A few years ago I bought these very compact tail clicky 16340 lights at Fasttech to be used as a host (they were only available for a short period), decent quality and threading, standard 16mm board, 17mm driver and Omten1288 switch, and it is the most compact 16340 tail clicky light I know except for the EagleTac D25C, smaller than the Ultrafire 602C and Supfire S1.
I modded one to a UV-light at the time but, anticipating on new better 16340 cells to come hopefully, last week I modded it again to become my latest EDC, with a 3000K 95CRI SST-20 and BLF-A6 driver, output 440 lumen of nice incandescent quality light.
Of course it needed a lighted tail (I do them in all my lights if possible), and one of my wishes was to make a warm high CRI firefly tail at some point so here is my go at that. Does one need high CRI in light that you can hardly see?
I tried warm white 0603 leds from ebay before, but even the ones that promised the cosiest light of all were horribly yellow. So when ordering at RS-online anyway, I added a few Cree XD16 2700K 80CRI to my order, those should have at least a few magnitudes better tint than the ebay leds.
On the BLF driver I soldered a 470 Ohm bleeder resistor. I chose a tail ring with 2 strings of 3 leds (one of PD68’s designs, this one), one that could be sanded back to 14mm to fit in the small host. I soldered 3 of the XD16 leds in place with a 8.9 kOhm resistor behind the string (left the second string of three leds un-occupied for now) and assembled the tail to see how that looked. The lighted tail had an ok colour but did not emit quite the cosy warm light intended. It was 0.01 lumen and just enough for my spectrometer to extract the colour data:
The specs of the XD16 were at least confirmed, 2616K (through the tailcap) and 83CRI 
. But the spectrum shows a steep dive after 620nm, for cosier higher CRI light more red must be added. I had 645nm quite deep red 0603 leds that should do that job so I added the second string with 3 of those, behind a 2.2 kOhm resistor (you need some power to see deep into red).
After assembly it looked nice, but the red was not really apparent.
It did help the tint and spectrum though, but the 645nm leds (from Osram btw) look more like 660nm, perhaps the very low current gave that shift. I was at 2400K and 88 CRI, still slightly above the BBL.
I wanted to get under the BBL, and the 660nm was too far into the invisible part of the spectrum, so I swapped the 645nm leds for (more visible and more efficient) 633nm 0603 leds, still with the 2.2 kOhm resistor.
The red peak is at 640nm which is a nice add-up to the XD16 spectrum but it is too high, too much red, very atmospheric but not intended. The spectrometer software even refused to provide a CRI for this tint. So I changed the 2.2 kOhm of the red string to 4.7 kOhm.
That looked more balanced. And bingo:
The red peak has halved, that brought the colour temperature back from 1600K to 2000K, the tint closer to the BBL but still nicely under it for that good mood, and the CRI sky-rocketed to 95.2, with an R9 of 97.7. And indeed when the illumination is compared next to a 2000K R9050 E21A flashlight (a modded Tool AA) on a white wall, the tint looks exactly the same.
So mission accomplished, a very high CRI 0.01 lumen lighted tail 
. Not entirely useless, it does make a very nice night stand. Btw, the current that it draws is 0.52 mA, which drains a 700mAh 16340 battery in about 2 months.
Now lunch, after lunch I will add a led-test of the XD16 2700K 80CRI.
Ok, the test. I had these XD16 leds anyway, and it is quite an insult to use these high power leds at .2 mA as I did, and the XD16 has only been tested once before (a nice test on his own custom board by EasyB, here) so I did a output test, same method as always, a description of the method can be found in my Cree XP-L test via my signature link. The led was reflowed on a Virence 16mm high performance ledboard. In the graph I added test of a few relevant ‘competitor’ leds with similar die size, although there are too many differences between the leds (domed/dedomed, different tints/CRI, die sizes not exactly the same) for a good direct comparison.
Without a long discussion, the XD16 is a very nice good performing led (mind that the led at test is a 2700K 80 CRI sample, a cool white one has 20% more output, bringing it at 5A very close to the famous dedomed XP-G2 S4 2B) that perhaps did not get the attention on BLF that it deserves (the beam it produces in a reflector light is very nice for example). It probably handles more current than the Nichia E17A and E21A because unlike those it has a ceramic base that sheds the heat from the middle of the die better than the Nichia’s that lack a substrate.
(Stereo pic)
