Hello again. Here is a "silent PC planning" that uses much of what has been discussed in this thread. I'd like to read your opinion on it.
An easy, yet expensive way to reduce noise even further is to buy a fanless PSU. I am an advocate of fanless PSUs. Of course I don't have empirical proof or user testimonies to judge upon, but I still really believe that the huge price you pay for those fanless units goes into extra-good build material and components, along with good design. Guarantee periods are 2 year in general, like any other fanned PSUs, so manufacturers are confident on these products. Experts like MikeC do not like them, because they say they fail more easily, mainly because of the high temperatures they develop... as I said, I think the money you pay for them goes precisely into solving that. Since a fanles PSU provides no airflow, I think they should be removed out of the case: since they are noiseless, there is no noise concern in doing so.
So the PSU fans goes missing in exchange for a lot of money. Noise floor is now conditioned by case fans and HDs.
Has anybody already tried to remove the HDs out of the case? It can be done if you are using SATA, since data cables are quite long. There are also ATX power cable extension, if they were needed. If you take HDs out, you'll surely need this:
http://silentmaxx.net/catalog/sm_hard_d ... losure.php
This HD enclosure must be the best around: no HD fans are needed, HDs are completely enclosed, and a rather big heatsink sits on top of them. In principle, you should get 0dBA, extended service life (because of reduced HD temperatures), and strong EMI shielding, so there should be no noise concern with this, neither. Oh! This is important: SilentMaxx's enclosures cost 50? a pop.
Well, if HDs are removed, there is less heat inside the case, and that's good. Also, seek/read/write noise are also out of the (sonic) picture, and the noise floor is determined only by the fans (of course, as silent PC freaks that we are, our northbridge is passively cooled, and so is our good-but-not-too-powerful VGA card).
Now we have the CPU, chipset and VGA as main heat sources, and it seems clear that hot air from the CPU could be also removed from the case with a sucking HS, which could allow us to make away with the case fans.
Here comes our little blow vs suck discussion again. Since arguments in favor and against have come from many corners, I'll try to wrap up my small conclussion.
If we use a sucking HS+ducting with the CPU, we get even more heat out of the case, making it easy for us to kiss goodbye the case fans, but that comes at a price: as Rusty has pointed out, sucking HSs are not as efficient as blowing ones, and you can see it in the SPCR HS top rank and reviews (though the new alphas have not been tested yet). There is also the fact that you're using hotter air to cool the CPU. Finally, if you take into account that ducting reduces a fan's airflow (and we won't be using any further fan in the duct exhaust), you can conclude that, in order to achieve equivalent cooling, you need higher rpms sucking than you do when blowing. Thus, the noise advantage of removing case fans could simply be not worth the effort.
On the contrary, ducting into the case fresh air and blowing it on the HS has allowed people around here to reduce CPU fan voltages to 7V, while keeping good temp levels in the core. This, even with the airflow restriction that a duct poses.
Well: the only solution I see, that combines the possiblity to use outside air to cool the CPU, while more or less keeping the ability to exhaust the hot air stream to the outside, is the use of a tower HS (e.g. Aerocool HT-101). So, you put a duct from one fan hole to the intake part of the HS, where a 7V Panaflo pushes air through the fins and heatpipes. The air gets through the HS, exits, and moves further towards the upper side of the case. There, it can be fastly exhausted through convection. The mobo and graphic card don't even touch it.
NOTE: in my current PC, the exhaust air of the CPU HS hits directly the northbridge HS. Of course, my northbridge temp sensor gets crazy about it. That's why I think it is important to be able to direct the hot air stream where it can be exhausted quickly.
The problems of this: I don't know how good tower HSs are when compared to the Swifttechs and Thermalrights. They have 6 heatpipes, so, damn, they should be really good at sucking heat out of the CPU. The fins and pipes are huge, so a small airflow should make a lot. And don't forget there is no dead-spot problem with this HS. In spite of all this, there is a point that worries me: tower HSs are far cheaper than the Swiftechs and Thermalrights. It could be nothing important, but price uses to be a measure of performance, which would mean that tower HSs are not that good after all.
Also there is the size problem: more than 500gr hanging perpendicularly of the mobo means the case has to be handled with care. Other HSs don't have this problem.
People, your turn. Do not be harsh. Be tender. I know that all this is just too obvious...
EDIT: we have also learnt here at SPCR, that 2 slow-spinning fans might do the work of one fast-spinning fan at lower noise levels. Could it be possible that two 5V-80mm panaflos mounted on the tower HS could make the job of a 7V-80mm at lower noise levels? And what about one single 5V-92mm panaflo instead of the 7V-80mm?
And one last thing: in the ducted tower HS + fan setup described above, do you guys think that removing the fan from the HS and screwing it as a case fan, then ducting it to the HS would do any harm to the airflow? (independently of the voltage used) I know noise would increase because of suck turbulence, but the weight supported by the mobo would decrease this way.
EDIT2: Jeeesus, I am forgetting radial cooling (Zalman 7000). Definitely a hot air spreader! We don't want that, do we?
)