I just want to openly express my appreciation for the work you do, Hank. Thanks a lot!! 
Now that is something I would buy. I have enough multi emitter anduril lights to last a lifetime. Okay well, maybe not a lifetime but, enough that I had to start giving them away.
Is the E21A version going to have four E21As spaced far apart like the other available LEDs, or clustered in the middle to use an optic designed for a single emitter like the Virence boards do?
Is this new light?!! I just saw it few seconds ago!! E21a and mule???!!!! I donāt see it anywhere?? Help please
The e21a and mule arent released yet, there is sst20 and xpl hi currently
We have managed to put 16 pcs E21 LED into the KR4 MCPCB as a mule.
16 E21sā¦now thatās tempting. Will you be using a buck driver for it?
If TKās sustainable output test is actually of the 95 cri sst20ā¦Thatās amazing!
It will use 4*E21 LEDs, with carclo 10622/10623 optic.
No, the sizes are not the same. The FW clip is too small.
It is not really like anything elseā¦ itās kind of a new thing entirely, which combines the good parts of several other lights. Itās smaller than a D4S but offers similar performance. Itās regulated and more efficient like the K1, but adds direct-drive for bursts of turbo. It has a tail switch like a FW3A, but is more robust. It has the features and beam profile of a D4, but is more refined.
There is no PWM below 5 Amps. It has no amc7135 chips. With default settings, it has no PWM on any level, since the ramp ceiling is set at 5 A and the next step up is turbo (direct-drive). It is only between those two levels that any PWM is used, but the ramp doesnāt go there unless the user reconfigures it.
Yes, it seems to get more efficient as voltage drops. As for overall efficiency, if my runtime tests are accurate, it seems to be about 20% more efficient than a D4. At least, I tested both with the same exact battery, and the KR4 had 20% more area under the curve. It is possible that I made a measurement error, but itās definitely more efficient and the question is mostly a matter of how much. Maybe itās only 10% or 15% more efficient.
The only code Iāve ported to this driver is Anduril, so I doubt it will have any other stock firmware options.
However, even though you hate Anduril, you are in luck. It is open-source and uses the FSM UI toolkit, so you can make whatever interface you want. There are a few example UIs to start from, tooā¦ ZebraLight, Olight, Wernerās UI, and two simplified versions of Anduril, plus compile-time options to enable/disable many of the featuresā¦ and sort of a guide for how to make your own.
If you can represent the UI you want in the form of a flowchart, or as a list of states and the button events which move between states, it should be pretty straightforward to create. Just make sure to get a flashing kit with it, and then it can have any interface you want.
Iām confused too. Itās definitely SST-20, and it looks like 4000K, but itās performing like a low-CRI version instead of a high-CRI version. Itād be nice to get some confirmation on what it actually is.
Itās hard for me to gauge the CCT of SST-20 LEDs, because they all look yellow-green to me. However, the beam looks almost identical to another SST-20 4000K 95CRI light I have, and looks similar to a XP-L 4000K, and does not look like any 4500K or 5000K lights Iāve compared it against.
But in my calibrated lumen tube, Iām seeing numbers way higher than a high-CRI SST-20 should be capable of. Even with an old cell from 2014, itās measuring 3840 lm at start. With a newer cell, I got 4044 lm. So Iām not sure whatās going on. It shouldnāt be able to do that.
FWIW, the way I scaled the runtime graph was: I turned the light on at its default level and measured it in my light tube. It came out to 197 lm. Then I moved the light into position for runtime testing, started the test, let it run a few seconds, then double-clicked for turbo. Afterward, I scaled the graph so the first few seconds hit exactly 197 lm.
If you look at anything brown you can clearly tell the difference between 70 and 95 cri. With low cri it will look yellow/green.
Iām in for an E21a version if it uses a buck or boost driver 
I think the E21a under those optics will be amazing. Much better intensity over the other larger LEDās.
Itās 5A constant linear driver, without direct drive.
TK WROTE
QUOTE. The only code Iāve ported to this driver is Anduril, so I doubt it will have any other stock firmware options.
However, even though you hate Anduril, you are in luck. It is open-source and uses the FSM UI toolkit, so you can make whatever interface you want. There are a few example UIs to start from, tooā¦ ZebraLight, Olight, Wernerās UI, and two simplified versions of Anduril, plus compile-time options to enable/disable many of the featuresā¦ and sort of a guide for how to make your own.
If you can represent the UI you want in the form of a flowchart, or as a list of states and the button events which move between states, it should be pretty straightforward to create. Just make sure to get a flashing kit with it, and then it can have any interface QUOTE.
That would be a NO then ,
Ok thatās not a flashlight for me ,I think Iāll leave it .
Thank you.
I love Anduril and Noctigon/Emisar, perfect combination!
Thx TK for clarification. Was about to pull the trigger but the more i look at that clip the more seems to me someone took it from no name Aliexpress light and put it on Noctigon...
Iāll wait for E21 or maybe TiCu (if any at all), not to think about that clip.
True, it makes things look yellow/green. But so does high-CRI SST-20. However, I tried some other color testsā¦ Mystery SST-20 vs FA3 SST-20 (4000K, 95 CRI).
On a white wall, both look very similar. Both are yellow-green, but the mystery SST-20 looks slightly more yellow-green.
Shining at a bright red object, both look good but the FA3 looks more vibrant.
Shining at a pink/magenta object, the mystery SST shows pretty accurate color while the FA3 sucks the blue out and makes it look like a salmon color.
Shining at blue objects, the mystery SST looks blue but the FA3 looks sort of sea-green.
So theyāre definitely different. The mystery SST has weaker red and stronger blue rendering, which suggests lower CRI and/or higher CCTā¦ but aside from the white wall comparison (where the mystery light looks warmer), it looks like it produces more accurate colors than the high-CRI FA3.
Once Iām done testing everything, I may just add a minus-green filter to it and run a no-FET version of the firmware. It looks pretty decent after adding that, and doesnāt suck the blue out of things like my FA3 does.
Or I could order a XP-L HI 5000K version. Those always look good.
Wow, thatās heavy. 25 g more than already heavy D4 (gen 1)? It is about as large as FW21 and actually weighs more, even with cell inside. I appreciate innovation and Iām very curious about the efficiency but itās not as space-efficient as I would hope for.
I donāt know if this is news or not, but Hank now has J5 output bin SST20s in HB4 FB4 tint. Iād assume thatās what is in the KR4
As for the E21A version, Iād hold off until there is a buck or boost driver available. Maybe loneoceans could work with Hank and adapt his new lume1 driver for this?
Very interested and excited for the E21A MCPCBs though. Will you offer these as components?
A series quad E21A would be a perfect application for a 12v boost driver.