Here's some information that some of you may find interesting.
At Pass Labs we use Bergquist Silpads for thermally conductive
insulators on TO3P (Plastic packages) but not on the TO3
packages because they tend to shear and fail with the TO3.
This is not the case with the TO3P because of the smaller single
hole and the transistor's plastic insulation around the mounting
hole.
We've been happy with the Silicone pads in this application as
they are less messy and perform nearly as well as the
traditional Mica and thermal goo. As far as I can make out,
they have lower thermal impedance than claimed. However this
is assuming a certain degree of smoothness of the metal, which
is simply the flat surface of an anodized extrusion, and has a
somewhat polished look to the finish.
More recently with the First Watt product, the heat sink has
been bead blasted to make it more spiffy, and you'd think that
the slight dullness of the surface wouldn't make a difference.
In fact it made quite a bit of difference with the silicone pads -
apparently something like another 50% thermal resistance, and
I ended up using Mica and goo. The figures I get are 1.5 deg C
per watt with mica, and 2.1 with silicone. This is obtained by
running the devices at 20 watts and measuring the temperature
of the top of the case versus the heat sink right next to the
device.
If we take the ratings of the device itself, we see internally we
can expect .83 deg/watt, and so we see that Mica and goo
end up about 1/2 the total thermal resistance. If your surface
is not flat and polished, and/or you're pushing on the power
ratings of the device, consider good old messy mica and goop.
At Pass Labs we use Bergquist Silpads for thermally conductive
insulators on TO3P (Plastic packages) but not on the TO3
packages because they tend to shear and fail with the TO3.
This is not the case with the TO3P because of the smaller single
hole and the transistor's plastic insulation around the mounting
hole.
We've been happy with the Silicone pads in this application as
they are less messy and perform nearly as well as the
traditional Mica and thermal goo. As far as I can make out,
they have lower thermal impedance than claimed. However this
is assuming a certain degree of smoothness of the metal, which
is simply the flat surface of an anodized extrusion, and has a
somewhat polished look to the finish.
More recently with the First Watt product, the heat sink has
been bead blasted to make it more spiffy, and you'd think that
the slight dullness of the surface wouldn't make a difference.
In fact it made quite a bit of difference with the silicone pads -
apparently something like another 50% thermal resistance, and
I ended up using Mica and goo. The figures I get are 1.5 deg C
per watt with mica, and 2.1 with silicone. This is obtained by
running the devices at 20 watts and measuring the temperature
of the top of the case versus the heat sink right next to the
device.
If we take the ratings of the device itself, we see internally we
can expect .83 deg/watt, and so we see that Mica and goo
end up about 1/2 the total thermal resistance. If your surface
is not flat and polished, and/or you're pushing on the power
ratings of the device, consider good old messy mica and goop.
Interesting,
I never bothered measuring the difference but it seems obvious to me that the good old grease is doing a much better job especially with non flat surface. I really don’t mind working with it. Whenever I open an amp and see the grease sticking out of the transistor surface (Peavey is a nice example) I always consider it as good craftsmanship. Perhaps that’s only my conservative opinion.
I recall reading something about thermal isolation that had the best of both worlds. Must have been an old Elector magazine.
It appeared to be some kind of silpad that had the ability to become softer when heat was applied. The principle was to burn in the transistor or mosfet or whatever had to be isolated and after a period of time turn on the screws to tighten the contact with the heatsink.
/Hugo
I never bothered measuring the difference but it seems obvious to me that the good old grease is doing a much better job especially with non flat surface. I really don’t mind working with it. Whenever I open an amp and see the grease sticking out of the transistor surface (Peavey is a nice example) I always consider it as good craftsmanship. Perhaps that’s only my conservative opinion.
I recall reading something about thermal isolation that had the best of both worlds. Must have been an old Elector magazine.
It appeared to be some kind of silpad that had the ability to become softer when heat was applied. The principle was to burn in the transistor or mosfet or whatever had to be isolated and after a period of time turn on the screws to tighten the contact with the heatsink.
