There is some work left to do regarding the likes of 'the torch'.. it's a 12-cell 623 IIRC.. and with GP2000s.. which hold voltage better than anything i've seen.. that math worked out to a nominal voltage closer to 1.25V and that puts Vbat at 15.0 and Vlamp about 14.6.. re-rates to 3600 torch lumen.. but that is an enthusiastic rating.. when comparing a direct-drive 64623 from 14.4V source (4D LiON).. the 64625 as 12.5V.. which re-rates to 2700L.. is seemingly brighter in a head-to-head comparison.
Delvance made a 13-cell version of the torch concept
here which is going to be far brighter than the torch, and even brighter than the 65W IRC lamp that re-rates to something like 3550 Lumen even when re working the calculator to be 20% less than it's supposed to be.
In a 2D size.. the torch is the reigning king of output for now.. with nominal output about 1000L more than a USL... just plain stunning. I'm sure that the color is a little whiter on the USL..
The Mag625 is a LOT whiter and will look brighter than probably even the torch, not to mention with regulation it doesn't dim..
The upcoming IRC based lights i'm working on.. they have CCT similar to the USL.. but the lamp with the same output of the USL takes under 100W to run vs the 120+ for the USL lamp to get the same output.
It would be neat to have a chart comparing the output of the various mag mods.. specifically of interest.. grouped by size.. 3D, 4D, 2D etc..
One of the hosts i use for the IRC is a 5D light.. it's enormous.. but.. with up to 90 minutes runtime.. specifically.. 93, 66, 49, 37 for the 25, 35, 50 and 65W lamp.. it will output almost identical output to 'the torch' but for 37 minutes!
I'm working on shoe-horning all that power into the most dense solution possible. Anticipate more movement on that once we finish the nanos.
-awr