Thermal conductivity of silicone

[MacGyverS2000]

I intend to encapsulate some electronic components (4 resistors) in silicone. Power dissipation for each will be in the area of 100mW. The material I intend to use has a thermal conductivity of 0.2 W/mC (they specified it in C rather than K, but no biggie).

While looking at charts on the web, I came across a few interesting figures...the thermal conductivities of Aluminum (typical material for heatsinks) at 205 W/mK, ice at 1.6 W/mK, and air (0 degrees C) at 0.024 W/mK.

If I'm reading that chart correctly, the silicone will conduct heat away from the resistors better than cold air, correct? If that's true, then this is the stuff I need to be using rather than letting the components hang in the open air.

 

[sreid]

Electronics in free air is cooled primarilly by thermal convection. This removes much more heat than thermal conduction through the air alone. Silicone will be an insulator. A good book on the subject is "Cooling Techniques for Electronic Equipment" by David Steinberg

 

[IRstuff]

Consider it as a thermal resistance network. Any additional series resistance will potentially increase the junction temperature. Rjc+Rc+Rcs+Rs+Rsa vs. Rjc+Rc+Rca
j=junction
c=case
s=silicone
a=air

The only time the series resistance works in your favor is with a heat sink, since the series resistance of the aluminum is mitigated by the fact that the convection coefficient of the heat sink is significantly better than that of the IC package itself. Thus, the net series thermal resistance is lowered, even though the immediate resistance is increased somewhat.

TTFN

 

[MacGyverS2000]

I figured I wouldn't be getting a free lunch out of this one

As I mentioned before, each package will be dissipating around 100mW from a standard surface mount package (0805 size). While not a painful amount of heat, it can still be a bit uncomfortable. The silicone should insulate the "user" from direct contact anyway, but I'm more worried about the effects of the resistor's health due to a lack of heat dissipation.

Can either of you provide an idea and/or web source that would help me determine the derating needed for the resistors? The ones I'm currently using (in open air) are 1/4W, but if I need to derate them further I need to know now.

 

[sreid]

First, I believe that most 0805 size SM resistors are rated at 125 mw. Second, I (and I believe many others) generally don't use resistors abvove 50% of their power rating except in very specific circumstances. The reasoning is as follows;

1) Derating to 50% is the rule for many companies.

2) I don't know how the maximum power rating for a resistor is determined by the manufacturer but I suspect it is for one resistor mounted on a vertically mounted PC board with the resistor attached to fat traces.

2) My experience is that resistors running at or near their power rating are quite hot, hot enough to eventually scorch the the PC board and hot enought to eventually crack the solder pads from thermal cycling.

3) The thermal environment for the resistor is not exactly known. It may be located near another hot device or shrouded from air flow. In your specific case, the four resistors are going to be right next to each other.

 

[MacGyverS2000]

My mistake...the 0805's I'm using are for BJT base drive (yes, they're 1/8W). The four resistors in question are actually 1206, 1/4W. Since I'm only using <100mW on each (in free air), I have properly derated them. Things change, obviously, when I go planting them in a few millimeters of silicone.

I suppose I could move up to 2010, 1/2W devices, but would the minimal change in surface really make that much of a difference when still buried under silicone? I would prefer to stay away from the larger (and more expensive) packages unless I have to. The resistors will reside on oppositie ends of the board in pairs, so there will be some heat sharing between two devices, but other than that they will be relatively free from other heat-producing components.

 

[sreid]

You are probably just fine with the 1206's at 100mw each even under some silicone insulation especially if you can hook up the resistors using, say, 0.050 traces for couple of inches to act as additional heatsinking.

Exactly why do you feel the need to cover the resistors with silicone?

 

[MacGyverS2000]

LOL, actually, the traces ARE 50 mils (what are the chances, right?) This circuit will be out in the environment and possibly abused, so it needs all of the protection it can get. Potting it will help with the elements getting to the circuit and hopefully provide a modicum of physical shock protection.

Is your recommendation to stick with the 1206's from experience? I will be running some experiments on my own to make sure it's OK, but I'd like to run as few tests as possible. I also feel it will be OK, but with this kind of money at stake (going into production), I can't afford to make a mistake (literally)...lots of tests to follow.

Thanks!

 

[sreid]

How much to thermally derate electronic components can be an endless discussion. I work in power electronics and we often use components to 75 or 80% of their ratings worst case because we have to to be cost competetive. But for low cost items we try to stay under 50% to limit the number of stressed parts. Having a derating requirement also makes you check that you are not over dissipating some part.

Some engineering organizations follow the rule "It's too hot if you can't put your finger on it." In your case, going to 8 paralled 0805 resistors would only be, say, 4 cents in parts cost.

You also seem to be saying that you are going to "Conformal Coat" the PC board for environmental reasons. May I suggest Humiseal, it is the most commonly used conformal coating.

http://www.humiseal.com

 

[MacGyverS2000]

OK, a bit more on my specific application. This is a flex circuit, so I can't spread too many components over the surface, else flexibility is sacrificed. I have chosen to put a few components at either end of the circuit and place a small rigid piece underneath them for structural rigidity.

Also, the conformal coating needs to be transparent as there are LEDs involved. One of the engineers at Dow Corning has suggested a few silicones to try, but I'm currently having difficulties getting one of their local reps to return my calls. I would like to pick up some sample silicones to try before committing to a production run. I figure a test run at max current for 24-48 hours should help me determine if the heat is going to be a problem.

I'll be sure to check out what products HumiSeal has to offer.

 

[Istre]

An additional resource for those who are unfamiliar with the hand-waving and praying that thermal systems often induce is a book titled "Hot Air Rises and Heat Sinks: Everything You Know About Cooling Electronics is Wrong" by Tony Kordyban. It's not a rigorous treatment of cooling electronics, but rather a collection of anecdotes from a veteran thermal engineer working around other engineers that didn't have a complete picture of thermal engineering. It nicely outlines some of the pitfalls and unknowns of the field that must be considered, all in a humorous manner.

Anyone trained in thermomechanical or thermal engineering knows that there is rarely a single answer to any cooling problem, as there are a million variables (conduction vs. convection, free convection versus forced, radiative concerns, fluctuations in ambient temps, time constants, fluctuations in device performance over time, etc., etc.) that can make a rough rule-of-thumb estimate wildly divergent from what happens in reality. This is why you see the huge derations and factors of safety in these designs. This book just might make you think of something you've overlooked, and at the very least, it should make you chuckle.

 

[MacGyverS2000]

I like books that show real-world incidents/stories...I'll have to pick it up.