Sticking a thermocouple junction to a FET case so as to measure its temperature

Thanks i take it this is something to do with "cold junction compensation", or thereabouts.
I think we will strip our thermocouple measurement ends of plastic sheath, then kind of twist the two metal ends so they are together, then solder them so they stay together....but the other day, i had a real problem trying to get a blob of solder to stay on them. I hope they havent changed the metal types?
I am speaking of the metal ends that do the temperature measurement.

Type K (NiCr/Ni) wires can't be well soldered with usual flux. You need either special stainless steel solder or phosphoric acid as flux.

Spot welding is the preferred method, brazing with silver alloy works too.

Ouch!

No don't do that. The voltage is produced by electron migration between the two dissimilar metals. If you solder them you add an alloy between them. It will still produce a voltage but it will be unpredictable. The wires have to be spot welded or fused together with no intermediate material. Bear in mind we are talking about microvolts being generated into a high impedance load, you need to amplify the voltage considerably to make it usable so adding a few uV in here and there is out of the question. Besides, the types of wire will not bond to conventional Sn/Pb solder. It's been a while since I worked with thermocouples but to bond them to a PCB I remember having to use a special solder and connect them directly to the input pins of an AD595 Cold Junction compensated amplifier. It might have been an Al based solder or pure Pb, I'm afraid I can't remember exactly. I recall the direct soldering to the pins was essential to avoid a temperature difference between the IC and the solder joints or it gave a wrong reading.

Brian.

Thanks for the answers, we will use a off-the-shelf thermocouple reader to measure the thermocouple temperature.
We are still in some debate about the method of fixation of the thermocouple junction to the FET...."thermal glue" or "super-fast glue"

I am referring to the 4th thread of this
https://www.edaboard.com/threads/184341/

….The thermal glue that I used to use is now obsolete, will it be OK to just glue the thermocouple end to the FET case with Blak-Tak super glue (Loctite 4105)? After all, if the thermocouple junction is against the FET case then the thermal conductivity of the glue isn’t really going to matter…there's not going to be a massive temperature difference across a few micrometres of low thermal conductivity glue.
Proper thermal glue also takes hours and hours to cure, and often the thermocouple “falls off” whilst curing due to tension in the thermopcouple lead.

Loctite 4105 datasheet:
**broken link removed**

Also, what do you think is T(jc), ie to the plastic bit of the case, for a typical T0247 FET?

Betwixt I am grateful for your comments about thermistors…..i like the idea and will seek out some suitable types for looking in to…..what I like about thermistors is that they are electrically isolated from the T0247 metal case even if the thermistor body is glued to it…..this will avoid the problem that we sometimes get with thermocouples whereby the noise of being glued to a metal switching node, say, actually feeds noise into the thermocouple measurement equipment and even off-the-shelf thermocouple readers suffer considerable resultant inaccuracy.
As discussed, here, thermocouples can be glued to the plastic case of a T0247 but then the reading is likely to be anywhere from 10 to 20 degrees or so lower then the junction temperature of the FET, which gives poor accuracy.

No don't do that. The voltage is produced by electron migration between the two dissimilar metals. If you solder them you add an alloy between them. It will still produce a voltage but it will be unpredictable. The wires have to be spot welded or fused together with no intermediate material.

Click to expand...

.
Thanks, the problem is that we cant afford a spot welder, and cant afford to buy thermocouples where they are already spot welded. As we discussed, soldering is not optimal for joining them (wont make the thermocouple junction), and in any case, as discussed, regular Pb/Sn solder doesn’t even work in joining the thermocouple wires. The solder simply wont "take" to the wires.
As such, what do you think about simply twisting the thermocouple wires tightly together, and then glueing this “twisted end thermocouple junction” to the D2PAK metal tab?
Also…
To Glue …or Not to glue…….
We wish to measure the case temperature of our D2PAK FET on our PCB. The contractor says that all we need to do is rest the thermocouple junction on the tab of the FET and simply measure the temperature like that. However, surely a dot of super glue is needed (Loctite 386) to glue the thermocouple junction to the metal tab? If no glue is used, then there is slight airflow over much of the thermocouple junction region and this will result in a reading that indicates its less hot than it actually is?

You need either special stainless steel solder or phosphoric acid as flux.

Click to expand...

Thanks..regarding use of phosphoric acid flux to facilitate soldering together of k type thermocouple wires when using standard 60/40 Sn/Pb solder for electronics, do you know of any cheaper vendors of phosphoric acid flux?
The cheapest we can find is £32 /bottle...
https://www.somersetsolders.com/phosphoric-acid-flux-for-stainless-steel-soldering/p291
..we would tightly twist the thermocouple wires together before soldering them, since we would want to minimise the situation of another (third) metal junction which would otherwise make the readings more inaccurate.

cant afford to buy thermocouples where they are already spot welded.

