• Hello ES! We could use some help to get us past the finish line on building the new knowledgebase for the forum.
    Can you donate? Please see our fundraising page. Thank you!

thermal glue - does it work well for adding heatsinks?

neptronix

Administrator
Staff member
Joined
Jun 15, 2010
Messages
22,769
Location
Utah, USA
So i have a lightest.bike mid drive and i finally found it's limit of about ~1200w continuous.
It's a super efficient motor, just has a super thin case, and i think a little heatsink slapped on the side where the stator conducts the best to the case might result in a noticeable bump to continuous wattage.
The only problem is that i don't have spare metal to add bolt holes to fix the heatsink on that side.

Was wondering if thermal glue was a thing and found a few on google shopping, such as:
https://www.amazon.com/GENNEL-Conductive-Silicone-Adhesive-Compound/dp/B072MSXHJD

I'm wondering how a thermal glue would compare to a thermal paste - in terms of heat transfer?
I'm noticing that they say not to use it on a CPU, so to some degree, i imagine it is not good as thermal paste.
 
If it kinda sucks, maybe the correct tactic is to make a ring of some super strong epoxy around a large amount of thermal paste for best results instead.
 
I don't know how thin the case is that you want to attach a heat sink. But if you can do it carefully, what if you drill and tap small holes directly into the case , corresponding to mounting holes of your chosen heat sink? Then all you'd need is simple thermal paste
 
23years ago I glued ferrites to a watercooled aluminium plate.
It was for a 30kW inductive charger prototype.
I can't remember the number but it was an adhesive from Dow Corning.
Thermal paste can conduct more heat.

Nowadays they use double sided adhesive thermal pads to fix mosfets on heatsinks without screwing or clamping them.
 
If it kinda sucks, maybe the correct tactic is to make a ring of some super strong epoxy around a large amount of thermal paste for best results instead.
I'll let you know in a few months when I finally get around to adding the heat sinks to my motor. I plan on trying them without anything additional to secure them (safety wire, etc.), because I'm planning on them being removable or pried off in case of a spoke break.
Seems like the same stuff, but this states:
  • Strength of connected buildings: 55 lb.
 
I don't know how thin the case is that you want to attach a heat sink. But if you can do it carefully, what if you drill and tap small holes directly into the case , corresponding to mounting holes of your chosen heat sink? Then all you'd need is simple thermal paste

Really thin and with very little spare metal hanging around.
I'd need to use 2mm screws :/ kind of not appealing.

So far the best i can find for a thermal glue is:
https://www.amazon.com/dp/B00HKJ6QHK/ref=cm_sw_r_cp_api_i_EtIhBbPTMZPSE?th=1

Rated at 1.4W/(m-K)
Whereas a good thermal paste is 10x that.

Nowadays they use double sided adhesive thermal pads to fix mosfets on heatsinks without screwing or clamping them.

I quickly googled these and see that they can be in the 6-10W/(m-K) range which is pretty impressive.

In this case it might be feasible to wire wrap so that the adhesion isn't challenged by all the vibration the bike experiences

Not optimal, but not bad solution. Maybe the same idea applies where i cut a pocket in the pad and fill it with the most conductive paste money can buy.
 
It's a super efficient motor, just has a super thin case, and i think a little heatsink slapped on the side where the stator conducts the best to the case might result in a noticeable bump to continuous wattage.
If you think it will survive your conditions, something with the thinnest fins and the most of them, should provide the most surface area for getting rid of the heat. If the fins are splayed out radially it helps, vs all being parallel.

If you need something tougher, then thicker fins spaced farther apart, or even ones iwth pegs vs fins (for more surface area).


Back in the old computer-overclocking-for-gaming days, there were some decent thermal transfer silicones used for mounting liquid-cooling-system's heatsinks to the electronics side of harddisks. I can only imagine they've gotten better in the couple decades since.

It's likely that the ones we used were by the big PC heat-managment companies like Thermaltake, etc., but I don't know if they bought and rebranded what they sold, or if they designed and made it themselves.

The two biggest characteristics to be concerned with are the thermal conductivity (which needs to be high), and the curing volume-change (which needs to be negligible). (thermal volume change, too).

The former for obvious reasons, the latter because you don't usually want mechanical stresses added to a system from expansion, or the heatsink to pull away from the cooled device from contraction.

