Awesome. I didnāt know it would have a E21A version. Iāll probably have to get one of those. 
The FW3A holds all its pressure on its inner tube while the outer tube can move a bit. It is sensitive to very small manufacturing errors. If it has the wrong amount of pressure, it doesnāt work right. Also, loosening the head doesnāt actually cut powerā¦ it just disconnects the switch.
The KR4 is built more like other lights, keeping all the pressure in the outer tube. Inside it has a spring which transfers the switch signal, and it stays connected even if the other parts are loose. Power is disconnected by loosening the head or tail, but the switch stays connected.
Iāve only gotten to test one, but it has been very reliable.
Also, the KR4 can safely have its tailcap removed. It does not fall apart like the FW3A tailcap.
Yes, it is pretty grippy. The narrow part of the tube is covered in small grooves which provide friction. It is not the rough rock-like finish of the D1S. Instead it is more like the machine marks of bare metalā¦ but more pronounced to provide more grip.
About the clip, itās pretty decent but I would probably change the design a little. Itās pretty comfortable to use and it lands at exactly the right place, but the tongue sticks out a bit far. So Iād probably make the tip shorter and shallower.
Or perhaps add an option for a deep carry clip, as you said. Iāve always liked the style Olight used in ~2016 or so, with most of the clip straight and parallel to the body, then a dip and a small tongue at the end. I like this shape (but as a ring-in clip instead of a clip-on clip):
The KR4ās performance is impressive enough that I thought I made a measurement error and had to triple-check the results. It can sustain much higher brightness than a D4 or FW3A. In my testing, it performs more like the larger D4S. However, itās not like any of those because its runtime graph is almost completely flat. It varies by less than 0.3% regardless of battery voltage.
With mild fan cooling, I measured the sustainable brightness (at 45 C) of a D4v2-XPG2 at about 630 to 690 lm. The larger D4Sv2-XPG2 measured at 1030 to 1160 lm during its stable regulation phase. Both had sort of a sawtooth pattern while the FET output gradually falls and then thermal regulation bumps up the power to keep the temperature steady. Meanwhile, the KR4-SST20 held rock steady at 1100 lm.
This is despite the KR4 using a high-CRI LED, as far as I can tellā¦ while the D4 / D4S did not.
Additionally, the KR4 is regulated up to about 5 A, which on my prototype is about 1750 lm. The D4 is only regulated up to 350 mA (about 140 lm).
