Issac,
With increased area, you also get a decreased thermal resistance. That's why the Luxeon V thermal resistance is lower than the Luxeon III. Likewise, for the warm Luxeons vs warm Luxeon IIIs.
Some time ago, I used the thermal resistances of the Luxeon V and III to calculate what the die->submount thermal resistance is for each Luxeon III die, and the thermal resistance from the submount -> slug.
Lemme see if I can dig that up:
Here are snippits from that post:
[ QUOTE ]
However, the thermal resistance (j-s) of an L5 is only about 1/2 of a 1W...why?
Theoretically, it should be 1/4.
It's possible that there is some base thermal resistance for the submount, based on the thermal resistance of the submount, plus thermal resistance for each die (junction -> submount)
This base thermal resistance would be nearly identical between the 1W and 5W (actually, a little worse for the 5W), since the submount area is the same between the two.
For a Luxeon III, if we say we can go from 13C/W to 8C/W, but some part has to be divided by 4 (because of the junction area/number of solder blobs being
multiplied by 4), then that's Rj-s = Rbase + (Rj-sub)/(# of dies)
[/ QUOTE ]
so using the Lux V and III for the above (the good ole system of equations):
8 = rbase + (Rj-sub)/4 <--- L5
13 = rbase + (Rj-sub) <--- L3
Then Rj-sub = 6.7 C/W
Rbase (sub-slug) = 6.3 C/W
Likewise, we apply the same forumla to the warm Lux I/III - there isn't a silicon submount, but we'll say that Rj-sub is the thermal resistance from the junction to the joint between the die and slug.
17 = rbase + (Rj-sub) <--- L1 warm
6 = rbase + (Rj-sub)/4 <--- L3 warm
Then,
rbase = 2.3C/W
and
Rj-slug joint = 14.7C/W
by quadrupling the size of the die, that Rj-slug gets reduced to 3.7C/W, for 6C/W total.
That 14.7C/W also includes the die attach method, so any thermal resistance in that gets reduced to 1/4 as well when the die size increases. This is not true for the "cold" luxeons, because the silicon submount->slug is the same size/attachement method between the Luxeon III and V.
So, one way that Lumileds could reduce the warm K2 thermal resistance to 12C/W is through improving the die attach method, and also improving the slug thermal resistance (with a beefier slug). But, there is still a ton of thermal resistance in the junction->die attach path that looks like it's simply a result of the materials/structure themselves.
Notice that the cold K2s thermal resistance is only reduced to 9C/W. That could have been accomplished mostly by improving the submount->slug and slug thermal resistances (which is probably easier than improving the junction-> submount thermal resistance), but I'm sure improvements in the junction->submount path have been made as well.