to3 heatsink

alo
could someone tell me how to put the mica and washers in a to3 heatsink?.
i guess it goes like this:
mica on top of the board,then heatsink on top of the mica,then 2n3055 inside the heatsink,then from under the board comes the 2 washers inside the screws holes,then i put the 2 screws,then the 2 nuts from the top.
i fell so idiot asking this,but i realy don`t know,because i`ve never done this before.
best regards
pedro

Be sure to smear the mica on both sides with a thin, uniform layer of heatsink compound (silicone with zinc oxide). The stuff is messy, so make sure you're not wearing your favorite shirt.

The purpose of the mica and the fiber washers is to prevent any electrical continuity between the transistor case and the heatsink. Verify that there is no continuity before you solder in the transistor and power up the circuit for the first time.

Just a note, some power transistors can be very leaky from collector to emitter, to the point where you might think you have a short, but it's just the transistor.

Also, they make thermally conductive pads nowdays that do not require grease.

Thermalloy makes kits with these pads.

If you're reading a couple of ohms between C and E with the transistor out-of-circuit, you ain't got a "leaky" transistor, you got a short!

As for the "squishy" insulators, I don't like 'em. They just don't seem to conduct as well as the good old mica and grease. I admit that I don't have scientific evidence to back that up, so you can take it (or ignore it) as just the opinion of a guy who used to fix a lot of power amplifiers.

I have been using the "squishy" pads for 7 years now with no comebacks.
I was skeptical at first also, but now, I will never go back to that greasy mess.

They might not conduct quite as good initially, but over time I think they are much better, as the white grease tends to dry out and lose its' conductive properties.

Aslo, the pads do not compress as much as the grease. You may ship a unit with the transitor torqued down, but that grease will creap, you lose 1 mil, and your pressure goes down. This can be verified by un doing a transistor that has been bolted to white grease for many years.

Good point. The grease does turn into white crud cake after several years. Then again, the squishy insulator could never fill in all the surface imperfections of the transistor case as well as the grease does. Short answer: I had better results in the short term with mica and grease than with squishies, but I can't say for sure which is best in the absolute sense.

> the squishy insulator could never fill in all the surface imperfections of the transistor case as well as the grease does

Filling "all" the roughness is not the point of grease. Grease is a crummy conductor. It can't carry much heat up from the valleys. We just use it because it is better than nothing (air). Engine oil, even without the zinc-oxide (or silver) filler of Thermal Grease, is a lot better than nothing; that's my trick for light duty work.

The practical goal is to surround the high spots of the surface with something better than air, so points-contact changes to blobs-contact. The spongebob stuff can do that.

I'll raise my pet point: be sure the sink and transistor are FLAT. I have had warped sinks where the device touched on two corners and not in the center where the heat was. I have not seen a warped TO-3 but I am sure someone makes TO-3s out of old bumpers.

There IS data on all insulators, not excessively biased by manufacturer self-interest. (If they claim it is better than it really is, stuff melts, customers scream.) The better sponges are competitive with greased mica/mylar, though the deciding factor for many situations is mess and creep.

Note that a GOOD grease/sponge joint is MUCH better than any but the best heatsinks. If you have a 2"x2" sink, you don't want a loose transistor, but you don't have to go crazy on the mounting. 0.2degC/W versus 0.3degC/W in the joint is moot on a 2degC/W heatsink. When you get into gitar amps with sinks that are significant but not excessive, then the joint may be the difference between a good gig and a re-repair.

alo
thank you all for the help.
best regards
pedro

another info-particle- i ran out of heat sinks one day and decided to save a buck by chopping up some black anodized Al for some DIY sinks. Well, the old ones were bare AL, and the new anodized ran cooler. There seems to be less cooling whin mounting to a chunk of Al with oxide on it.

PRR's remarks about flatness should be laminated and given to all mechanical engineers working with heatsinks.

