You’ve given me more to think about — in particular on the use of different fan design.
The highest temps, whilst running on mains, I’m generally able to get the CPU; GPU Heat Pipe(1); and Skin Proximity is whenever I’m compiling or running macports upgrades. In a room averaged at 24–25°C, the highest figures I tend to see, respectively, are 67°C; 71°C; and 41°C. Sometimes, I can get the former two to peak, briefly, at 68°C and 72°C, but I have never been able to get the Skin Proximity (the sensor touching the underside of the bottom case) to exceed 42°C. This is good because it’s at higher temperatures when extended skin exposure can result in second-degree burns.
I have noticed, unsurprisingly, how all of these peak sensor temps are affected for each degree of change in room temperature (i.e., a 2°C-cooler room yields sustained peaks of 65–66°; 69–70°C; and 39°C, respectively).
Our differing goals for the dissipation of the heat sources (the CPU and GPU) are interesting: I approached the task to make having the bottom case not burn my bare legs whilst using it as a laptop. This meant getting the passive thermal dissipation to do more work. Your approach has been to improve on the active cooling by doing a slight re-design/augmentation of the fans.
(Sidebar: I think I have one of those unibody MacBook fans from a parted A1342. I should have another look at it.)
