Storage server HDD noise attenuation with scientific principles

[sxr7171]

I posted this over at SilentPC Review but it seems most of them are fumbling around with "brute force techniques":

I see a lot of posts here about 1 or 2 drives and most of the posts are oriented towards hushing seeks.

The requirements for home server hushing are completely different.

Firstly the number of drives involved tends to be 4-24 drives. So there the approaches need to be different for this sort of endeavor.

Secondly a WHS or other NAS software does not really cause drive seeks so much and the main source of sound is simply having 4-24 drives spinning. It sounds like a bunch of fans more than anything.

It's also subjectively not a terrible sound, but nonetheless I do not want to hear it.

Currently I have 8 drives sitting in an Antec 300 case and the whole thing placed inside a kitchen cabinet. The case fans are on low and are silent from within the cabinet.

But the HDDs themselves are audible since apparently the design of my kitchen cabinet is not completely sealed. That leaks a little bit of sound which I can hear. I could seal it up, but I'm looking for a solution that acts on the drives themselves like a hush box because I am planning on making the 8 drives -> 16 drives.

Does anyone here have experience with this sort of thing. My thoughts currently are to use a wood or MDF box heavily lined with acoustic foam or neoprene. I guess a scientific way to do this is to analyze the frequencies emitted by these spinning platters and finding the correct foam density and thickness to properly attenuate that frequency band. I don't have the measuring equipment to do this so I'm wondering someone has such data already which would be much appreciated.

Inside such a hush box we could devise an elastic suspension strategy for the drives and cooling strategies. Now 40c is perfectly acceptable for HDDs per the 2007 Google HDD study (Google it if you are not familiar with it).

I think the fan openings are the points most susceptible to sound emission so maybe some kind of flow path that is foam lined would be good perhaps inspired by the design seen in corner horn and transmission line speakers. If we go even deeper into it we could even come up with a design that provides acoustic bandpass attenuation as seen in bandpass subwoofer modules. Those are designed to pass only certain frequencies like for example 50-120Hz and the rest are attenuated by 36dB per octave.

I would like to apply more acoustic science towards this goal than my limited knowledge of sound attenuation would allow. Towards that end I am seeking feedback from members of this forum who have knowledge in these areas. As a group it would be good if we devised simple implementations that go beyond simple "brute force" sound attenuation by acoustic foam. We must target the offending frequencies in a scientific way. I also think this could end up as a project that could be an article for by the SilentPC review principals that takes PC sound attenuation to the next level. We could devise hush boxes for entire PCs based on the frequency spectra generated by fans placed on heatsinks.

I know this is a tall order but if any place has the people with the expertise I think it would be here where silencing a PC has been an art form for many years. The typical constraints of pre-manufactured PC cases do not apply to us and we can definitely taking quiet computing to the next level. We can succeed where the case manufacturers have failed us.

We could start with this article:

http://www.sciencedirect.com/scienc...serid=10&md5=5686c47c4724a5b59ac32317dc504d2b


If anyone is in university maybe they could get read access to that article and if anyone here is an engineer they could understand it far better than I could.


Here's another one:

http://www.samsung.com/global/busine...oiseGuard.html

Any engineers here know if we can use this second calculator to design a vented enclosure:

http://www.mh-audio.nl/Bandpass.asp


I put:

vent diameter = 12cm (for 120mm fan)

tuning frequency = 45Hz

enclosure volume = 120l or 4.24 Cu Ft.

3 vents (for 3 fans)

I got vent length = 33.3cm or about 1 foot long.

Could we use the same size front and rear chambers and place our 8 or 16 drives in the middle divider where the speaker driver would be? We could devise a wooden frame lattice from which to suspend the drives with elastic. 3 120mm quiet fans would provide sufficient cooling and those could be attached to the vents with some neoprene or maybe suspended in front of the drives.

Some information about acoustic mufflers:

http://en.wikibooks.org/wiki/Engineering_Acoustics/Filter_Design_and_Implementation



Now I suppose this is highly technical enough that really very few people understand this stuff and I'm not one of them. Another thought I have is active noise cancellation of HDD platter noise with a microphone and 3 small speakers placed right behind the drives. I couldn't find any working software to try it out.

What are your thoughts on these approaches? Could I design an acoustically sound enclosure that allows proper ventilation while filtering out frequencies related to HDD platter noises?

 

[Red Squirrel]

I find the best method for noise reduction is putting servers far from the office.

 

[sxr7171]

I find the best method for noise reduction is putting servers far from the office.

I know, but this one is kept inside a kitchen cupboard and I can still hear it a little. I am quite anal about sounds since the living room is also my listening room. I can pick up on the slightest sounds when listening to quiet passages.

