Over on Reddit, I made an observation to Zak (he posts as zak.wilson here) about the efficiency of an Acebeam he reviewed, and he countered by pointing out some of the more efficient lights he had tested were much better than the one we were discussing at the moment (based on an approximate calibration method, I believe):
These were just from high CRI lights that he reviewed. He’s tested a couple low CRI Thrunites and Olights that were in the ballpark of 200 lm/W.
Lumen per watt is a bad measure of efficiency as it’s mostly dependent on the emitter which is used. Also, it depends on what current the emitter is driven at. An underdriven flashlight will be more “efficient” everything equal.
Unfortunately I don’t have that many flashlights especially from majors brands like Acebeam, Olight, Fenix, Thrunite… etc to compare them.
Zebralight gets often parroted as ”very efficient” but that depends on the model, the SC64 (XHP35) is not that good, and the H/SC53 is very bad (it’s an AA light though, it’s bound to be less efficient), the SC700 is excellent.
I have a H600II and a H600 IV opened so I’ll try to measure those too.
The Skilhunt H04/H03 is often recommended as a higher value headlamp compared to stuff from Sofirn for example, thanks to it’s buck driver, but while it’s better than FET direct drive and to a lesser extent linear driver, its efficiency is not good for a buck driver.
The Convoy XHP35 driver has been measured here : Convoy XHP35 Driver Analysis / Testing / Schematic
(Modified to 6V and RPP FET bypassed so slightly better numbers than stock). And it’s very good, most of their switching driver looks to be using good synchronous converters and good components (low RDson RPP FET, decently sized inductor) so it looks to me that their offering are pretty good.
Only synchronous buck or boost drivers have highest efficiency. But its technical data and ordinary potato user dont care about it, so manufacturers advertising lumens only.
EK4 uses a buck driver (edit : or boost maybe if the LEDs are wired in 4S), given the high power I think we would expect good efficiency at sustained output level, but ultimately we can’t really know without measuring or at least seeing what the driver looks like.
D4V2 uses a linear driver (not sure if they still use the older FET+7135 at all), sustained output driver efficiency is going to be around 75% in average.
It’s hard to measure efficiency with a single metric, but from a driver standpoint, boost/buck are most efficient with everything else (FET, 7135, or equivalent) less so. It also depends on the emitter. High CRI low CCT are less efficient, and also depends on how it’s driven (direct vs. refulated). So far, lights with xhp70.2 behind boost or buck drivers have been most efficient in my experience. I’ve had good results with XPG3, and XP-L2 as well. Cyansky, Skilhunt, Thrunite, Acebeam, and Convoys newer ully regulated boost/buck drivers are all very very good. I haven’t tested it personally but the Lume1 driver in the fireflies E07x is superb (but spendy)
You can have a 6000lb SUV with a V8 thats efficient.
Most high MPG cars are small. If I have to give up size to gain in MPG , that does not make it efficient. I paid for greater mpg by giving up size.
With electronics, efficency has to do with how much energy is lost to heat. So if you have a 1000lm light that gives 1hr run time compared to a 1000lm light that gives 45 min runtime but runs much hotter, that light would be less efficient.
If you compare a 1000lm light that runs for an hour vs a 2000lm light that runs for 2 hrs, you really dont kbow which one if more or less efficient.