SBT90.2 @ 38A? 
Yes, driving the single spot at 38A, but reducing after some seconds then to ~32A.
I implemented this reducing mechanism for all channels: The very high power is reduced slowly to a level that is safe to operate.
Further, power is reduced if battery voltage drops too low and also with temperature rising.
But due to the 3kg heavy-metal, it can take multiple 100W, or even >1kW rather long.....
And main reason to build it: The building/design itself. Wanted to see what ist possible.
But I certainly also use it: I go out with the dog in the dark (no longer dark now ;-) ), and sometimes do some Jogging/Running in the dark.
Wow! Incredible work. 
Looks like a new pocket EDC! 
sick 
Germans tend to exaggerate engineering sometimes…
Very nice. 
Wow, very impressive power and integration of all the electronics!
How do the buttons control all the power converters? There must be some computer/MCU in there?
What made you decide on such a high current, 38A, for the sbt90.2? That is most likely above the max output on the output/current graph.
How are your batteries arranged in terms of series/parallel? What made you choose two different cell types?
Wowie! Lots of work -> lots of fun 
Yes, there is a ATxmega128A4U MCU controlling everything (polling buttons, LED converter, charge converter, balancing, display, temp-sensors, ....).
I will do some more measurements (Candela) in the next days, but until now it seemed to be that at least for short time operation brightness reaches max at 38A (~560.000 cd for the center sport; ~520cd coming from the four SBT90.2 behind the 30mm TIRs).
Batteries are arranged 4s3p, resulting in 16.4V full voltage and 4+4+2.6 Ah capacity (measured around 150Wh energy; implemented and automatic battery calibration cycle, so that display now shows items like "runtime left").
That’s an incredible feat of engineering and design execution. Looking forward to seeing some video in action. How do you manage the cell balancing of the vtc5A vs the 40T? Are they on separate balance circuits?
There are always two 40T and one VTC5A in parallel, and then these four blocks are in series. So each parallel block is balanced as one cell. I had to choose this battery combination to have enough space available for the electronics. Initially I wanted to use 4x 40T in parallel with 70C LiPo, but due to savety and expected life span I went for save cells only. The pack is now capable of doing around 1.5kW for several seconds, and it does around 1.8kW in Flash Combat Mode. Plenty of runtime with 150Wh....
Absolutely incredible!! Have you thought of building an even more powerful version with this bad boy?
“ CBAK Energy Technology, through its BAK Power division, announced that from 2023 it will also produce 25Ah/1kW peak power density 4680 cylindrical cells on a large scale”
Thanks for taking the time and effort to build your beast! I am super impressed with your advanced knowledge and skills.
Being able to see your work makes life exciting!
WHERE IS THE CLIP??? 
Good job!
Bleah, THOR mjolnir
Welcome to BLF Holzleim. 
Very complicated and expensive light you are building
@Photon Master: interesting cells, good amount of energy per volume. But already now the battery pack delivers enough power to very quickly heat up both the lamp head (>1 kW being dissipated here on ALL_FULL button), and also the DC/DC converter (LED driver) section. Although efficiency being at ~93%, that still means ~100W peak power being dissipated on the driver PCB.
I checked that my chosen principle of conducting the heat originating from the MOSFET, the power-diode (boost converter) and the Induktors to the backside of the PCB, and then to the aluminium of the lamp body. Put a dense array of vias underneath the components, being filled with tin this conducts pretty good (PCB is only 0.8mm thick). Not suited for automated assembly, but for manual work good.
Originally I hab planned to go for 4x LiPo in parallel with 4x 40T, but for safety and reliability reason I then changed to save cells only. The LiPo initially chosen would have offered endless power: 357A continues, 714A short term current, meaning ~ 5000W continues power
But I made some bad experience using LiPo, lost a big lamp ~10 years ago due to LiPo cell fault:
@vit55: "Bleah, THOR mjolnir"....what does that mean? Interesting that you mention Mjölnir, that is exactly the name I chose for my device. Will engrave that later into the metal on several locations. I thought of this name because especially the tactical flash-combat mode is very impressive: Use currently around 1800W total power for flashing (2000W would be possible, can be unleashed later per SW update), not measure yet, but should be >120.000 Lumen during flashes. So the name "Mjölnir" came into my mind....
Will do some beamshots and/or video clip one weather conditions here are better, currently we have continues rain....
@Th558: Thanks for asking this, did some measurements: You were right, the sweet spot for short term (some seconds) seems to be at ~34A; Changes SW now to begin at 34A, then drop down to ~30A over some seconds. I just implemented the 38A hardware-wise, to have enough headroom to max out the sweet spot per SW.
@CNCman: Thanks for the welcome words. Yes, several 100€ accumulated for material, beginning with the SBT90.2 being rather expensive. But that is ok, I will not build such a big one every year....took me already many hours of development and building time...
Super impressive !
Wow I want one so bad… (: wish I had your knowledge and could build something like this myself. The heat issue makes perfect sense…
Awesome! Extremely skillful and well executed build. 
I like the way you mounted those XHP70.2 without a PCB.
You should really join the 9th BLF Old Lumens contest this year.
What a great build
What I think is that
The SBT in the middle is useful
the four others in the 30mm reflectors are not so a good choice,
cause of many heat.many amps.less throw cause of diameter and so on.
Maybe I am wrong. 
But good skills and well done,thanks for sharing
regards Xandre