Click to expand...

You get welded wire thermo elements with connector for < 10 EUR at catalog distributors.

As we discussed, soldering is not optimal for joining them (wont make the thermocouple junction).

Click to expand...

I've read this, but it's not true according to my understanding of thermoelectric effect. Placing a third metal between the two thermo element wires doesn't change the voltage as long both junctions have the same temperature.

Thanks,

And do you have an answer for our contractor who insists on just resting thermocouple junctions against a D2PAK drain tab, rather than glueing it?.....

To Glue …or Not to glue…….
We wish to measure the case temperature of our D2PAK FET on our PCB. The contractor says that all we need to do is rest the thermocouple junction on the tab of the FET and simply measure the temperature like that. However, surely a dot of super glue is needed (Loctite 386) to glue the thermocouple junction to the metal tab? If no glue is used, then there is slight airflow over much of the thermocouple junction region and this will result in a reading that indicates its less hot than it actually is?

treez said:

And do you have an answer for our contractor who insists on just resting thermocouple junctions against a D2PAK drain tab, rather than glueing it?.....?

Click to expand...

The basic idea is (i) the two metal junction must be at the same temp (ii) what is important is that there be electrical continuity.

You can join one wire at the tab at one point and another wire at another point on the tab and the result will not differ IFF the whole tab at one const temp.

If the wires are reasonably fine they will present a very small thermal load and you can just touch them to the tab to read the temp

If you place the thermo couple on the drain tab, you usually need high voltage isolation. Or a temperature meter that can accept superimposed switching voltage.

"Loctite 368" is no active Loctite product, typo?

"Loctite 368" is no active Loctite product, typo?

Click to expand...

Thanks sorry i meant Loctite 382. The point is, that if the thermocouple junction is just resting against the metal drain tab, (instead of being glued to it), then there may be air currents flowing over the thermocouple junction which will slightly cool it and give a slightly inaccurate reading. As such, a small spot of glue is needed around the thermocouple end to attach the thermocouple end to the metal drain tab.?

If glue is not used, then you have just a point contact of the thermocouple end to the metal drain tab, and the air that is slightly flowing around the rest of the thermocouple junction will act to cool it, and give inaccuracy in reading?

Loctite 382 and activiator 7455...
https://uk.rs-online.com/web/p/cyanoacrylate-adhesives/1987144/

treez said:

TIf glue is not used, then you have just a point contact of the thermocouple end to the metal drain tab, and the air that is slightly flowing around the rest of the thermocouple junction will act to cool it, and give inaccuracy in reading

Click to expand...

You have a valid point.

I suggest some rule of thumb: use or abuse it at your own discretion.

1. The error due to the heat conduction by the thermocouple wire will be determined by the temp difference (between the ambient and the tab) and the relative heat capacities (of the thermocouple joint and the tab).

2. The best results will be in the steady state; you will need proper care if you want to measure rapid changes.

3. When considering air currents and cooling due to air currents, please remember that the layer in touch with flat surface is always at rest. When fixing the contact point of the thermocouple, please ensure that the surface remains as flat as possible.

4. Turbulent and laminar flows offer different cooling.

Thanks, and do you have any thoughts on using a dot of glue to glue the thermocouple to the metal drain tab, as opposed to just resting the thermocouple end on the metal drain tab of the FET.
We ask because the Loctite 382 glue and activator 7455 costs £30 for 25ml bottles....because of this huge price, we would like to evaluate not using glue at all....instead just kind of resting the thermocouple end on the metal drain tab of the FET.

treez said:

I must admit, we are looking for the cheapest superglue, and the cheapest activator which will immediately (within a few 10’s of seconds) cure it….but the entire Loctite website does not allow you to find what this combination is….Do you know it?

https://www.loctiteproducts.com/epoxies.shtml#multi-purpose

We must have the glue and the activator in separate tubes/bottles because otherwise you just end up with the applicator nozzle getting jammed.

Click to expand...

1.

I agree about JB-Weld (see post #2). I used JB-Weld years ago to repair a broken aluminum valve cover on our Toyota. It's like steel epoxy in two tubes.

2.

Silver solder works on some substances that won't take lead solder.

3.

The flux can make a difference in effectiveness. Years ago an aluminum welding kit came out containing aluminum rods and special flux. The inventor claimed the flux was the key to success.

treez said:

Thanks, and do you have any thoughts on using a dot of glue to glue the thermocouple to the metal drain tab,....

