Everything You Know Is Wrong February 2003

Answers to those Doggone Thermal Design Questions

By Tony Kordyban

Copyright by Tony Kordyban 2003

Dear Tony,

I have a question about the proper way to attach thermocouples to electronic components.  Is one type of adhesive better than others?

We recently had a difference of results between our Product Safety Lab and our Thermal Test Lab. The difference was about 10 degrees C when they were measuring the same power supply component.

The Thermal Lab uses thermal epoxy, while the Product Safety Lab uses superglue, which most product safety labs in theUSseem to use.

What has been your experience attaching TC’s to components with different types of adhesives and how much of a difference can be expected? 

All Gummed Up in Wrigleyville 


Dear Gum,

On the wall of my lab is a dog-eared list of general rules, procedures and precautions.  Number 7 reads:

For maximum repeatability of thermocouple measurements, make them only once.

It is  a joke appreciated only by those who struggle to measure component temperatures with thermocouples.  There are lots of sources of error in measuring component surface temperatures with thermocouples, so I am not surprised when you say that two labs measured the same component temperature and came up 10 degrees C apart.  Some common sources of error are:

  • using a too-heavy gage wire thermocouple, so that it is nearly as large as the component
  • routing the thermocouple wire away from the component so that it acts as a heat sink
  • using the wrong calibration curve for the type of wire
  • mistaking stray voltage from the live circuits that get onto the thermocouple wires for temperature signals

If I had listed a dozen more, in order of importance, I probably would not get to “Failing to use thermally conductive adhesive to attach the thermocouple to the component surface.”  I think of it as a secondary kind of source of error.  If you do everything else right, you don’t need thermally conductive adhesive.  If you make other mistakes in attachment, thermally conductive adhesive won’t be enough to overcome it.

That seems to go against our thermal intuition.  To get the best temperature measurement, we want to have the best possible thermal contact between the thermocouple and the surface of the component.  Wouldn’t a thermally conductive adhesive naturally give a much better thermal contact than a typical superglue, which is a thermal insulator?

Figure 1. Glue hold the thermocouple onto the component, but also holds it away sometimes.

The truth is that “thermally conductive” adhesive falls into the range of what we would normally think of as thermal insulators, too.  What they should call them is “not as much a thermal insulator” adhesives.  Look at the properties in the data sheets.  Superglue is essentially a plastic when it hardens, and has a thermal conductivity of about 0.2 Watt/meter/degree C (W/m/C).  The thermal epoxy that I use for attaching heat sinks to components has special fillers to boost its conductivity — all the way up to 0.5 W/m/C.  I have seen some adhesives with  conductivity as high as 1.0 W/m/C.  That is 2 to 5 times higher than superglue, but it is a long way from what we usually think of as a good conductors, like aluminum (180 W/m/C).

That by itself is an interesting revelation, but not convincing.  5 times the conductivity might be important.  Let’s see how the difference in thermal conductivity leads to differences in temperature measurement.

Figure 1 is a close-up view of your thermocouple glued to the top of your component.  I was going to say it is not drawn to scale, but I don’t know.  Maybe you use really big thermocouple wire.  Anyway, the goal is to get a reading as close as possible to the surface temperature that would be there if your thermocouple and blob of adhesive were not there.  To do that you would make the bead and wires and blob of adhesive as small as possible, so they don’t interfere with the normal heat transfer that goes on when you are not measuring.  But at whatever scale you do your measuring, you end up with a bead glued onto the component surface looking something like this.

Now we can write the equation that tells us how the temperature you measure (the temperature of the thermocouple bead) differs from the surface temperature of the component.

T component  – T bead  = (Q t) / (k A)         where

Q is the heat flowing from the component into the thermocouple bead
t is the thickness of the adhesive layer
k is the conductivity of the adhesive
A is the effective cross-sectional area through which heat flows to the bead from the component

Area (A) is some value between the contact area of the blob of glue with the component and the surface area of your thermocouple bead.

The most important things I wanted you to glean from this equation is that the temperature difference is directly proportional to to two things:

Q/A, the heat flow per unit area from the surface of the component, and

t, the thickness of the layer of adhesive between the bead and the surface.  Ideally, this thickness should be zero.  The bead should be in direct contact with the component, and  the adhesive should merely trap it in place.  In practice, t is some not zero, but obviously the smaller the better.

My point is, that if you have made t as small as possible, literally zero, then the conductivity of the adhesive makes no difference at all.  And if Q/A, the heat flux, is small, then the conductivity of the adhesive is not important either.

