Thermocouple attachment on the heatsink / case

[homoly]

Hello,

I have done some measurement of case temperature of 500A SCR comparing on thermocouple placed in the hole drilled inside the case and one attached on the case.

The probes were J-type soldered to cable lug attached to case with superglue ( it was told to me the usage of some pad is best for the thermal transfer ). The measured difference was around 6.5C between two probes at 70C on case TC. After I have isolated the cable lug with TC thermally with paper tape from outside the difference have decreased to 4C.
I am not fully concern about the super glue usage. Is it some better way to attach the TC to the metal surfaces ( maybe without cable eye, which cause some problem with its thermal inertia )?

I would appreciate also some links refering to this problematic.

Gabriel.

 

[itsmoked]

Spot welding is sometimes done.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[homoly]

Keith,

Thanks for the advice. I think in the case of aluminum heatsink this would be a benefit and would give better thermal bridge like the insertion of the TC in hole, however the spot welding of the aluminum is a little bit tricky according my colleagues.

Unfortunatelly ,for me this is also destructive method as the quality dep. would not allow to pass me such heatsink to customer. And also I would not risk to spot weld to the case of thyristor .

For the heatsink I think I could try to put mica pad or silicone grease under cable eye ( and I will try to try some thinner copper pad like cable eye ) and put a plastic plate from outer side with 2 holes and I will screw the plate through hole to the heatsink and thus make a tight connection of the cable eye+TC with the heatsink.Maybe also add paper or plastic tape isolation additionally.

For a semiconductors cases I think I could find some material which should be applied from outer side, should be thermal isolation and after hardening on air should shrink and thus press the TC pad against the case. But at this moment I have no idea what this material should be

 

[mcgyvr]

I just use thermocouple securement paste (this is the best I've found)

http://www.productsafet.com/products/tempa.html

and a piece of kapton tape to hold the tc bead in place as the paste hardens..

Omega has a 2 part paste but it's not nearly as easy to work with and doesn't hold as well as the stuff in the link I provided..

 

[homoly]

Thanks mcgyvr,

This is something I was looking for and hoped it exists . I will try to get a sample of this material and maybe I will come back with the results later on.

Of course I will welcome also other ideas.

 

[IRstuff]

> Everything you add will have thermal losses, particularly something with low thermal mass and poor thermal connectivity with the substrate being measured. You must have a solid thermal connection between the TC adn the substrate, otherwise, any thermal losses will manifest as a thermal drop across the connection joint.

> You need to minimize the losses introduced by the TC. Some sort of thermal insulator around the outside of the TC to prevent convection loss to the air. If you think of everything as resistances, you can see that surface vs. thermowell introduces additional losses that affect the measurement.

TTFN

FAQ731-376

 

[itsmoked]

homoly; I am not following your last description but do not put anything between the SCR and it's mounting plate or it, then you, will regret it.

Is this just a one off test or a continuous production test?

Most things you contact the SCR will cause multi-degree changes - Heisenberg Principal with a vengeance.

Use a non-contact thermocouple. They are quite accurate. It will let you skip all the mounting headaches and issues.

http://www.mcgoff-bethune.com/manufact/exergen/Precalibrated/precalmodels.htm

Omega.com has them too.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[homoly]

Keith,

It was not my intention to put the plate under SCRs mounting plate, just to put a mica plate between the case ( outer side of the SCR case - perimeter, not the part mating directly with the heatsink ) and the cable lug when the TC is soldered. I admit it was written in the confusing way

Anyway thank you for the tip, i will have a look at this non contact TC method.

This test would be used on "slightly" modified units, the prototype would be measured still in case/heatsink mating area with drilling.

Lets say for slightly modified units the -3,4C would be OK.

Gabriel.

 

[mcgyvr]

Be very careful with IR/non-contact measurement devices. If you don't have the correct emissivity value for the material you are measuring you can be off way more than 3-4 deg C.