Adding fins to a C8

I had this idea for a while now and finally got around to making it happen. It’s all relatively easy lathe work.

First, I severed the head where the bottom of the star sits, leaving it a little long so it can be sanded to it’s final length later. Fortunately, everthing ran true with the battery tube clamped in a collet and the head screwed on. Not having to deal with a jawed chuck for this particular operation was nice! Next, I cut the remaining threaded portion of the head down to a wall thickness of less than a millimeter (1 inch total diameter).

For the heatsink, I used 1.75” diameter x 6” aluminum bought on ebay (2 for $10) and bored into it with a .75” drill bit. Then, I widened it to about .0003” (just a guess) less than 1” to give the previously saved threaded section enough interference to be pressed in permanently. (Cutting new thread’s isn’t a option for me at this time, nor is it necessary. Plus, I’m trying to keep it simple.)

The fins are cut with a .090” tool leaving them .085” thick. Varying the depth over it’s length should insure better thermal transfer to the lower fins than just a thin sleeve.



The pill gets modified to screw in from the battery side and the LED gets mounted directly to the finned heatsink topside. Installing the driver and soldering is done after what’s left of the pill gets screwed in. It shouldn’t be much more difficult.

Still not finished but this is how it’ll look.

I need to make a yoke to secure the top half of the head in place. I could just use screws alone but it wouldn’t be very durable.

Having screws in the head also means I can quickly convert it to direct drive by simply removing a makeshift insulator and connecting the (-) to ground.

The yoke is made and now things need to be drilled and tapped. I’m using six 3mm screws which I want evenly spaced so hopefully it can be assembled in any position.

Inside the thin wall looks a little rough because I hit a nasty resonance on the last pass. Using a long boring bar and not taking time to support the piece very well are to blame but it wasn’t the critical side so I didn’t care. On the positive side, it gives it that original Chinese quality look. :wink:

I used a Sharpie and etched 6 arcs at a radius of 15mm carefully using my caliper.

Then, I measured and scribed across each arc at 15mm to get my centers.

Using a pin vise rather than a punch to start the hole helps center the bit more precisely.

After the holes were drilled to 2.5mm, the yoke was super-glued to the heatsink and used as a guide to start holes into the heatsink. The holes in the yoke are drilled to >3mm and heaksink drilled/tapped after the the parts are separated and cleaned with acetone.

In hindsight, I should’ve measured the screw head diameter before I bored the inside diameter. I thought 5mm would be enough. Doesn’t hurt anything but looks funny.

It’s all done except for the electronics. For test fitting, the LED is just held in by pressure from the reflector.

Wow. That looks great. That is a massive heatsink.

Heavier than I expected. If someone was to hand it to me when it was freshly machined, I’d think it was SS. I’d like to make one of copper but I think it’ll be about 3 times heavier. Copper is better in small doses.

Nice machine work lightme. What running gear are you putting in this light?

Looks good lightme. Do you use CAD software for the drawing above?
Looks like with a little mill work you could add a momentary switch in that heat sink and a dr.jones driver.
Nice job!

Thanks. For now, I’m gonna run what I got so I can make comparisons to my C12s. Qlite or Nanjg (3A) and a XML2 T4 5B1. It wouldn’t surprise me if the LED runs cooler and the exterior hotter. I think that’ll determine what eventually gets used.

No CAD, just Paintshop Pro (it was free).

Would that allow me to vary the brightness in small increments? I haven’t been paying attention to fancy drivers.

Finally, the yoke is made and now things need to be drilled and tapped. I’m using six 3mm screws which I want evenly spaced so hopefully it can be assembled in any position.

Inside the thin wall looks a little rough because I hit a nasty resonance on the last pass. Using a long boring bar and not taking time to support the piece very well are to blame but it wasn’t the critical side so I didn’t care. On the positive side, it gives it that original Chinese quality look. :wink:

I used a Sharpie and etched 6 arcs at a radius of 15mm carefully using my caliper.

Then, I measured and scribed across each arc at 15mm to get my centers.

Using a pin vise rather than a punch to start the hole helps center the bit more precisely.

