Pretty impressive.
That circuit board looks like a 2 sided metal core board.
12W isn't too much to dissipate across that much surface area.
I would like to know the junction temperature those LEDs are running at.
Those LEDs are cheap Chinese imitation LEDs.
They don't have the same quality control as Cree, Philips, OSRAM, etc.
133 lm/W is quite high, and that's just for the emitters themselves, not factoring in power supply inefficiencies. They're even claiming 150 lm/W on one model. Looks like it's got a fairly lousy power factor though, just a bit below 0.6 while running at close to 220VAC. One frame shows 214.5V, 0.1A, 12.3W, PF=0.577.
I also typically see higher lm/W values on cool white colors, in the 5000K-6500K range; looks like these are 4000K, neutral white.
And of course, there's color rendering quality, measured by CRI. (0=monochromatic/utterly horrible. 100="perfect," a blackbody emitter, such as the Sun or an incandescent lightbulb.)
Typically, higher CRI comes at the cost of luminous efficacy. Warmer color temperatures also typically depress the efficacy.
Example: The Luxeon Rebel.
LXW9-PW27. 2700K, 90 CRI. 120lm typical.
LXW7-PW40. 4000K, 70 CRI. 180lm typical.
LXW8-PW40. 4000K, 80 CRI. 170lm typical.
Warm color temperature + good colors = lower lumens.
Cooler color temperature + lower-quality colors = more lumens.
Unfortunately, more lumens is usually what sells more lightbulbs. So you get people buying bright LED lights, but then getting turned off on LEDs because of lousy, blue-tinted light.
I'm also curious as to why some of the bulbs are painted black. White has a habit of being a
tiny bit better than black at optimizing light output. :hmm:
Maybe that was just a quick test build.
But hey, if they've found a way to do this, using some extraordinarily-efficient LEDs, and can produce a good-quality product, great. I'd love to know what they're using for emitters and power supplies though. Cree's got some things that are pushing 200 lm/W, though some of those things they've got look like 1st-gen Luxeons from Philips.
The performance of the NanoLight is not affected by frequently turning on and off.
Good inrush limiting on the power supply? A lot of switching power supplies I've seen don't appreciate being turned on and off frequently.