/Hugo
I forgot to mention it, but I believe goop may be even more
crucial to thermal coupling for insulated plastic packages like
some of the chip amp packages.
By the way, we use Thermalloy's "Thermalcote" (a fine Texas
company), although the computer crowd raves about "Arctic
Silver" (if you can afford it)
crucial to thermal coupling for insulated plastic packages like
some of the chip amp packages.
By the way, we use Thermalloy's "Thermalcote" (a fine Texas
company), although the computer crowd raves about "Arctic
Silver" (if you can afford it)
Hi Nelson,
I agree with your assessment as I've found the same thing. When repairing an amp with a rough heatsink I smooth it with 600 grit sandpaper and a milled block (flat). It does make a difference. Yes, you can see the differential temperature drop with a thermocouple thermometer.
Aluminum chips from the heatsink can pierce the sil pads
, so watch for that. Same goes for solder balls.
-Chris
I agree with your assessment as I've found the same thing. When repairing an amp with a rough heatsink I smooth it with 600 grit sandpaper and a milled block (flat). It does make a difference. Yes, you can see the differential temperature drop with a thermocouple thermometer.
Aluminum chips from the heatsink can pierce the sil pads
, so watch for that. Same goes for solder balls.-Chris
Nelson, which silpads were you using? the better ones are filled with boron nitride and are supposed to be equivalent to an alumina insulator, though they still may fall short of mica and goop on a bead-blasted surface. It may be Chomerics rather than Berquist that makes this particular silpad variation. A lot of times I put a thin film of goop on a silicone insulator anyway just to hold it in place while I'm trying to assemble things - best of both worlds...
There is a metal insulator. I saw it in Rockford Fosgate car amp. The insulator is a metal sheet, but somehow it doesnt conduct electricity. What is this called? Who sells them?
Also I wonder where to buy insulator sheet that is not small-cut. In size of A4 paper or bigger. Where to buy this un-cut insulator? The one I saw is like plastic, with yellow or orange color. Not mica, not that greyish-rubber like insulator.
Also I wonder where to buy insulator sheet that is not small-cut. In size of A4 paper or bigger. Where to buy this un-cut insulator? The one I saw is like plastic, with yellow or orange color. Not mica, not that greyish-rubber like insulator.
I can absolutely assure you that sandblasting a surface would increase its thermal resistance as I've also done this before and had to go back and finish machine on a mill that blasted(pun intended!) surface. What some may think by sandblasting the heatsinks mounting surface would be that its flatter and smoother... but not true! Blasting alumnium creates a pebbled/ textured finish that can be most uneven in itself since most blasting is done by hand, not cnc. It will definately prevent good contact between two parts because of all the tiny raised almost microscopic dings in the metal. Its great for decorative purposes though and the resulting matte finish is very nice looking after anodizing although its a champ at holding dust and lint and a pain to clean easily.
To lower the thermal resistance in the surface of an alumnium sink one should finish grind that alumnium surface on a surface grinder to a very good, almost mirror like finish. If you are the patient type that has lots of time to spare this can also be done by crefully drawfiling with a flat file across the surface. In doing this you would lower the thermal resistance quite a bit and the sil pads would be even more effevtive.
Mark
To lower the thermal resistance in the surface of an alumnium sink one should finish grind that alumnium surface on a surface grinder to a very good, almost mirror like finish. If you are the patient type that has lots of time to spare this can also be done by crefully drawfiling with a flat file across the surface. In doing this you would lower the thermal resistance quite a bit and the sil pads would be even more effevtive.
Mark
Yes; Nelson was making funny by calling it "goop." It's probably usually called "heat sink compound" or "thermal grease"
Here's one page from the Digi-Key catalog:
http://dkc3.digikey.com/PDF/T042/0564.pdf
In addition to a flat cooling surface, it helps to 'clean' the mounting surface of the transistors as well. Very often, the plastic around the edges is just a tiny bit higher then the metal surface. I use 600 or 800 sandpaper (wet) and polish the entire mounting surface of the transistor to the point that the copper underneath the plating shines just before mounting the devices.
