{"id":514,"date":"2013-01-26T16:32:30","date_gmt":"2013-01-26T22:32:30","guid":{"rendered":"http:\/\/tonykordyban.com\/?page_id=514"},"modified":"2013-01-26T16:32:30","modified_gmt":"2013-01-26T22:32:30","slug":"everything-you-know-is-wrong-august-2003","status":"publish","type":"page","link":"http:\/\/tonykordyban.com\/?page_id=514","title":{"rendered":"Everything You Know Is Wrong August 2003"},"content":{"rendered":"

Answers to those Doggone Thermal Design Questions<\/strong><\/p>\n

By Tony Kordyban<\/strong><\/p>\n

Copyright by Tony Kordyban 2003<\/p>\n

Dear Thermal Consultant without Equal (or cost),<\/em><\/p>\n

\"phase_change\"<\/a>My problem is a little weirder than the ones you normally deal with.\u00a0 I use a special component that not only needs to get rid of a lot of heat in a small space, but its temperature must be 75\u00b0C plus or minus 5\u00b0C, or it just won’t work.\u00a0 When the ambient is hot, it needs to get rid of lots of heat, but when the ambient is cold, I’d like to reduce the heat loss so it can regulate its own temperature.\u00a0 Why add a secondary heater when it generates lots of heat already?\u00a0<\/em><\/p>\n

I know I can’t use a passive solution like a heat sink.\u00a0 I need some kind of temperature controller, maybe a thermo-electric cooler.\u00a0 But that is expensive, eats power, and takes up space I don’t have.<\/em><\/p>\n

I was looking at this chart from a brochure for a Phase Change Interface Material.\u00a0 It seems to say that when it heats up to a particular temperature, the thermal resistance goes way down.\u00a0 This sounds like just the thing I need.\u00a0 When the temperature goes up, the PCM changes resistance, and lets more heat out to the heat sink.\u00a0 When the temperature is low, the resistance is high, and keeps more heat in the component.\u00a0<\/em><\/p>\n

Can this really be the answer to all my prayers?<\/em><\/p>\n

Sounds too good to be true in Puget Sound<\/em><\/p>\n

 <\/p>\n

Dear Sounds,<\/p>\n

I am glad to find a fellow skeptic.\u00a0 I won’t play the lottery because I know that if I win the big prize I’ll be hit be lightning on the way to collect it.<\/p>\n

There might be such a “thermal valve” material out there in the universe someplace.\u00a0 No physical law prevents such a thing from existing.\u00a0 It’s just that the Phase Change Materials used for heat sink interfaces don’t do what you think.<\/p>\n

\"melting1\"<\/a>They do change thermal resistance when they melt, but only once.\u00a0 They don’t change back to high resistance when they solidify again.\u00a0 Let me explain why.<\/p>\n

This picture greatly exaggerates the\u00a0 nooks and crannies on the surfaces of the heat sink and the component.\u00a0 It is those mismatches between the two surfaces that create the high thermal resistance in the first place.\u00a0 The solid surfaces would barely touch each other at all.<\/p>\n

This is what it looks like when you first put the cold, solid pad of Phase Change Material between the surfaces.\u00a0 It does nothing to improve the thermal resistance, and might even make it worse by adding another layer between them.\u00a0 There are lots of air pockets separating the heat sink from the component.\u00a0 Tiny pockets of air are pretty good thermal insulation.<\/p>\n

Don’t ignore that little yellow arrow on top of the heat sink labeled “pressure”.\u00a0 For the Phase Change Material (PCM) to do its magic, the heat sink has to be free to move, and there has to be some downward pressure.\u00a0 The next picture will make that more clear.<\/p>\n

\"melting2\"<\/a>In this picture the component has been powered on for a while.\u00a0 It has heated up above the melting point of the PCM, and the now grease-like material is free to be smooshed.\u00a0 The pressure on the heat sink squishes it down until it makes hard contact with the surface of the component.\u00a0 The PCM wets both surfaces, flowing into many (probably not all) of the tiny crevices and pores, displacing the air from the joint.\u00a0 It is this wetting of the surfaces which lowers the<\/p>\n

thermal resistance, not the conductivity of the PCM itself.\u00a0 You might even say that the conductivity was “immaterial”.\u00a0 Ha.\u00a0 Ha.<\/p>\n

What happens between the first picture and the second picture is what leads to that sudden drop in resistance in your chart.\u00a0 But what happens when you shut off the power and the component cools below the melting point of the PCM, and the PCM re-solidifies?<\/p>\n

I’d show you a picture of that, but I don’t have to because it would look exactly the same as the previous picture.\u00a0 The PCM solidifies, but it doesn’t pop back to its original shape.\u00a0 It stays tucked into all the nooks and crannies, and the heat sink stays in contact with the component.\u00a0 The only difference is that the PCM can’t flow, until it melts again.<\/p>\n