I'm wondering how a thermal glue would compare to a thermal paste - in terms of heat transfer?

That's dependent on the specific glue...and how you use it. Most pastes are better than glue, but:

If you use it the same way you would use thermal paste, it doesn't matter much how good it is, because you are only using it to fill the tiny imperfections between the flat surface of the HS and the flat surface of the cooled device. The rest is direct metal-to-metal contact.

All the thermal transfer material is intended to do is fill the air gaps, which have a pretty awful thermal conductivity, so you could literally use toothpaste and get better than that. :p


If you use the paste or glue in place of metal to metal contact, then the entire thermal load passes thru the paste or glue, so you only have whatever it's thermal conductivity is.

So the more metal-to-metal you've got, the better heat evacuation you get.


I know that lots of systems don't use it this way, but you'll get much better results if you do.

Oh, also, if you use paste rather than metal contact, then if (as usual) the carrier flows away or evaporates over time, then you have air gaps and particles, so much worse thermal transfer as the system ages. The glue would have less of this problem depending on it's expected volume change rates / conditions.

I'm noticing that they say not to use it on a CPU, so to some degree, i imagine it is not good as thermal paste.
Well, it's partly because the thermal paste is probably much better at heat transfer, but It's also that you can damage the CPU (or it's mounting socket, or even the motherboard) trying to remove the heatsink, and you usuually can't access the CPU-mounting-hardware without removing the heatsink, so gluing the HS on the CPU can make a system unserviceable.
 
Sounds like the pad would be the best way to go for adhesion.
You're right, thermal paste could be a problem in my hot deserty environment if it were not in an airtight pocket, which i can't guarantee unless i'm using RTV on the edges, which would handle expansion and contraction a lot better.

Apparently there's some RTV's that are designed to have OK thermal conductivity, close to the glue i was thinking.
https://docs.rs-online.com/d8cf/0900766b80f8df6e.pdf

Maybe the best outcome could be had with thermal RTV on the outer edges. I can't get good metal to metal contact that way though. This might render the paste not useful.

Thinking about all these things, i think just a thermal pad is the winner here
 
I quickly googled these and see that they can be in the 6-10W/(m-K) range which is pretty impressive.

In this case it might be feasible to wire wrap so that the adhesion isn't challenged by all the vibration the bike experiences
That's enlightening. I should have done more research. In my case, I got super thin self adhesive thermal tape to place on the edges of the heat sink, then a glob of thermal paste in the middle, the thought being the tape will serve to make up for the curved contour of the motor and flat heat sink, as well has holding the sink in place while the thermal glue sets up.
I did a calc at one point to figure out the thickness that would work with the curved surface.
Maybe I should use a couple of thickness of tape and forget the glue.
 
That's enlightening. I should have done more research. In my case, I got super thin self adhesive thermal tape to place on the edges of the heat sink, then a glob of thermal paste in the middle,
Keep in mind that anything between the metal surfaces is much less conductive than the metal.

I don't recall the actual numbers, but aluminum to aluminum direct contact should be in the couple of hundred range (copper twice that? steel half that?), air is less than one, and most of the pastes were around ten.

So the non-metal-contact areas could have as little as 1/20th (or worse) heat transfer, vs the m-m ones.

That's aside from the problems you get as the carrier flows away or evaporates over time.

Thinking about all these things, i think just a thermal pad is the winner here

see above.....


As long as you're not depending on this extra heatsink, and it's just a bonus exit for heat, then it doesn't "matter", but if it's a critical thing to keeping the system cool, you'll probably want the best conductivity you can get.

That might include convexing / curving the mating face of your heatsink to match the contours of whatever it's intended to mate with, if possible.
 
Arctic Silver used to offer a 2 part thermal epoxy. I had some left over I went to use more recently, but the hardener turned to a rock on its own, and when I searched to buy more of it, I couldn't find it, or it was ridiculously expensive, cant remember.

I recall one guy saying that it worked better than actual rear view mirror adhesive, for that specific task, though I never tested its adhesion personally, can onky say it has not failed in any application I used it. Generally epoxy is great at adhesion to a properly prepped surface

I Wound up using 24 hour JBweld instead. I was considering using some powdered aluminum or copper to mix into it to enhance thermal conductivity, but didnt bother. I weigh my epoxy parts a and b with epoxy. Judging blob sizes for equal parts A and B is highly inaccurate, and the closer it is to 50/50, AND thoroughly and evenly mixed, the harder and stronger it cures, with higher adhesion too.

