Lack of thermal paste in controller.

[Mark42]

I opened up my 48V controller and found that there was very little thermal paste between the metal bar the transistors are mounted on and the aluminum case. There were 4 little spots of paste along its entire 6" length. How disappointing. So I wiped off what was there and coated the metal bar and inside the case with a thin layer of silicone thermal paste and tightened down the screws connecting the case to the metal bar. I could see just a tiny bit of paste squeezed out between the mating surfaces inside the case, and that is just what you want to see. Then put on the end cap and was done.

I thought about using thermal tape, because it works so well on video chips in laptops, but there is not enough space for tape, so thermal paste it is.

This is a simple and easy thing to check if you have the common aluminum fin extruded controller. Even if you are not a techie, its hard to screw this mod up!

I have been building PC's and repairing laptops for years. I can not express how important thermal paste is for proper heat transfer from device to heat sink. Just a $2 tube of thermal paste from RadioShack will go a long way toward maximizing heat transfer to the aluminum case. I'm not sure exactly how much improvement this will give, but it was obvious that the factory did was far from what should have been done.

Excuse me if this has been covered before. I'm a noob.

 

[Punx0r]

I have been told that when thermal paste is properly applied it is so thin you can barely see it - the objective being to fill low spots on the surface, not cover the entire surface with a layer.

I cannot verify this one way or the other, but it would be interesting to know for definite.

 

[Mark42]

Punxor, you are correct. I smear a layer thin enough you can see through it. That is usually sufficient for heat sink mating surfaces. When tightened, excess squeezes out.

 

[amberwolf]

You don't want enough on there for excess to squeeze out. If you have that much, you have WAY TOO MUCH on there.

All it is there for is to fill the air gaps / scratches in the surfaces, because air sucks as a heat conductor worse than thermal paste does.

You want METAL to touch METAL, you do not want thermal paste blockign that transfer of heat. It does not conduct heat nearly as well as direct metal to metal contact will.

Anywhere you have thermal paste between two metal surfaces taht would otherwise touch, then when the paste dries out, now you have NO contact and just air, and a very poor heat path vs the metal to metal contact.


If you want the technical details and math, you can check out various sites with thermal paste and lapping technique testing articles from CPU overclocker forums and websites; there have been some good ones over the years on Tom's Hardware and Anandtech, among many others.

 

[Mark42]

Don't get too upset about paste squeezing out the sides. When you separate the parts, you will see it was full contact and only imperfections of the surface are filled, and that is typically in very high tolerance devices like CPU's in computers where surfaces have been polished. This controller doesn't even come close to those tolerances. The metal bar is plated and a rather rough surface compared to a CPU or transistor, and the extruded case is just a mass of grooves. So there is a lot of imperfections to fill on these non-precision devices.

Point being, most of these controllers could benefit from a fresh application of thermal paste if mine is any representation of the norm.

 

[friendly1uk]

The 20A controller in the box like bottle batteries I got had no paste at all. In a situation where all the semicondutors were closely mounted on a small plate, that should transfer the heat to the sole plate which is also small in comparison to other controllers. That is the worst case I have seen. I added some of course, and lapped it in well rather than just letting compression squeeze it out. Happy in the knowledge that it would be hard to make it worse lol

 

[Mark42]

HaHA! You are beating the results of low cost assembly over seas. LOL!

 

[Overclocker]

You don't want enough on there for excess to squeeze out. If you have that much, you have WAY TOO MUCH on there.

All it is there for is to fill the air gaps / scratches in the surfaces, because air sucks as a heat conductor worse than thermal paste does.

You want METAL to touch METAL, you do not want thermal paste blockign that transfer of heat. It does not conduct heat nearly as well as direct metal to metal contact will.

Anywhere you have thermal paste between two metal surfaces taht would otherwise touch, then when the paste dries out, now you have NO contact and just air, and a very poor heat path vs the metal to metal contact.


If you want the technical details and math, you can check out various sites with thermal paste and lapping technique testing articles from CPU overclocker forums and websites; there have been some good ones over the years on Tom's Hardware and Anandtech, among many others.

on a microscopic scale there is VERY LITTLE metal-on-metal contact that's why TIMs (thermal interface materials) are used.

with low viscosity TIMs the excess will simply flow out with sufficient clamping force. this produces the most consistent results. if you apply too little you risk having air gaps.

with higher viscosity TIM or PCM (phase change material) the application of heat may be required to obtain the optimal bond line thickness

 

[amberwolf]

with low viscosity TIMs the excess will simply flow out with sufficient clamping force. this produces the most consistent results. if you apply too little you risk having air gaps.

Well, what I said in my quoted post is based on my direct experiences with many experiments on CPUs and PSUs and harddisks in my computers, years back, trying to make a truly silent PC for audio / studio recording use (after having read various test articles on methods and types of paste, lapping methods, etc).

I never went to extremes on lapping anything, just basics, but I did find that using only enough to fill visible scratches/etc and leave all metal exposed to view (for instance by applying it with a creditcard-type of plastic card), gave the best heat trasnfer over the time I used things before trying other experiments/methods, based on temperatures reported by the boards' sensors on programs that can report htem in Windows.

I wouldn't call my testing precise in any way, but it was consistent in the results, in that whenever I used thicker paste than that, regardless of methods I used to try to squeeze the excess out when I did try that, it would never do that and end up drying out in weeks or months, and begin to quickly lose heat transfer ability then.

Perhaps if I could have applied sufficient pressure without damagng components (board, socket, etc) I could've gotten better results?

 

[izeman]

I always apply an amount big enough to cover the whole surface slightly. If you put the controller in the box and secure all screws a little of the high viscous paste is squeezed out. When you remove the controller from the case you will realize that there is almost no paste left and only scratches filled with it. So the remaining film must be super thin.
Results are great. I run a 9 fet at 45a constant for 20min and the controller stays below hand warm.

 

[Mark42]

I've put about 50 miles on the bike so far. Longest trip was 28 miles going 15 mph or less most of the time. I did stop to check the wires a few times during the trip and the controller was just warm to the touch and its inside a saddle bag with the battery. Either I'm not stressing the controller, or the case is wicking away the heat well. Either way, its performing good.