Seeking advice on Pill/Heat Sink to be constructed from 1/8" x 40mm dia. copper plate

Flashlight Modding and DIY Parts
1

Please kindly take a look at the following sketch:

I could use 1/4” plate, but it is more expensive, and using 1/8” provides me with greater flexiblilty (in case I need more “depth” or “height” for the Driver), and is also easier to shape.

Right now, I am planning to have two “solid” discs, and two discs with a 22mm “donut hole” for the Driver.

I am a little worried about poor heat transfer between the discs, but I have found some good (trustworthy) thermal grease. Both Dow Corning TC-5022 and Shin-Etsu X-23-7921-5 appear to be “excellent” as far as “Thermal Interface Materials” go.

(Dow claims thermal conductivity of 4.0W/mK for their TC-5022 which is supposed to be very high, but I really don’t understand what that number means. All I know is that apparently most other “high end” CPU greases are making outrageous claims, but don’t actually measure up despite their cost. One company claims 8.7 W/mK, but when tested by the US National Renewable Energy Laboratory, it was found to be only 0.94 W/mK.)

Anyway, I am planning to use a thin layer of good thermal grease and bolt the assembly together with small flat heat screws and nuts.

I don’t wish to get into any large scale soldering, but I may end up using some very soft solder around the outer edge of this assembly to help achieve a good tight fit between it and the host body.

Anyone see any inherent problems with this design?

Any and all comments and suggests will be greatly appreciated.

2

Why don’t you just solder the plates together? Gain the space of those screws back to copper thermal path material and have a solid bond with better thermal path between them.

And wouldn’t one plate cut out for the mcpcb be sufficient? You really don’t need any indention it’s just there to help in centering. If you can cut a thin line in the top surface where the mcpcb is centered, you could easily re-flow it to your stack as well.

Can you say what you plan to run on this? And at what power level you’re aiming for? What it’s going in also has a large bearing on the end result of course.

I’d say you’re off to a great start, looks like you’re having fun with it anyway and that’s the main thing. :slight_smile:

3

For one thing, I have never soldered anything like 1/8” copper discs together before. I am not sure exactly how to go about it, nor whether or not it can even be done with my current soldering equipment.

Another thing, unless you can somehow “slather” the entire surfaces with solder, it seems to me that you are almost certain to have air trapped between the layers. As I understand it, this is very bad… Therefore the need for thermal grease (I think).

But, I am certainly open to suggestions. That is why I posted the sketch. Can you tell me how best to solder such discs together?

4

I have had cause to use copper mcpcb’s with the dielectric layer sanded off, stacked 5 high. I just put a dollop of solder paste between each layer and put them on a steel plate on my stove top. When the solder melted, the pieces mated together beautifully. I removed them from the plate with hemostats (essentially needle nosed pliers) and set them on a heat sink to cool, but before they could I pressed down in the center to make sure the solder layer was as thin as possible.

They will self center if left alone, in my case above they were contained inside the head of a flashlight with superb anodizing.

Solder paste is really nice stuff to use in many cases. When re-flowing a copper star into a brass or copper pill I also re-flow the emitters at the same time. I do this with triples and quads. Everything settles into place at once and it’s very neat, usually. Large pieces of copper like yours are difficult to solder to with an iron as they absorb heat faster than the iron can make it. Because of this, it can even be difficult to solder wires onto the star pad. So it’s a good idea to place some solder paste on the pads when re-flowing the emitter so it will be there already and easy to make the connections.

You could always make the stack with 2 of the screws as you’ve designed, even all 4, but put the solder paste between the layers and heat them, even with a torch if need be. The clamping action from the screws will squeeze out the excess solder when it melts and you’ll have a one piece robust heat sink.

5

Thanks for the detailed information on solder paste and how to use it.

I must get some and “play with it.” It certainly sounds like the “way to go” with my Pill/Heat Sink construction. I am still checking into it, but it certainly seems that solder paste has vastly superior thermal conductivity when compared with even the best thermal grease. Thanks for the great advice.

6

Look up Old-Lumens on this forum. He’s got a lot of great information that you would love, from soldering copper discs together to hand filing and re-flowing, he’s one of the Master’s around here that get’s everyone revved up to do things. :wink: (follow the link, check out his posts, and have fun!)

About a year ago, I had no clue how to do this stuff. Old-Lumens challenged us all to do something by hand, be creative, build a light from scratch using hand tools. He also challenged the guys with machines. I did something I didn’t think I’d ever be able to do and that taught me to dive in, go for it, and I’ve been mod crazy ever since.

Images deleted, out of context…Norm :stuck_out_tongue:

7

I skimmed or skipped some stuff, but I’m just here to agree with DBCstm: soldering together is the way to go. Using a thermal paste is full of potential problems. The screws and clamping are a pain. Some thermal paste can dry out over time. You’re just as subject to air gaps with the TIM-and-clamp method as with soldering, IMO. You could even have thermal pumpout or something where the TIM migrates out from between the surfaces.