I have some double sided thermal tape but dont find its adhesion impressive.
 
Last edited:
I don't recall the actual numbers, but aluminum to aluminum direct contact should be in the couple of hundred range (copper twice that? steel half that?), air is less than one, and most of the pastes were around ten.

So the non-metal-contact areas could have as little as 1/20th (or worse) heat transfer, vs the m-m ones.

Wow, here i am trying to figure out if i should have 1/20th of the conductivity or 1/200th of the conductivity..
I have the advantage of a flat surface area, but i don't have a lot of it.

I think i'll come up with some 4 bolt clamping system for the heatsink; there are 2 areas where i could bolt to the drive, and for the other 2 bolts, i could clamp on both sides to make work.

Unless there's some modern miracle of materials engineering we don't know about that would me the job much easier.
 
Screws were mentioned, but it seems more like a job for rivets.
 
Screws were mentioned, but it seems more like a job for rivets.

Unfortunately those wouldn't work for the difficult area where the stator is attached to the case ( this is an inrunner ).
But it would be possible to make a kind of clamp.
 
Yeah here's a visual of the drive. That area with the sticker is where i detect the most heat. The idea was to take a ~10mm tall heatsink, cut it in a crescent shape so that the idler doesn't hit it, and cool most of that area.

yee-jpg.372321


Maybe due to the limited amount of space, and some hope of getting proper metal contact, i might want a watercooling block instead
 
You could go for a round cpu heatsink, perhaps something like this one or others similar to the Zalman Flower, then simply trim away any bits that would interfere with moving parts.
1753062376126.png
I'm not sure how well the regular airflow past the bike would work with it's shape; you'd probably be better off with cutting up one wiht pegs that allow airflow from any direction to help carry heat from exposed surface area.
1753062708221.png

I found htose just with descriptive searches for "round cpu heatsink" and "round heatsink with pegs"
 
I weigh my epoxy parts a and b with epoxy. Judging blob sizes for equal parts A and B is highly inaccurate, and the closer it is to 50/50, AND thoroughly and evenly mixed, the harder and stronger it cures, with higher adhesion too.
Epoxy parts a and b have different weights per volume. So mixing them by weight you are probably off by around 15-20%. Epoxy is supposed to be mixed by volume.
 
That area with the sticker is where i detect the most heat.
The question is how much heat does that specific area radiate? If it feels warmer than other areas, it radiates more than other areas, but still very little compared to the overall surface area that radiates heat. Just looking at it, it might be 5% of total. Say the whole motor generates 200W of heat, you are trying to improve 10W of that. A heatsink won't do much overall, and the type of adhesive you use is basically inconsequential.
 
Epoxy parts a and b have different weights per volume. So mixing them by weight you are probably off by around 15-20%. Epoxy is supposed to be mixed by volume.
It depends on the specific epoxy. My one surfboard laminating epoxy is 100:45 by weight, 2:1 by volume, my 5 minute epoxy is 100:97 by weight, 1:1 by volume.

Ive made far.more mistakes with epoxy using volume than I have weight
24 hour JBWeld is 1:1 by volume, and weight.
 
49.7C over what ambient? 25-30C? That's only 20-25C over ambient then. That's very little heat coming through that area. I'd estimate around 10W radiating through an area of around 10 square inches.

With a CPU that's usually around 100W from under 1 square inch. Here thermal paste makes a difference, but with a CPU there's more than 2 orders of magnitude more heat flux density than in your scenario.

No matter how big a heatsink is or how good the thermal paste it won't ever radiate more than what is coming through. If you bring that down to 10-12C over ambient, which is very optimistic, that's still just a few watts.
 
That was at 30C ambient but i was also running 300w less peak power.
Now 70C in the same condition, so i'd like more cooling, even if it's just 10w.
 
As many have pointed out most thermal adhesives are pretty bad but I do recall seeing Tech Ingredients develop this awhile ago, Thermal Epoxy - 10ml + 3ml | tech-ingredients There is a whole video on how it works and even how to make it if you want, if I recall it performed quite well in comparative tests. Still as with any thermal application check to make sure both the body and heatsink are as flat as can be and ideally remove the paint/anodizing.
 
Back
Top