A project for Dell had a bad reputation for quality precisely because my initial specification for the power IC clamping and the thickness of the material clamped to was disregarded by mechanical and manufacturing people who should have known better. The material bowed and the parts repeatedly went into thermal protect when pushed and eventually blew up. Most customers didn't push them so most of them didn't, and sometimes the clamping was "just right" and the thermal contact not so bad, but it was extremely frustrating.

Having any surface on aluminum other than bare metal helps radiative cooling immensely, but since this is a fourth-power-of-absolute-temp law you have to be pretty toasty before it makes much difference. Alumina or other refractory aluminium compounds are not all that great thermal conductors, but enough anodizing to protect and insulate a surface is mighty thin.

FWIW, due to the restricted size of the 'Neve' BA283-card I'll be
literally approaching this from a different side: mounting a sink
on top of the 2N3055, it goes something like this:

Had a few of these around, it's some 7.3K/W and doesn't use any
isolation/grease/mica. At an estimated dissipation of 2W for the 2N3055 in that circuit I expect it'll be enough cooling.

Only drawback I can see is that you can't check as easily that it's
the correct flavour of 2N3055 when it's time for pron-pics :wink:

Any horror-stories of those sinks welcomed of course, one could
better just know it.

im not using any insulator for my heatsink...im not getting any shorts between transistor case and C, E.

Am i missing something here?

Should i be using insulators or is it ok the way it is?

/J

[quote author="Jonkan"]im not using any insulator for my heatsink...im not getting any shorts between transistor case and C, E.

Am i missing something here?

Should i be using insulators or is it ok the way it is?

/J[/quote]

That's for the conventional way of mounting, right ?
It could be OK like that but note that your complete heatsink will then be at the collector-potential. Better have it at ground-potential, less risk when a screwdriver needs to fly around in the box.

Anyone else using those op-top-of heatsinks ?

Haven't used them. But bear in mind in your app the 3055 is being run way way under its rated dissipation, so a little help is all you need. The top-of-package approach is not very effective in reaching the chip---which is centered in the middle of the lozenge-shaped bottom piece. Some Q's have some copper there in some form, maybe a transitional slug internally, maybe a sandwich with steel. Don't know anymore about the exact 3055 construction---but the point is the thermal R junction-to-case is higher if you are sucking heat off of the top instead of the bottom surface.

Again, a few watts, big deal. The bare case itself would be alarmingly warm to the touch but would probably run indefinitely at 2W. I'm not recommending that, just noting that a metal case has to be damn hot before there's a real problem with decent silicon run otherwise within ratings. It's often the thermal cycling issue with really big temp excursions that leads to failure, which argues for leaving equipment on if feasible.

[quote author="bcarso"]Haven't used them. But bear in mind in your app the 3055 is being run way way under its rated dissipation, so a little help is all you need. The top-of-package approach is not very effective in reaching the chip---which is centered in the middle of the lozenge-shaped bottom piece. Some Q's have some copper there in some form, maybe a transitional slug internally, maybe a sandwich with steel. Don't know anymore about the exact 3055 construction---but the point is the thermal R junction-to-case is higher if you are sucking heat off of the top instead of the bottom surface.

Again, a few watts, big deal. The bare case itself would be alarmingly warm to the touch but would probably run indefinitely at 2W. I'm not recommending that, just noting that a metal case has to be damn hot before there's a real problem with decent silicon run otherwise within ratings. It's often the thermal cycling issue with really big temp excursions that leads to failure, which argues for leaving equipment on if feasible.[/quote]
Thanks Brad,
You're right that this sink is a less optimal way of attaching. Must add that the sinks I have are actually half of a complete set: the usual square plate with fingers on four sides and the depicted thing on top. But on those BA283 PCBs there's no room for the complete thing, unless I want to mount it way above the PCB.
A quick & dirty calculation estimates the dissipation of the 2N3055 in that circuit as being just over 1.5W:
(24V - 80mA*(12+47)) * 80mA

So let's say it's under 2.5W

Might be a good idea to put at least some additional metal under the TO-3 as well, we want this to be an assemble-&-forget thing.

Regards,

Peter