For the past couple of weeks, I've been conducting some cooling experiments, just to see if some ideas work better than others.
One of these ideas is using two 120mm x 25mm fans on a Thermalright Ultra 120 Extreme in a push-pull arrangement.
I experimented with various pairs of fans with varying CFM ratings. Moreover, I also experimented with differing fan speeds, with one fan on constant speed (basically full speed, plugged into a motherboard fan header) and its partner hooked up to a fan controller, to both fans in the pair controlled by the fan controller on one channel (spin-ups and spin-downs synchronized).
My informal testing revealed that no matter which fans were paired, and no matter how much CFMs were moving through the fins, the temperature readings were more dependent on the ambient temperatures than anything else. The hotter the ambients were, the higher the core temperatures.
There was one obvious effect and advantage to a push-pull setup, though: Cool downs were substantially accelerated. Within three seconds of terminating a full-load stress test program, the temperature readings settled down onto "idle" conditions.
Here are the various fans and configurations tested:
Silverstone FM121 120mm x 25mm (paired) - By far, the noisiest pairing
Thermaltake Blue LED A2018 120mm x 25mm (paired) - Very effective
Thermaltake Blue LED A2018 120mm x 25mm (push) + Silenx Ixtrema 120mm x 38mm (pull) - Noisy because the Thermaltake at full-tilt is quite loud
Silenx Ixtrema 120mm x 38mm (push) + Thermaltake Blue LED A2018 120mm x 25mm (pull) - Not very effective in terms of temperatures
Thermaltake A2029 120mm x 25mm (paired) - A little whiny, but is the most effective by a shade over its Blue LED sisters
I've included some pictures, which show how to transform a single-fan Thermalright Ultra 120 Extreme to accept two fans: Zip-ties!
Hope someone finds this useful.