Heat dissipation in a circuit 1

Heatsinking is definitely a must. Or at least heat spreading.

I've got a camera in the lab that just happens to consume almost exactly 6W in a case that's a 1.5" cube. It's barely warm to the touch.

TTFN

MacGyverS2000,

It is not the watts alone that matters, it is the watt density. For the same watt output, increasing the surface area lowers the watt density. I do not know the specifics of what is tolerable, but I do know one application formula that I use all the time to explain a common practice on space heaters.

Lets say I want a strip heater in a box, and have 120V to power it. It I buy a 6" x 2" (12"sq., 2 sides = 24sq&quot strip, the smallest is rated 100W at 120V. I then have 100W over the 24sq" surface, or 4.17 W/sq". I know from experience that this is hot enough to cause at least 2nd degree burns. If I take a 6" x 2" heat strip rated at 250W, 240V (1.04A), the watt desity is 10.4 watts/sq". The surface is potentially hot enough to cause 3rd degree burns. The resistance of that heater is fixed at 231 ohms (R=E/I). If I connect the same heat strip to my 120V supply however, the wattage drops to 62 watts (P=Esq/R). So at the same 24sq" surface, the watt density drops to 2.6 watts/sq", mildly warm to the touch. So from this I know that 2.6W/sq" is a tolerable level. So with 8W to dissipate, if you can do something to increase your surface area to roughly 3sq", your in!

Hope that helps.

Quando Omni Flunkus Moritati

These posts are highly reassuring for future tests, then. I'm basing my past experiences on a (roughly) 1.2W dissipation in a 2W resistor. This measures 1/2" long by 1/8" diameter, so I suppose you could say the Watt density was quite high. After running for 10-15 seconds, it would definitely be too hot to touch any longer than a quick tap.

I'm looking at using a 12V regulator on a 12.5-14.5V supply line pulling 3A, give or take. So, I'm now left wondering how warm the plastic box will get if I use a small heatsink on the regulator. The power dissipated is about 5-6 times the resistor from the past, but the surface area of a TO-220 package combined with a small heatsink has got to be 10-20 times more.

Does this sound about right to you guys?

Can you make one surface of your box aluminum? The problem with plastic enclosures is the dissipation qualities of the surface. Some approach being insulators rather than dissipators of internal heat. You can get resistors that already are imbedded in a heat sink, but if the heat is still trapped inside your plastic box without vents, you still have a problem.

Quando Omni Flunkus Moritati

No chance of aluminum. I will run some tests in the coming weeks to see just how hot a TO-220 packaged regulator will get when dissipating 7-8W. If it's warm, but not hot, I'll probably go with that method and make sure the case has vent/holes in the necessary spot.

I decided to get a feel (no pun intended) for what kind of Watt density the resistor had. If my calcs are correct, it was something in the area of 20W per square inch. The typical heatsink for a TO-220 package is in the area of 2-3 square inches, so I'm hoping that will be enough.

Does this sound about right to you guys?

A small TO-220 heatsink usually have a thermal resistance of something like 10-15 deg C/W. (Centigrade !)

Using such a thing dropping 8 W, should give you a heatsink temperature close to if not above that of boiling water !
That is, if you do not loose any energy in transporting the heat from the resistor/regulator to the heatsink.

A regulator (LM317 or sim.) in a TO-220 package (1.5 Amps max.) usually have something like 3 deg. C/W (junction to case) and requires you to work with an internal (junction) temperature below 125 deg C. (approx. 255 F)

If we say ambient temp. is around 40C (100 F)
(I understand we may be dealing with something for a car
then you need to look for a heatsink no smaller than 7 deg.C/W. And it will still be like putting your hand into boiling water !

A heatsink at 1 deg.C/W (quite big !) and an ambient temp. at 40C should give you a surface temp. arround 48C (118 F)
Closing in on the 'pain threshold'.

With an 8 W drop you would probably have to go to a TO-3 housing. (Internal thermal res. at maybe 1-2 deg.C/W)
I have seen 5A regulators in that housing.

The efficiency of a heatsink can, however, be dramatically enhanced by using a small fan to increase the flow of air over it. If air is allowed in and out of your plastic box that is !


Oops. Just came to think about it. Most positive linear regulator chips require a minimum voltage drop of 2.5 V in order to work propperly !

I believe I have stepped away from the linear regulator option. Min voltage drop (on a reasonably priced unit) is generally around 0.8V+, so that alone drops the idea. A fan is not really an option, even though air is allowed in/out of the case.

Thanks for the info, guys. I'll keep plugging away at ideas, and if any more questiosn pop up I'll be sure to post them.