Sound: It varies. The fan specs on the web are often questionable (deceitful) and the rated sound (even if accurate) is only free-air. Most of the noise comes from mounting so for any particular, specific fan implementation there will be a different threshold of what user perceives as too loud.
Do not focus too much on db, for a "sound freak" it is necessary to use a fan controller or other speed reduction method so you aren't really choosing based on that. If for some reason you aren't as concerned about noise level as lowering cost or space by not using a controller, I suggest using an inline resistor to limit voltage, and/or choosing lower RPM.
You need choose an RPM that produces acceptible flow rate. Many will claim they can undervolt and their "temps" dont' go down much but they're not really measuring important temps, only things like CPU or so-called case temp, which is actually (usually) a different chip. That doesn't tell you the temp of most of the comopnents on a HDD PCB, or the capacitors, or in the power supply, all over the video card except maybe the GPU, etc.
Point is, there are too many variables to break it down into a simple "buy this" answer, BUT I'll still try to provide as much help as possible with the following suggestion:
Carefully choose your case with the idea that it needs to accomodate as many fans as possible. That doesn't mean you "Must" put a ton of fans in, only that you have this fan-expansion option.
Plan on running all fans (except maybe video card, PSU exhaust, and high-speed CPU with mid-grade or lower 'sink) under 2100 RPM, even better under 1700 but it gets harder to do it with a packed system.
The kye is to run fans as slowly as possible because this gives the best noise/flow ratio, until you get down to a point where they're beginning to stall and at that point the flow ratio to noise is worse again. This is typically below 6.5V for most fans.
If X # of fans in this configuration aren't keeping parts cool enough, assess the options for adding fans, whether you can, whether their contribution will be of benefit to the hotter parts (varies per case and parts inside case), and how this addition effects case pressurization to the extent that it does alter airflow. For example, on a case with only a rear PSU and lower fan exhaust and passive intake, it can be a bad idea to add only another fan to the side panel as an intake because it reduces flow through the drive racks.
When you find it isn't feasible to add more fans and parts are still too hot, THEN you need to look at increasing fan speed some, above the minimal threshold for stal mentioned previously.
So if you have a large case and thus options for more fans, you might start out with a low-speed, RPM a little under 2500 @ 12V. If you have a smaller case, you might shoot for a more median-speed fan. The idea is not that this is the resulting speed of the fan, but rather it determines the range of RPM adjustment provided by a fan speed controller, that there is no way you're going to need 2500 RPM fans in a large (decently designed) case that allows more fans, but in a small case where you can't even fit a 120mm, maybe not even 92mm, you can expect the speed to be a little higher and there is no need to be able to use a controller to turn those 80mm fans down to 1200 RPM because they're simply not going to be sufficient at that speed.
Now about brands. Much commotion is made about brand X fan spec and db:noise ratios. On decent quality fans, the flow:noise ratio is very close to the same, so that is not really a parameter that needs as much consideration as fan construction. Rigid fan frames, like the aluminum ones or clear ones, and a few fiberglass, have the potential to transmit more vibration to the case and thus are best avoided for lowest noise, particularly on medium or low quality fans which may have lesser balance.
PSU exhaust fan is ideally dual ball-bearing, far more reliable for that location. Other case fans will produce less objectionable, lower-pitch noise if sleeve bearing but choose a quality brand to get a bearing that doesn't wear out so fast, like Panaflo, Papst, some Sunons and some Deltas. Lesser quality new fans' noise ratings, if/when accurate, might degrade more over time, and/or require more frequent relubrication.
The quick and short answer is: buy fans with a fan manufacturer's name on them. For example, Coolermaster is not a fan manufacturer, nor is Thermaltake, but occasionally Coolermaster sells decent Deltas or Thermaltake Panaflo, but not always. In a smaller case with smaller fans expect to run them at slightly over 2000 RPM whether it be by default fan spec or with a controller reducing the default spec. WIth larger case, aim for sub-1700 RPM (default or after controller reduction. These are not hard rules,
I haven't even mentioned ambient temp or thermal rise, you have to take temps and see what the result is, and often a case needs some carving done to it for optimal flow, or sub-topics like filtering (which requires fans with higher pressure potential (typically thicker).
To some extent you should also, always choose the thickest fan possible for any use so long as it's in a same/similar fan line so you don't have a higher RPm necessary for smooth operation (for example, some Sunon 120mm x 38mm fans cannot run quietly because below a voltage where the turbulence is heard, the fan already begins ticking loudly from stalling.
Shortest answer possible: Buy a panaflo "H" speed in every size you might ever need up to 92mm and "M" in 120mm, as spares. When you're building the system us a fan controller to find out how fast it can run before it gets too loud, THEN buy a fan with that parameter at 5V, 12V, or a common resistor value like 47, 68, 100 Ohms. Get name-brand sleeve-bearing fans that match this.
Ultimately, systems keep getting hotter but people keep getting more and more critical of noise. SOmething's gotta give. That something,is usually carving up a case for more fans or intake/exhaust area than was intended by manufacturer of case. On the other hand, it is up to the user to decide how long the box needs to run. If parts failure within 4 years is ok, temps are not as important as a box that might run for a decade or more. Personally I always shoot for a decade, since even long after I'm done with any given parts combo, now more than ever someone else would find it more than sufficient for their office/email/websurfing/etc.
Facebook
Twitter