Let’s try an example with some real numbers to get a feel for this.  I have a 10W component on a circuit board.  5W flows into the board, and the other 5W flows out uniformly from the top surface, which is a 40 mm square copper heat spreader, into the air.  That is a heat flux of about 3100 W/m2.

I attach my thermocouple bead with a blob of super glue that has a diameter of 2 mm, and the thickness of the glue between the bead and the copper heat spreader is a 0.12 mm (about 0.005 inches).  Assuming that the presence of my blob of glue does not change the uniform heat flux (good enough just for this exercise), then the heat flowing into my bead is about 0.0098W.

From the equation above, the temperature difference between the bead and the surface of the heat spreader is about 1.9 degrees C.  If I had used a thermal adhesive with conductivity of 0.5 W/m/C, then the difference between the heat spreader and the bead would have been 0.75C.  The difference between the two measurements would be about 1 degree C.  If I need more accuracy than that, I should force my thickness of glue to be smaller.

Maybe in your application the thermally conductive adhesive would help.  This simple analysis shows you the things to look for to help you decide.  High heat flux is one of them.  But you can reduce error more by controlling other things in your gluing:  make the bead tiny, and make the layer of adhesive between it and the component as close to zero as possible.

I use a form of superglue most of the time.  Because it hardens fast, I can make sure the bead stays where I want it, right on the surface.  If thermal epoxy takes 5 minutes to harden, how do I keep the bead from floating away from the component surface during hardening?  I have to sit there for 5 minutes holding the bead in place.  That is 250 minutes of holding when attaching 50 thermocouples to an assembly, telescoping a 2 hour chore into 6 hours of tedium.  Tiny improvements in accuracy don’t seem so important after Hour 3 of breathing epoxy fumes.

Oh, and by the way, I guess I should have asked you one important question before I got started on all this analysis.  You said the two labs had a 10 degree C difference.  But which lab had the higher reading?  If the difference is due entirely to the type of adhesive, then the lab using the thermally conductive adhesive should have the high temperature.  If it is the other way around, then you’d have to chalk up the difference to something else, like the phases of the moon.


Dear Doc Thermal,

Our Product Safety guys measure component temperatures to get through UL approvals.  I need to measure component temperatures for Performance Evaluation, so I thought, maybe I could just use the safety report and save time and money on all that repeat testing.  But when I compare past safety reports to my own, it seems their temperatures are often much, much higher than mine.  I figure that if anybody is making a mistake in measuring, it must be the one with the lower temperatures.  What am I doing wrong? 

Anxious about Performance

Dear Anx,

First off, thanks for the promotion.  I’m not a Doc.  I don’t have a doctorate or PhD degree.  You can call me Master, if you like, but it won’t improve the quality or authority of my answers.

As far as your discrepancies with the other lab, go talk to the people in in Product Safety and find out.  They’re not dangerous — they’re in the Safety Department, after all.

My guess is that you are measuring the same components, but under very different conditions.  UL doesn’t care if your electronics actually work.  They just don’t want it to kill anybody or start a fire under a fault condition.  So that is mostly what they test — fault conditions.  They measure temperatures of transformer coils while the output of the power supply is shorted, for example.  Or the temperature of a fan motor with the rotor blocked.

You, on the other hand, are looking at component temperatures under actual operating conditions.  Component power is usually lower under normal conditions, which might explain the wide differences between your two reports.

Or maybe they just use a different kind of glue.


Isn’t Everything He Knows Wrong, Too?

The straight dope on Tony Kordyban

Tony Kordyban has been an engineer in the field of electronics cooling for different telecom and power supply companies (who can keep track when they change names so frequently?) for the last twenty years.  Maybe that doesn’t make him an expert in heat transfer theory, but it has certainly gained him a lot of experience in the ways NOT to cool electronics.  He does have some book-learnin’, with a BS in Mechanical Engineering from the University of Detroit (motto:Detroit— no place for wimps) and a Masters in Mechanical Engineering from Stanford (motto: shouldn’t Nobels count more than Rose Bowls?)

In those twenty years Tony has come to the conclusion that a lot of the common practices of electronics cooling are full of baloney.  He has run into so much nonsense in the field that he has found it easier to just assume “everything you know is wrong” (from the comedy album by Firesign Theatre), and to question everything against the basic principles of heat transfer theory.

Tony has been collecting case studies of the wrong way to cool electronics, using them to educate the cooling masses, applying humor as the sugar to help the medicine go down.  These have been published recently by the ASME Press in a book called, “Hot Air Rises and Heat Sinks:  Everything You Know About Cooling Electronics Is Wrong.”  It is available direct from ASME Press at 1-800-843-2763 or at their web site at http://www.asme.org/pubs/asmepress,  Order Number 800741.