Thermal Paste In Between Star and Heatsink

Hi all!

Sorry if this has been discussed in depth before (please point me to any thread). I have seen some people suggesting thermal paste (not adhesive) in between the star and heatsink followed by epoxy on the outside. May I ask if anyone actually have done this and have done any test on it? Is it possible without any screwing down involved? I am thinking to apply the paste, clam the star and heatsink and epoxy on the outer edge.

Also, would like to know which type of epoxy you used.

Thank you in advance.

I use Arctic Alumina it isn't expensive, IMHO using the right solution is always easier rather than trying to cut corners.

This is a topic that has been covered in the past, opinions vary.

Norm

Norm said:

I use Arctic Alumina it isn't expensive, IMHO using the right solution is always easier rather than trying to cut corners.

Norm

Click to expand...

I agree with you. But unfortunately, (good) thermal adhesive is not easily found in the place I live. I have bought a set once and it went bad in less than a year.

Where do you live?

Arctic aluminal thermal paste + epoxy = thermal glue?

Norm

Norm said:

Where do you live?

Arctic aluminal thermal paste + epoxy = thermal glue?

Norm

Click to expand...

Thank you! Exactly what "yellow" said in the third post. Did any of the members have a discussion about it? Or proved that it is doable?

I use system3 t88 epoxy mixed 1:1 with zinc oxide powder for my thermal glue. Performance is excellent.

Actually, will any expoxy work? I have seen one that states it will soften at 60 degrees C though.

All epoxies soften somewhat with heat. Some soften with time as well. 5-min is the weakest. I use t-88 because it is structural and has good properties. Its main use is laminating aircraft parts.

Thanks for all the help above! Have two small questions that probably do not need to be on a thread on their two.

I powered the LED (XM-L2) briefly and noticed there is a yellow tint light on the outer edge. The center looks fine though. Is this normal? This is without any reflector/lens. Just the heatsink, LED and wires. Secondly, can alcohol be used to clean the dome of the LED? Looks like there are some specks of dust or glue on it. And the dome feels soft.

I would not worry about what the LED looks like. Shine it on a wall and judge the color that way. Also, alcohol is proper for cleaning the LED, as per CREE:

"Don't use unspecified chemical liquids to clean the LED; the chemical could harm the LED. When washing is
necessary, please wipe the LED with alcohol at normal room temperature and dry at normal room temperature
for 15 minutes before use."

Thank you for the clarification.

I have two different setups:
1. XM-L, Aluminium star, Arctic Alumina Thermal Adhesive, Aluminium heatsink
2. XM-L2, Copper star, some thermal compound and epoxy outside star, Aluminium heatsink

I injected 3Amps directly to both LEDs and used a cheapo infrared thermometer and sweep around to find the highest temperature. Both are hitting 90+ and 100+. Am I doing something very wrong?

Edit: Forgot to mention temps are in degree Celsius.

Can we assume that your copper star has the LED directly soldered to it in all 3 positions? That is the only way to get the copper advantage. Also, I'm not sure you can accurately measure junction temperature with an infra-red thermometer. The reading beam is too large and imprecise. What you want to look for is how the LED handles higher amps - and for an XLM on copper you need to go higher than 3. Google up the XML on copper lumen chart from "Match" and you will see much higher power levels before the advantage of copper is obvious.

schizeckinosy said:

Can we assume that your copper star has the LED directly soldered to it in all 3 positions? That is the only way to get the copper advantage. Also, I'm not sure you can accurately measure junction temperature with an infra-red thermometer. The reading beam is too large and imprecise. What you want to look for is how the LED handles higher amps - and for an XLM on copper you need to go higher than 3. Google up the XML on copper lumen chart from "Match" and you will see much higher power levels before the advantage of copper is obvious.

Click to expand...

The LED + copper star I bought is those usual Noctigon ones and are soldered and so, I am guessing it is a "yes" to your first question. What I am really worried is the heat transfer from the copper star to the heatsink. The heat is too hot to touch and I am guessing it is a good sign that the transfer is working. But this test is done in an open air environment and I am also guessing that after inserting to the Maglite body, the whole setup may be "suffocated" and temps will rise even more (although I also wonder the heat travelling to the body of the Maglite will help enough).

schizeckinosy said:

I use system3 t88 epoxy mixed 1:1 with zinc oxide powder for my thermal glue. Performance is excellent.

Click to expand...

Where do you get zinc oxide powder?

Sorry, I missed this post. I use pharmaceutical grade powder (for skin creams, etc.) I get it from suppliers through Amazon. There are several.

For sure you should firmly attach the heatsink to the mag body, wither mechanically or with a press fit and thermal paste to transfer the heat to the Mag and thence to your hand.

ExtremeSeries said:

The LED + copper star I bought is those usual Noctigon ones and are soldered and so, I am guessing it is a "yes" to your first question. What I am really worried is the heat transfer from the copper star to the heatsink. The heat is too hot to touch and I am guessing it is a good sign that the transfer is working. But this test is done in an open air environment and I am also guessing that after inserting to the Maglite body, the whole setup may be "suffocated" and temps will rise even more (although I also wonder the heat travelling to the body of the Maglite will help enough).

Click to expand...

schizeckinosy said:

Sorry, I missed this post. I use pharmaceutical grade powder (for skin creams, etc.) I get it from suppliers through Amazon. There are several.

Click to expand...

So, I have been thinking of this route. How far does a pound of zinc go? What is your mixing ratio?

I am guessing original 6900 psi jbweld (works on plastic too) with %20 to %50 by volume of zinc mixed in.

Maybe mixed outside, if zinc as bad as lead for kids.

Also, to note, I ran across jbweld steel reinforced epoxy yesterday. But pulling up the msds, only %1 ferrosilicon. So, at first glance, not enough metal to do anything other than a marketing label on package. But see below on Actic Silver/alumina numbers..?

schizeckinosy said:

Also, I'm not sure you can accurately measure junction temperature with an infra-red thermometer. .

Click to expand...

you can't really mesure die temp. the closest you can do is to mesure the star temp, and not with ir termometer. you need contact probe. than you can somewhat guesstimate die temp.i usually go by, die temp is about 20c higher than star (dircet connect) temp. i might be wrong with 20c, but i have not fried a single led using this number yet. when heatsink, gets 75c i trim down current or shut of led.

and i would recomend if there is no right paste available, just polish heatsink\star, and press them hard using screwes, or by pushing reflector down, you will have better heat transfer, with no paste and polished\pressed hard together parts than if you used some diy paste and rough surfices that are loose.

Arctic Alumina:

Aluminum Oxide 10 mg/m3


Artic Silver 5:

Silver 0.1 mg/m3
Boron Nitride 10 mg/m3
Zinc Oxide 10 mg/m3 (dust)
Aluminum Oxide 10 mg/m3 (dust)

If memory serves, 1 cubic meter is 1000 liters, and 1 liter is 1000 milliliters. Maybe 1 gram of zinc is 50 milliliters of epoxy? Seems like a tiny amount.

degarb said:

So, I have been thinking of this route. How far does a pound of zinc go? What is your mixing ratio?

Click to expand...

I use about 1:1 zinc oxide to epoxy. So far a pound is lasting forever!