Regards,
Andreas
Regards,
Andreas
If you have the possibility after mounting the transistor to apply so much pressure that the grease that´s too much will be squeezed away, that´s OK.
Usually you shouldn´t use more than you need.
It obviously is only effective filling the mountain-like surface.
A whole layer of it won´t do any good.
When I repair TV´s for example there are only cheap clamps that hold the transistor. Applying too much of that goopy stuff isn´t a good thing and I can assure you that I did it wrong in the beginning.
Jens - used to overgoop transistors and send them to semiconductor-heaven
In addition to surface finish, the holes for TO-3 devices may
have to have the edges beveled on the mounting side to
prevent shearing of the silicon insulators. Because of that,
I've usually used mica or thin plastic insulators which have
a better shear strength. Berquist comes to mind for
insulators. I'm not fond of the mess that thermal compounds
make, so I'd like to avoid their use too.
Beryllium oxide insulators used to be popular, but there
may be restrictions on their use now; aluminum oxide is
nearly as good. You want radical? How about diamond/
silicon carbide composites? Gotta find the link for this,
but it's a real product!
Popular among CPU overclockers, Artic Silver, a non-conducting silver paste might also be helpful. Instructions caution that
it will increase capacitive coupling, if that's an issue; they
also have non-silver products and other enhancements in
the form of variable-size particles that pack more tightly to
increase thermal conductivity.
Another possibility is simply increasing the thickness of the
heatsinking anodizing and mounting the device directly, using
only conductive compound without an insulator. This is
risky but probably gives the best thermal results short of
direct metal-to-metal contact and an electrically isolated
heatsink. You'll need to consult your heatsink supplier to
see what special needs they can accomodate; it may be
easier than you think.
And, what everybody else said, too.
have to have the edges beveled on the mounting side to
prevent shearing of the silicon insulators. Because of that,
I've usually used mica or thin plastic insulators which have
a better shear strength. Berquist comes to mind for
insulators. I'm not fond of the mess that thermal compounds
make, so I'd like to avoid their use too.
Beryllium oxide insulators used to be popular, but there
may be restrictions on their use now; aluminum oxide is
nearly as good. You want radical? How about diamond/
silicon carbide composites? Gotta find the link for this,
but it's a real product!
Popular among CPU overclockers, Artic Silver, a non-conducting silver paste might also be helpful. Instructions caution that
it will increase capacitive coupling, if that's an issue; they
also have non-silver products and other enhancements in
the form of variable-size particles that pack more tightly to
increase thermal conductivity.
Another possibility is simply increasing the thickness of the
heatsinking anodizing and mounting the device directly, using
only conductive compound without an insulator. This is
risky but probably gives the best thermal results short of
direct metal-to-metal contact and an electrically isolated
heatsink. You'll need to consult your heatsink supplier to
see what special needs they can accomodate; it may be
easier than you think.
And, what everybody else said, too.
What's cookin?
I can't help but think what happens in real life. Generally, assemblers just want to stick the transistor on the heatsink quickly. That's where sil pads come in. Grease and mica take too long, but are great in service (where many techs overtighten & cut through sil pads). Using metal oxide as your insulator turns the unit into "factory service only". Or let the destruction begin!
On a DIY level you can get esoteric if you want. In manufacturing I'd suggest keeping device dissipation within reasonable limits by speading the power around more outputs.
-Chris
I can't help but think what happens in real life. Generally, assemblers just want to stick the transistor on the heatsink quickly. That's where sil pads come in. Grease and mica take too long, but are great in service (where many techs overtighten & cut through sil pads). Using metal oxide as your insulator turns the unit into "factory service only". Or let the destruction begin!
On a DIY level you can get esoteric if you want. In manufacturing I'd suggest keeping device dissipation within reasonable limits by speading the power around more outputs.
-Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.