It is not a matter of impressive but a incremental improvement. To the comment about energy required to manufacture ... almost nill compared to what they collect. The energy to manufacture the rest of a concentrator assembly is way higher. For regular panels, they take about 2 years to recover the energy required to manufacture. For these I expect much less.
To the comment about the number of suns and clouds ... normally these are used for very sunny areas, think desert, but the efficiency does not suddenly drop to 0 below 400 suns, it just drops.
In terms of cooling, most concentrator solar panels use passive cooling. Absolutely you loose efficiency due to higher temps, but still very high overall.
No. 5.5mmx5mm is a fingernail clipping. It obviously can't yield anything bigger, or it would be bigger. Call me (I'll be at the rest home, peeing in a bag by then) when it's on a 6 inch wafer and I can focus 400 suns on it. I'll get rid of the heat, give me the 'lectricity...it'll have to be for less than $1 a watt, by the way, kids.
If you want to see a real improvement in solar efficiency check out the solar Australian company K2 who in collaboration with Mears technologies have developed a technique for 60% increased efficiency.
Any advance is good, but they fail to mention many details.
CPV usually needs aggressive cooling to allow the Concentrate part to work, and Tj is not mentioned.
I think they need to extract ~23 watts of head, from something the area of a pencil-rubber.
At 1000x, that's also very focused, not so tolerant of cloud.
Mirrors + Tracking + Cooling, likely dominate the costs, so the 41 or 43% detail is moot.
Since cooling is costly anyway, they could be smarter to look at adding Solar-Thermal into this mix ?
concentration factor vs normal sun illumination at earths surface, as in a parabolic mirror of 100 cm^3 focused on 1 cm^3 would be 100 suns. Only question is, what is 1 sun... avg over entire earth for a year?
Blog Doing Math in FPGAs Tom Burke 24 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...