ATX/BTX PSU fan configurations

[LongRunner]

Most people tend to prefer large top-mounted fans but it's not as simple as "bigger fan = better". That's mainly because larger fans can't fit on the rear panel of an ATX/BTX power supply and compromises have to be made.

(When referring to the fan orientation, "in front" means the side with the "cutting" edge of the blades, that the fan sucks air from, and "behind" refers to the opposite side, which usually holds the stator and label.)

One of the compromises is obvious - a top fan takes up space inside the PSU which could otherwise be used for bigger heatsinks. The idea of course is that the better noise/airflow ratio of the bigger fan will make up for that - but the question is, is the bigger fan actually better in practice???

To start with, obstructing the front of the fan increases noise far more than the same obstruction behind the fan. I tested this myself with an (80mm, 2700RPM) NMB 3110SB-04W-B40. I didn't notice an increase in noise putting a wire grille behind the (otherwise in free-air) fan (though it no doubt reduces airflow) but putting the same grille in front of the same fan made it much louder.

So the PSU fan conundrum is much less clear-cut, noise-wise, than the size of fans elsewhere in the PC. My take on the basic configurations is this:

80mm rear fan
+ fewer obstructions in front of the fan (generally)
+ lower fan cost
+ allows bigger heatsinks
- lower flow capacity

80mm top fan (I'm sure there were a few PSUs with this)
The worst of all, with no benefits and all drawbacks.

92mm top fan (well, I haven't seen it recently but it was semi-common on some old PSUs)
Better than 80mm top fan but I still think it would be worse than an 80mm rear fan. I provided a poll option in case you prefer it anyway.

120mm top fan
+ higher airflow
+ marketing advantage
- grille placed against the front of the fan
- takes up space above the board
- part of the fan may have to be blocked off

140mm top fan
Even higher airflow, but retains the disadvantages of the 120mm fan and is the most expensive.

[c_hegge]

Neither. 92mm front fan, like on my DPS-750CB

On a PSU that actually fits in most cases, I like a push-pull 80mm fan configuration (like on the Antec SP series), failing that, either a front or rear mount 80mm fan. No preference between the two. The actual airflow that is available to the PSU's internals isn't much different between an 80mm fan and a 120mm, because much of it either has to be blocked off, or just goes straight out the back, as well as the fact that re-directing the air 90 degrees creates additional resistance and turbulence inside the PSU.

[LongRunner]

Push-pull fans would only help with static pressure capacity, but that may be a benefit all by itself in a crowded PSU.

The True330 from the PC I also got that ST3120026A from has a 92mm top fan and an 80mm rear fan, in case you're curious.

I forgot that PSUs with large top fans often have part of the back of the fan blocked off. I've edited that into my previous post.

Most current PSUs would be fine with an 80mm fan but they won't sell that way because of how caught-up the public is in the "bigger fan = better" mindset.

I think you missed the part about obstructions in front of a fan making it noisier, so a front fan wouldn't be too great in that regard either.

[Wester547]

I prefer the old 80mm rear fan approach. It's ultimately easier to direct airflow that way, and the capacitors will be in closer proximity to the PSU fan. The Delta PSUs that I see last ages with Ltec capacitors? Those have 80mm fans (often Sunon or Delta) and they blow out a ton of air even being temperature controlled. The forest of wires that tend to cover the capacitors are much more trouble in bottom-mounted 120mm, 135mm, and 140mm fan PSUs since the fan will have a lot more trouble piercing through that. The advantage of PSUs that employ 120mm+ PSU fans, though, is that those fans can spin at lower RPMs to achieve the same amount of airflow and noise levels. The 92mm front or rear fan sounds like the best approach to me as well - significantly more airflow without other disadvantages. The push-pull 80mm configurations.... if a power supply needs two 80mm fans to keep it cool, it must not be efficient to begin with, like those Thermaltake 430Ws (HEC builds?) that had two 80mm fans in that very configuration.

I agree with c_hegge's assertion. I think the real advantage of bottom-mounted 120mm-140mm PSU fans is the airflow it provides for the computer as a whole (that is in unison with the case fan[s] and CPU fan, graphics card fan, etc), not the power supply.

[LongRunner]

I don't think relying on the PSU fan alone to ventilate today's power-hungry PCs is a good idea anyway.

It's not only OEMs that designed their own PSU form factors. Antec made one for enthusiasts (CPX) with a 120mm (!) front fan.

I think dual-fan PSUs were marketing more than anything else. Super Flower even had one with three (!) fans (all 80mm, one each on the front, top, and rear, the front and top ones being thinner than usual).

[c_hegge]

^
They also had one with 4 fans. Two 80mm (Front & Back) and two 60mm fans on top.

[LongRunner]

I compared a Te Bao Metallic Plastic (a crappy brand, I know, but most physically similar fans have similar airflow/noise characteristics) 120mm fan running at ≈1700RPM with the 80mm NMB I already mentioned. Both fans have 7 blades.

Half of the back of the TB fan was already covered by a (screw-mounted) plastic sheet. I mounted the original wire grille on the front of that fan and the same grille I used in my previous experiment to the back of the NMB.

They seem to move a similar amount of air. The NMB is of course higher-pitched but the TB fan is, in this configuration, much louder.

I think that about says it. IMHO, mounting any grille to the front of a fan is heading for acoustic disaster. I'll wager that the H750P-00 reviewed here would be even quieter if the fan was rear-mounted.

A CPX unit with a rear-mounted fan would likely be awesome, but out of the poll options, I'm going with the first.

I expect that if people are better informed about this, 80mm rear-mounted PSU fans will return to status.

(In case it wasn't clear, this is about cooling/noise, not just airflow. You can always just use higher-speed fans if noise isn't the issue.)

[LongRunner]

Of course, I eventually had to compare 80mm rear vs. 92mm top. The 92mm fan doesn't have a plastic sheet blocking off an entire half of it, but still has the whole grille thing. Like before, I used wire grilles, and for fairness, mounted one to the back of the 80mm fan.

This time around, it's the same NMB compared to a "Maxima" CH90S12M (≈1900RPM?) from a Seventeam ST-250GK (it's the only fan in that PSU). The NMB fan moves more air with less noise (if I actually matched the airflow, I'd almost certainly get a much bigger noise difference). Interestingly, with no grille, the Maxima fan is actually the quieter of the two.

I haven't yet done 80mm rear vs. 140mm top, but I have a bad feeling that it still wouldn't break even.

Now all we need is to convince the major PSU manufacturers to bring back the old-style fans...

[shovenose]

If i were to start a PSU company I would use one, two, or three (!) 80mm fans in many of the ATX PSUs.
The larger units like 1500W would have three 80mm fans. One on the back, one on the front, and one in the middle to keep the air moving through the heatsinks and to ensure that it doesn't overheat should one fan fail.

[LongRunner]

...one in the middle...

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.

...and to ensure that it doesn't overheat should one fan fail.

Aside from the common 2-wire and tachometer fans, there's a type of fan with a "failure warning" output. See here for more info but in short, the fan grounds it when running normally, and floats it if stopped. That would be perfect for the application. You could also install an LED on the back of the PSU, which can flash different codes under failure conditions.* (Some PSUs e.g. Hipro HP-D3057F3H already have an LED but it doesn't have as much functionality.)

*i.e.:
A code for fan failure (as mentioned)
Another for overloading
Another for short circuit
Etc. for more serious PSU failures (of course, the most serious ones will prevent the LED lighting at all...)