I assume 170W/mK AlN PCBs are still costlier than 10Āµm oxidized Al?
ADDED:
BTW, I donāt see a need for that at the moment but out of curosity I searched for 250 W/mK AlN PCBs and found none. Would you mind pointing us at the manufacturer?
Yes. e21A mcpcb for LT1 please.
It is a pity that the Optisolis does not go down to 2000K, it makes just more sense to use a handful of midpower leds for a lantern than the current 8 high power leds. But changing the lantern to 2700K/6500K Optisolis (I would choose 2700K/5000K btw) for me would probably not make enough difference to go through the trouble and cost to do the mod.
Some users would also not like the fact that the tint would be above the BBL at the cool end of the CCT ramp.
I think you know, but others may not, that Optisolis is designed to mimic standard illuminants. At 5000K and above, the standard illuminants are based on various daylight conditions, which have 0.002 to 0.003 positive DUV.
some new and improved E17a and E21a boards are available on Virence.
Anybody from Vancouver, Canada thinking of ordering so we can group shipping together?
Hurrah ! Two new copper MCPCB ! Compared 
Excellent. Iām planning an E17a RGBA build with a floody optic (either S2+ and Kaidomain driver or Skilhunt H03RC hosts). The E17a board should help reduce beam weirdness.
I would also like to know if a modified version to fit Armytek headlamps will be offered again.
Compared to older copper boards it has a much better dielectric layer.
Compared to alu boards:
- worse dielectric (but still good)
- better base board
- slightly higher price (though that may be due to the alu one being phased out)
For some lights I think copper would be better but for others alu still seems better.
So how does max amperage compare for the 17 and the 21 board? It has become a bit overwhelming.
Thanks Johnkey, I borrow your pic hereā¦
The copper board works unexpectedly way better than predicted. From many tests performed in BLF, we knew the tighter the LED pitch the hotter it gets. Not only because the copper trace surface area needed to shed the heat from the LED base reduced, but also due to the āphoton cross talkā as mentioned by Nichia in their application notes. The LED takes and recycle the photon from the neighboring LEDs and the stored photon, instead escaping as light flux, trapped and heats the phosphor layer. The same phenomenon known as back scatter lights in undeveloped light guide.
Previous 16mm x 1,5mm nanoceramic aluminum PCB (VR16SP4) maxed out at 11,2A. And that was with LED pitch 0,4mm. The other 21mm x 2mm (0,25mm LED pitch) nanoceramic PCB (VR21SP4) peak output maxed at approximately same 11A but itās thicker and larger in diameter. New copper based VR16SP4 v.2 with 16mm x 1,5mm (0,2mm LED pitch) maxed at 11A! I underestimated it, and expected no more than 8A. At first I couldnāt believe it, because VR16SP4 v.2 also has reduced copper trace area for mounting clearance. And then, it uses āstandardā IMS technology with more than double (50Āµm) the thickness of the nanoceramic MPCB NC MCPCB (20Āµm). The thinner the dielectric layer the faster the heat moves away from the LED base to the heatsink.
Below you can see the big penalty in output just by closing the gap down to 0,1mm in VR16SP4m. With wider LED pitch, quadtrix E17A should be able to reach at least 7A instead of 5,75A. But this reduction in output means wider optics compatibility and more importantly, more control over beam shapes. With VR16SP4 v.2 you canāt go too narrow or there will be ādonutā hole in the middle of the beam. And OP/textured reflector or microlens (honeycomb/peebled/textured) TIR has to be used for narrower optics. VR16SP4m can use lighter texture OP reflector or finer microlens surface TIR. It also produce higher intensity beam (brighter hot spot).
Same 1000mA total input. Note how E17A have much higher luminance for farther throw
Note:
Due to complication in maintaining relatively constant and cool MCPCB temperature using total loss water cooling, clamping jig, etcā¦ I could not directly measure the output. Ouput measured once at 100mA briefly to prevent heating the LED and the rest of the output numbers derived using the plotted lux measurements (TASI 632A). Measurement done in my DIY light box using Maukkaās calibrated lights.
[Clemence]
Didn't see that coming...
This is even better than i was thinking it could be!
Clemence, thanks for all this test data. It definitely provides confidence in your boards.
Will you be offering those optics through your store, or can you share the source (especially for the microlens version)?
Also, will you be offering either of the VR16SP4 boards trimmed to fit in Armytekās as you have in the past? Thanks!
Indeed.
Nowā¦any news about 3x3 boards? 
Savage....
I think Iāll skip that for thisā¦
[Clemence]
Clemence, you caught the wave... 
Interestingā¦but too large for the 20.6 mm diameter light that waits for it.
ADDED: though without the corner LEDs the following 12 would have just 0.42 mm larger diameter than 9ā¦.
ADDED: About 10-12A when driven off 16340? A hottie.
Too sleepy to continue, will post the beam shots tomorrowā¦
[Clemence]
^ That looks like exactly what Iām planning on doing: RGBA quadtrix (in an S2+). 
Iām hoping to find a better 4-channel driver than the single mode per channel model that kaidomain sells, but even if not, I will enjoy experimenting with these.
Clemence, is that 4x4 still going to be 35mm? If so, it wonāt be able to fit in any flashlights 