Currently there are no special modes, just constant output modes.
Turbo can definitely be disabled upon compilation.
Richard might be able to comment on the lowest output.
I'm hoping it goes lower than the Qlite. The moonlight mode in my 5.8A triple XP-G2 is brighter than I expected. If that was lower, the spacing between the next mode would be much better.
The lowest output level is highly variable depending on a number of factors... you could pull out 5 drivers and 5 XM-L2s and get different minimum levels due small production variance. Temperature also affects the minimum level. My DMM isn't very accurate when measuring moonlight current levels so I can't give you any definite mA numbers. Hard to get much lower than the default qlite moonlight or NLITE moonlight level driving an XM-L2 with any sort of guarantee that it will work in a particular setup.
However, I haven't tested it yet but I think that I may be able to go lower with the triple XP-G2s, especially with the 4 or 8 added 7135s. We may be able to bring moonlight back down to a similar level to what you would get with only 8 7135s.
The lowest mode is always tricky, especially when using the driver with different emitters. It uses the same PWM frequency as the NLITE driver. I had a light running with firefly output, but once I left it on for a few minutes that mode no longer worked (has been mentioned on this forum, that it's something to do with the 7135's).
Thank you both for your responses.
My interest is achieving modes slightly brighter than moonlight in various steps. This is a light which will be used at night for walking around the house. I’d like to have a couple of modes brighter than moonlight but lower than the 60ma low.
Kodachrome, that should be doable.
Great! Thanks!
I’ll start figuring out the modes. I think I might base it on a drive with six chips removed. 700ma output should be a lot easier to get spacing from.
PS. I cant wait to start playing with the multicolor led P60 you made me. Now where is that mailman??? I might have to go looking for him :bigsmile:
Awesome. I could definitely get onboard with that. If the triple S3 gets a max of about 1200 lumens (which I’ve heard may be a conservative estimate), and if the power/output curve is pretty close to linear (since it uses PWM), I’d love something like moon/1%/4%/20%/100%, low to high with no memory. That’s roughly ~0.5 lm, 12 lm, 48 lm, 240 lm, 1200 lm.
BTW, what does the step-down step down to? Is the mode lock time measured while it’s on or while off?
I should have been clear that I was only thinking of the triple xp-g2. I've heard the same thing about how low it can go with a single emitter. I need to get some of those high CRI xp-g2's so you can build me another triple. That'll be very nice if you are able to manage getting the moonlight mode back down.
Is the floodier 10509 an option too? I'm craving more flood than the 10508.
I think 1200 lumens is extremely conservative.
I just added the 10509 as an option. I am almost caught up enough to start offering these again, just need to finish a few Small Suns today and tomorrow then I will have them back up.
Working on getting all of the firmware options up too.
3 XP-G2 R5 2B’s on a triple Noctigon at 5.23A are making 1542 OTF on start-up and 1487 at 30 seconds. This is in an L2m shorty running an AW IMR 18350. 
This IS with the help of 3 ounces of copper. 
OK, well at the 5.6+ that most of these are making with 8 chips and a 20R I think that 1600+ is a safe estimate. I don't have any way to measure actual lumens. I do have a light meter and my eyes, but no integrating sphere or any way to know how it is calibrated. Is 1200 pretty close for 4.4A?
I have two light meters, but unfortunately I don't have an IS either. I was going to make a pvc IS like a few members here have, but my nearest Home Depot doesn't have the parts. I would believe 1600 lumens. I haven't compared it directly against my Eagletac MX25L3, but that's what I'm reminded of. It's amazing to have so much light in such a small package.
3.75A makes 1160 out the front at 30 seconds
It is on-time memory, but we were playing with adding a capacitor for off-time memory. Not totally reliable yet.
It starts the step down around 3v, cutting the output by half after flashing 3 times. After 3 seconds it will check the voltage again, and if still below 3v it will keep going. Once it gets to the lowest mode (moonlight if enabled, or whatever your lowest programmed mode is), and it gets under ~2.9v, it will flash 10 times and then turn off. It goes into a power down sleep mode, so it shouldn't consume much energy. The voltage levels (3v and 2.9v) are options that can be adjusted at the time of flashing the program. Seems as though the ADC voltage monitoring is not very accurate though as the reference voltage used in the MCU of 1.1 can be between 1.0v and 1.1v.
Any other ideas on the best way to do this?
Ah, okay. I was thinking the step-down was to avoid burning up in the highest mode. Like, step down from 100% to 50% after 60 seconds.
Low-voltage step down / shutoff is good too, though it’s rare that I let my cells get that low.
On a different topic, given the lumen estimates of similar lights, I think I’d have to either adjust the output levels or the number of extra chips… even 1% of ~1600 lm is a bit high for a “low” level. What kind of scale is it using to set modes, or what is the smallest step size? I’m guessing it’s probably not decimal percentage, since base-10 is awkward on binary processors… Perhaps I should just go grab some published firmware and read it. I really need to get myself some firmware flashing tools and mess with it myself… got a bunch of odd ideas in mind that I seriously doubt anyone else cares about (like a variable-speed motion-stopping strobe light controlled via electronic switch).
In any case, I was thinking that, instead of 1% - 4% - 20%, it could be more like 2/256 - 8/256 - 32/256.
Oh whoops, you were talking about turbo ramp-down. Sorry about that ;) Right now turbo is programmed to just step down one mode. So if it's L-M-H-T, then it will drop back down to H.
The Qlite is 2%, and I think that works fine for the low mode. So 32 lumens if light output was linear, but I'm pretty sure 2% is based on current draw...or close enough. According to the Cree tool, it should be around 50 lumens. I used it for reading my Kindle on my last camping trip. I'll be happy enough if the moonlight is moved down a bit. If I were to use the STAR firmware and all six modes, I'd start with the Qlite modes...
5mA - 2% - 25% - 100%
Moonlight with 5.8A is too high, so I'd move that down so it's similar to a stock Qlite driving a single emitter. Replace the 2% with a 1.5% and a 5%. That's 5 modes and very similar to what you described earlier. I suppose if I really wanted another mode, I'd add a 50% mode that should put out around 1000 lumens and would appear nearly identical to 100% with much less current and heat. I probably wouldn't do that though. I don't like having a hard time distinguishing which mode I'm in. I want it to be obvious.
2B+3C+5A1 would probably be nice too. I have some 7C4s to replace the XTE but they're only Q2 bin, 5A1 is R4.
Even a mix like 1A+2B+3C/D is good, all the stray undertones of each LED get cancelled out by the other two. Looks a lot like the MTG2, in that there's no stray blues/greens visible.
If you wanted to go steampunk hipster retro incan style, 3C+5A1+7C4 would work.
I'm tempted to order a third one with dedomed XP-E2's.