Thanks, I’m happy someone else agrees, I actually felt kind of anal by pointing it out because I figured most readers understood what he meant, but it was my own slow working brain that had a duh moment when I first read his post, took me a few minutes to realize he was referring to USB cells :person_facepalming:
Ahah, so both we have slow working brains mate, worry not 
I was also confused at first, but then perceived the intention! :+1:
Anyway, I don’t think that USB rechargeables will fit, and mostly, they probably won’t deliver the maximum energy that this light needs to be “a hot rod”! 
I received my Emisar D4 with High CRI Nichia from Hank. My first one got lost in the mail, and he was very professional in getting me a replacement. I am very happy with his service, and I am happy with the light overall, but my only complaint is that the lowest lumen setting is not low enough for me. I like the sub lumen settings of the Zebralights, and the Emisar won’t go that low. Are there any plans to change this in the future?
TK has said there’s hardware limitations to achieving the lower sublumen levels.
I’m repeatedly hearing that Emisar plus Zebralight is a favourite combination. I can see why, I’ve arrived at the same conclusion. I just don’t have a Zebralight yet.
Welcome to BLF!
The D4 can actually go lower, but the way it’s implemented in the stock firmware is very sensitive to voltage and emitter type, and would risk not emitting any light at all on some units if lowered any more. But a different method can be used which makes it more stable, more efficient, and less variable per unit. To get it lower, you’ll need to reflash the firmware… which requires some soldering and some extra tools.
The stock firmware has a lowest mode of about 0.5 lm, which uses about 7 mA and has a runtime of about 18 days.
What I’m running on mine (Anduril) has a lowest mode of about 0.2 lm, which uses about 1.7 mA and has a runtime of about 70 days. Plus, it has a whole bunch of extra features. Its thermal regulation isn’t currently as aggressive though, so it’d require a bit more manual care to avoid overheating on turbo.
Mtnelectronics have d4 in stock… 113 available…
http://www.mtnelectronics.com/index.php?route=product/product&path=80&product_id=855
Are you sure the lowest is 0.5 lumens on stock firmware? Seems higher than that. Out of curiosity, how does Zebralight get theirs to go so low?
The stock D4 lowest mode varies quite a bit from unit to unit, anywhere from ~0.2 lm to ~1.2 lm. And this varies quite a bit with voltage.
Zebralight seems to have a separate power channel or something for moon. I think. I’m not sure. However, it’s actually not as efficient as one of BLF’s recent designs. The BLF Q8’s lighted button can put out two levels of light, and both have longer runtime per mAh than Zebralight’s lowest firefly modes, even though the Q8 button emits more light. This is because the Q8 button is basically just a battery and a resistor and a LED. Zebralight’s solution, whatever it is, is more complicated than that.
Correction: My Emisar D4 Nichia is running at approximately 0.5 lumen on the lowest setting. I have my Zebralight’s lowest setting at 0.01 lumen, which is why the Emisar seems so bright in comparison.
I find the D4 moon(or what ever you want to call it)mode pretty decent. But then i just dont care for sub lumen levels at all, inc ZL’s 0.000000001lm type outputs(ok a little exaggeration).
Generally i like to see well, even in the middle of the night(limited uses here though), sub lumen i just cant see well enough. I prefer a few lumens for example to light a few steps up. If i want some ambient light/mood type thing, the HDS candle mode(9 clicks from off) is pretty awesome. If i tend to tail stand a light, i always use a few lumes, D4 case i ramp it up a little to light up the corner .
I think i am 1 in a 1,000,000 
Interesting, I hadn’t heard about a HDS candle mode before. I think that’s now four different implementations I’ve seen, all of which seem completely different from each other. The HDS one looks particularly spastic, like it’s always wobbling and never settles.
Anyway, super-low modes are nice, but they generally require extra hardware to make them feasible. The D4 doesn’t have any extra hardware for that, so its lowest modes are pushing more general-purpose hardware to the edge of what it can do. All I did with Anduril was find a way to make that edge a bit lower, more efficient, and more stable.
Is it possible to use high frequency PWM specifically on that single 7135 chip, let’s say on a ~1 lumen output to generate a stable sub lumen output of say ~0.01 lumen?
Yes and no. The faster it goes, the most voltage-sensitive it is.
0.01 lm, yes.
Stable, no.