What is the thermal resistance with re-solidified PCM?\u00a0 The PCM brochure probably doesn’t say.\u00a0 Most users don’t care what the thermal resistance is when the circuit is turned off.\u00a0\u00a0 My guess is that the joint thermal resistance is pretty much the same as when it is melted.\u00a0 The contact resistance stays the same.\u00a0 Maybe the conductivity of the grease is slightly different from the solid version, but probably not by much.\u00a0 It’s not as if it changed to a real liquid, just a gooier solid.\u00a0 If the pressure is good, and the surfaces fairly flat, then the thickness of the material in the joint is very small anyway.\u00a0 In my opinion, the change in joint resistance would not be measurable.<\/p>\n

To solve your real problem — how to make a joint with reversible thermal resistance, all I can think of is that you could attach your heat sink to a motor.\u00a0 When the component is too hot, press the heat sink down hard.\u00a0 When the component is too cold, reverse the motor and pull the heat sink away.\u00a0 A small air gap should allow your component to heat up again.\u00a0 Use a temperature sensor to control the motion of the heat sink.\u00a0 Maybe the interface material should be a foam rubber pad instead of PCM, so it won’t get pumped out of the joint over time.<\/p>\n

Do you think a small stepper motor, some levers and cams, and a controller would cost more than a thermoelectric cooler and its controller?\u00a0 Let me know if my idea works out.\u00a0 I want to be listed on the patent (but not on the lawsuits.)<\/p>\n

\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014\u2014<\/strong><\/p>\n

Dear Tony,<\/em><\/p>\n

I agree with you regarding the need for a large (relative to the electronic box) environmental chamber to simulate ‘still air’, and the use of solar lamps.\u00a0\u00a0 (See June 2003 article.<\/a>)\u00a0 One way around this is to use a very large cardboard box, like from a refrigerator, as your environmental chamber.\u00a0 Install the lamps as far above the test unit as possible. Be sure to use enough lamps to get 1,000 W\/m*2 at the test unit. Install heaters in the big box.\u00a0 By trial and error keep increasing the heat load (heaters) until you maintain 40 C ambient.\u00a0 Be sure the thermocouple measuring ambient near the test unit is shielded from the lamps above. You may have to vent the box by cutting holes here and there to maintain the 40C condition. A somewhat crude approach, but I think it would simulate the environmental conditions better than one could in an environmental chamber.<\/em><\/p>\n

Mr. Natural from Sun City<\/em><\/p>\n

 <\/p>\n

Dear Nat,<\/p>\n

I have seen the kind of “home-made” environmental chamber you talk about in various electronics development labs over the years (although I have never seen one equipped with solar load lamps — those are pretty expensive to be tossing into a refrigerator box held together with duct tape.)\u00a0 It is possible that they were just as good at representing a customer’s environment as an official “environmental chamber.”\u00a0 A refrigerator box can be hard to control, leaky, inaccurate, give unreproducible results, and might even be a safety hazard.\u00a0 But it has the undeniable quality of being cheap.<\/p>\n

I have amused myself over the years during boring meetings by trying to design a workable environmental chamber that could test electronic equipment under “still air” conditions.\u00a0 I wasn’t even trying to solve the problem of producing a realistic solar load.\u00a0 I was thinking about indoor equipment which is supposed to function in so-called natural convection or still-air.<\/p>\n

Everybody has their own idea of what “still air” means.\u00a0 It means that you don’t have fans blowing at your equipment.\u00a0 Enhanced air flow would cool the electronics.\u00a0 But other than the negative definition of “no forced air blowing on the equipment”, how should a “still air” environment be defined?\u00a0 The purpose of such a definition is to come up with a way of testing a product, so that when a customer uses it in a place where the air isn’t blowing around, it still has a pretty good chance of working thermally.<\/p>\n

I never did come up with a good way of defining a “still air” environment.\u00a0 One starting point would be a large room that has no heat sources in it, and no ventilation.\u00a0 In such a room after a long time there would be no air currents, and the air would eventually all settle down to the average temperature everywhere.\u00a0 That would be “still air.” Is there such a room in real life, except in an abandoned building?<\/p>\n

What happens as soon as you put heat generating electronics into the room?\u00a0 The heat\u00a0 from your box starts a plume of warm air rising for the ceiling.\u00a0 That generates a complex, slow, 3-dimensional circulation pattern throughout the room.\u00a0 The air is no longer still, and no longer all at one temperature.\u00a0 And unless you start removing heat from the room somehow (ventilation, conduction), the average air temperature is going to go up, and keep going up.\u00a0 That heat removal mechanism is going to create flow patterns and temperature gradients of its own.<\/p>\n

If you try somehow to stifle these flows and gradients, then you are blocking the very mechanism of natural convection which is supposed to cool your product.\u00a0 That doesn’t seem like a very fair test!<\/p>\n

So, given that a “still air” chamber must have some air flow and variation in air temperature from place to place, how much flow and how much temperature gradient can there be and still qualify as a “still air” environment?\u00a0 And if the temperature is not uniform, where do you define the “ambient” temperature?<\/p>\n

I never came up with an answer to satisfy myself.\u00a0 At least not a single answer.<\/p>\n

I imagine that the “still air” environment is very different for each of these examples:<\/p>\n