TK's Emisar D4V2 review

I understand that, I should have clarified everything else about the design being equal. I do believe the XHP70.2 is more efficient lumens per watt. Of course driver design will impact heat output just as much as anything.

For Cree XP-L HI they claim Maximum Efficacy at Binning Conditions (lm/W) 136 lm/W. That is compared to 181 lm/W for the XHP70.2. I am not sure exactly the conditions and of course that ignores driver efficiency.

4 X-PLs (yes, X-PL HDs, but its not a huge difference) is effectively an XHP70 though, and the XHP70.2 in the SC700 is high-CRI so almost certainly much lower efficiency than 181lm/W.

As far as I can tell, the bulk of the difference is due to differences in efficiency… mostly in the LED and the driver. More efficient LED plus more efficient driver equals higher sustainable lumen output at the same temperature.

How much higher though?

Looking at some of maukka’s old tests, he measured up to 140 lm/W on a ZL SC64c with a high-CRI LED… and only 93 lm/W on a FW3A with cool white XP-L HI. That suggests potentially as much as a 50% increase from driver differences alone, even before giving ZL a more efficient LED.

But the XP-L HI’s beam looks gorgeous, while the ZebraLight’s XP-L2 beam (despite being high-CRI) was rather ugly. Because ZebraLight’s LEDs seem to be chosen based on what has good specs on paper, while Emisar’s LEDs are chosen mostly based on what people request and what looks good during use.

The host definitely factors into the results too, but for lights of similar size and mass I’m guessing it’s a smaller factor than the LEDs and drivers.

It’ll be interesting to so what kind of numbers the D4/KR4 mules get.

Yeah, but they all have eyes to see what the beam looks like, pattern and tint. Choices are made because of financial reasons mainly, for example green tinted LEDs are easy to find so they are pumped into flashlights. That is why I have 95CRI LEDs with greenish tints, the paper says great, visual experience says not great.
XHP50.2 XP-L2 XP-G3 are not really LEDs that beam and tint snobists should buy anyway.

Hank doesn’t offer XP-L HD in the D4, so I am going to run with XP-L HI. Because Firelight2 was talking about being able to run his Zebralight for a significant amount of time at 1000 lumens let me do some comparisons.

Texas Ace tested a XHP70.2 that was 80 CRI. I can’t find any tests for the 90 CRI version, so I am going to run with his test. At 1319 lumens it used 6.9w which is 190.3 lumen/watt. He did state in his tests between a low CRI version and 80 CRI there was little difference at lower power and 8% difference at high power. I think that makes sense because of the lower Vf.

Texas Ace also tested a XP-L HI 1A V2 which is 70 CRI. At 327 lumens it used 2.2w which is 145.4 lumen/watt. So if you had 4 in parallel it would be around 1308 lumens using 8.8w. Obviously many other variables come into play and this isn’t a completely fair comparison. Those numbers suggest up to around a 30% difference. But if there was only a 10-15% efficiency gain if you stack that with a much more efficient driver then it really adds up.

I personally can not stand the XHP70.2. I use my flashlights inside mainly, and I am one of the worst white wall hunters.

As a daft question time, but where do you get hold of the latest firmware?
i tried the link in Toykeeps signature, Flashlight Firmware Repository in Launchpad
but that just leads here:


Here: Index of /torches/fsm

Brb while I go buy a ZL.

Those are some awesome videos.

Be careful with that stuff, im fairly certain AS5 is electrically conductive.

Yeah, sorry, the way Launchpad is organized can be a little strange at times. The source code is there, but I don’t normally keep the compiled versions there. If you’re looking for a pre-compiled .hex file, I occasionally put builds at the link SammysHP shared. Those are not always production-quality builds though, so some caution is advised.

The 2020-03-18 set has received a relatively large amount of testing though, so those should be pretty safe as long as you calibrate the temperature sensor and don’t keep pushing it back up to turbo while it’s already hot.

Mass should not have significant effect on sustained performance.
Surface temperature integrated over surface area is roughly what determines heat extraction efficiency.
Mass may be used to improve heat transfer or to increase are but it’s a very non-linear relationship and so mass is a very bad predictor of host thermal performance.


Does you D4 have the screws to hold down the PCB? The early versions did not have 2 screws, that is why I am asking.

The screws have no real effect on thermal performance. This is because the early model without the screws still had pressure from the optic on the star to get a tight fit with the shelf.

The purpose of the screws is to prevent the star from rotating to protect the driver wires from torquing.

Maybe Loneoceans buck-boost for the next Noctigon
He’s close. 3A regulated plus open FET, in addition to RGB aux LEDs. Hm, getting a driver from him will put my FW3A even higher in the EDC rotation… I’m so glad the o-ring mod made that light useful.

I’d be cautious about what’s used in threads. If the threads are used at all frequently, most compounds we’re talking about are abrasive. I can definitely see arguments for thermal management vs. wear though, especially considering that most lights are rarely opened from the head.

AFAIK very few manufacturers have stepped into that ballgame. Zebralight, obviously already mentioned. Nitecore has at least a couple as I recall. The Lumintop EDC05C is one piece. I know there’s others I’m missing.
That said, it definitely makes for modding obstacles. Even initial assembly is tougher. I wouldn’t mind seeing it, though. If there’s somebody that can make a light that I’d trust to be good enough to not be opened, it’s Hank.
As for the on-MCPCB temp sensor, that’s easy.

Keep in mind that a lot of thermal compounds will solidify also. Not sure how long it takes.

I doubt that the pressure from the optic is as good as the pressure from the screws, some people even reported having no pressure on the MCPCB from the the optic.

Before it was proven with tests that hard pressure on the MCPCB to the integrated shelf made a difference, in general, I am not talking about D4 tests.

Good point. When I had my D4v2 it had screws in the MCPCB.

My TO50R has screws as well and seems to have have good thermal path from the emitters because the heat can be felt around the entire head of the light very soon after turning it to high or turbo.

I have many D4 of various different models including 2 of the first generation model without screws.

The head heats up at about the same rate in all of them. The screws seem to make no difference at all.

If I recall correctly, the screws were added because people were moving the O-ring on the original D4, which made it so there was no pressure on the MCPCB… and then people were finding that their lights would overheat and damage the 219C emitters.

The screws hold down the MCPCB tight even if other parts are moved (or removed, like for a mule).

ToyKeeper, any thought to adding a Beacon Cfg 2 that you could set N for duration of light on while in beacon, not just the interval between on but the time in which it’s on for beacon? I prefer a shorter burst of light for beacon than default but could see where some would want it up to a couple seconds.