Kapton Polymide Film insulation

[jamo96]

Hi all

I have some Kapton Polymide Film and would like to use this as insulation for the FETs in my 24 FET infineon controller, rather than the silicon pad.

I have Kapton® HN 100 gauge film (25.4 μm). Just wondering if I should do 1 or 2 layers? It feels pretty thick and tough. Cutting it with my albeit a little blunt utility knife (box cutter for those from the US) it seemed very tough!

I am thinking that doing 2 layers would cancel out some of the benefit of using this type of insulation over silicon (adding extra thickness).

I have bought a 12 gauge piercing needle (outside diameter ~2.8mm) to punch the holes in the film for the FET mounting hardware. I intend to lightly sand the fets and heatsink with some very light sandpaper to blunt any sharp edges.

 

[heathyoung]

This is the MO of a lot of manufacturers of the cheaper speed controllers - rightly or wrongly.

 

[jamo96]

This is the MO of a lot of manufacturers of the cheaper speed controllers - rightly or wrongly.

What do you mean by this?

 

[heathyoung]

Sorry - MO - modus operandi - method of operation - its standard practice, I'm not sure of the thermal transfer ability of the tape, but it will hold up to high voltages quite well IF it is genuine kapton - the cheap copies sometimes don't meet expectations (ie. are orange sticky tape, or not much better).

 

[amberwolf]

I intend to lightly sand the fets and heatsink with some very light sandpaper to blunt any sharp edges.

Just don't sand the flat areas, only the edges/corners, and only if you have to. You want as smooth a metal contact as possible--air is a terrible heat conductor, and that's what will be in the scratches from sanding, unless you also use thermal paste (whcih is a much poorer conductor than the eixsting metal). Also, thermal paste usually eventually dries out, contracting in size and leaving airgaps, whcih is why it is best used to *only* fill the scratches to eliminate air.

This is why some overclockers of CPUs and such polish (lap) their heatsinks' contact patches and the same area on the CPU, so that much better thermal transfer can take place without having to use thermal paste. (assuming that the surfaces are truly flat and are parallel).

Regarding thermal transfer ability of Kapton tape: AFAICR it's a lot better than those gray rubbery thermal pads, which is the usual other option. Dunno if it's better than mica, which is another possible option for electrical insulation but thermal conduction. There was a thread (controller mod thread?) discussing the properties of it; searching on "kapton" in ES search should find it.

I wonder if there is any easy "physics rule" that would allow us to find good materials for this purpose, simply by knowing that it must easily pass particle motion/kinetic energy (heat) but not electrons?

 

[jamo96]

Just don't sand the flat areas, only the edges/corners, and only if you have to.

Just on that note. I have been inspecting a spare 4110 fet I have and it doesn't really look like it has any sharp edges. The edges on the side are rounded off.

 

[bigmoose]

The kapton you have is 0.001 inches thick in American speak.. which is what TI recommends in the following paper I attached. You may enjoy reading it. It has about everything you will need from an engineering sense. It is based on a TO-3 style transistor from the old days, but the general principals apply to everything.

If you want to flatten the heat sink, consider "lapping" it on a flat surface. Plate glass is amazingly flat. Get a sheet of say 400 grit wet and dry silicon carbide sandpaper, squirt some water on it to stick it to the glass, keep the surface wet and flat sand your heat sink against it to get it really flat. The other thing is to mind the threaded holes that will hold down your FETs. They should be slightly counter sunk, and do this before you lap the heat sink.

good luck!

 

[CamLight]

I wonder if there is any easy "physics rule" that would allow us to find good materials for this purpose, simply by knowing that it must easily pass particle motion/kinetic energy (heat) but not electrons?

Might be. But, I'd hazard a guess that just about any material that could potentially be used as an insulator has already been tested (and results are available)?

IMHO, thickness is more important. Think about how incredibly good an insulator (thermal and electrical) most of the existing FET-insulating materials are and you can see how important thickness is. It's hard to beat using just about any material if you can find it in <1mil thickness.

