Thanks. I just put the data in excel, and in single log itās a declining line. Not sure if itās supposed to be a straight line in single log to be visually linear; I have to look that up.
There are a few different theories about what makes the best āvisually linearā brightness curve. The one I showed was a cube-root method, as recommended by CPFās āselfbuiltā.
However, if you donāt like the cube-root curve, the firmware repository has a ramp calculator to generate other curves. It includes square root, cube root, fifth root, natural log, and base-two logā¦ or itās pretty easy to add others by changing only two lines of code.
I expect it will probably be reasonably easy to reflash the D4; just take out the driver and use a SOIC8 clip with avrdude on the MCU. So you should be able to change the ramp shape or change the interface entirely, if desired.
^ Great, thanks! Interestingā¦ I donāt find myself reflashing firmwareā¦ or tinkering with driversā¦ Who knows in a couple of yearsā¦
Same mistake again. You are not able to give lumens because you never know actual conditions that influence on output.
Of course smooth ramping is very usefull for flashlight which cant work longer than 30sec from start 
.
And no steps is great achievent when any level will loose several % of output every minute :person_facepalming: .
Just imagine that youve bought car, and seller claimed that it can go 100mph. But after some time you realize that this can be only reached with 95-100% tank load, with 50% tank load it cant go faster than 70mph. And there is no document which will show this feature, no method to calculate actual values.
In this analogy (it is always carsā¦) it is easy to see where you and some others differ from reality: you demand another safe and road-legal car (another Zebralight) while Hank now offers a drag racer (D4).
It is just different people who buys the two types, no reason to argue. It is Hank who made this choice, Iām sure also for commercial reasons, but I can just imagine that as many others he is attracted to small wow-lights as well
Iām really looking forward to the D4 
Im not with them 
. Im just trying to prevent some problems caused by claiming values that are not always repietable.
Drag racer dont need 18650 cell. Just imagine how many times you can turn it on (100%), wait until you cant hold it and then turn it off. New aspire 18350 capacity is more than enough for such using.
And thatās why some of us are getting 18350 tube too
As stated, itās an estimate. The actual output is complicated, but the estimate gives people a rough idea of what to expect.
You might have a point if the light only had āoffā and ādirect drive 100%ā modes. Fortunately, there are 149 other levels in-between to cover a wide variety of situations. Unfortunately, exaggeration and polarization are almost never helpful, so I hope that can stop.
Levels 1 to ~64 are current-regulated with long runtimes, and are quite useful on a small light. Above that, the levels are partially regulated and partially direct-drive. The runtime graphs should look similar to the BLF A6, where high modes sagged over time, medium modes sagged but only a little, and the entire bottom half of the modes were very close to flat.
This method gives low and medium modes which are pretty stable and practical, while also giving good āwowā modes. So it can be practical or fun depending on what you want at the time. Is it a problem to have some fun modes?
Reviewers didnāt generally measure the lower half of the modes, because the runtime was too long, but the graphs for those are mostly just a flat horizontal line.
On a FET+N+1 driver, the regulated portion goes much higher, like 3A instead of 0.35A. So it can be regulated up to about 1000 lumens instead of just ~150. This doesnāt matter much to me because I normally use less than 150 lumens, but itās nice to have sometimes.
You mean, like how the M43 can run at 100% without losing any output over time? Letās find a runtime graphā¦ (red line)
Maybe car analogies arenāt very relevant here.
The M43 does a pretty good job with power and thermal regulation (though the thermal regulation is a bit bumpy). However, virtually every flashlight sags quite a bit when running at 100%. And thatās okay. Some manage to get a flat graph on medium modes, which is nice. Others still sag a bit over time, which is not as good but usually not noticeable by eye. And both the M43 and D4 have regulation on low modes.
They are both good lights. I donāt understand why a few vocal people are upset about the D4. I expect Iāll enjoy mine quite a bit.
I love those Aspire cells, it made all my 18350 lights worthwhile, it is my perfect size flashlight but the battery capacity always made me have mixed feelings about them.
That said, I think I will probably mostly use 18650 cells in the D4 because my current 18350 EDC is about the perfect version of the 18350 format, a Jaxman E2L shorty (bistro driver with R9080 Nichiaās), for me the D4 will not beat that one, but with 18650 battery, from the lights that I have, it will probably be the best in its class.
Runtimes couldnt be similar to blf A6 cause bigger current with same cell size.
Im not against fet, but small edc should have one regulated mode that gives such output that host can dispread in air. In this case, fet+4 amc with one separated amc could be great achivement in comparing with A6 driver. Right now (three years after) we cant see big difference in fet drivers.
Im dont have m43, but I think it is able to give near 100% output with 50% charged cells. Graph that you posted shows thermal regulation.
Can anyone confirm if the Emisar D4 will be sold on intl-outdoor? Or will it be sold exclusively to ME?
Hank not reply my question, from US the price of shipping is prohibitive for Europe.
Compared to the BLF A6, runtimes on the Emisar D4 ā at the same lumen level ā might actually be a bit longer. Four emitters are more efficient than one, producing more lumens per Watt. This may vary by emitter type though, so measurements are needed.
At maximum power, runtimes will be very short. This is normal for āhot rodā lights. If it pulls 15 amps from a 3Ah cell, it could drain the cell in 12 minutes under ideal conditions. During actual use it should be significantly longer though, because thermal step-down will reduce the power to safer levels.
I have a M43, and it does not deliver 100% output with 50% charged cells, even right after turning on. It does as well as can be expected, performing admirably, but no high-power light can be expected to do that. With current battery technology, it is like trying to squeeze blood from a stone.
Perhaps new battery tech will make this a thing of the past.
I think Hank plans to sell it at his site soon, but there are some details to work out first. He noticed something heās not happy with in the first batch, and wants to change it. If things go well, it could maybe be available in a few days. Or it could take longer. I should find out more soon.
My order says āout for deliveryā so it should be here later today. Iāll have a review up as soon as possible.
Thanks for the information ToyKeeper, I am sure that many people will be interested in UE for this light!
(I keep my eyes in your review) 
But you cant manage lumen level. You can just choose pwm %.
3sĆ8AĆ·4P*(Vout/Vin)Ć1/85%=11Ć8/4Ć1Ć1.2~25W
Most high-drain cells will easily carry this power, there might be too rough cell voltage drop protection.
Phew! Review posted:
Now I can finally go get lunch.
(and then try to get the driver out so I can make a mess of the internals)
Iāve taken to doing runtime tests with the flashlight immersed in water to test what the driver can really do without the interference of that pesky thermal regulation. This, of course has very little bearing on any practical use case, but there do exist some fairly strong boost driver lights that have flat output if they donāt overheat.
On the other hand, ToyKeeperās review demonstrates that the D4 puts out considerably more light on a ~50% battery than the 1500 or so lumens of a Wizard Pro.
If you will repeat this with D4 and it will pass it well, this will be best waterproof test.
LOL, good point. Itās hard to compare 1500 lm with regulation to 3800 lm without regulation. Or 2100 lm on a half-full 18350 cell without regulation. In this case, even a half-full cell with a fraction the capacity is putting out more light. Apples and oranges.
Iāll have to see if I can simulate that Wizard Pro by setting the thermal ceiling low enough to limit it to 1500 lm, and then see how flat the graph is. Itāll be a good way to make sure the regulation isnāt bumpy.
I will be getting a D4, and I will be testing it under water, but we all know about what to expect. Testing boost drivers is more interesting because they canāt all maintain flat output. They do all stay close enough to flat that you wouldnāt notice if you werenāt doing an instrumented test though.