Given the traditional mica thermal pads and thermal paste for isolating FETs to heatsinks...
Any reason one couldn't cut their own out of the clear plastic sheets like you sometimes get at a craft store or with notebooks? Is that a common thing?
I've also recently used silicon pads without grease. No muss, no fuss, but I assume some paste would help conduct heat better regardless. Yes/no/I'm a fool?
Any reason one couldn't cut their own out of the clear plastic sheets like you sometimes get at a craft store or with notebooks? Is that a common thing?
I've also recently used silicon pads without grease. No muss, no fuss, but I assume some paste would help conduct heat better regardless. Yes/no/I'm a fool?
Vinyl sheet is an insulator, both electrical and thermal. Very bad idea.
The correct material is the Sil-Pad, No paste. Paste does not help as the pad is designed to get into the micro-crevices. If it is there, you know it is correct and can count on the transfer parameters. ( Berquest used to have a very good tutorial on-line)
If you can tolerate the heat sink being live, the very best is a graphite pad as like the Sil-Pad, it gets into the surface irregularities.
Mica with grease can performed as well as a sil-pad, but only when new and only if done exactly correctly. They degrade over time as the thermal cycles pump the thermally conductive oxide out from under the part leaving just the oil which is no t as good. We wrote an engineering spec to ban mica and Kapton with grease in favor of sil-pads or graphite. Lesson learned from Failure Analysis. Heed if you wish.
The correct material is the Sil-Pad, No paste. Paste does not help as the pad is designed to get into the micro-crevices. If it is there, you know it is correct and can count on the transfer parameters. ( Berquest used to have a very good tutorial on-line)
If you can tolerate the heat sink being live, the very best is a graphite pad as like the Sil-Pad, it gets into the surface irregularities.
Mica with grease can performed as well as a sil-pad, but only when new and only if done exactly correctly. They degrade over time as the thermal cycles pump the thermally conductive oxide out from under the part leaving just the oil which is no t as good. We wrote an engineering spec to ban mica and Kapton with grease in favor of sil-pads or graphite. Lesson learned from Failure Analysis. Heed if you wish.
I've bought sheets of polyamide (Kapton) to cut my own insulators. I still used paste of both sides though.
Mylar and polyimide (kapton) can be used, and are sold in precut shapes, but they have poor performances, similar to mica.
Thermoplastic polymers like PVC suffer from creep, and are totally unusable.
For ordinary jobs, silpads are OK, and if you need better use keratherms or special ceramics, but they are difficult to use properly: you need good surface conditions, controlled tightening torque, etc
Thermoplastic polymers like PVC suffer from creep, and are totally unusable.
For ordinary jobs, silpads are OK, and if you need better use keratherms or special ceramics, but they are difficult to use properly: you need good surface conditions, controlled tightening torque, etc
Kapton is not the best thermal conductor by any means BUT can be had in very very thin sheets, which can offset that.
Being very strong and tear proof, if safe to use that way.
Always with grease, of course.
Being very strong and tear proof, if safe to use that way.
Always with grease, of course.
Attach FET to a large block of copper. Then insulate that from the heat sink, enjoying the benefits of dividing the thermal resistance by the bigger surface area.
Not always. Now if you used a graphite pad between the FET and the copper, that would work, but you still need to couple the copper. Every transition is a loss. A large block of copper is expensive. Sil-pads are not.
I have seen some attempt to hone and polish packages and heat sinks thinking better coupling. We even had an engineer claim if you did so, the anodizing would be a sufficient insulator. He was wrong. It was an attempt to cover up a grossly over-margin design.
This is a well described engineering problem. I suggest going to the material suppliers and hit the books to understand the details, get the equations ( pretty simple if I remember) and the actual data on the different methods. I was temporarily an expert on this, but that was 40 years ago when I was in industry. I remember the take-away which is what I suggested. If a sil-pad is there, it is correct. Of course always use a spring of some sort to mount to the heat sink and never try to use the hardware stack as your current path.
We would reject any design that required better performance than a sil-pad.
Another problem with any of the grease based solutions is the grease migrates out and attracts dust, insulating the package from airflow.
Here's a picture of an old Adcom 5400 amp which I recapped: Note the two round circles of oil originating at each of the two diode bridges, showing that the oil runs out way farther over time (20+ years in this case) than the oxide particles, which more or less stay in place.
But in general for output transistors I used Keratherm Red pads, like those from our store:
https://diyaudiostore.com/collections/frontpage/products/keratherm-transistor-insulators
Attachments
Yes, it is grease which flows away under pressure and thermal cycling, not the solid oxides as somebody suggested earlier.
Part of my manufacturing policy is to always keep my brand Guitar amps running, for 50 years now, and when I replace 20/30/40 years old TO3 transistors, I typically find dry caked ZnO between surfaces and a ring of grease and dust/lit around, where it does not help at all.
Oh well.
Part of my manufacturing policy is to always keep my brand Guitar amps running, for 50 years now, and when I replace 20/30/40 years old TO3 transistors, I typically find dry caked ZnO between surfaces and a ring of grease and dust/lit around, where it does not help at all.
Oh well.
If you are determined enough, you can make your own thermal compound, or even your own specialized silpads.
https://www.diyaudio.com/community/...supply-truly-diy-friendly.330193/post-5612240Why would you do that? It is unpractical, messy and complicated for a one off, but it has some advantages: as a compound, if you use a 2-component RTV as a binder, there is no risk of silicone oil migration; it will remain in place for life (shelf life of the quantity you prepare is limited to an hour or two obviously).
The filler can be much better than ZnO: MgO or alumina powder for example.
Silpads can be produced in any size, shape and thickness, meaning you can make the breakdown voltage arbitrarily high.
Finally, it is cheaper than commercial solutions (but it is much more work).
It is not possible (easily at least) to match the performance of the best commercial solutions, but you can easily exceed what standard silpads offer
https://www.diyaudio.com/community/...supply-truly-diy-friendly.330193/post-5612240Why would you do that? It is unpractical, messy and complicated for a one off, but it has some advantages: as a compound, if you use a 2-component RTV as a binder, there is no risk of silicone oil migration; it will remain in place for life (shelf life of the quantity you prepare is limited to an hour or two obviously).
The filler can be much better than ZnO: MgO or alumina powder for example.
Silpads can be produced in any size, shape and thickness, meaning you can make the breakdown voltage arbitrarily high.
Finally, it is cheaper than commercial solutions (but it is much more work).
It is not possible (easily at least) to match the performance of the best commercial solutions, but you can easily exceed what standard silpads offer
There is a thermal glue available here for bonding LEDs to heat sinks, used a lot for driver on board circuits.
It seems to be an adhesive with heat sink filler, sort of silicon with some zinc oxide (or whatever is used for the solid part of heat sink compound).
Pretty cheap, 50 cents in China for about 100 grams, and $1.20 here in retail, only thing is like all RTVs, it starts to cure once you open the tube, you have to finish it in a month, or about that depending on the ambient temperature and so on.
It is a single component adhesive, used directly, and if tightened before curing, will form a compliant filled joint with no air gaps.
For LEDs, we just put a blob, and rub it in circles, to spread it out. Takes a few hours to full strength, so we assemble etc. next day.
Just pressed firmly, so a thin layer stays in place, about what you would use to form a thin air tight layer.
No fasteners used for LED, but for transistors and so on the hole matching, and heat sink support would need fasteners or springs.
So far, it has held up for five years in some lamps we made.
You have to break it off, so use a thin layer of silicon release spray on one side if you intend to remove it in the future.
And no, it does not separate out to grease/oil and filler....at least not yet.
It seems to be an adhesive with heat sink filler, sort of silicon with some zinc oxide (or whatever is used for the solid part of heat sink compound).
Pretty cheap, 50 cents in China for about 100 grams, and $1.20 here in retail, only thing is like all RTVs, it starts to cure once you open the tube, you have to finish it in a month, or about that depending on the ambient temperature and so on.
It is a single component adhesive, used directly, and if tightened before curing, will form a compliant filled joint with no air gaps.
For LEDs, we just put a blob, and rub it in circles, to spread it out. Takes a few hours to full strength, so we assemble etc. next day.
Just pressed firmly, so a thin layer stays in place, about what you would use to form a thin air tight layer.
No fasteners used for LED, but for transistors and so on the hole matching, and heat sink support would need fasteners or springs.
So far, it has held up for five years in some lamps we made.
You have to break it off, so use a thin layer of silicon release spray on one side if you intend to remove it in the future.
And no, it does not separate out to grease/oil and filler....at least not yet.
Last edited:
I wonder how heat conductive metal filed epoxy would be?
Some grades are available here with steel or aluminum powder, intended for repairing castings (blow holes, voids etc.)
Strong enough to drill and tap.
Some grades are available here with steel or aluminum powder, intended for repairing castings (blow holes, voids etc.)
Strong enough to drill and tap.
Just got a paper from EEWorld Online about compliant "fillers" by FujiPoly.
Metal filled epoxy might be leaky as the voltage goes up. Plus, it would make rework really hard. Most of them are pretty brittle. ( If duck tape won't fix it, JBWeld will)
Metal filled epoxy might be leaky as the voltage goes up. Plus, it would make rework really hard. Most of them are pretty brittle. ( If duck tape won't fix it, JBWeld will)
Does JB Weld have a metal or ceramic fillled grade?
The idea was a heat conductive, compliant and durable bond, as most amps do not blow the output devices for years.
Thermal glue, as above, does work.
Fasteners do come loose in some years of thermal cycling, if not properly installed.
The idea was a heat conductive, compliant and durable bond, as most amps do not blow the output devices for years.
Thermal glue, as above, does work.
Fasteners do come loose in some years of thermal cycling, if not properly installed.
Last edited:
Well, I have.
In one of our periodic Financial meltdowns (we are going through one right now, our Economy Minister is battling IMF in Washington and our President is visiting Putin and Xi Jin Ping, obviously asking for help) there was NO thermal paste available anywhere (shops tightly close doors and sell nothing, fearing new product price will be twice what they sold it for) so I bought one kilo of Zinc oxide, and proceeded to make 2 kilos of the stuff.
This was in 2008 and haven´t used 20% of it yet, go figure.
Yes, it is advantageous, especially if you use alternative ingredients: for example MgO instead of ZnO, or two-component RTV instead of silicone oil.
This allows you to improve the thermal performance, or the durability (it becomes a thermal glue).
You can also mix silicone oil and RTV in the right proportions to end up with a kind of thixotropic paste, which remains storable and applicable, yet doesn't creep or decompose like ordinary comound
This allows you to improve the thermal performance, or the durability (it becomes a thermal glue).
You can also mix silicone oil and RTV in the right proportions to end up with a kind of thixotropic paste, which remains storable and applicable, yet doesn't creep or decompose like ordinary comound
Or a piece of aluminium. The sil-pads are a major bottleneck, so even a few mm thickness of live metal should easily spread the heat, allowing a sil-pad with many times more surface area to be used.