I've seen a good number of opinions on having fans on intake vs. exhaust for a radiator, but not much actual data. I have a Phobya 200mm radiator blowing intake in my Corsair 650D case with a AP-181 Silverstone in pull. In the top of the case is an XSPC EX240 with two GT 15's mounted beneath blowing exhaust in push. I was debating if the setup would show improved results with those AP-15's reversed to blow intake in pull. I ordered a DEMCIiflex filter to go on the top of the case. I didn't want to start sucking unfiltered air inside that rad.
This filter is pretty slick because it's magnetic, and looks pretty nice there, maybe better than the stock look of the case. As long as I was changing things a bit, I also got a Phobya 120mm red LED PWM fan to replace the stock Corsair rear exhaust (it was the last stock fan in the case). I have to admit I did this mostly for bling to match in with the red theme, but also to see if it was any quieter.
Here's what I wanted to find out:
1. I had tested the pressure with a simple smoke try at case openings, and the original setup was definitely giving me a negative pressure, so dirt would start accumulating. To what degree would reversing the top fans cure this?
2. What impact would having intake airflow on both radiators have on actual cooling performance?
The answer to the first question is simple, intake on both gives me the expected positive pressure, so it should keep my internals cleaner over time.
The second question is not quite as easy to answer. First let me show you the comparative numbers at 3.8ghz, 4.8ghz, 4.9ghz and 5.0ghz.
The tests were run with all fans and the pump churning full speed. The first thing that surprised me was how for the most part, temperature deltas between the two fan mountings were small or non-existent. The idle deltas with intake when overclocked were 1C lower for all three multipliers. On optimized default stock speed the default was actually 2.5 degrees higher on intake. On load (prime95 blend) the stock clock ran one degree cooler on intake, 4.8ghz and 4.9ghz were unchanged, and 5ghz was one degree higher.
My interpretation of the results is that:
It's probably invalid to generalize the data to other setups with different cases, fans, radiators etc. For those using the same or very similar setups, intake or exhaust on that top radiator will make minimal difference in cooling performance. I'd speculate that in this case there is some offset from the theoretical gain of passing colder outside air through the radiator rather than case air. By having positive pressure in the case there is probably some resistance created that decreases what might have been a gain. Also, an additional air filter adds resistance. The net is pretty much a tie as to either method.
So do I think it was a waste of time and money to add the filter and go intake? No. I'm actually OK with minimal temperature change while gaining a cleaner positive pressure case. The filter mesh actually "sucks down" to contact the case mesh holes above the radiator.
Btw, I do think the Phobya fan is quieter, and it certainly looks pretty cool. I went purely for performance on the radiator fans but that is fine, since they really don't show anyway in a 650D case window. With the red LED Phobya and the red LED's I put on the Raystorm block it's starting to look pretty wicked. I am expecting two more red case lighting toys today to finish off the effect.