I just bought some fujipoly ultra extreme thermal interface material. This is for computers, but Fujipoly has a business relationship with Osram Opto for thermal interfaces for Osram’s commercial LED applications. Rated at 17 watt/mk thermal conductivity which is higher than any thermal compound or adhesive that I am awake of. It’s unfortunately 0.5mm thick and has peel off plastic on each side, but is only slightly tacky. I’m sure the led pcb will have to be screwed down to secure it and compress the fujipoly for better heat transfer. Anybody used this stuff before for anything, and what did you think of it? My goal for this test is to slightly overdrive an IR led on a noctigon. It has to be electrically isolated from the pill so figured I’d try this to see if it could be the best thermal option to isolate it.
I plan to test the fujipoly against maybe a berquist star and led, and maybe a cerakoted pill face for isolation. See my post 2-3 down for further explanation.
I’ve used them, “squooshed” between the toroid and the driver board, on some buck drivers, to try to spread the heat. While I can’t say for sure, I think it helped in that before doing that, I lost at least one diode that went up in smoke.
I think that’s not great! Comparing watts/mk between two materials is fine, but to determine fitness for an actual implementation thickness also needs to come into play, right?
Watts/mk is a measure of heat transmitted through a “unit thickness”, so to compare two TIMs the based on w/mk you must use the same thickness. If the thickness is not equal, whichever TIM has a lower thickness gets an advantage in proportion to the difference in thickness. Thinner is better.
17 W/mk is pretty sad when you have to use a layer that’s 0.5mm thick! (That’s extremely thick for a material which is much less thermally conductive than aluminum, copper or even anodized aluminum) It would be better to either use a layer of anodized aluminum or Arctic Alumina IMO. (or use a host where you can go body-positive, or use a host where you can just isolate the body electrically)
Whatever you’re planning for the IR LED, be sure to lookup djozz’s testing. Djozz has tested both a cheap 3535 IR and an expensive LED Engin IR emitter. There’s no point in going beyond that, since djozz demonstrates clearly where you’ll start losing output even with a large heatsink.
Wight, I don’t completely disagree with you. I’m doing this as an experiment. This product is made to go between cpu’s and heat sinks., and 0.5mm is the thinnest they offered. Heat transfer should be the same throughout whether it’s 0.5mm or 10mm. A copper pcb should transfer heat at 401 w/mk which is the same as a copper slug. I don’t see alumina as a good heat transfer interface no matter how thin. It seems it would act as more of a roadblock. A copper pcb transfers at 401. The copper pill transfers at 401. The AA transfers at <4. I imagine it being like a highway, there would be a long red light at copper highway and AA road. The heat cars end up backing up at the intersection waiting to cross through the intersection. This is going to happen whether AA road is a 2 lane road or a 6 lane road. Now I do understand that the smaller the AA road is the less traffic and the faster the copper hwy cars can pass through. The next red light is fujipoly lane. It is 6 lanes wider road. It has more traffic due to the size but could possibly a faster red light. This is what I want to test.
Unfortunately, the osram IR leds on noctigons have to be electrically isolated from the pill. I could buy a different pcb and flow the led which I’ve never done but may. I don’t want a reversed polarity light even though it’d probably be easier. In the dark and hunting, I want to easily change batteries in the dark without having to remember that this one goes in differently. I’d have to increase the thickness of an AA layer to ensure electrical isolation. Others that have done this say that increasing the thickness of the AA is a led killer due to heat. I’m not super overdriving the Led. I plan on pushing it at 1.4a. Everything I’ve read says AA is not a good choice for this application, so I’ve kinda crossed it out from the start. I saw this and found no info about it and flashlights. I figured I’d try to screw the pcb down over it and take a temperature reading. Then possibly try doing the same with the led on berquist star and measure the heat. I don’t know how much the fujipoly will compress. I don’t know how it will perform. No way to know but to try it and see. If it sucks between the pcb and pill, I can always use it on drivers, etc. It won’t go unused either way. Another option I thought about is to test and see if cerakote will electrically isolate. If it does, I could cerakote the top of the pill, use AS5 and screw a noctigon based LED on it and take a temp. All this for the sake of just screwing with stuff to see what it does.
I reread and edited my initial post to make more sense of what I’m doing.
I think that Arctic Alumina adhesive will produce a thinner layer than this stuff, I used it in a mod specifically to electrically insulate a DTP-copper board (Sinkpad in my case) with an Oslon IR-emitter from the flashlight body.
It performs more like an insulator than a thermal conductor. “Thermal conductivity: 1.3W/m.K NOT to be used as a substitute for thermal paste on CPUs, Video Card GPUs, or other applications where thermal paste should be used.”