Thanks for explaning the process. So it’s really more complicated to do properly. Likely beyond my skills and my simple equipment…
I wish that Sofirn would have use a thicker wire for their bypasses and solder more properly from the factory… Should likely send them a friendly ‘complaint’ about this issue. Since it happened on both my SP32A and C8F (the tail-portion bypass got cut, likely due to inability to handle that much current on Turbo mode?)
I’m glad my light meter wasnt imagining things then re my 18650 version being brighter than my 21700 version.
If the Sofirn 21700 cells that I ordered additionally do not help the situation I will definitely be in touch with Sofirn for a friendly word especially as I am still having crashing issues (although it hasn’t been as bad the last few days) as I bought the 21700 version of this light plus 2 21700 cells as it was supposedly rated with a much higher output 
All the best
Moley
Let some more people get it so we can figure out the problem. It might be a simple fix.
I think someone needs to try doing as JasonWW suggests and remove tail cap and just firmly press a stiff wire to negative battery and the non anodized end of tail thread.
If that doesn’t show any difference then I guess checking/comparing driver wires (diameter/length) and components (FET) would be the next step. I only ordered hosts which haven’t arrived yet so I wont be any help.
Would a metal paper clip suffice?
a scrap of copper wiring would probably be best. like the stuff you find as a “ground wire” in a light fixture or scraps trimmed from household electrical wires
Or a copper coin. 
Silver dollar!
Might be difficult to angle the larger coin into the battery tube to bridge the cell to the edge of the tube.
@contactr, never use a paperclip.
Trust me. You do not want molten steel on your hands.
Woohoo! My 21700 C8 host made it to the US! It’s in customs at LA, maybe another week or less…
Hi everyone,
Just an update…
My 2 x 21700 Sofirn branded cells have arrived and I have duly tried those in my C8F-21700 and I am still getting the same range of figures as I was previously (235-240 for 21700 version light, and 275-280 for 18650 version of the light). This is the really disappointing bit for me as the 21700 version is supposed to be around 3500 lumens and the 18650 around 2500.
Now, whether I would be able to “see” the extra 1000 lumens or not, I didn’t expect it to come in at less than the 2500 lumen version 
Especially with the added expense of buying the Sofirn 21700 cells too.
Tracy from Sofirn has been good enough to get in touch with me so hopefully I/we may be able to get some further info.
All the best,
Moley
Sofirn told me that in their tests they were getting around 3500 lumen. So they are stumped.
I still don’t think anybody has tried to simply bypass the entire tail cap with a piece of copper wire and measure the lumen output (like a ceiling bounce) to see if it jumps up or not. This is really easy to do and it eliminates the rear spring, rear switch, threaded contacts, etc… It can really help us to diagnose the problem.
Ps, mine is still being shipped so I can’t do anything to help, but make suggestions.
Hi everyone,
OK, so I bypassed the tail switch with some copper wire and took a Turbo reading again using a Sofirn 21700 cell and got around 245ish which I guess would be in the margin of error to be an equal result of a max of 240ish with the tail switch in place.
It was as an experience trying to hold the wire in place on the cell and the battery tube whilst pushing the battery in firmly AND operate the side switch whilst trying to bounce the light off of the ceiling
Hope this helps,
Moley.
Thanks. That does help. It means there’s nothing wrong with anything in the tail cap area. We can look elsewhere.
There must be something in the head/driver area. I think now we would need to inspect the driver. Maybe Sofirn switched to a different FET that is not so efficient? Maybe there is bad contact between the driver and body? We will have to see.
I got my 21700 host today, waiting for some specific emitters to get here…
My build is already in state of repair: one Luxeon V led decided to quit lighting for half of the die. I’m sure it is the one that I re-positioned too much during reflow, the solder already began to look brittle and I should have replaced the solder or add extra flux. But I’m out of Luxeon V’s (and money ) unfortunately.
I thought about waiting for the Oslon Blacks but decided against it with the way this MCPCB is screwed to the reflector and such. So I sliced domes off Samsung LH351D 80 CRI 5000K emitters and built an Anduril driver for it. The Anduril firmware, to my experience, can be finicky about the switch in use and that’s the case here, actual working conditions are intermittent with a quick blink of the emitters the norm about 80% of the clicks. When it engages it ramps and works fine, but it’s not at all reliable. So these switches must have some peculiarities to em.
Beam profile looks really neat from these sliced Samsungs, works nicely. I’m using an iJoy 21700 3750mAh cell with one of Blue’s large springs cut off about 3 coils from the top. No bypass in it as yet. 20ga leads to the board with an SIR404DP FET. Left the spring bypassed tail switch section stock, again, for the moment.
Charging it up, will see if I can get it to work on the light box and get a read on it…
Oh yeah, the two white wires are the switch leads, the black illuminates a green emitter on the switch board and the red illuminates a red emitter, I didn’t try hooking either up as I really don’t care that much for an illuminated switch. (didn’t know where to put them on the Anduril driver anyway. lol [I used the D4S Anduril variant] )
I wonder if it uses a common ground or common positive? Did you happen to take a picture of the switch?
Could this be related to the other intermittent problem reported earlier in thread?