just had a thought about why the heatsinks cool better than their size would suggest - the middle (main?) HS is bolted to the PSU case (the side that the cables come out of) with some TIM between the HS and the case. Perhaps by using the metal of the case as an extended HS, smaller HSs can be used? It's not something that I've seen in other PSUs reviewed here - perhaps another example of it's nerdiness?!
I confirm your suspicion ; for calculating the efficiency of this PSU, I had to use a very accurate thermometer (accuracy : 0.1°C), and the idea is simple. If your PC has only one airflow that enters in your case, refreshes your components, then enters from one entry in the PSU, and leaves the case, you can easily calculate the efficiency.
Let's say the temperature at the entrance of the case is T1, the temperature at the entry of the PSU is T2, and the exit temperature is T3. So the efficiency (I had to write a lot of lines on my forum of PCSilencieux, and it's in French, so it would be quite long to translate it...) is simply (T2-T1)/(T3-T1).
For my first attempt to measure the efficiency, I had the following temperatures :
- T1 : 23.7°C
- T2 : 41.7°C
- T3 : 35.2°C
If you make the calculus, and if it were real, you'd be a multi-trillionaire, and your name would be in the memories for the whole millenary, because I had the incredible efficiency of 156.5% !
The error was in the protocol of test ; to measure T2, I stuck the thermometer's probe on the case of the PSU, which is hotter than the air's temperature I should measure...
This type of engineering is quite interesting, since it explains how FSP could drastically reduce the PSU's weight, while not overheating it. But since the PSU's case drives a part of the heat outside the PSU, there are a few Watts which go directly in the PC's case. It's not dramatic, since with an 80%+ efficiency there won't be a lot of Watts, but I'm thinking of a way to reduce this effect.
A simple of way of improving the heat extraction is to simply use some adhesive thermal epoxy that glue additionnal and bigger heatsinks directly on the original and tiny heatsinks. Of course, the ideal would be to use copper heatsinks, and to find a shape that is a good compromise between less drag and more heat exchange surface.
If I could make copper heatsinks with the shape of blades guiding the airflow to the exit, but which microscopically look like some active coal (It's the material used in gas masks to filter the air.), I think I may cool passively an overclocked Pentium Extreme Edition in the Sahara !