I work in batch. Those 101pcs in the box are tapped head unit. Only MCPCB and soldering left to do.
[Clemence]
Thanks for the update. Looking good. 
UPDATE 200712: Progress
[Clemence]
Why not indium for the wire pads ? Cost ? Worse wetability ? Joint strenght ?
I used indium yesterday and I had trouble making nice solder joints with it, I had less trouble with SnPb (but it gets hot enough to melt the indium beneath the LEDs due to the crazy heat sinking of the mcpcb).
Edit : I have some rose’s metal, I could try to make a mix close to yours.
I often use 60/40 or 63/37 on the wire pads no problem. The key is to use adequate soldering iron with as big tip mass as possible for quick heat build up. Heat the MCPCB outside the host whenever possible and insulate it’s base. Then make a quick work on the joint.
I don’t like to use Indium for wire joining because it’s too weak and soft. A light pull with your fingers is enough to break the weld.
[Clemence]
UPDATE 200802: Color Mixing Problem
Tested some lights and found problems in mixed configuration. The 16mm lens can’t mix the color properly, resulting in odd looking beam. Only E17A module combined with 70° lens has acceptable beam (still not perfect to my eyes). Those with single color selection don’t have this problem.
The number in each corner represents the LED configuration on the MCPCB. Distance to the wall = 20cm. Shot with ultra HDR to mimic real human view.
The solution are:
- Change the module to E17A. Not perfect, but usable with the widest optic. The picture above shows 3000K + 4000K E17A module, not too visible due to relatively close color difference. In person, you can still see the faint orange diagonal band across the beam. But those with wider range, for example 2000K + 6500K, would be visible. Please make the change no later than 200810, otherwise I will solder the LED as ordered.
- Use special Gaggione color mixing TIR lens. Tested, and the beam was perfect! But unfortunately I can’t show the picture since the only sample left was destroyed when I tried to trim it to fit inside H04RC. I already ordered 50pcs medium and 50pcs wide. This lens is very expensive for the size but it’s the ultimate cure for the problem. Trimmed and modified to fit H04RC I ask for $3,7/pc.
You can also get them yourself here: Gaggione 16mm color mixing optic
Regards,
Clemence
Thanks for the update, Clemence. Was the Gaggione TIR lens you tested the medium or the wide variant? Is it safe to assume both will work equally well at blending mixed color temperatures?
It was the medium (LLC22M). Will update the result when I get them both. The packet shipped from France on Friday and already transited in Singapore today.
[Clemence]
If you want my opinion, I have been using my H04 with 2000+3000 E21A since February and yes the CCT shift is noticeable when shining on a wall, but in real use I rarely notice it, and I use it a lot, I love this thing. I can’t speak for larger CCT difference.
While I wouldn’t say no to a truly uniform beam, I would be interested to know the optical efficiency with the Gaggione TIR, would it be worse or better ? I dont want to loose output for an uniform beam, I say that as someone who will generally try to eliminate tint shift (like green corona) by slicing or diffusing, with output loss.
I’m from France and I’m disappointed to see that shipping still cost 15€ for me… I should check what other interesting things they have.
Nothing much to do this morning so I measured the CCT shift, perpendicularly along the 2000K (red dots) and 3000K axis (blue dots) so where the shift is the most prominent, at output level 3/5 with the default optic :
Does an even mix using E21A have this problem as well? e.g. 2x 3500k and 2x4000k since the example shown was unevenly mixed.
Yes, but 3500K + 4000K should be minimal.
I see. Updated mine just to be safe. Thanks for the update Clemence.
What a nice dilemma. I went for 3000K + 4000K. Now I wonder should I opt for 3500K insted? Maukka test showed that the 3000+4000 combo sits lower of the BBL line. Decisions decisions…
In this extreme super rosy mix (2000K + 6500K) the color mixing lens is needed although E17A module was used.
[Clemence]
Wow just watching this groupbuy happen is so dramatic and suspensful, first I was really bummed to miss out on such a custom gem, but now I see how much careful tinkering goes into it and I am almost happy I missed out because I wouldn’t dare touching this beauty without wearing silk gloves 
So if we want the new lens solution, we just PayPal you an additional $3.70, correct?
Yes, but please hold on until I fully tested the new optic. Should be here in 1-2 days.
[Clemence]
Will the new optic affect the measurements in your e17A/e21a reference thread?
I used white reflective tube without reflector nor lenses. The beam/tint shots made partially integrated