4x4 E21A would be better 
Clemence, would taking pre-orders for future products in your store help with speeding up development time? If so, I would gladly pay ahead for some 3x3 E21A along with Carclo triple compatible e21a boards.
Just a thought. Excited for what your doing! Keep up the good work! 
Jared, it will need hundreds of pre-orders to fuel a single MCPCB model.
[Clemence]
Max current? voltage? Diameter?
[Clemence]
I asked the designer of it here.
Total current is 4.9A, so 0.6125A per LED (two strings of four LEDs are in parallel, so there needs to be a bit a tolerance for the max current per LED). Generelly it's very low powered.
I bet quite a few people might be interested in this. The interest list for the lantern alrrady has over 2000 entries!
For lantern with no directional beam needed, I suggest 757 LEDs. Depending on the space available you can populate lots of them to get very high efficiency at much cheaper price. The options are.
——-
A.
NF2W757GV3F1:
- R8000: 2000K, 2200K, 2700K, 3000K, 3500K, 4000K, 4500K, 5000K, 5700K, 6500K
- R9050: 2700K, 3000K, 3500K, 4000K, 4500K, 5000K, 5700K, 6500K
Note: $0,1/pc. Very efficient at extremely cheap price
——-
B.
NF2L757GF1:
- R95: 2700K, 3000K, 3500K, 4000K, 4500K, 5000K
NF2W757GF1:
- R95: 5000K
Note: $0,3/pc. Priced very close to Optisolis, CRI slightly higher than R9080 unlike Optisolis, they can be mixed freely
——-
C.
NF2L757GF1 (Optisolis):
- R95+: 2700K, 3000K, 3500K, 4000K
NF2W757GF1 (Optisolis):
- R95+: 5000K, 6500K
Note: $0,5/pc. CC and WW can’t be mixed in parallel. Currently, the highest possible CRI for LED
——-
D.
NVSLE21AT
- R9080: 2000K, 2200K, 2700K, 3000K, 3500K, 4000K, 4500K, 5000K, 5700K, 6500K
NVSWE21AT
- R9080: 5000K, 5700K, 6500K
Note: $1/pc. The most efficient and expensive compared to A ,B, and C. Light quality slightly lower than B
[Clemence]
It should be noted, for this purpose, that it’s max 2.34A (7*7135) per channel and that is with all solder bridges connected. The factory default will likely be even less:
You can add/remove the solder bridges to enable/disable current. He could even add more LEDs to reduce current per LED as long as it’s symmetrical.
No point doing A) since it will come with Samsung LH351D 90CRI 3000K and 5000K
C) wont be a problem I guess cause it has 2 separate channels
To increase current : 7135 piggyback 
I would be interested in a new board but only if it is E21A because the E21A comes has the famous 9080 2000k option that can mimic candle light. Since the other options only go down to 2700K, it doesn’t add much more to the stock 3000K to 5000K LH351D 90CRI, despite the Optisolis has superior color rendering.
Exactly! The 2000K 90CRI+ with tint on BBL (!!!) variant is the main reason. 6500K for the other end of the spectrum makes it "perfect" and the tint more reddish.
Cost will be high either way because of shipping, the board etc.
Just be careful, this has positive and negative reversed on the silk screen. It is common positive but common negative.
Just wanted to share. I put a single 6500k optisolis into an ultrafire u-f10 which is a cheap but very high quality aspherical. The goal was to have a perfectly uniform wide beam: more uniform and more concentrated than that of a mule. Works GREAT
A: probably not enough Rf (and R9) for extra ‘warm’ CCTs, otherwise similar to 351D offerings (more available CCTs though)
B: would be very tempting for making a tint mix between 2700K and 4500K
C: too bad, that no tint mixing is possible between ‘L’ and ‘W’ group - some people would find 2700K to 4000K tint mixing a bit narrow
D: probably the best option - would let very useful tint mixings to happen (more than one) - a mood variant with 2000K to 3500K, and a base variant with 2700K to 5000K
If DBSAR would make Sofirn to have Nichia variant of BLF LT1, that would be awesome (probably I would just buy those).
I think its better to just make the lantern as designed. We can always start a GB for this MCPCB with poll later. I need at least 300pcs committed order to make it
Is the lantern still far from realization?
[Clemence]
@clemence, no, it’s not.
It’s actually rather close to production. DBSAR received it, and he is currently doing testing, and will be making videos about the final prototype after a few days.
Tint mixing should still work ok since the “L” and “W” are driven by separate banks of 7135 regulators.
The firmware corrections to maintain constant brightness as the tint is adjusted will probably be off, but that should be a minor drawback.
No change is going to be made to the factory MCPCB at this stage. It will ship with Samsung LH351D’s.
A group buy list would be a good way to handle this proposal. It seems prudent to wait until the lantern actually starts shipping to start the list though, so potential buyers can decide if they are happy with the stock emitter.
Yes, it makes sense to wait until the lanterns are being shipped. I think that this is definitely worth it though. It gives owners a large flexibilty regarding the CCT range of the lantern and Clemence might be able to sell a lot of LEDs (and pcbs).
Maybe Clemence might work with DBSar to post the group buy in the OP of the BLF LT1 thread so people know about it. Better yet, if he can get the groupbuy news included in the PM when they notify the people in the interest list when the lantern is available. With 2,000+ buyers for the LT1, we should be able to get 300+ orders for the E21A board.