LM3886TF Heat Sink Insulation

[Andrew Eckhardt]

I mean if you think you're going to run one channel that close to the edge you're better off with a parallel set or bridge parallel.

 

[AndrewT]

You get about 2 degrees per watt out of the epoxy coated case. If you get .15 additional out of some paste with 1 inch pound on a tab mounted screw, and .12 out of a clamping bar with 5 inch pounds on each screw, what happens to the 2 degrees C of the plastic?

the extra 1C/W of the TF is a problem and it shows as increased operating temperature.

To ignore potential improvement in Rth c-s is burying your head in the sand.

 

[Andrew Eckhardt]

Of course.

.03/2= 1.5%

Two LM3886TF's, with one #4 screw each, equals 50% reduction in thermal impedance and half the temperature rise at power.

Or you could just turn down the signal on one of them by .27dB (SPL). Temperature rise-wise, same thing as adding a clamp.

I like to fiddle as much as anyone but usually demand that I get somewhere fer it.

 

[random007]

Thanks for this thread. Just bought a set of LM 3886s: the source was excruciatingly cheap from China but didn't specify they'd be F. Oh well. At least I don't need to find a suitable insulator. Will use a bar and some decent heatsink compound. Tick. Nice picture. I like the challenge of a parallel amp even if I'd never be brave enough to try it. I like the look of the insane input filter cap even if you can mount a tiny 100pF filter cap (silver mica if you're into that kind of thing) across the + and - input for the same yet more real-estate friendly low-pass effect. I think that why it's there (but am prepared to be proven wrong). In any case, thanks for the example: for what it's worth, I'll be using it.

Suggestion: check out a good computer enthusiast store (locally, PC Case Gear) for specialist heatsink compound vastly longer lasting and more heat conductive than cheap-as-hell silicone/zinc white paste. Generic silicone paste is very likely to solidify and loose thermal conductivity after a few years and isn't that good to start with.

[Edit for clarification]

 

[random007]

Andrew E and Jack allude to the TF package as not having a heat dissipation issue, because the TF is electrically isolated.

It might be misleading because the allusion doesn't happen. All I said is that high force isn't going to drop the thermal resistance of the case. You get about 2 degrees per watt out of the epoxy coated case. If you get .15 additional out of some paste with 1 inch pound on a tab mounted screw, and .12 out of a clamping bar with 5 inch pounds on each screw, what happens to the 2 degrees C of the plastic?

If you're that concerned about reliability and thermal resistance, you're definitely better off with the metal tab, and almost better off buying more of them, since they don't cost much more than a few manually drilled holes in a piece of metal.

You could always take some sandpaper to it and sand it back (with some glass sheet and a bit of 200 grade wet and dry) to the metal! I'm not going to do that any time soon but if your $20AU ($15US, 7euro) TF chips are borderline overheating, what the hell, why not?