Heat sink paste
- Thread starter DavidAD
- Start date
Help Support Candle Power Flashlight Forum
Mr Happy
Flashlight Enthusiast
I think it depends whether you are trying to fix (glue) the LED in place or not. If you use epoxy it will be permanent, but it will also provide the mechanical stability needed if there is nothing else holding the LED on the heatsink.
qwertyydude
Flashlight Enthusiast
- Joined
- Aug 10, 2008
- Messages
- 1,115
I use arctic silver then run a bead of superglue around the edge of the star. While it sets I make sure to press the star down with weights or clamps. Works great, maximum heat transfer that way, no epoxy to impede heat flow.
depends on the LED. On one with leads like a SSC-p4/p7 yes.
On ones surface mount ones, no since you cant reach the leads, and even if you could you could overheat the epoxy as you heat up the carrier as you solder. The worst situation would be you just weaken the epoxy as you solder on the leads afterwards, and then it comes off later while its operating and overheats quickly.
- Joined
- Jul 15, 2007
- Messages
- 591
Wow- this is a great idea. Does it hold up well to repeated thermal cycles?
qwertyydude
Flashlight Enthusiast
- Joined
- Aug 10, 2008
- Messages
- 1,115
Yes I have it on a light I use regularly for over a year now. The reflector doesn't press the led down so it's up to the super glue to hold the led down and it's holding up just fine. But you have to apply the compound very carefully so it doesn't squish out all the way and prevent the glue from sticking. Oh and let the glue dry for at least 24 hours so the fumes don't deposit on the inside of your reflector and ruin it. I put it inside my computer case which always stays on and is toasty warm so all the glue dries.
Last edited:
ateallthepies
Enlightened
Does nobody use Arctic Silver thermal adhesive?
I used this on my Tiablo build.
Steve.
I used this on my Tiablo build.
Steve.
45/70
Flashlight Enthusiast
I could be wrong about this, as it's been a while since I was researching all this, but as I remember, the epoxy version is the only one that is 100% non conductive. So, if you're attempting to electrically isolate the emitter, star, or whatever from the heatsink, it's the only one of the compounds containing silver, that can be used. That, and the significantly stronger bond, are the reasons why I use the epoxy.
Dave
qwertyydude
Flashlight Enthusiast
- Joined
- Aug 10, 2008
- Messages
- 1,115
The arctic silver paste and epoxy are both non conductive. The difference is that the epoxy has less capacitance but at the frequencies that pwm we use may be, capacitance will not cause any conduction at all. This only applies to high frequency traces the arctic silver may go over. It can cause the high frequency signals to cross. But either one is bad for high frequency electronics use, but that doesn't affect anything except in the megahertz region or higher. Also if you have a star already you have to assume that it can be connected to ground with no issues as many lights use screws to hold down the star and they'll simply short the star to ground anyways.
The epoxy impedes heat flow. It's thermal conductivity is hard to compare within the product line and I think they do that on purpose. The epoxy is listed as 7.5 W/mK and the paste is listed as 350,000 W/m^2 but converted it is 9.0 W/mK so the paste greatly outperforms the epoxy.
The epoxy impedes heat flow. It's thermal conductivity is hard to compare within the product line and I think they do that on purpose. The epoxy is listed as 7.5 W/mK and the paste is listed as 350,000 W/m^2 but converted it is 9.0 W/mK so the paste greatly outperforms the epoxy.
45/70
Flashlight Enthusiast
Thanks for the detailed explanation, qwertyy. I knew there was some reason I picked the epoxy over the others. I guess maybe it was the wrong reason though.
Sometimes I really would prefer not to have to "freeze pop" to disassemble projects. It'd be nice to be able to just pry apart components, although I have no doubt that the epoxy holds things together better under harsh conditions. I've pretty much moved beyond the "show and tell" stage of the flashlight/torch aspect of this hobby and actually use my lights. This subjects them, unfortunately, to drops bumps etc.
I still would like a less permanent solution though, for alpha and beta testing etc. I've been using a Mil spec white paste for this purpose, however it definitely doesn't hold things together all that well. Unless the reflector pushes against the LED, for example, I don't trust it.
Dave
Sometimes I really would prefer not to have to "freeze pop" to disassemble projects. It'd be nice to be able to just pry apart components, although I have no doubt that the epoxy holds things together better under harsh conditions. I've pretty much moved beyond the "show and tell" stage of the flashlight/torch aspect of this hobby and actually use my lights. This subjects them, unfortunately, to drops bumps etc.
I still would like a less permanent solution though, for alpha and beta testing etc. I've been using a Mil spec white paste for this purpose, however it definitely doesn't hold things together all that well. Unless the reflector pushes against the LED, for example, I don't trust it.
Dave
Mr Happy
Flashlight Enthusiast
Less permanent solution: ?
http://www.luxeonstar.com/Thermally-Conductive-Adhesive-Tape-for-High-Power-LEDs-s/210.htm
Justin Case
Flashlight Enthusiast
- Joined
- Mar 19, 2008
- Messages
- 3,797
Arctic Silver epoxy thermal conductivity is listed as "greater than 7.5W/mK". The problem is that this could mean 7.500001 W/mK, 8 W/mK, 9W/mK, .... Who knows.