I tried to move it to a closet on the other end of the apartment but the problem there is that there is only a light bulb outlet and that would be ungrounded. The only other option is the bathroom!

 

[Swivelguy2]

Okay, the abstract (too lazy to log in to my library and/or read the whole thing) says that the main source of HDD noise is from air motion and is mostly in the range 2-8kHz. In air at sea level, that's a wavelength of 0.17 - 0.04 meters.

Are you going to make your sound dampening 17 centimeters thick? I think not. In any case, this theory doesn't even matter. The bottom line is still that thicker insulation will block more noise at all frequencies than thinner insulation. That's why brute force is in style.

 

[DominionSeraph]

I find the best method is to not buy the wrong drive.
Get Caviar Greens.

Could we use the same size front and rear chambers and place our 8 or 16 drives in the middle divider where the speaker driver would be?

LOL no, that's not how a bandpass box works.

What are your thoughts on these approaches? Could I design an acoustically sound enclosure that allows proper ventilation while filtering out frequencies related to HDD platter noises?

Cardboard + duct tape.
If your drives aren't already rubber mounted or suspended, just throw a couple layers of duct tape against the sides and remount them. Works pretty good for 2 cents worth of tape.
Your case sides act as a sounding board. Duct tape a sheet of cardboard to the inside of both sides.
Your air ports allow sound to freely escape. Make ducts out of cardboard and duct tape.
Here, I MSPainted the rear duct for you.

There. 20 cents worth of materials, and your computer should be waaaaay quieter.
If you want to throw some money at this, get some Quick Roof from Home Depot. It's the poor man's Dynomat.

Oh, and talk about picking the wrong case. Does that thing have anything BUT holes?

 

[sxr7171]

Okay, the abstract (too lazy to log in to my library and/or read the whole thing) says that the main source of HDD noise is from air motion and is mostly in the range 2-8kHz. In air at sea level, that's a wavelength of 0.17 - 0.04 meters.

Are you going to make your sound dampening 17 centimeters thick? I think not. In any case, this theory doesn't even matter. The bottom line is still that thicker insulation will block more noise at all frequencies than thinner insulation. That's why brute force is in style.

No doubt I am prepared to go for brute force as well and I mean I am willing to build a cabinet and line it with either 3" thick acoustic foam or 1.5 inch thick neoprene. But there has to also be some more intelligent adjuncts to use as well like a helmholtz resonator design.

Anyway this would be my source for both types of foams and its pretty darn inexpensive considering.

http://www.foambymail.com/neoprene.html

http://www.foambymail.com/Wedge.html

 

[sxr7171]

I find the best method is to not buy the wrong drive.
Get Caviar Greens.



LOL no, that's not how a bandpass box works.



Cardboard + duct tape.
If your drives aren't already rubber mounted or suspended, just throw a couple layers of duct tape against the sides and remount them. Works pretty good for 2 cents worth of tape.
Your case sides act as a sounding board. Duct tape a sheet of cardboard to the inside of both sides.
Your air ports allow sound to freely escape. Make ducts out of cardboard and duct tape.
Here, I MSPainted the rear duct for you.

There. 20 cents worth of materials, and your computer should be waaaaay quieter.
If you want to throw some money at this, get some Quick Roof from Home Depot. It's the poor man's Dynomat.

Oh, and talk about picking the wrong case. Does that thing have anything BUT holes?

Some great inexpensive ideas there. In fact I have already stuck cardboard to the insides of the case. I know that case is more holes than steel. Poor choice for a home server but it was cheap. It was great for my SSD running overclocked gaming PC though.

I will try the ducts on both the top and bottom as well as duct the sides of the drives. Also I like the quick roof idea as well. I have nothing to lose in any case since that case is as good as disposable the way I look at it.

I was thinking of getting the Silverstone FT01, but I was wondering that since every case needs air holes sound will come out, but you answered my question with the a simple way to put ports at the inlets and outlets. I was wondering what I could do if I had to design a case myself of wood. I could build the ports into the inlet and outlet areas to get the attenuation I want. But this is a great quick fix and it may get the job done with less time and effort. But I also wonder what I will do in the future when I need 16 or 24 drives. It would be good to learn some basic theory now so I apply it later when my needs grow.

 

[canis]

I posted this over at SilentPC Review but it seems most of them are fumbling around with "brute force techniques":

I see a lot of posts here about 1 or 2 drives and most of the posts are oriented towards hushing seeks.

The requirements for home server hushing are completely different.

Firstly the number of drives involved tends to be 4-24 drives. So there the approaches need to be different for this sort of endeavor.

Secondly a WHS or other NAS software does not really cause drive seeks so much and the main source of sound is simply having 4-24 drives spinning. It sounds like a bunch of fans more than anything.

It's also subjectively not a terrible sound, but nonetheless I do not want to hear it.

