Danged nice! Thats what tools and that thing the brain box surrounds are made for!! 
Dan.
Now that’s what I call a serious amount of skill. You start making them for the forum, I’m in for several 
$35 for a heat-sink to go under an aluminum star. If we could get so lucky it’d only be 35. Budget gone wrong.
beautiful work! I second the request for how you cut out the square, must have been a very small end mill!
Tom E - the AMC7135 chips themselves have a rudimentary overtemp function, which shuts them down individually when they go over a certain temperature. 10W+ in such a tiny host is going to make that happen in a hurry - if the housing is getting very hot then it’s just a case of there not being enough surface area and airflow, if it’s not then you need to add thermal paste to the various junctions.
Thanks Matt! Then wonder what Erik and moderator007 did to get their's working at 3.5A...
Oh! Just thought, the 7135's on the bottom may be contacting the copper disc I added, if so, they could be getting very hot! When I did the depth measurements, it was very tight in there so that could be the problem - damned if you do, damned if you don't. I'll have to check that out carefully, maybe remove the copper disc or raise the driver position somehow. The outer host gets hot, but not too hot to handle at 1 minute or so when it shuts down.
I totally knew I saw it before. :D
Here: https://budgetlightforum.com/t/-/11766
and here: https://budgetlightforum.com/t/-/8515#comment-186787
Its good to have some variations and documentation.
+1, big time!
Even if you don’t have a lathe, you could probably get decent results by pasting a board to a copper disc and drilling a hole or two through both at the center pad for some solid copper wire and soldering that in place. Not as good but better than before.
That's what I was think'n - simple small diameter copper rod maybe, drill a hole to match thru the star, but I'd like to use one solid rod thru the pill top and into copper discs below. Not sure how it will all be soldered and in what order. Best way maybe would be to reflow the emitter last, but that means maybe a hot air method.
Thats what I actually did here:
You just need some decent drills. I reflow soldered everything, works like a charm. As good as solid if you ask me.
I'm not sure exactly how you did the reflow on a cooktop - did you place the entire copper heatsink on it? The heat got up that high through all that copper I guess?
I can cut these on my CNC in fairly short time... What are they worth and IS there a demand?
Dan.
Glass cooktop, yes. It took a little longer than usual but it worked. I just couldn't touch it for minutes after. ;)
Thanks for all the great comments guys. 
I milled the center out using a milling machine with a DRO and one of these. 1/16 carbide 4 flute end mill.
!
!
If it wasn’t for the DRO it would make this task more difficult. The DRO allows me to travel precisely the same distance down each side. The end mill leaves little round corners on the aluminum pcb that had to be filled square. This is a pic of a amc7135 with the end mill so you can see how tiny this is.
!
!
When I dreamed this up it was my own idea. I have seen several people solder a led to a piece of copper with a small pedestal to rest the center section on. Soldering the wires to the bottom pads on the actual led in mid air. I had never seen what I have done anywhere before until today. And its just my luck someone else has beat me to it. Some minds think a like I guess OldLumens and jamio.
This thread is pretty cool too.
Dude, I think there are probably a good number of people who know how you must feel right now. To invent something and then find out that someone else also invented it before you sucks. But I for one, can totally tell you are being sincere and I’m amazed by what you did here. Thank you so much for thinking of it, executing it so artistically, and documenting it so well for us. You are truly an inspiration.
Thanks for your kind words.
Before installing the copper I did a little test on the Xiaozhi with all the mods but the copper. Current draw was 3.48v measured at the tail. I used a wallbuys light meter at one meter from the light with a fully charged Panasonic 3100mah.
Before the copper I got 5900 lux starting.
5400 lux at 1 minute
5300 lux at 3 minutes
After the copper I got 6100 lux starting
6000 lux at 1 minute
5900 lux at 3 minutes
5800 lux at 5 minutes
5700 lux at 7 minutes
5700 lux at 10 minutes
This is not a reliable test but I kind of wanted something to go by to see if the copper had made a improvement. I wish I had tested the before copper a little longer to see where it would have been at 10min. I had to stop the test at 10min with the copper because the Xiaozhi was becoming blistering hot >) . I could definitely tell that the copper was absorbing more heat and transferring it to the light. It was beginning to concerning me with a 18650 li-ion setting inside a Xiaozhi frying pan, so I stopped at 10min.
Nice results! The copper is doing the job quite well.
Your results may not be comparable to anything else, but relative before and after, they seem fine.
Awesome build, can’t believe I almost missed it. Thanks for sharing.
This is very precision work moderator007, I like it a lot. Well done!