Noctigon copper mcpcb mod for P60 setups (improving thermal pathway)

Flashlight Modding and DIY Parts
1

The benefits of direct LED to copper mounting is well documented here on the forums. Copper mcpcb’s without a dielectric layer such as the Sinkpad or Noctigon do help tremendously in allowing LED’s to be overdriven. Overdriven LED’s means more heat and that heat needs to be transferred to the body of the flashlight host in an efficient manner.

P60 setups typically use a brass pill, which is a bottleneck in terms of thermal conductivity compared to the rest of the metal used in the thermal pathway to the body of the flashlight host. Brass has roughly half the thermal conductivity of aluminum, while copper has roughly twice the thermal conductivity of aluminum. Disregarding thermal interface materials such as thermal paste and epoxy the thermal pathway of a P60 setup is:

LED -> MCPCB -> Brass Pill -> Aluminum Reflector -> Flashlight Host

If the Brass pill is a bottleneck what can be done? There aren’t any widely available pills made of better materials such as copper or aluminum. Here is my solution.

My mod helps transfer heat from the copper mcpcb to the aluminum reflector directly without going through the brass pill. The amount of surface area where the small copper strips come into contact with the reflector isn’t that great, but it’s definitely better for heat transfer than going through the brass pill. Heat will still be transferred to the brass pill, just not as much as before with a regular setup.

Here is how I did the mod. First step is to scrape away the dielectric layer of the mcpcb on the edge and to determine how thick the dielectric layer is.


1.58 – 1.3 = 0.28mm thick dielectric layer

Next up is to measure the distance to be bridged between the mcpcb and the reflector. FYI I used a Fasttech XP SMO reflector (SKU 1208506).

—>
—>

Distance between the reflector and mcpcb with dielectric layer: 4.84 – 3.6 = 1.24mm
Distance to be bridged by copper strips: 1.24 + 0.28 = 1.52mm

Here are the copper strips for the mod that were from a piece of copper pipe. I filed them down just a tad bit too much (0.07mm), but in the end it worked out just fine.

Copper strips and LED reflowed onto the mcpcb. The reflow process added 0.08mm in height between the copper strips and the mcpcb. The distance bridged by the copper strips + solder paste is 2.83 – 1.3 = 1.53mm.


Here are some assembly pics of the dropin. Note the copper strips are above the top lip of the pill. Ceramique thermal paste was used between the mcpcb and the pill, while Fujik was used on the copper strips and on the threads of the reflector. Everything was screwed together tightly and allowed to cure/dry.


FYI the driver used is a Kaidomain v2 8xAMC7135 (3.04A). A XP-G2 driven at this level definitely needs a better thermal pathway; hence the reason for the mod to the Noctigon mcpcb being done. Last thing to do is wrap the dropin with some copper foil and I’ll be good to go.

2

Did you notice a difference?

3

Cool idea! Would work nicely when combined with this to get the heat from the reflector into the body:

Make the spacer ring just tall enough that the head/bezel doesn't bottom out when screwed down, so you get max clamping force at the reflector/spacer contact point.

4

simplec6:

I don’t have a similar pill setup without the mod done to do a direct comparison. I guess I could do a quick 5 minute runtime test with my lightbox to measure how much output sags.

comfychair:

Hmmm that looks awfully familiar to what I did with my Solarforce P1 a year ago. :wink:

5

Nice mod lilkevin715. You have put quite a bit off work into this mod. Well done.

6

Nice mods. :slight_smile:

does anyone make an actual copper pill for p60 stuff?

or one that doesnt taper and is at least as wide as the reflector?

Brian

7

This is pretty cool. If I had the patience I’d try it out. Maybe when I have some free time I’ll give it a go.

8

The closest I have seen is one I built myself, with a copper spacer ring soldered to the base of the pill. But of course it has to be undersized to still be removable, so doesn't have as much surface area contact with the head around the perimeter as the C-shaped spacer ring. Six of one, half-dozen of another...

9

Its probably not much help but this is one I’ve done in the past. You may be able to get someone that lives close by to you to knock one up.

10

I’ve finally got around to doing some testing to see how effective the mod is. For testing I compared three different XP-G2 dropins as I’ve built as listed below:

XP-G2 R4 4C @ 1.7A with Aluminum mcpcb

XP-G2 R5 3D @ 2.4A with Noctigon mcpcb

XP-G2 R5 3C @ 3.04A with Modded Noctigon mcpcb

The output of each dropin will obviously be different. What I am more interested in is output loss over time. FYI Each dropin is built in the same manner (Fujik on the reflector threads, reflector wrapped with copper foil) with the main difference being the mcpcb used. To measure output loss I used my lightbox (details here) and took measurements every 30 seconds in a 5 minute runtime test. I did not run the test longer than 5 minutes as the lights were not cooled externally in any way (no fan for airflow).

Here are the test results (click on the images to enlarge). Lux readings from my lightbox have been converted to Lighbox Lumens (LBL for short) and the percentage loss from initial turn on has been calculated.

Here is a graph of the LBL loss percentage over time of the three different mcpcb setups.

Output of all three mcpcb setups sags the most in the first minute of runtime, which is to be expected. After the first minute of runtime looking at the data and graph we can see two trends. The Aluminum and Noctigon (unmodified) mcpcb setups continue to loose output as time passes by. However the output of the modified Noctigon mcpcb setup levels off. This can be attributed to the better thermal pathway of the noctigon mcpcb mod; heat is transfered faster to the body of the flashlight host. If the runtime test was longer the output of the modified Noctigon mcpcb setup would have most likely sagged more eventually.

So to answer simplec6’s question (and my curiosity) the mod definately helps. Visually I can’t tell a difference, but that is why I tested output with my lightbox.

11

Thanks for posting your results lilkevin715 You have really excelled with the modded Noctigon.

12

Would be interesting to see 3,04A before and after mod. If you have one stock available, please make a comparison some time in the future. :slight_smile:

What I find interesting with your numbers are the fairly large sag on aluminium mcpcb, even though its “only” at 1,7A (I know 1,5A should be considered max), but it truly shows the advantage of the Noctigon@ 2,4A. More power, less sag.

These are numbers that show the P60 haters, and the “P60s should not be used at over 2 amps” people, that it can be no issue. No surprise for me, my eyes have told me this a long time ago.
But the performance gain with your mod was way more effective than I would have imagined! Great work, and thanks for sharing!