Currently I have 8 drives sitting in an Antec 300 case and the whole thing placed inside a kitchen cabinet. The case fans are on low and are silent from within the cabinet.

But the HDDs themselves are audible since apparently the design of my kitchen cabinet is not completely sealed. That leaks a little bit of sound which I can hear. I could seal it up, but I'm looking for a solution that acts on the drives themselves like a hush box because I am planning on making the 8 drives -> 16 drives.

Does anyone here have experience with this sort of thing. My thoughts currently are to use a wood or MDF box heavily lined with acoustic foam or neoprene. I guess a scientific way to do this is to analyze the frequencies emitted by these spinning platters and finding the correct foam density and thickness to properly attenuate that frequency band. I don't have the measuring equipment to do this so I'm wondering someone has such data already which would be much appreciated.

Inside such a hush box we could devise an elastic suspension strategy for the drives and cooling strategies. Now 40c is perfectly acceptable for HDDs per the 2007 Google HDD study (Google it if you are not familiar with it).

I think the fan openings are the points most susceptible to sound emission so maybe some kind of flow path that is foam lined would be good perhaps inspired by the design seen in corner horn and transmission line speakers. If we go even deeper into it we could even come up with a design that provides acoustic bandpass attenuation as seen in bandpass subwoofer modules. Those are designed to pass only certain frequencies like for example 50-120Hz and the rest are attenuated by 36dB per octave.

I would like to apply more acoustic science towards this goal than my limited knowledge of sound attenuation would allow. Towards that end I am seeking feedback from members of this forum who have knowledge in these areas. As a group it would be good if we devised simple implementations that go beyond simple "brute force" sound attenuation by acoustic foam. We must target the offending frequencies in a scientific way. I also think this could end up as a project that could be an article for by the SilentPC review principals that takes PC sound attenuation to the next level. We could devise hush boxes for entire PCs based on the frequency spectra generated by fans placed on heatsinks.

I know this is a tall order but if any place has the people with the expertise I think it would be here where silencing a PC has been an art form for many years. The typical constraints of pre-manufactured PC cases do not apply to us and we can definitely taking quiet computing to the next level. We can succeed where the case manufacturers have failed us.

We could start with this article:

http://www.sciencedirect.com/scienc...serid=10&md5=5686c47c4724a5b59ac32317dc504d2b


If anyone is in university maybe they could get read access to that article and if anyone here is an engineer they could understand it far better than I could.


Here's another one:

http://www.samsung.com/global/busine...oiseGuard.html

Any engineers here know if we can use this second calculator to design a vented enclosure:

http://www.mh-audio.nl/Bandpass.asp


I put:

vent diameter = 12cm (for 120mm fan)

tuning frequency = 45Hz

enclosure volume = 120l or 4.24 Cu Ft.

3 vents (for 3 fans)

I got vent length = 33.3cm or about 1 foot long.

Could we use the same size front and rear chambers and place our 8 or 16 drives in the middle divider where the speaker driver would be? We could devise a wooden frame lattice from which to suspend the drives with elastic. 3 120mm quiet fans would provide sufficient cooling and those could be attached to the vents with some neoprene or maybe suspended in front of the drives.

Some information about acoustic mufflers:

http://en.wikibooks.org/wiki/Engineering_Acoustics/Filter_Design_and_Implementation



Now I suppose this is highly technical enough that really very few people understand this stuff and I'm not one of them. Another thought I have is active noise cancellation of HDD platter noise with a microphone and 3 small speakers placed right behind the drives. I couldn't find any working software to try it out.

What are your thoughts on these approaches? Could I design an acoustically sound enclosure that allows proper ventilation while filtering out frequencies related to HDD platter noises?

Hello. What eight hard drives are you using? Do you live in a studio apartment?

 

[Red Squirrel]

I know, but this one is kept inside a kitchen cupboard and I can still hear it a little. I am quite anal about sounds since the living room is also my listening room. I can pick up on the slightest sounds when listening to quiet passages.

I tried to move it to a closet on the other end of the apartment but the problem there is that there is only a light bulb outlet and that would be ungrounded. The only other option is the bathroom!

Hmm guessing since it's rented you're not allowed to run new electrical right? I would just run a 20 amp drop and you'd be all set.

 

[sxr7171]

Hmm guessing since it's rented you're not allowed to run new electrical right? I would just run a 20 amp drop and you'd be all set.

Yeah I wouldn't spend any money on improving someone's apartment. But with Dominion's ideas I have really cut the sound levels to an acceptable level. I just need to add some quiet fans to get my HDDs down a little. Maybe its better to just put a fan in the cabinet to help exhuast the hot air since the case fans do pull enough air by themselves. Just the air in the cabinet gets a little warm. There is a gap in the cabinet and I might just jury-rig a cardboard vent to a 120mm Gentle Typhoon.