After the holes were drilled to 2.5mm, the yoke was super-glued to the heatsink and used as a guide to start holes into the heatsink. The holes in the yoke are drilled to >3mm and heaksink drilled/tapped after the the parts are separated and cleaned with acetone.

In hindsight, I should’ve measured the screw head diameter before I bored the inside diameter. I thought 5mm would be enough. Doesn’t hurt anything but looks funny.

It’s all done except for the electronics. For test fitting, the LED is just held in by pressure from the reflector.

Before I put an XML2-T4 in it, can someone tell me if higher current makes an LED appear much cooler? Would a U3 be too blue?

I don't think high amps like 4-6A (resonably high) will make a LED appear to be a cooler tint. U3's, like XML U3? Well, U3's have limited tint, just cool white - haven't heard of any legit U3 neutrals.

Love the way your C8 came out! Similar, sort of, to a L4 but maintains the classic C8 look at the head.

Nicely done! Have you ever considered using a copper core? All you need to do is press fit it in a bore in the aluminum that is .001" under the size of the copper. 5/8" copper rod works nicely. If you don't have a press a BFH will do. Just make the bore for a sliding fit, then go to town with a 2# or 5# hammer. The copper will mold itself into the cavity nicely. Then machine as required.

I didn’t want to use solder between the star and heat sink which I understand is what gives the copper core an advantage. Without solder, the heat has to transfer across two heat spreaders. The transfer between the copper and aluminum won’t be as bad as between the star and copper because of it’s greater contact area but a continuous solid metal is still best. There was a test done with annealed copper that showed that no matter how hard you press it against another surface, it couldn’t match the transfer rate of continuous metal. Soldering a star to copper is a different story because you eliminate the bottleneck by making the small restrictive heat spreader (star) move heat a lot faster to a much larger heat spreader.

Even without a soldered star, I think if done right, copper may help move heat quicker away from the source outward and spread down a thin wall of fins without having to make the light longer by using a 5/8” diameter rod to get the same surface contact. But without solder, wouldn’t the star’s bondline always be the bottleneck?

I have thought about this quite a bit. Looking at the emitter, substrate, star, pill and head as a heat transfer system the area of the substrate is the most limiting factor. I think this is why the MT-G2 can survive much higher amperage than other LEDs.

MT-G2 has an 8.9 x 8.9 substrate giving it 79.21mm2 surface area.

An XM-L2 is 5 x 5 yielding 25mm2 surface area.

The XP-G2 is tiny at 3.45 x 3.45 for 11.9mm2.

So the XM-L2 has a bit over twice the surface area the XP-G2 and the MT-G2 has a bit over three times that of the XM-L2.

That is why I think the die size is the limiting factor for current.

Interfaces are always an impediment to heat transfer when compared to homogeneous material. However, not all interfaces are created equal. A press fit between two smooth surfaces has a pretty small thermal resistance. Whereas thermal resistance for two rough pieces pressed together would be higher. If you were to press fit two well machined pieces of T6061 aluminum together and compare the heat transfer between that and a solid piece of T6061 aluminum the single solid piece would transfer heat better. There wouldn't be that much difference though.

Using a copper core does have the disadvantage of the press fit interface, but the advantage of the copper's properties should out perform solid aluminum as long as the copper core is large enough (diameter). You can look at a variety of CPU heat sinks to see hybrid copper/aluminum examples. My son's Xbox 360 has one and I'm pretty sure my i7 cpu does as well. Intel wouldn't do it if it wasn't advantageous. I think the main reason they use aluminum at all is to save $$$ in material costs.

Bucket, I agree with everything you said, it’s all about having sufficient surface contact area between materials. The right amount or more is better but not enough and your light might end up being bigger and heavier with worse cooling. I think my heatsink is too short to get enough contact with 5/8” copper rod to come close to the transfer rate of aluminum alone. A solution might be to use pipe threads and use a good TIM.

I can always add a copper core later but it can’t be reversed so I’d only do it for the big advantage of soldering the star. Otherwise, all the poor heat transfer between metals will negate the good transfer rate of the copper to some extent and gains will be minimal at best.

The only thing I know for sure is that a soldered star is best so copper is necessary if you want the best. For the most efficient compact heatsink, aluminum shouldn’t be used at all.