I stopped doing spring bypasses on my lights after I got my shipment of low-resistance Be-Cu springs.
I just realize one of the batteries was a Sofirn 2200mah and not the VTC5D. I remeasured using 3x VTC5D and got 12,400 lumens at 1s. Better but still less than what I expected.
That’s pretty good. You are getting number close to ANSI.
Lumens at 1s are Ahh! and all, but are not really relevant and no one uses that measurement. I think we’d all like to see something more realistic once the output settles a bit. Maybe 15 to 30 seconds or maybe Turn on and 20 seconds so we see how it drops.
Firelight, even the Blue springs benefit from spring bypass. I chopped a D Maglite and used the Q8 reflector with it’s board and XP-L HI emitters, counting on just the springs the output seemed low, a single tail spring bypass gained 500 lumens, I removed the one on the driver and used a solid copper contact. Now the light can make the expected 5000+ lumens on a hot cell.
The collection was joined by Emisar D18.
I will say briefly, I like it very much.
Turbo withstands 30 seconds. When setting the temperature threshold to max, after 6 minutes I was able to warm it up to the temperature of 75 degrees Celsius. Beam profile similar to M43, looks reminiscent of a bejspol stick 
Amazing that you got so much lumens out of the 95 cri sst20. Its just a little bit less than my xpl hi version.
Maybe do him cleaning the contacts. I undressed the whole thing, cleaned the contacts, the tube, the brass ring, everything I could. Then I re-greased the threads, twisted and just started to play. The result did not surprise me because I was counting on those 10000. It surprises me a bit, how poorly XP-L HI is. Sky Lumen did not squeeze out too much. Maybe it’s a matter of optics and poor fitting to XP-L HI?
I will be looking at my friends D18 later today, expect to find gains but where and how much remains to be seen. It’s on the mail truck , looking forward to it.
He’s so excited to see what I do to it he didn’t even measure it on his new light box! lol So I have to get readings with a few different cells first, then I can start looking for the improvements…
I did clean the contacts with alcohol but saw no gains. Sad but at least the xpl hi 5d tint is beautiful as usual
I wonder how it would be a direct comparison of D18 with SST-20 4000K vs XP-L HI 5D
I have the impression Hank should’ve went with a dual spring design on this one.
Dual BeCu springs would be nice as we wouldn’t need spring bypasses anymore at that point.
Having only 3 cells power it is limiting it.
So we need to do these things to get as much power out of it:
- Figure a way to bypass the springs/add a small inner BeCu spring for even higher output.
- Use higher efficiency optics if possible.
- Clean everything well.
- Use very high drain cells like the 20S/24S/VTC5A.
Dang it Komeko, you make that black one look SO appealing! I might just have to go ahead and pull the trigger on one of my own!
Wonder if I could squeeze 2A’s in there? 3V XHP-50.2 2a’s… 
My XP-L HI 5D measures 12,400 lumens while yours measure 10,528 lumens on your SST-20.
Does yours come with spring bypass? Mine doesn’t but I was reading some have spring bypass from the factory.
3V XHP50.2s seem a bit risky here.
You’d be going from:
18 LEDs —- equivalent 72x LEDs.
Original springs. I want to have some fuse. I do not see them as a limitation of power. My measurement is 45A, or 15A for each cell. I use samsung 30Q with the top button, bought on BG. They are slightly longer than those with a flat top, so they also compress the springs more tightly. It can also affect the result. I am very pleased with the efficiency and overall quality. It is a pity that the LEDs are not aligned with the switch. Fortunately, only a little so it does not bother me.
Risky? Aw come on, the lab IS in the dungeon basement for a reason! 
Doesn’t the MT-G2 have 72 dies? Don’t I have more than one light running multiple MT-G2’s? Didn’t Richard put 50’s in his 18 emitter TRJ20? Who’s gonna do it if I don’t? Sometimes ya just gotta tear the box up and move on… 
Edit: I might not even be able to move up to the larger footprint as space constraints might limit the possibility, so for the time being I will stick with the XP footprint and see where I can take it. W6 LH351D emitters on stand-by.
@Dale, the difference being that the MT-G2 has 72 small dies.
XHP50.2s are 4x XP-G3s on a single die.
I wouldn’t want 72x XP-G3s in such a sized light.
Regulation would be great, but with a FET driver, dual BeCu springs/bypasses? No bueno.
Yes, I also got the XPL HI’s running on 3x 30Q’s, bit disappointed in the output, was expecting much more, should have stuck with the stock emitters and saved myself $50