I'll never understand, with all of the work everyone does to find a better way to cool stuff, that Type 3 hard-coat anodizing isn't used more often. Like most other materials, it's a really good insulator of heat and voltage but is put on so incredibly thin (less than 0.5mil) that it ends up being superior to just about any other method/material. And it's incredibly tough.

Just find a plating company that does Type 3 and you can usually get your piece(s) tossed into another batch for not-so-many-$$. I'll never use another method/material again.

But, if you must use a material, find 0.5mil Kapton without adhesive. Heck, any plastic is OK if you can get it thin enough and not damage it during installation.

 

[CamLight]

The kapton you have is 0.001 inches thick in American speak.. which is what TI recommends in the following paper I attached. You may enjoy reading it. It has about everything you will need from an engineering sense. It is based on a TO-3 style transistor from the old days, but the general principals apply to everything.

If you want to flatten the heat sink, consider "lapping" it on a flat surface. Plate glass is amazingly flat. Get a sheet of say 400 grit wet and dry silicon carbide sandpaper, squirt some water on it to stick it to the glass, keep the surface wet and flat sand your heat sink against it to get it really flat. The other thing is to mind the threaded holes that will hold down your FETs. They should be slightly counter sunk, and do this before you lap the heat sink.

good luck!

Great advice! I'd recommend continuing on to 1200 grit though once the 400 grit leaves a uniformly sanded surface on the heat sink (i.e., no unsanded spots left).

Jamo96...
One layer of that tape is more than enough for insulation. Don't use two layers.

 

[Lebowski]

How do you use this stuff ? I know it's for isolation and good thermal conduction, but
do you still need to apply heat paste ? For my 6 FET TO247 build I orderede some of these:

http://www.digikey.com/product-search/en?x=0&y=0&lang=en&site=us&KeyWords=ber169-nd

and I'm wondering whether I need to lather up on heat paste in combination with these ?

 

[bigmoose]

You do not need paste with the Bergquist pads.

 

[CamLight]

You need paste only for hardcoat anodizing, mica and the plastic films (Kapton, etc.).

 

[hillzofvalp]

Would it be worthwhile to to a finishing pass with a aluminum shell mill to my 18fet heatsink bar?

I assume that, if going with kapton, you must apply thermal past to both sides, correct?

Is it worth using silver CPU compound on the isolated side of the kapton, or too big of a risk? I have some laying around

Also, is it only necessary to beef up the two main battery legs? Don't need to do phases?

 

[CamLight]

Would it be worthwhile to to a finishing pass with a aluminum shell mill to my 18fet heatsink bar?

I assume that, if going with kapton, you must apply thermal past to both sides, correct?

Is it worth using silver CPU compound on the isolated side of the kapton, or too big of a risk? I have some laying around

Also, is it only necessary to beef up the two main battery legs? Don't need to do phases?

1) It all depends. What are the surface finish specs for the existing bar and how much smoother can you get it with the mill?

2) Correct, Kapton tape requires paste on both sides. ONLY just enough to form a tiny bead all around each FET once fully tightened down. And use the correct tightening torque if using TO-220-cased FETs. Overtightening the mounting screw with these FETs can actually lift up the body of the FET as the tab deforms, ruining the thermal contact to the heat sink.

3) Not sure what risk you're referring to. Use any decent paste you want but just use a thin film of paste applied to the sink and the FET.

4) I'll other folks answer this one.

 

[jamo96]

I did get around to replacing the silicon insulation with kapton film.
I also did a few other mods...

Still have a few things to tidy up wiring wise. But all works well, fans move plenty of air!!
There is space for a terminal to be added for a fuse.
Controller is being used on a motenergy pmac me907.

[youtube]kvDHB19b3tI[/youtube]

Fans are a bit noisy but work well.

 

[jateureka]

nice work. What sort of power are you running? any links to your build?

 

[jamo96]

nice work. What sort of power are you running? any links to your build?

http://endless-sphere.com/forums/viewtopic.php?f=34&t=40631

Looking at 10-15kw peak power. I bought a CA though I didn't get time to install it or do any actual testing with the controller on the weekend. Batt current limit at the moment is 200A. Battery voltage is 55v (14s lipo).