12A+
Sounds like an arc welder to me rather than a flashlight if run too long. However Richard estimated to me in a PM that my modified SupFire L1 is drawing 25 Amps or so when using Sony VTC4 batteries running on the max output that he tested. IMO truly outrageous. If they could be left on Turbo continuously some of these modded lights could suck down a set of batteries in almost no time at all. Waiting to get it back from Richard.
The PWM can still be detected, but it’s nearly impossible to see during regular use. My standard test is to rapidly wave a thin, stiff, white sheet through the beam and look at it from the side to look for individual “frames” while it moves. A business card or playing card can work, though what I’ve found the most effective is a pupillary distance ruler which came with some glasses I bought. Here is what it looks like on a light with 188 Hz PWM and a 11% duty cycle:
On lights running at 4.5 kHz, I can still see the PWM in reflections and such; every time I move my eyes I see discrete frames reflecting back at me. But at faster PWM speeds like 18 kHz, I don’t see it at all during normal use. The only way for me to see it then is by using the card-waving test. That speed produces “frames” too narrow and too close together for me to capture on a camera though… on a pic like I showed above, there would be a new frame every pixel or two so it’d look continuous. (BTW, this method requires a camera which exposes the entire image at once, not the moving-window-blind method the iPhone uses)
Fast PWM looks great, but it’s less efficient than slow PWM due to the ramp-up and ramp-down time for each frame. I think I might make a firmware soon which uses slower PWM on levels under 10 lumens, to increase efficiency on those lower levels.
Wouldn’t that accentuate the effect? I would think that, efficiency aside, lower output requires higher pwm levels than higher output to remain undetectable. Maybe I’m misunderstanding this.
I get 12.04A from a rested Efest 35A Purple cell, in the light. (Yes, the 18650 sticks out the back of the Cypreus EDC, but it sill shows a tail reading just fine. 
)
On your last point - I suspect that point of view is part of why RMM isn’t selling triple XP-L lights. From the front line reports it sounds like “I don’t want to limit this” simply doesn’t make sense here. Tons of high end cars artificially limit power and are still more impressive than their less powerful cousins. Fighter jets are the same way. Without the limits they would not function.
That said, maybe an immediate ramp-down rather than timed step-down would be beneficial. JonnyC describes a method of starting at a PWM level of 255 (full on) and then ramping down to 140 over the course of 30s in the OP of this thread. That sounds like a great way to get the blinding “wow” factor and then rein it in without anyone noticing ;-). The time period may need to get closer to 10-15s though. I’m not really sure how long I can hang onto an 18650 tube light running at 12A…
EDIT: spelling. reign =/= rein
Not really. It’s harder to see PWM at “moon” levels, so it’s okay to run those a little slower. I’ll have to do some testing to see where the sweet spot is; where it should switch to faster PWM.
In any case, I’ll probably only reduce it to 4.5 kHz at the moon level, which cuts the ramping loss by 75% and is still fast enough most people can’t see it. Maybe I can increase the “moon” runtimes from ~50 days to ~75 days. Looking back at my efficiency measurements though, I doubt there will really be much difference at a “low” mode of 5 to 10 lumens. Moon is the main spot where the efficiency drops.
At the very least, slower PWM would probably allow more fine-tuned changes in the output level on moon modes. The last light I tested went from like 0.14 lm to 1.14 lm just by changing the PWM value by one, with nothing in-between. I’d kind of prefer if it were closer to 0.3 lm instead.
HOLY GARBANZO BEANS!!!
That would probably work remarkably well and be difficult to see by eye.
Good thinking.
This.
Also, bump for 3k lumen PocketFire mtn edition.
I think that the picture pretty much says it all! A very nice light from all outward appearances, but appearances can be deceiving... 
I got this one from a member on the forum to hopefully mod but ended up just doing a UI upgrade instead.
I bought that SRK POS from an Amazon seller unfortunately. Externally looked good and actually cost about $10 more than a internally much better SecurityIng bought a bit later for under $30. The SecurityIng at least has a proper LED platform that attaches to the body for cooling purposes. Have to go back and see if I can give it a one star review that mentions the seller by name.
Tonight I’ll be soldering the MCPB for the triple XPL to the copper spacer of my Cypreus. I have soldered the spacer to the pill and dedomed the emitters. Still seeing a triangle of spots after testing to see that they work. I will have to do the paper sack test after completing.
Link
I hope that the optics won’t melt like when I did my triple XML that had the really poor thermal pathways or connections. I have a spare just in case!
I hope all goes well with the move, Richard. I’ll be placing another order when you get back up to speed in your new location.-Rick
Nice!
I ordered a custom copper Cypreus 18650 kit. Mine will have a shortened head specifically designed for a triple along with a shortened pill. Entire light should be about 100mm long, instead of the 130mm for the stock kit.
When it arrives I’ll put a triple XPL inside along with FET driver.
I should order a new driver from Richard for this. I’m thinking off-mode memory with 3 modes: moonlight - 20% - 100% with 30 or maybe 40 second stepdown.
Or maybe I should go with 4 modes: moonlight - 10% - 40% - 100% with 30-40 second stepdown.
I remember reading many years ago about how cool LEDs ran for their light output. Even referred to as “cold” light sources IIRC. Now we are pushing power levels to the point that any glitch in cooling can cause catastrophic failures including “melt downs”. We are mounting the LEDs direct to copper heat sinks for better heat conductivity and pulling every trick except mounting them on Peltier coolers. Not using the last due to their power requirements and getting hot in the non cooling surface area too I expect. How times and LED power levels have changed.
I recently bought a NOS 128 LED shower head light just to remember what LED light technology used to be. Definitely a shelf queen. Six D battery requirement and a head that looks like it is off of a toilet plunger!
OTOH, showering with a waterproof LED light can be fun. Especially if it has motion-stopping strobe modes. Water looks cool when you only see it in narrow timeslices.
Would a light with low frequency PWM at intermediate outputs give a similar effect?
Is there any consensus as to what the best batteries are for current medium to heavy current draw flashlights with good capacity versus performance considered?
Based on capacity versus current rating I like the Panasonic NCR18650BD Protected 3200mAh batteries for protected ones as the base battery is rated for up to 10 amps continuous output. If you can use them the unprotected flat top versions of the same basic battery are good as unprotected cells it looks like. Richard does not list a button top unprotected version yet.
For an unprotected button top battery the “Panasonic NCR18650PF 2900mAh Button Top Unprotected ” ones look good unless you are trying to get the ultimate out of a unregulated high current light in which case any of the 20 Amp+ continuous output batteries from Efest, Sony, Samsung or LG that Richard offers should so far as I know provide comparable performance.
Am I missing anything? Comments please!