Soldering SST90 to Copper Disk with Indium

[Mick]

As those who have worked with the SST/R90's know getting 20+ watts out of a contact area of ~46 mm^2 is a challenge. The Luminus Devices star and other metal circuit board designs are a good answer if you can incorporate the form factor. For my helmet light design I decided to use a copper disk between the SST90 device and the aluminum housing.

Indium solder has a conductivity of ~70 w/mk compared to 63/37 lead tin at ~50 w/mk or Bismuth tin at ~20 w/mk. I use this Tix solder.

Indium can NOT be soldered to copper directly due to the formation of a brittle alloy. It is recommended to plate the copper with nickel first. This is the nickel plating kit I bought from Caswell.

The first step is to sand the copper disk flat. I use 320 W/D paper taped to a milled iron platen. Sand unti all the machining marks are gone.

Next Clean, clean, clean. Do the water sheeting test. Don't touch the mounting surface from this point on.

Now I plate the mounting area with the Caswell kit. Keep the tip wet and in motion. You don't have to rub, just go around the area in a continuous circular motion. This is the result ... as best as I could photograph it.

Now I solder the power wires to the SST90 with lead free solder. This solder melts at about 220 degC so the wires won't fall off when I use the Indium solder that melts at 135 degC.

The jig is machined from glass filled teflon and can take ~275 degC.

I now flow the Indium solder on the SST90 thermal pad. I calculater the length of solder wire necessary to coat the area with 2 mills (0,05 mm). This was a guess and seems to work out good. I see a little extra solder at the ends after reflow. I bend the solder wire sharply at 1/2" so I know how much to lay down. The complete thermal area should be covered with solder. Now I load the prepared SST90 into the second jig.

The hot plate is turned on and set for 180 degC and the copper disk is placed on top of the SST90 in the jig. The jig is then flipped and set on the hot plate.

The reflow took almost 5 minutes so I have to try to speed that up but no harm will be done at 180 degC. Some thermal compound might give me a better coupling to the hot plate but would make a mess. I did not wait long enough before moving this assembly and it skewed. I just reheated it and it straightened out. ( no picture proof ... sorry).

This is a 3D cross section of my helmet light showing the SST90 / copper disk mounting.

I will be inserting thermal sensors into a prototype light to measure the heat flow as best I can. The plots should be interesting ... stay tuned.

 

[65535]

Something about the heat sinking requirements and head mounted make me think of other issues before proper mounting of the led.

Can you say ouch?

I will say though, you have taken a very professional and quality approach to this.

Very meticulous.

 

[archer6817j]

That's very nice work! I'm looking forward to seeing this build come together. Always surprised at how many machinists are around here

Any guesses as to what the temp difference should be with the higher conductivity solder? I don't me to restart the copper/alu/brass thread!

 

[AnAppleSnail]

That's very nice work! I'm looking forward to seeing this build come together. Always surprised at how many machinists are around here

Any guesses as to what the temp difference should be with the higher conductivity solder? I don't me to restart the copper/alu/brass thread!

In theory it'll be:
~70 w/mk compared to 63/37 lead tin at ~50 w/mk or Bismuth tin at ~20 w/mk.

"(70* W) / thickness in meters" hotter than the heatsink with Indium
"(50* W) / thickness in meters" hotter than the heatsink with Lead tin
"(20* W) / thickness in meters" hotter than the heatsink with Bismuth tin

So proportionally speaking, it'll face 71% of the temperature difference as lead tin, and 28% of the temp difference with bismuth tin. How much is the difference? Well, if the tin is pretty thin, it won't make much difference. At 9A I think it takes 1.5mm to make a 1*C difference.

 

[ddv]

That's a heck of a set-up you've got there. I really like the jigs. Where do you buy glass-filled teflon? I've never heard of it before. How did you cut the squares? Obviously not broached -- tiny end mill? Dremel router bit?

Very nice work.

 

[PhotonFanatic]

McMaster Carr carries the glass-filled teflon--I'm sure there are other vendors though.

 

[wquiles]

Awesome post and pictures - thanks much for putting this together and sharing it with us

Will

 

[Epsilon]

Very nice job .

Mounting/soldering the led directly to the copper disk is always better than using thermal adhesive.

When using a star, how is the led mounted to the star? Soldered? This increases the contact surface area to the heatsink ofcourse, but every junction is one to much .

 

[65535]

Very nice job .

Mounting/soldering the led directly to the copper disk is always better than using thermal adhesive.

When using a star, how is the led mounted to the star? Soldered? This increases the contact surface area to the heatsink ofcourse, but every junction is one to much .

Any star mounted LED worth it's weight is soldered.

Not always, but always should be.

This is truly a very nice execution of mounting an LED.

You don't even want to compare the thermal conductivity numbers between this job and normal thermal adhesive.

 

[Rej]

Just came across this post and .....Great work Mick....any updates on your build??
Always nice to see a 'craftsman' at work!

 

[Microa]

Very interesting tutorial. Good to learn different approach. Thanks for sharing.

 

[hellokitty[hk]]

Very nice work!

How do you prevent the wires from touching the copper plate?
When I imagine doing something like this, I thought that you'd have to cover the wires with kapton.

 

[gadget_lover]

He milled slots that were deep enough that the wires were not touching anything.

Daniel