I just spent yesterday trying to silence my home-built HTPC case further. I've had a semi-tonal noise (or perhaps best described as a sort of medium-to-lowfrequency pink noise) annoying me for a while, and decided to try to do something about it with padding.

And when I opened the side panel... the tonality disappeared.

My first thought was "ok, standing waves, fine, padding fixes it."

And then I held a big block of padding against the open side, and the tonality reappeared!


After experimenting back and forth, I could confirm that it was the two otherwise-silent input fans (750 rpm) that started creating this sound as soon as airflow through the box got more restricted.

I don't believe it's a case of insufficient exit surface area. I've got 2 x 1.3 x 55 cm + 2 x 1.3 x 37 clearance all around that sideplate (~240 cm2), PLUS the PSU fan sucking air out (albeit at low RPMs).

The only explanation available to me is that vortices form around the blades as the required pressure increases. And the difference in noise is big. The fans go from inaudible at 1m to somewhat-annoying at 2m.


So, this leads me to the question: How relevant is the testing done by SPCR? Is SPCR's tests in fact driving fan makers to develop fans that are silent in free air flow, rather than with some pressure?

Perhaps you should design a new fan test platform? Such as a fan blowing into something PC case size, constructed out of sound absorbing foam to not add resonance, with some strategically placed air exhausts. (NOT one big hole directly in the air path!)

What a thread title !!
No, it's not flawed. It tests only one scenario, is of high quality, and results can be compared.

Have you tried your setup with only one fan ? The tonality could be the result of both fans' noise signature adding up. Try with fans at different speeds.
Are your fans decoupled from the case ?
Also, do your fans have enough breathing room close to them ? When I used 92mm Noctuas, they had a noise to them when the grill was too close to them. If I move the grill away, it would disappear.

Which fans are you using ? Maybe they are not the best choice for your setup (higher static pressure ?).

The title was quite serious. If it leads to the development of fans that are only silent in free air, there is a serious problem with it!



Tried only one fan. Tried different speeds. They are completely decoupled. LOADS of breathing room (no grill at all!). Using a pair of S12Bs. Higher static pressure model sounds wrong for PC case fans?

In all my test scenarios, noise went up substantially when I held my foam block against the side.

How's the airflow affected ?
I would assume that such a change in the noise pattern would cause some airflow change...
What's your case btw ?

My case is a rectangular wooden block 55 x 37 x 17 internally, lying down. Two fans in one of the 55cm sides. The top is raised 1.3cm.

Again: No top, no noise. - Top, noise.

What if you raise the top 5 cm just as a test ?
1.3 cm is not very much space, and I wouldn't be surprised if you got a lot of turbulence (might cause the noise).

I have experimented with the top at different angles and heights just as you suggest (ruling out standing waves) and just as you say the noise recedes with bigger space.

Again: free exit area is 240cm2 (more than 2*6^2*3.14) AND there is a PSU fan sucking out too.

But the sound is definitely coming from the direction of the fan blades - which makes sense - where is the bigger air speed differential? Fan blades or somewhere else? =)


I still believe that resistance testing of fans is an untried idea, and that it merits testing. There might be surprises in store!

There might be... Next comes the question of how ?
Just one thing to keep in mind : the only way to get a super quiet computer is to have as little restriction on the airflow as possible (in and out of the fans), because that allows you to run the fans as slowly as possible and as efficiently as possible (best noise/cooling ratio).
The CPU heatsink test that the reviewers here run is designed to address some of your concerns about noise in a more restrictive environment.

There are many threads that address both those topics all over the forum. They make for a very interesting read.

Quite so. But what then is a sane ratio? My input area is what.. 210 cm2? My free exit area is 240 cm2. And on top of that I have the PSU fan assisting with sucking air out (with its own standard size exit on top of those 240 cm2).



Mmmm yes and no. Those are typically high static pressure fans - not at all what we shop for case ventilation. And they're mounted straight up against heatsinks which create their own turbulence. That's not very useful in determining what airflow fans to shop.

If we suspect the noise is coming from turbulence, you should check the airflow and all obstacles inside of the case. Avoiding dead corners, harsh angels and such.

Pictures might be a great help.

To your testing question: You have a non-standard case. So you would need non-standard tests. We should find the cause of your noise problem first. If we know, what caused this problem, we might be able to relate it to the testing method. As long as we're guessing spcr fan testing is totally fine.