Click to expand...

Most epoxy glue takes hours to set but the time can be reduced if you increase the hardener proportion (double the amount and it will set in 15-20 mins with reduced strength).

My suggestion: just use any common superglue (instant glue; cyanoacrylate based) but if you want epoxy (most contains lots of filler) you use a heat gun and it will set in 30 sec with some reduction in initial strength.

If you still want to solder the thermocouple end, electroplate the weld point with copper or tin (only the tip) and then you can solder it as usual.

Most epoxy glue takes hours to set but the time can be reduced if you increase the hardener proportion (double the amount and it will set in 15-20 mins with reduced strength).

Click to expand...

Thanks, we intend to use Loctite 382 with 7455, and it sets in seconds. We can hold the thermocouple against the metal drain tab, and it will set like that.
**broken link removed**

If you still want to solder the thermocouple end, electroplate the weld point with copper or tin (only the tip) and then you can solder it as usual

Click to expand...

Thanks, I thought we could just use standard electronics 60/40 Sn/Pb solder and phosphoric acid flux?


_________________________________________________________________________________________________
******* To Glue, or Not to glue? ********

Our contractors don’t want to use glue to fix the thermocouple end to the metal drain tab, they say its allright to just rest the thermocouple end against the metal drain tab. What are your thoughts on this?

Our contractors don’t want to use glue to fix the thermocouple end to the metal drain tab, they say its allright to just rest the thermocouple end against the metal drain tab. What are your thoughts on this?

Click to expand...

You didn't yet comment on the high voltage capability of your thermo couple measurement circuit. Most people will prefer an isolated thermo couple mount.

Thanks, I thought we could just use standard electronics 60/40 Sn/Pb solder and phosphoric acid flux?

Click to expand...

It works for me. You can also consider to crimp the thermo couple tip into a small wire-end sleeve and solder the latter.

You didn't yet comment on the high voltage capability of your thermo couple measurement circuit. Most people will prefer an isolated thermo couple mount.

Click to expand...

The FET is a linear regulator FET in an offline linear regulator which feeds from the rectified DC bus, which doesnt have a capacitor on it.....The drain tab can go up to 373V.

Do you see problems in just resting the thermocouple end on the metal drain tab instead of glueing it?

Some of the instant glues (cyanoacrylate types) do not wonk above 200C- they may come off. All the epoxy glues are thermosets- they will not melt.

Use a battery operated temp unit to avoid the high voltage problem. (see #36)

Our contractors don’t want to use glue to fix the thermocouple end to the metal drain tab, they say its allright to just rest the thermocouple end against the metal drain tab. What are your thoughts on this?

Click to expand...

For metals, both electricity and heat are conducted by electrons (mostly) and if the electrical connection is good, the thermal connection should also be.

QU: Our contractors don’t want to use glue to fix the thermocouple end to the metal drain tab, they say its allright to just rest the thermocouple end against the metal drain tab. What are your thoughts on this?
ANS: For metals, both electricity and heat are conducted by electrons (mostly) and if the electrical connection is good, the thermal connection should also be.

Click to expand...

Thanks, but if the thermocouple end is not glued to the metal drain tab of the FET, then might there be air currents around the thermocouple end which will cool it down and give a falsely low temperature reading?
…..Whereas if the thermocouple end is glued to the metal drain tab of the FET then at least there are no air currents running around the metal end of the thermocouple, as the glue molecules are stationary, unlike air.

As such, my question is.....is it acceptable to merely rest the thermocouple end against the metal drain tab of the fet when doing a temperature measurement? (as opposed to actually glueing the thermocouple end to the metal drain tab)

treez said:

As such, my question is.....is it acceptable to merely rest the thermocouple end against the metal drain tab of the fet when doing a temperature measurement? (as opposed to actually glueing the thermocouple end to the metal drain tab)

Click to expand...

This is a difficult question to answer; it depends.

If the wire is fine and the tip is very nicely jointed, yes, you can use this tip to measure the temp of the tab. The temp of the joint is what is recorded and not the temp of the wire which gets cooled by air flow.

Your original pic (#1) shows a nice weld connecting the two wires. If the weld point is less than 0.2-0.3 mm dia, I guess the error will be less than 1C (just a guess) and is satisfactory for your purpose.

You can just test it by touching the thermocuple tip with a glass test tube containing some ice. It should not take more than 1-3 sec to get the temp.

About air currents: they are present with and without the thermocouple bead touching the tab. The disturbance caused by the thermocouple wire may be small.