Are You On Pot????? (potting and heatsinking with epoxy and silicon carbide)

any material other than air is much better at moving/absorbing heat, especially stagnant air inside a pill that isn’t moving…air does “ok” if it is moving, but if it is just sitting there it really does no good at moving heat away from an object

http://www.ioffe.ru/SVA/NSM/Semicond/SiC/thermal.html

with mixing epoxy and silicon carbide, it still averages out much better than stagnant air

I actually put a blob of silicon over the chip on my TP4056 charger module…what I noticed was that instead of the chip getting so hot you can’t touch it at 1A draw, the entire board is now quite warm but not crazy “burn your fingertip hot” when air was the only medium to move the heat from the chip (sure if they had it soldered to a surface mount vias or whatever it would be better but the IC in that package pretty much just sits above the board and doesn’t transfer heat very efficiently), sure there is the issue of it maybe keeping the heat in, but when it diffuses the heat over a larger area, and allows the heat to soak away into the board it lowered the temperature (I’m sure there is a laundry list of calculations and maths involved…but it breaks my brain)…all fine and dandy until you put the daggum battery in backwards (no reverse polarity protection = kaput IC…dangit)
That is pretty much my real world experience with conformal coating/potting/heatsinking…

It’s actually really interesting stuff. It’s man made now and it’s considered a ceramic. It has very high thermal conductivity. On the chart I was looking at it was the best of the ceramics by quite a bit. I can’t find the chart right now but this site is pretty good shows it’s properties. Really neat stuff.

It probably would, Also I know that you can coat the emitter first with wax so that it will peal off. Comfychair does that I think. I would do that if I had an application where I was going to be messing with the driver later. But for me, I’m done once I finish. Some people of course want to be able to flash the driver later to change modes and stuff like that.

Great idea thanks for sharing. I’ve got two drivers I’ve held back on.
This also looks like a solution to a too thin LED shelf. Straws to protect the shelf holes then pot the under side to a more desirable thickness. Maybe cap off the underside with an Aluminum or Copper plate.

Wow, this is really helpful - 18sizfifity, thanx! I really should buy some silicone carbide and whip up a batch. This just may solve some problems - I'm thinking the hot J18's I got with the beast drivers, most of which out there seemed to have died, but I use my J18's very sparingly...

Thanks Tom and I never got to try it on that driver because mine died. It’s one of the first things I thought of when I started using this. Dang wish I had used in on that driver. Especially when I killed one driving 3x MT-G2. The light was sweet while it lasted and if I had potted it well it might have lived.

You’re welcome and I bet it would work.

So, does it not do that if you use only the JB Weld and leave out the powder?

What about potting drivers that don't generate any significant waste heat? Does it make those better, too?

I fully understand that this powder had good thermal properties, but what happens to those thermal properties when it's broken down into fine particles, and then mixed into an epoxy? Where the particles are no longer in direct contact with each other, but all have gaps inbetween filled with the epoxy? Why would the thermal properties be better than just the epoxy alone, since the heat has to jump the epoxy gaps between the silicon carbide particles to get where it needs to go?

Yeah not sold on potting dough so much as using strait copper, which does have a very high heat transfer. Like create a copper sheet to contact areas of the driver that need cooling like the fet, it should be enough. I’d be scared this would trap heat. But fifty is using copper in the potting, that may be why it draws heat and works instead of insulating it.

It’s interesting that’s for sure :slight_smile: I’m going to stick with copper and aluminum, know what works. Coolabs works great too just not good on strait aluminum. You have to solder a thin copper sheet to the aluminum pill then use the coolabs on the mcpcb and copper sheet to get the best heat transfer (imo). Then you could still replace the emitter and not worry about the tim breaking down over time.

I can’t wait for my stuff to get in really.

I don’t really know, all I know is that it works great. You could maybe ask why the guys that make arctic thermal how it works. They use it in their AS5 and alumina in Arctic alumina which has good thermal properties but not as good of thermal properties as silicon. Also Arctic alumina has been tested side by side with AS5 and it isn’t as good.

Also it is easier to work with than simple JB weld because if you add enough it becomes a putty form.

When possible I do use copper but if you have an air gap you need to fill that in with something or it acts as insulation. Also you can’t use copper in places where it can cause an electrical short. There are other advantages as well. This includes making the driver shock resistant and corrosive free. Copper is great but if it’s not making contact then it’s not going to transfer heat. Also on a driver the chips are surrounded by air and that air works as insulation, potting with anything will at least get rid of that insulation. Think about putting down a sinkpad. Direct copper but unless you remove the air between it and the pill it’s going to be trouble.

For those doubting the effect of these compounds.

It is nothing new. The electronics industry has been using similar or exactly the same compounds since it became an industry.

They work. And better than just straight epoxy.

A quick google search would set you up with more litterature of lab tests from many different compunds than you probably care to read.

It probable works. Vinz used diamond dust with his potting material and claimed great thermal transfer. A modified style might evolve where copper is bonded to the hot components. Then potted with JB Weld and Silicon Carbide.

That makes sense. Can give it a try.

For the same basic reason its faster to drive your car to work 20 miles away, than it is to walk, even though you have to walk from your front door to your car, and from where you parked the car to the front door of the office. For that matter, its why having more metal in the heat conduction path is good for heat dissipation, even though the heat still ends up being transferred to the air.

I’m thinking of trying this with some sort of silicone as the binder so I can remove it. I think a mix of grits is optimal, though I don’t know the optimal mix. If it is all fines, then there will be a lot of binder between all the fines to hold them together. If it is all coarse, then there will be big gaps filled with binder. With the right mix the coarse stuff should be packed as tightly as possible, with finer grit filling the gaps, and then even finer grit filling the remaining gaps, etc, and just enough binder to keep it together nicely.

I guess the way to do it is start with a known volume of coarse, then add a finer grit until the total volume starts expanding, which means the gaps are filled. Then add an even finer grit until the volume just starts expanding again, etc.

BTW, According to its MSDS, JBWeld is 5-10% powdered iron. They claim it doesn’t conduct electricity, but it sounds like it might conduct, slowly, under some conditions, or at least some people on the Google think it may have added slow short to some of their circuits.

Here you go, Lapidary Diamond a couple of large and small grit sizes to help fill in the space. I have lost the link to the research document but lapidary diamond is synthetic and because of the structure of its crystal being a tad bit different from real diamond, it conducts heat better than real diamond. (At least the synthetic diamond used for the engineering paper) I was going to try but did not have a good recipe for the epoxy, until now…

Have you gotten a chance to try it yet? I was looking at some materials’ thermal conductivity properties here and it seems to me that maybe a high concentration of diamond in epoxy might be a better heat conductor overall than pure copper! Do you think that’s possible? Especially, if it’s true that the lapidary diamond is an even better conductor than real diamond, it should do quite well if you can get enough diamond mixed into the epoxy. What do you think?

Edit: For example, if diamond has a thermal conductivity of 1000, and epoxy is negligible (0.35 though, if ya want to know) then a 40% mix would have a thermal conductivity of 400, right? Or am I doing the math wrong?

How much more light comes out the front, comparing a hot driver (one that truly needs cooling) potted with plain JB Weld vs. same driver potted with JB Weld + diamond/silicon carbide?

Who's got the time and resources to answer another innocent comfy Q? Smile Sure is a simple question... Wink

I'm assuming this JB Weld and fine grit silicone carbide harden as much as JB Weld itself?

If it hasn't been tested there's no way to know if it helps or not. I'm not a fan of doing extra stuff that has no benefit, sorry but that's getting too close to superstition for my taste. :shrug: