Effect of natural versus forced convection on thermocouple reading

[chadupperbody]

Hi.
I'm designing a box to do thermal testing on some electrical products at elevated temperature (85C) under natural convection. Up till now the products have been tested under forced convection, i.e. free within a forced convection environmental chamber. However this environment isn't representative of what they'd experience in real-life, i.e. a static air enclosed environment where the only cooling mechanisms are natural convection and radiation.

What I've done so far is place a plywood box into a forced convection environmental chamber. Thermocouples have been placed in air near 5 faces of the box, both inside and outside - 10 total. These thermocouples are there purely to measure the air temperature and see if it's possible to obtain a steady temperature within the box. Once this is done, I'll introduce the products and see how that works out.

My question is: Given enough time to reach steady state, will the thermocouples on the inside of the box read the same temperature as the outside of the box?
Or will the fact that one set of thermocouples is in moving air, whereas the other is in static air result in a difference in readings?

Also, now that I've explained what I'm trying to do maybe someone has come up with a similar problem in the past. So another few questions...
- Should the box have high (metal) or low (nylon) thermal conductivity?
- Should there be holes in the box to allow some airflow through the box, i.e. at a level that doesn't introduce forced convection as a significant heat transfer mode?

Thanks.

 

[IRstuff]

If it's a fully enclosed box, then it'll be pretty close. A temperature delta must require a heat flow and if there's nothing inside the box, the source of heat is from outside, and the only issue would be that one face of the box is probably sitting on the floor of the chamber, while the other 5 sides have access to air flow.

Is the intended application location actually a wooden box? If not, then you may overstress the hardware, given that the intended location is probably inside a metal box, which has a certain level of heat transfer through its walls. Unless you're doing a stress test, you need to account for the differences. Holes might do that, but you need to run the calculations.

TTFN
faq731-376

 

[ko99]

Hi Chad, I've run into this problem before. Forced-convection environmental chambers give misleading results. Attempts to fix them with blockages, etc, often only make things worse. I suggest you consider a natural convection thermal chamber.

Another option, which I've resorted to a few times, is to create a CFD model of the DUT inside the chamber and compare the results to a CFD model of the DUT in it's worst-case environment. Then you can be smarter about making modifications.


ko (

http://www.MSthermal.com)

 

[tkordyban]

I have had to deal with this problem a lot over the years. It is true that the circulating fan in an environmental chamber can actually cool a test unit that is supposed to be tested in "still air", and reduce its temperature enough to make it appear to pass the test. Without the circulating fan, the test unit fails.

My solution to this problem (I have never found a source for a "natural convection chamber") is to use the biggest chamber I can find, much larger than the test unit. Put the test unit inside, run it for a long time at the desired temperature, then turn the chamber controls off. If the air volume in the chamber is large enough, its temperature will not change too much. Allow the test unit to come to a new thermal equilibrium while the chamber circulating fan is turned off. That temperature reading will be as close to a "still air" environment as you can get.

Ideally the chamber would be gigantic. I hear the Pontiac Silverdome is not being used for anything these days, and is climate controlled...