But if we take the conservative value of 7.5 W/mK, we can estimate the thermal resistance for AS epoxy, compare it to AS5 compound, and examine the impact of any difference for the example of a Cree MC-E driven at full power (about 10W).
The slug area of the MC-E is 5.4mm x 2.6mm, or about 14mm^2. Let's assume a bond line thickness of 0.005", or 1.27x10^-4m. If we use Arctic Silver 5 (thermal conductivity of 8.89 W/m-K), we get a thermal resistance of 1 K/W. For AS epoxy, we get 1.2 K/W. At 10W of power draw by the MC-E, probably about 80% of that goes to waste heat, or about 8W.
Thus, when we use AS5, we get a calculated temp increase across the bond line of about 8C. For AS epoxy, we get about 10C.
But if we take the conservative value of 7.5 W/mK, we can estimate the thermal resistance for AS epoxy, compare it to AS5 compound, and examine the impact of any difference for the example of a Cree MC-E driven at full power (about 10W).
The slug area of the MC-E is 5.4mm x 2.6mm, or about 14mm^2. Let's assume a bond line thickness of 0.005", or 1.27x10^-4m. If we use Arctic Silver 5 (thermal conductivity of 8.89 W/m-K), we get a thermal resistance of 1 K/W. For AS epoxy, we get 1.2 K/W. At 10W of power draw by the MC-E, probably about 80% of that goes to waste heat, or about 8W.
Thus, when we use AS5, we get a calculated temp increase across the bond line of about 8C. For AS epoxy, we get about 10C.
...never thought about electical conductivity from thermal paste through the pill:naughty:
How's about mixing arctic silver and arctic epoxy the same parts?
Have that read as suggestion in a german forum for modding a romisen rc-c5 to xp-g. Sounds nice for easy later swaps and I'm definitely planning to do this with mine also.
How's about mixing arctic silver and arctic epoxy the same parts?
Have that read as suggestion in a german forum for modding a romisen rc-c5 to xp-g. Sounds nice for easy later swaps and I'm definitely planning to do this with mine also.
Steve-at-Springboard
Newly Enlightened
Re: Stars
Just a footnote to one of the posts above - not all stars can be grounded. Depending on the construction, the slug may or may not be electrically connected to the aluminum in the star's PCB. Some lights will use non-conductive screws to hold the star against the heat sink. Also protects against shorting out traces on the top side of PCB. Be careful! Read the specs!
Just a footnote to one of the posts above - not all stars can be grounded. Depending on the construction, the slug may or may not be electrically connected to the aluminum in the star's PCB. Some lights will use non-conductive screws to hold the star against the heat sink. Also protects against shorting out traces on the top side of PCB. Be careful! Read the specs!
maxspeeds
Flashlight Enthusiast
Anyone use arctic alumina ceramic thermal compound? How does it stack up against the silver version?
qwertyydude
Flashlight Enthusiast
- Joined
- Aug 10, 2008
- Messages
- 1,115
Arctic alumina is thinner, it's used when the capacitance that Arctic Silver 5 can be a problem like on high frequency traces on computers. It has a conductivity >4.0 W/mK so does not conduct heat as well as Arctic Silver 5 which is around >9.0 W/mK.
45/70
Flashlight Enthusiast
Re: Stars
Yeah, in the Lux/Lux III days, there wasn't much of a problem with this, as the star was at negative (-) potential. Nowadays, especially when mounting emitters directly to the heatsink, you can run into problems.
Dave
Yeah, in the Lux/Lux III days, there wasn't much of a problem with this, as the star was at negative (-) potential. Nowadays, especially when mounting emitters directly to the heatsink, you can run into problems.
Dave
MikeAusC
Enlightened
I wouldn't say it greatly outperforms the epoxy - as Justincase's example shows - the LED will be 2 degrees cooler !
In reality, there will be a lot of direct metal-to-metal contact, so the amount of heat flowing through thermal compound will be even less. All the explanations suggest the compound is only used to fill in the airgaps caused by roughness of the metal.
I always lap the back of the MCPCB and the top of the heatsink with 1200grit paper to minimise air spaces.
qwertyydude
Flashlight Enthusiast
- Joined
- Aug 10, 2008
- Messages
- 1,115
A properly lapped surface will actually run better no matter what the compound is. The theoretical 2 degrees isn't that much better between the Arctic Silver epoxy and Arctic Silver 5 and even compared cheaper compounds Arctic Silver only performs about 4-5 degrees better. On my computer I lapped both the CPU and heat sink and saw about a 15 degree reduction in temperature. Yep this means taking out the $400 CPU chip and literally sanding it down and doing the same to the cpu cooler but it makes more difference than any heat sink paste can. I have to assume it also works the same way for led's.
