So how did they break the 251 lm/w theoretical maximum?
http://www.cree.com/news-and-events/cree-news/press-releases/2013/february/276-lpw
Didn’t you say yesterday something like “if we assume everything as real, nothing would be invented”?
Well, that’s not the actual quote, but maybe you remember… I’m lazy to search 
Edit: you did!
Cool. Really, I thought 251 lm/W was a hard theoretical maximum like Carnot efficiency, or something?
Also, why were they going after an efficacy milestone with a 4400k emitter? Not that I’m complaining, as that’s the CCT sweet spot for me.
That is still correct, if 254 lm/w were good enough, they would not have worked on 276lm/w
That reply was in response to someone who said something is good enough. I’m not saying 251lm/w is good enough, i’m saying its quoted as the most lumens possible in classic physics. If we could get 700lm/w i would be extra happy!
You can’t trust a physicist…
I thought the theoretical maximum was around 300-320 lm/W? In which case going from 250 to 270lm/W would reduce heat sinking requirements by around 30%…. wonder how long before such LEDs enter mass production (if ever)?
As for why 4400K: my guess is they are going for a minimum CRI of 80? I guess it doesn’t really make much sense to claim records with LEDs that have a low CRI.
The theoretical maximum is for pure white light, following a black body radiation. If you go for monochrome light it is possible to reach 683 lumen/watt at 555 nm (This is something between green and yellow).
I.e. by reducing the CRI it is possible to go above 251 lumen/watt.
I suspected something like this, i figured 251lm/w was based on a weighted RGB output.
So this also suggests you can’t calculate efficiency based on lumens per watt (to calculate heat output from the led)
Right. Lumens is based on a human perception of brightness. Different frequencies are percieved differently even though the energy is the same.