 

[sxr7171]

Hello. What eight hard drives are you using? Do you live in a studio apartment?

Unfortunately some Seagate 1Tb and 1.5TB and one 2TB WDC Green Power. I plan to replace all to 2TB Green Power though.

But it's spindle noise that's the problem. I don't hear seeks at all. I think the cardboard solution has really done wonders.

 

[Modelworks]

Place the drive inside a container that contains no air, just a vacuum. Support the drive via a spring to dampen vibrations to the outer case. Of course cooling would be an issue

 

[DominionSeraph]

Place the drive inside a container that contains no air, just a vacuum. Support the drive via a spring to dampen vibrations to the outer case. Of course cooling would be an issue

Much bigger problem: Your heads would crash.

 

[lefenzy]

Much bigger problem: Your heads would crash.

Perhaps then, build a double-wall chamber with a vacuum between the walls. LIkely not practical though.

 

[bobsmith1492]

Perhaps then, build a double-wall chamber with a vacuum between the walls. LIkely not practical though.

The walls would have to be black so they would radiate and absorb the heat generated from the drive. Of course radiation is so inefficient, the inner box would have to get up to a few hundred degrees to remove several watts from the hard drive. Probably the power wires breaking the vacuum boundary would transfer most of the heat.

EDIT: Ok, now I'm curious; just how hot would a hard-drive-sized box have to get in order to radiate the heat from a hard drive?

Let's imagine a 1TB Western Digital Black drive:
http://www.wdc.com/en/products/products.asp?driveid=488

- Requires 8.4W while reading and writing
- Dimensions: 1.028 x 5.787 x 4.00 (inches)

So, a containing box would have a surface area of (1.028x5.787x2)+(1.028x4x2)+(5.787x4x2) = 66.418[in.^2]

Off to church... have to decode this info later: http://en.wikipedia.org/wiki/Thermal_radiation#Formula

Ok, if I'm reading this right, for black-body radiation, we can relate area, temperature and power transmission by:

P = (sigma) x Area x Temp(^4)

Solve for temperature T:

T = [P/(sigma x Area)] ^ (1/4)

where sigma = 5.670×10^(−8) [W·m^(−2)·K^(−4)]

So... converting 66.418[in.^2] to [m^2]:
66.418[in.^2] x (2.54 [cm/in])^2 x (1[m]/100[cm])^2
= 0.04285[m^2] = Area

Now: T = (8.4[W] * 1/(5.670x10^(-8) x 0.04285)[m^(-2)][m^2·K^4/W]))^(1/4)
=> T = 242.486K

So, the drive will have to be 242.5K above the surrounding radiation absorber. Let's assume the magic vacuum box is in a computer on a warm day with a temperature of 90F (32C).

32 + 242.5 = 264C

or 507.6F. Your hard drive would be glowing inside its perfectly black enclosure, which would also be glowing. Naturally, the power wires would help conduct away some heat, so it would probably be a good amount cooler. I doubt it would meet Western Digital's specification of 140F, though.

 

[sxr7171]

^^^Ha Ha. Great analysis. I guess a vacuum is not an option.

 

[C1]

I actually did something like this with my stereo hobby. I basically built a room inside of a room. However, some tips from my research might help.

- Ensure that the case is physically isolated from the home structure. A possible simple solution might be to rest the case on a small air mattress.

- Use of anechoic foam panel should be investigated.You will notice that is one way printer noise in offices were controlled (ie, putting the whole printer into a separate box lined with anechoic foam paneling.

- There also should be available off the shelf sound canceling units. These have a small microphone feeding an amplifyer that inverts picked up sound & puts that energy out thru a small speaker to cancel energy frequencies.

- Brick walls (because of their mass) really stop sound. So lining the cabinet with bricks (or double sheet rock) & anechoic foam might actually do the trick.

 

[silverpig]

Just get an extension cord and put the PC in the other room.

 

[C1]

Damn! Why didnt I think of that?

 

[Red Squirrel]

lol sometimes we don't think of the simplest of things.

Just make sure you get a heavy duty 12awg one given it will be in a permanent situation.

 

[Modelworks]

Just get an extension cord and put the PC in the other room.

No NO NO. It cannot be that easy. For it to work properly you must transmit the power from a power station several feet away into a coil tuned to the specific resonant frequency. It wouldn't hurt to throw in something nuclear while you are at it.

 

[C1]

Actually, if you dug thru the house's slab & installed the system in an underground chamber (similar to that for a large in-floor safe, I'll bet you'd never know it was there. Or just buy a large Diebold safe (the kind on wheels) & install the PC in it. Besides stopping noise, you will be assured that your data is safe (sic).