I see where your reasoning is coming from. But at the same time it strikes me as easy for someone with access to an anechoic chamber and SPL measurement equipment to take the existing fan testing platform and make it blow its air into a taped-together box of padding foam with a handful different fans and see if there is a difference between different fan makes. In my mind it would be a highly interesting experiment!

If different fans behave mostly the same, then my hypothesis was wrong, and we are a bit richer in knowledge. If they are shown to behave differently? Well then maybe it's time to invest more than some packing tape and a few hours in the problem (like for instance trying to figure out what a good standard platform would be, which is admittedly hard).

The total surface area that you have available for exhaust does matter, but the shape of the area matters a lot too (and maybe more) !
Check this out if you haven't already : http://www.pugetsystems.com/labs/articl ... -Noise-107

I should perhaps mention that one of my tests was to completely cover about two thirds of the side and leave about one third open. Exit diameter easily in excess of 500 cm2 but not in line with the airflow.

I do not see a connection between the shape of my exhaust (not surprising to me, 1.3cm wide is quite a bit!) and the noise level.

Your case has a noise problem. So it might be related to your exhaust shapes or the like.

http://www.diracdelta.co.uk/science/sou ... ource.html

Again, as long as we don't find the reason for the noise in your non standard setup, we won't even bother thinking about different fan testing methods.

Seems like you build an instrument that resonates with the air-stream. Have you tested what happens when you add more weight to the case?

1cm solid wood is plenty heavy. And still - playing by ear - if I stop one fan, the sound seems to stop coming from that direction (but keeps coming from the other fan).

Maybe I should add a picture or two. (Yes, the top is actually the table. As I said, 1.3cm gap.)

Both fans are now suspended on standard silicone plugs. (In the picture only one is)

During my noise tests, the fan filters are removed.

^^^
Sweet desk-bottom computer!

help us all understand the airflow in this case. I see what i think are 3 exhaust fans (including psu), but i dont really understand how the intake works...if its just a small space all around the to edge where theres a gap between the table's top, then i think your problem is an "air short". most of the air is coming in right next to the exhaust fans and looping back around, creating a vortex.

that's my best guess based on the pics, but i might not yet understand the build fully.

The 2 120mm fans are intakes (running around 750 RPM), blowing nice cold air at the CPU and passively cooled graphics card respectively.

The PSU is outtake, and obviously doesn't suffice on its own. Hence the 1.3cm gap all around for a surface area of circa 240 cm2. (Fans together are ~215 cm2 before subtracting the hub area)

Like I've previously explained, I've tried covering the top in different ways and different angles (especially making sure to cover the side where the fans are to avoid shorts) with a block of padding, and as soon as airflow gets a bit more restricted, the intake fans start producing a whooshy kind of sound that isn't there when running in free air.

Hence my original question about usefulness of also testing fans with a somewhat more restrictive airflow on the out side.

Sound waves are quirky and complicated. No general testing in one particular environment can substitute for trial and error testing for your particular situation.

Looking at your pictures:

-your intake holes have sharp angles

But i came up with another thought:

Decouple the top! Add some rubber between the top and the rest of the case. Make the gap 5-10 mm high, so you gain enough exhaust area between the case and the top. Judging from the pictures you did nothing to stop the hole wooden construction from transporting resonances. If your fans aren't decoupled enough, they create resonances because of their movement, not because of some airflow.

The top is the table. It's 100% decoupled. The gap is 13mm.

The fans are suspended on the silicone plugs they came with. (Originally I thought I could get away with foam padding + screws with silicone washers but THAT created humming)

There is no discernible noise while running in free air. Noise only starts happening when I start covering the top with a block of padding (which I hold in my hands, so 100% decoupled). This whooshy noise happens regardless of how I cover the top - leave gap, clamp it down over 2/3rds of the area, clamp down one side and leave gap the far side, whatever.

No but at this point of testing I'm thinking soddit & buy other fans to test with.

It's got to be the fans !! I've had the same problem with the 92 mm Noctuas I used to have : very quiet in free air but with an annoying noise as soon as the door of my Solo was closed !
I'm using a 120mm Nexus now and it's wonderful, highly recommended !

If what you are saying is that you are thinking about just trying some other fans. I think that is a good idea.

Take a look at the Scythe Gentle Typhoons. They, by the way, are designed especially for high static pressure operation, but do well under all circumstances. SPCR has warned about variances from fan to fan. 800 rpm Scythe slipstream fans can be inexpensive. As can the SPCR reference Nexus fans.