Finally: the solder is so superior to thermal paste I’d estimate that you could have 10% contact (and 90% air gap) and still have a superior connection.

8

Thanks for your post, wight. I started another thread because I felt my original thread was getting close to running its course. I figured a new thread about the pill assembly alone would allow a “fresh start” and possibly attract other opinions.

Can you (anyone else, feel free to chime in) recommend a particular brand of solder paste with high thermal conductivity?

I will address your questions about the “stock switch” assembly soon, back on my previous thread.

9

I wouldn’t worry about it, frankly. With a large and dependable interface, any old solder should do. While some are (much) better than others, I just don’t think it’s an issue here.

10

I’ve heard of guys using a grater to make dust out of their solder wire, then use that with some flux to make their own paste. They have claimed it works the same.

I use the name brand Kester EP256 and have heard of others using a “Mechanics” brand but have yet to try it.

11

(I use Kester EP256 as well.)

12

I only have tried the “Mechanics” brand. I felt that it had a lot of solder flux remains after reflow.

13

I use this also, for reflowing emitters.

14

One problem with soldered metal is eventual corrosion, from metal differences and remaining flux.

You can have boiling water handy and drop the metal into that as soon as it cools enough that the solder is solid, that blasts the flux off leaving bare metal.

For a much better copper-to-copper join, experiment a bit and see if you like “silver brazing” aka “hard soldering” — which requires getting the metal up to just barely red hot with a propane torch, but makes a better bond, jewelers use this method (and in fact construct fairly complicated things because silver braze wire comes in four or five different melting point versions so with careful heat control you can assemble something one piece at a time, each bond using a slightly lower-melting silver brazing wire).

Here’s a good discussion: About Solder | Nancy L T Hamilton

Look for the higher percent silver wire. This is ok in my experience:

Here’s a much lower melting silver solder from the same source:

Be wary that after a while the little plastic tube of flux will (a) harden if you failed to keep the cap on tight, and (b) split and coat your fingers with very nasty chemical if it’s more than a year or so old, so as with everything else, eye protection and care.

And with that, too, dunking the still-very-hot metal into boiling water will make the remaining flux explode off the metal leaving it close to clean, requiring just a bit of elbow grease.

15

I’d put the wire holes a bit further out and solder copper strips to the mcpcb solder points. One of the most common problems is solder bumps shorting to the reflector.

16

The part of the pill in front of the LED does contribute something to the heat path, especially if there isn’t enough space to make the pill as thick as you would like behind the LED. A thick or long pill is especially important if it does not screw into the body or fit it very snugly.

17

But only as saturation occurs.

I’ve seen a thermal imaging picture showing the heat path under the emitter and it’s almost identical to the angle of light coming out the front, a cone shape, giving the total an “X” configuration when looking at where the heat goes under the emitter and the light output at the same time (light above, heat below, emitter at the point in the middle).

So the mass to the sides only comes into play in a bottleneck situation where the heat backs up, unable to flow out as fast as it’s produced. This of course is far from ideal as we are wanting the heat sink to take heat away from the emitter.

18

That says four watts per meter degree kelvin. Since it is a temperature difference a degree kelvin (or absolute) is the same as a degree Celsius (or centigrade).
The heat flow in watts is 4.0 * the area / thickness * the temperature difference, in the above MKS units. The area is in meters squared and the thickness is in meters, so one m cancels out in the units for heat conductivity.
Here is data in Wikipedia:
Lead is 36 W / (m K), like stainless steel. Solder is similar to lead, I think. A few times as good as that paste.
Aluminum, cooper, silver and gold are around ten times as good.
Air is a thousand times slower, unless it circulates. That is why down, wool etc. make good insulators. They hold the air in place. If there is a temperature gradient and space, air does circulate. But between the star and the pill is too narrow a space.

19

Thanks for your very informative post. I had been looking mostly at a webpage that lists only metals, but the various different “units” were giving me difficulty when trying to compare them with thermal grease, despite a convenient conversion webpage.

I have since located the following graph showing both Thermal and Electrical Conductivity, courtesy of Cambridge University:

I believe that Kester EP 256 a mixture of 63% tin and 37% lead. As you can see on the above graph, tin is “better” than lead, but as shown in the chart you posted, even lead is “much better” than thermal grease.

Thanks again for your post. I am learning a lot about things I never needed to know before.

20

Interesting that tungsten alloys have fairly high heat conductivity and titanium way low. The thinnest wire available, at least at one time, was gold plated tungsten. Titanium must make good pot handles and such, except for its price and workability. Titanium is not suitable for high power flashlights.
“Lead free” 99% tin solder would be better than 63/37 where heat or electrical conductivity is an issue. Its higher melting point would take getting used to.