heatsinking

[Ted205]

I want to use my alu chassis as the heatsink for 2x lm3886. The chips sit on aluminum that is 1cm thick.

but the chip is getting too hot, the protection kicks in. The chips are the plastic coated type. Will metal backed versions help significantly ? Is the exposed metal the gnd for the chip ?


thanks

[jackinnj]

Use the formula from National Semi's website to calculate the required thermal resistance for the heat sink. For instance, if you have +/- 28VDC on the rails at load, you will need a heat sink with a thermal resistance of approximately 3.9 with the TF package. (The TF package has somewhat worse thermal characteristics than the TA package when you use the latter with a mica insulator.)

You can use this nomograph to calculate the surface area to arrive at the thermal resistance you need.

[AndrewT]

a flat slab of 10mm aluminium is just not enough to cool a pair of 3886 unless driving high impedances from a low supply voltage.
I recommend you double the heatsinking capability that National recommend for their maximum ratings.

Have you checked to see/measure if the chips are oscillating?

[Ted205]

well its an alu box 2 sides (with the chips) 1cm connected to the other 3 (5mm) by 4 corner posts. The complete box is supposed to be one large heatsink. Think i'm going to have to rethink it

[MartyM]

I agree with the above guy in that you need to double the size of the heatsinks shown in the design examples from National.

Alternately, you could do what I did, and instead of giving up of the large amount of space required, use DC cooling fans, slowed down to reduce noise.

But this is in a car amplifier application. I got surplus fans cheap from eBay and just didn't want to get more heatsinks.

[jackinnj]

Originally posted by AndrewT
I recommend you double the heatsinking capability that National recommend for their maximum ratings.

a 10mm slab is about .394 inches, so 12 sq inches (774 sq mm) should do the trick, barring the potential problem to which Andrew refers.

I've found that the National thermal resistance ratings are reliable -- could be that the chip is oscillating -- try putting a 200pF capacitor between the input + and - and see if the thermal issue goes away.

[jackinnj]

Originally posted by MartyM
I agree with the above guy in that you need to double the size of the heatsinks shown in the design examples from National.

Demonstrate that mathematically.

[ratza]

There's no need for a demonstration. It's a well known fact that there is no such a thing as a heatshink which is too big.

[AndrewT]

I can't demonstrate that because National choose not to divulge the way Spike protection operates when the chip is at elevated temperatures.
The heatsink data is derived for Tc=150degC, but the performance data is only shown for Tc=25degC. Therein lies the problem.
Basically the datasheet and application notes give sufficient clues to keep the chipamp cool to get good performance from it.

[Ted205]

does anyone know of a good resource for large heatsinks about 30cm x 70cm