I doubt very much that it's the fans creating the noise.

What you're witnessing is the effect of air resonance inside the box you've created. Every closed (or even semi-closed) space has air resonances. When a noise in or near the space approaches the resonant frequencies, it is amplified.

Closing your case (even with a block of foam) increases the air resonance peak level and moves the frequency to where the fans are making some noise, which then results in the tonal noise you hear. It's what enclosures for speakers exploit to extend & increase bass, it's what causes your voice to sound deeper when you sing in the shower stall. Even cupping your hands around your mouth while talking or making a shhhhhhh sound lets you hear the effect.

You could easily test this by making an audio recording of the fans without the top on, then blocking and turning off the fans, sticking a speaker inside the box (case) and playing that recording at a level similar to the original while closing/opening the 6th side. You should hear the same effect.

If my assessment is correct...

Solution:
1. Run the fans slower (changing both frequency balance & amplitude of their sound)... and/or...
2. make the exhaust opening a LOT bigger. I'd think of the airflow through your box more as a tunnel -- don't leave that 1.3cm gap. Close that up (if not completely, then mostly but keep it mechanically decoupled from the desktop) and make a hole perhaps 20 x 12cm on the panel where the PSU is sited (perpendicular to the fans) with an internal baffle along the side of the motherboard so the air from the fans has to flow through/around the CPU cooler before it can exit. But the mobo/fans area might still be too closed, leaving the air resonance effect, so a safer bet is to put the exhaust vent on the panel directly opposite the fans.

interesting...
MikeC, if I understand you correctly, adding a block of foam in the free spaces in the case might help, right ?

Yes, adding a block of foam in the free spaces in the case might help -- if its really acoustically absorbent. But opening the case up so it's not as enclosed will help more for sure because this eliminates the air resonance altogether (or brings it down to such a low frequency that it become inaudible/irrelevant). Generally, the smaller the volume of the enclosed space, the higher the frequency of the air resonance; larger volume means lower frequency. Non-parallel walls help spread the resonances so it's not as sharp and high in amplitude but doesn't eliminate it all together.

Mm, unfortunately the side directly opposite the fans is pretty much off-limits as far as major airflow goes. My projector sits there. I don't want to blow hotter air on it, and I don't want to suck in its exhaust either. (The PSU is far enough back that it blows air out way behind the projector - in the same area that the projector blows its hot air out so I figured I was fine there) Yes I realize that I'm already blowing some air in the direction of the projector with the 13mm gap up top, but that's as far as I'm willing to take it.

The fans are plenty slowed already, 750 RPM on one (Noctuas biggest resistor which claims to send it to 600 RPM but that's a lie) and now 800something on the other (fanmate). They were both 750 originally but running them out of sync reduced the noise a tiny bit (unless I'm just dreaming that, no SPL meter unfortunately - maybe it just changed the spectrum to be less annoying). Perhaps I should experiment more with slowing the 800 one to sub 700 instead, hrmph, bad me.

I still figure the next big step for me is buying new fans (if nothing else they'll have a different noise signature). If that doesn't work I might try a big wedge-shaped foam block on the side opposite the fans.


On a related note I'm really happy with the cooling situation in the box. The passively cooled graphics card sits at just over 50C while playing movies. The processor (old 84W TDP, no cstates or anything) idles at 37C with the old Zalman cooler fan running slower than the mobo can measure (900ish RPM guesstimate). The suspended hard drive is at 33C in its enclosure, clearly benefiting from the distributed airflow. I guess this means I could actually run the fans slower than I am, but the cooler fanatic in me warring with the silence fanatic in me!


I will keep on wishing I had an anechoic chamber so I could do a quick test of my hypothesis of different fans reacting differently to enclosed spaces. The hobby scientist in me wants to know! *snief*

It's an interesting theory but not one I would support after a dozen years of close experimentation with fans. On the other hand, I've given you all the info to test my hypothesis that it's the effect of air resonance in the enclosed space, not "pressure" on the fans. Let the scientist in you fly with that!

It may indeed be the case that you are 100% right and I'm 100% wrong about the root cause for the noise increase (I'll freely hand it to you - the odds are in your favor!). I would still be interested in seeing the difference between different fans - seeing as how they all have different free-air noise characteristics and so forth! Can fan A go from 12db to 16db in the same situation that fan B goes from 12db to 18db? Yes? No?

Speculation: A fan with a pronounced hump in the spectrum could behave differently from a fan with a flatter spectrum?