120mm fan mount for certain soft lights

Motivational note:
I have a ‘SAD lamp’, that I use as an always on desk lamp, based on a Ø18" softlight. (Chroma Sky Portal 480+810)

Since LEDs are only ~50% efficient, the panel generates waste heat; by my rough radiometric measurements, in the stock configuration, about 17W of is dissipated as thermal radiation from the front of the panel. By adding a fan to the back, this 17W can be reduced to to 11W.

By removing this 6W by convection rather than radiation, less thermal radiation hits you, and the apparent room temperature is reduced by ~0.2°C (for the head and upper torso only, perhaps half that, 0.1°C, for the whole body)*.

With my fan, the power consumption at the wall increases ~2.1W: 1.9W to spin the fan and 0.2W because the LEDs are now about 4°C cooler, so have a slightly (10–13mV/junction) higher voltage drop, consume more power, and are brighter. (They are also more efficient, especially the NIR LEDs; this is worth perhaps 0.15W)

0.1–0.05 °C/W compares favorably with other ways of cooling, like air conditioning, especially given the poorly optimized fan selection.

*at 24" distance, onto the head and upper torso.
Also note that the exact temperature depending on a number of assumptions, especially room air velocity… apparent room air temp, operative temp, globe temp… whatever you call it, its a bit hard to predict in the general case. I used 50 fpm in my calculations, which is a nice breeze typical of ceiling fans on low; at lower indoor air velocities, the effective change in air temperature is higher.

Note on fan selection:
I recommend a 24V fan, specified to provide like 40–100 CFM at 24V. Cooling performance will be similar anywhere within that range. 

Operating on the ‘reduced’ 15.5V available on the output pins, the speed, noise, and power consumption will be reduced; EG, a Noctua NF-F12-24V ought spin ~1300 RPM, draw 0.4–0.8W, and move 40 CFM.

(Pictured on my build is a BeQuiet Pure Wings 2 BL046 with an LM317 based voltage regulator to reduce the power supply's voltage to one safe for a 12V fan, and adjustable too. This is nice for testing but otherwise dumb; the regulator adds unneeded complexity and is an inefficient way to reduce a voltage, and the fan is both somewhat more powerful than really required (backside temperatures at 12V are reduced from about 7° to only 1°C over ambient) and not very efficient.)

Note on printing:
Posts are drawn very high (80mm height), as a guess at the optimal height for the fan given a fan-diameter:cooled-area ratio larger than typical in the literature.

However, diminishing returns sets in very early here (see EG 10.1115/1.3109244), so if space is a concern for you, simply stop the print when the height seems reasonable… I stopped at 40mm from the build plate, but paper referenced above, and others, shows good performance even lower (to at least 30mm, maybe lower). As a bonus: if stopping early, the infill will be exposed at the end of the posts, making inserting screws easier.

Print with support and with the posts pointing up. Posts and post-wings are kindof thin, so walls, floors, and roofs should be pretty heavy, like maybe 2.5/2/2mm, with infill at 35% or so.

Tested with a 0.6mm nozzle and 0.2mm layers; can probably be done coarser.

Wires are added in the manner documented here:
https://www.printables.com/model/486005-generic-np-f-battery-dummy