I used a combination of PWM and PFM to make a self-adjusting moon mode on my original Cypreus light, when it was a FET-only design. It adjusted itself from 8,000 to 2,000,000 pulses per second. But it still was too bright on a full battery, and too dim on a low battery. Kinda neat watching it regulate itself though, since it made an incredibly smooth sub-lumen ramp.
What I’ve found works better is slowing down PWM as much as possible on moon mode… not enough to look like a strobe, but enough to make the output more stable. If I recall correctly, Anduril adjusts its PWM speed from 4,000 to 16,000 pulses per second depending on the overall brightness level, to increase stability and efficiency at the low end. This also makes it much more consistent from one light to the next.
Cool, thanks! :+1: 
Does anyone know which type 7135 that is used on the D4 board. The 7135 on the BLF A6 FET + 1-7135 board can be run so low that it just barely glows in moonlight setting 1 of NarsilM. The reason I ask is because I just modded Astrolux E01 with a 22m FET + 1-7135 Mtn driver. I reflashed the attiny85 with NarsilMQ8. The led would not even come on in moonlight until I reached level 4. In ramping the floor level was not lighting up the led. The light was still on but there was no light. I could hold the button and ramp up a couple of levels then the led would light up and continue increasing until the ceiling. Going back down near the bottom of the floor the light would go out. I couldn’t live with this light being on at the floor level but no light coming out. I wouldn’t know if the light was on or off. I figured the difference had to be the 7135 since NarsilM always worked fine with the BLF A6 driver. I asked Richard and he agreeds it’s probably the 7135. So I hot aired the 7135 off the Mtn driver and Stripped a 7135 of a junked BLF A6 driver (mod gone bad) I had laying around. Reflowed the 7135 from the BLF A6 board to the Mtn board and everything works like it should now. Moonlight setting 1 is a faint glow. Ramps up and down just fine now, so that’s fixed.
.
So how low of a mode that can be accomplished with light, will depend on which 7135 that is used. Richard uses a 380 ma 7135. The BLF A6 board uses a 350 ma 7135 as far as I know. The A6 is on the left and the X6 is on the right both have a very low moonlight with NarsilM. You can see that little symbol (looks like a rainbow) to the left of the L7135. I don’t know who makes them but they work good for sub lumen modes.
.
The D4 uses the same “raptor claw” brand chip. So does the BLF Q8, IIRC. And that’s what we spec’d as a requirement for the FW3A.
The 380mA chips pretty much all have poor performance at the bottom end… but so do some of the 350mA chips. There are only two specific flavors I’m aware of which work well at the lowest levels, and both are 350mA varieties.
That’s not all there is to it though. Other factors matter too, particularly the type of LED. Like, XM-L2 will generally light up at levels which the older XM-L won’t. All this moon stuff is highly hardware-specific and tends to be inconsistent even from one identical item to the next.
The 7135 version with that “rainbow” logo is probably the best available 7135 design. You’ll normally see it referred to here on BLF as a “raptor claw” 7135, and they’re irritatingly hard to find a reliable supplier for.
It seems to be made by a company called “Nanjing Micro One Electronics Inc.” Their website isn’t up to much, but I did locate their 7135 datasheet - their part number is MEL7135 - on a different website that seems to be a Chinese component supplier. Be warned, the datasheet is decidedly flaky 
While we’re on the subject of 7135 logos, I hear that the “sailboat” 7135 made by ADDtek is one to avoid, and the generic 7135s with no logo are essentially a lottery.
Edit: I see ToyKeeper got in before me:
That’s interesting to know! What’s the other good version?
One of the unbranded ones that Convoy tends to use seems to work well. All it says is “7135” on one line, then “35B” or some other letter on the next line. I think I’ve seen A through F.
Thank you
I might have some of those lying about, chucked in a parts box thinking they weren’t as good as the “raptor claw” ones. That would be a nice surprise!
That’s a lot of good info. Thanks Toykeeper and Phlogiston.
The D4 uses the same (rainbow) raptor claw type 7135, good to know. :+1:
Is there any variation in the raptor claw brand that doesn’t give a good low or are they all the same besides the manufacture slight variances.
Mine has numbers at the bottom, right below the L7135. Is there any particular number code?
I got one more driver to give a 7135 super low transplant. I have a bunch of 7135’s, needing to find the better one to use.