Question about Thermocouples

[ryandias]

This is probably going to sound a bit Dumb, but how do you propperly apply a thermocouple to a device you want to get your temperature meaurements from?

Our test engineer left us, and I am trying to "fill the void."

I am trying to measure the temperature of a FET mounted on an IMS substrate. Three thermocouples will be use, on on the FET tab, one below the ims/fet, and one off to the side.

Suggestions were;
- with krazy glue
- with kapton tape (no grease)
- thermal adhesive (thermoset)

Is there a "propper" way to do this?

 

[IRstuff]

Thermal adhesive sounds the most plausible, although Krazy glue can work as well. The key is to minimize, but fill in, the gap between the sensor and the unit under test.

The tape would be a last resort, since it does not fill in the gap and can lead to a poor thermal conduction path to the sensor.

TTFN

FAQ731-376

 

[sreid]

The FET Tab is the Drain and is probably electrically hot. Ususally, thermocouples need to be electrically isolated.

 

[reidh]

IF you are concerned about transient temperatures, you need to ensure that the adhesive is thermally conductive.

Omega sells some thermocouple adhesive, Cotronics makes a high temperature thermally conductive adhesive that will work as well.

-Reidh

 

[tkordyban]

As in most engineering methods, there is no one adhesive attachment method that works best under all circumstances. The general idea is that you want the metal bead of the thermocouple to be touching the surface you want to measure the temperature of. Minimize the amount of adhesive between the bead and the surface, as in ZERO adhesive thickness. The purpose of the adhesive is to hold the bead in place, not fill in the gap between the bead and the surface. Thermally-conductive epoxy is still pretty much a thermal insulator, compared to metal (k -epoxy = 1 W/mK while k-aluminum = 180 W/mK).

I like to use instant adhesive, such as Crazy Glue (my preference is Loctite Tak Pak), because it hardens in less than a minute, so you can glue down the bead quickly. If the glue takes 5 minutes to harden, you have to hold the bead in place for 5 minutes, to make sure it does not drift away from the surface you are trying to measure.

Crazy Glue does not hold up under temperatures approaching 100 deg C, so if your coomponent will get that hot, you may not be able to rely on it. The adhesive will turn to powder and the thermocouple to break loose. For high temperatures, you probably have no choice but to use a high-temperature rated adhesive.

If you are mounting the thermocouple on a solderable metal tab, you could directly solder the bead to the tab. That will give you an ideal thermal connection.

Kapton tape (or other tape -- Kapton itself is not the problem) is not reliable. The adhesive loosens with temperature and time, and your thermocouple will come off during the test. It won't necessarily look loose. There may be only a tiny air gap under the bead. But that will be enough to mess up your measurement.

When attaching your thermocouple to live metal conductors, such as the tab of the FET, you must make sure you will not conduct that voltage into your temperature sensing device. Otherwise you might damage the meter. Most meters and data loggers have isolated inputs to reject common mode voltage, but this is something you have to be carefull with. I have blown up one meter this way. One way to protect your meter from dangerous voltage is to put down Kapton tape between the thermocouple bead and the metal tab. Then glue the bead to the top of the Kapton. There will be a small error due to the thickness of the Kapton, but it should not be too large if the Kapton is very thin.