a multi-channel and/or computer-based data acquisition system.

Answer 2

Welding to copper is not easy because copper has very high thermal conductivity. Aside from that there is no real issue. You could braze it, depending on what temperature range you need to withstand. This is also called "Silver Soldering" - melting point is in excess of 600°C so pretty high. Anyone can do this with a propane or MAPP torch and the appropriate flux and solder.

Soft soldering (what we do in electronics) doesn't work well with most thermocouples, but it does for coppper-constantan, so that's a possibility if you don't need to go too high in temperature (the thermocouple itself is usable to 400°C, solder melting point is typically lower). I'm not suggesting this, but you could even connect just a constantan wire to the copper and use a type T signal conditioner (not suggesting it because the copper you have may not be sufficiently pure to not require calibration, not because it won't work as a thermocouple- sometimes they add tellurium or other impurities to aid production procceses)

To avoid problems, you should use a measurement device that works with a "grounded junction" thermocouple. Any battery-powered device will be fine as will most industrial type thermocouple transmitters and signal conditioners (they should say they have "galvanic isolation"). It's also possible to use differential (instrumentation) amplifiers, but I would suggest avoiding it outside a very controlled environment.

Remember that thermocouple signals are relatively low level (tens of uV per K), and if you have a lot of EMI floating around you could have issues. Also the temperature of the junctions inside the measuring instrument must be measured to greater precision than the desired system accuracy because (to a first order) thermocouples measure temperature differences. That means that high absolute accuracy may be hard to achieve, but seeing tenths of degree C change in a lab environment is easy.

The best accuracy will be achieve by using a fine wire gauge since heat necessarily flows in or out of the wires. For most purposes you can say as fine wires as practical is best.

Answer 3

How hot is your sample going to be? For how long?

When testing mechanical seals, I have used a high temp epoxy to attach thermocouples to silicon carbide seal faces and test in applications over 550F. Temperatures during a seal failure would easily exceed 700F. Typically, a failure to get a temperature reading would be the result of insulation breakdown of the thermocouple wire and not the epoxy. I built my own thermocouple junctions and the epoxy also served to electrically isolate the thermocouple as well. The epoxy was not designed specifically for this kind of application and we were often operating beyond its published capability, however, in years of testing, I never observed a failure of the epoxy due to thermal breakdown or other heat effects.

Answer 4

Soldering the thermocouple might not be the best way to go as the joint created by soldering process will affect heating, and therefore your results. What you might considers is attaching thermocouple to the sample using thermoconductive glue/paste (preferably one which is electrically neutral)

Answer 5

I got an idea which probably might not work use a heat conducting rod solder to the sample and caliculate the heat near the sample and at end of the rod which might have cooled by the environment and near rod .then caliculate it on the rod to get the geat approximate by comparison.using weilding might cause potential differences in thermocouple as the electronic density of the sample might interfere