This was probably discussed before and I missed it, but the Optic in the D4’s has 2 small crack in the corner of one of the four lens. Noticed it on mine. Does not affect beam pattern. Contacted Hank and he immediately responded with pictures of individual optics and of a sheet of these Optics and all have the characteristic cracks which are just an unfortunate part of the manufacture process. Most here probably know this already, but just in case there is another old, memory-impaired member that was wondering the same thing, if I save even one person from having to send a repetitive email to Hank, this post is worth it. Hank has been very responsive to all my emails and am very impressed with him and the D4.
And many thanks, ToyKeeper, for all your efforts in thoroughly reviewing and championing this light to a v2 version so quickly and Tom E for this amazing UI! Really fun to play with and also to use as a real EDC!
Cheers!
Yup, it’s there, didn’t see it until you mentioned it and I looked for it.
I got a 30Q from BG as part of a BLF A6 bundle once and it took me forever to realize it had an impossibly clear and perfectly placed film on one of the contacts for shipping purposes. My vision is fine and I even looked and missed it several times thinking that could be the problem.
Jeroen, I do think the way you do all the time, but in case of the D4 just let it go. The theories are completely correct but in the practice of using the D4, a 20% lowered efficiency in the middle modes is not noticable. I took it out camping with a 30Q battery in it (admittedly mine has 3500K 90CRI 219C’s now) and what remains is that it is a convenient goodlooking compact flashlight with a great user interface and an incredable output range from sublumen moon to very very bright. I will seldomly use it over 350mA, so almost always it is as efficient as can be with a runtime that goes ways beyond my vacation.
I finally started on new thermal regulation code today. Currently charging up cells for testing, but the planned upgrades are:
- Try a predictive PID algorithm instead of predictive latch-lowpassed PD. More “I”, less signal post-processing. Adjust based on accumulated “D” instead of immediate “D”, basically.
- Round off the corners of the steps to make regulation less visible. (one more layer between target level and actual level, with a soft adjustment per clock tick when they differ)
At the very least, if it overshoots it should correct itself much faster. And the steps should be less noticeable. Hopefully it’ll also reduce noise in the runtime graph, but that’s optimistic.
Testing and tweaking always takes a while on these things though. It looks good in a simulation (even after increasing sensor noise and thermal lag), but real-world results may not match.
Once I’m happy with how it works, maybe I can finally take apart my green D4 and reflash it. It’s still running v1 code, since I only want to open it up once if possible.
(if this all sounds Greek, sorry, sometimes I slip back into my native tongue… at least I’m not speaking in S-expressions!)
Got my Solarforce clip yesterday and my copper washer from Oveready today. Dremeled out the inside of the clip and cut the washer and spread it out to fit snuggly in the cap. Very happy with this 4000K 219c modded green monster!
I actually just dremeled down the edges of the washer after I saw how sharp the ends looked in the picture.
Is this the $5 washer?
Yes. Was a $9.00 washer with shipping. But is working perfectly with no tailcap slippage.
Also, the clips shipping directly from Solarforce were not “branded” by them. So it doesn’t look like a Solarforce /Emisar light.
I have just ordered two clips from Solarforce, both have deep branded laser engraving. Can not polish off!! 
Hey, it’s a SolarD4orce!
May D4orce be with you. 
Sounds like divorce…
I think this one may fit too:
Hmmmmm. I’m not sure why my clip came unbranded and yours didn’t.
The other thing I thought was strange is my clip showed up to my New Hampshire USA residence 5 to 7 days after ordered. Fastest boat from China ever.
TK, will you also be doing reviews on the other upcoming Emisar lights?
TK, I have a question for you too. Is there intention to put your latest developments on the future Emisar lights? I mean both UI rewrite and thermal improvements.
I’ll probably review the D1 at least, whenever that’s available. I’m somewhat less interested in the D1S and D7 though, because they’re bigger and more specialized. Of course, that said, the D1S seems like a pretty good replacement for a C8 or X6 so it’ll probably be pretty popular too. And soda can lights are fun, even if they’re a bit much for my purposes.
I don’t know any details yet about future Emisar firmware. It depends on what Hank wants.
The new code is getting pretty close, at least. I did some thermal testing and tweaking today, and here’s how the result looked on the hottest most over-powered light I have: (D4 219c running at full power, ~20 amps initially)
The initial little dip at the top was me moving the light, but after that I held it pretty steady. There were also a couple times later I moved to switch hands, but I tried to keep the light from moving.
What this graph shows is that the adjustments are smooth and gradual now (hard to see without tools), that it reaches its stable state fairly quickly without wrong turns, and that it stays reasonably stable afterward. I had the temperature limit set fairly low to try to make it overshoot or bounce, but it behaved pretty well anyway.
What the graph doesn’t show is that the D4 219c still gets painfully hot during the first minute, even while constantly reducing power. I has way too much power for its thermal mass. So, on mine when I’m not testing stuff I set the default ramp ceiling to about half power, and the heat becomes much more manageable.
Also not shown: it reacts much faster to changing environmental conditions now. Get it wet or touch it with something cold, and it’ll raise the output pretty quickly.