ATX/BTX PSU fan configurations

[c_hegge]

You mean halfway along the PSU, hidden from view??? As with fans at both ends of the PSU, that would only help with static pressure capacity.

Unless you did this...

Although I guess it is possible with two fans

[shovenose]

Sort of. even if I do 4 fans, the ones inside the unit would be there to increase pushing power because a 2000W PSU has a lot of stuff in there.

I was also thinking a USB header and a piece of software that constantly runs and monitors the PSU. If a fan fails it would give a notification, and show like a picture of which fan has failed, as well as temps throughout the unit. That way the user can decide if the PSU needs to be sent in for repair or replacement under warranty. Of course if it gets too hot, it would just shut down. "PSU Shutting off in 60 seconds due to overtemperature detected, please save your work!" Heck, the damn thing could have ESR meters built into each cap to predict failure! I'm also thinking this is going to be a thouosand dollar PSU.

[c_hegge]

That would be a server unit, obviously.

[shovenose]

Or just a PSU for either really rich gamer people that want to brag about having a 2kW power supply in their computer, or for people that run folding@home or bitcoins or something.

[LongRunner]

What I'd like to see is engineers at the fan manufacturers testing the common configurations against each other, fan grilles, blocked-off sections, airflow diversion and all. Another thing not mentioned here before is that many of the big-fan units return a bit of the air through vents in the front panel, to prevent hot spots from forming - another of the deficiencies of the design.

If there's one way to convince people, that would be it.

[c_hegge]

I ran a little experiment with a Thermal Master TM-420-PMSR PSU (which comes with a 120mm fan). I loaded it up to 300W on the tester, with a thermistor from my Aerocool X-vision in each of the two heatsinks. I then used a molex-fan adapter to run the fan at 12V to prevent the fan controller from interfering in any way.

I got 49*C on the Primary Heatsink and 62*C on the secondary.

I then removed the PSU from its case and put it into that of a Bestec ATX-300-12Z (which uses an 80mm fan). The temps were now 47*C on the Primary and 68*C on the secondary heatsink (which was a bit off to the side of the fan).

I repeated the experiment one final time with the PSU in an Antec SP-350's case (with two 80mm fans in push-pull). The temperatures were then 52*C and 62*C - the hottest for the Primary side, but coolest for the secondary side. In this case (pun intended), the primary heatsink was directly in front of the intake fan's hub, but the secondary heatsink was getting some airflow from it.

The conclusion I drew from this is that 80mm fans can be slightly more effective than 120mm fans, but only if the heatsinks are not placed behind the fan's hub. What would probably change the game, though is if the heatsinks were taller. After all, we do now have some extra height to play with. having all of the fins at the same height as hubs on the 80mm fans probably doesn't help.

[LongRunner]

It seems that any feasible fan configuration will leave a significant "dead space" where there is little airflow. I wonder, though, if it would be possible to use a duct with baffles to try and spread the airflow more evenly. Of course, that would add to the cost (and probably the length of the PSU casing), and result in more static pressure for the fan to deal with, so it's not an easy choice to make.

An advantage of the front fan may be that the airflow exiting the fan is turbulent, and thus much more effective at heat exchange than the laminar airflow entering the fan. On the other hand, there's the tonal noise thing I already mentioned, but then there's also the whooshing that occurs in the confines of the PSU interior regardless of the fan configuration. Only testing will tell for sure. One difficulty may be quantifying the relative annoyance level of tonal vs. broadband noise.