Building a Dead Silent PC

Jouster writes "The folks over at HardCoreWare.net have finally lost it. They built a PC that's well over twenty times quieter than their comparison PC (40 dB versus 65). And it's no sluggard, either: P4 2.80 GHz, 7200 RPM hard drive and--get this!--an overclocked to the max GeForce4 Ti 4200! The only fan in the entire system is in the PSU."

Re:20 times quieter?

[Maniakes]

Decibels are a logarithmic scale. 80 dB is ten times as loud as 70 dB. 65 dB is actually 10*10^.5 = 31.6 times quieter than 80 dB.

Re:20 times quieter?

[Karora]

Please explain how something can be 20x quieter...

Decibels is a logarithmic scale.

So 10db is 10x (power), 20db is 100x and so on.

This might seem like they are understating the case, but in general 10db is considered to sound like "a doubling in volume" so a difference of 40db is roughly 2 to the 4th or 16x.

Fiddle with a few db more and you can easily justify "20x quieter."

Re:20 times quieter?

[Roosey]

It's a logarithmic scale. The volume of the sound goes up ten times for every ten decibels. Here's some math [wolfram.com] and a comparison chart [everything2.com].

Re:I have to wonder......

[Moofie]

Radiant heat transfer is on the order of 10-100x less effective than convective cooling (like with a moving airflow). Radiant transfer varies as the fourth power of temperature, and does not turn significant until you get to temperatures that would a) fry any component you can name and b) give you a really good burn when you accidentally brush it with your leg.

So the window, apart from looking stupid, doesn't really hurt anything. : ) 'Cept the faraday cage.

67C?

[Anonymous Coward]

That seems a little high for a CPU temp, Ive always heard that you wanna keep it under 60C at worst, usually under 50C though.

Of course, you can probably drop the temps by lowering the voltage to the CPU and underclocking it.

I don't buy that

[Anonymous Coward]

The scale is logarithmic because *hearing* is logarithmic. This '20x quieter' has more to do with sound pressure than what a human being would say.

Google Cache

[babycakes]

here [216.239.39.100]

What about fluids?

[kylef]

This is quite timely, with the release of Windows XP Media Center Edition [microsoft.com] creating the need for a silent box that won't detract from that $2000 sound system! They use mostly Zalman Tech [zalman.co.kr] heatsinks in this system for noise reduction, but I wonder whether air cooling will continue to be viable in the next few years? Fluid cooling systems were necessary for the Cray supercomputers of the 70s because of superior heat transfer characteristics. This may be the case in the near future if silicon densities increase much further. But fluid cooling is also good for silence as well. The LA-class fast attack submarine that I was on for a while used a chill water cooling system for the sonar display equipment. And you can imagine how much silence is paramount in that setting! (Of course, they also used magnetic rotating drums for display buffer memory, if you can imagine...)

Re:Old news

[Mike Schiraldi]

I'm having trouble with the math -- i know that it's a log scale, so:

10^3.1 = 1,258

10^4.0 = 10,000

10,000 / 1258 = 7.94, not 2512.

Am i doing something wrong?

Re:I have to wonder......

[Rolo Tomasi]

Semiconductors aren't that sensitive to moderate heat, in the way that it would destroy them; if they get too hot, you will just have system crashes, random resets, etc. (as if that isn't bad enough).

No, but heat kills bearings. So, the hard drive will probably be the first component to fail, especially with today's sensitive drives.

Another mid-term problem would be the electrolytic caps. They will dry up over time, which will introduce sporadic instability/reboots at first, and you'll be driven mad searching for the cause. Almost impossible to detect without special equipment, i.e. an ESR meter.

So, if you want a silent PC, you'd be better off buying cool components in the first place. It's just too much of a hassle (and expensive) trying to silently and reliably cool high power CPUs and graphics cards.

Re:Old news

[marauder404]

2,512 times quieter than the 65 dB machine:

10^3.1 = 1,259
10^6.5 = 3,162,277

3,162,277 / 1,259 = 2,512

Re:20 times quieter?

[packeteer]

This is also why radio antannea can be measured in decibles. They reproduce the signal X times as powerful. Thats why a the price differance between a 20 decible antannnnea and a 22 decible antannea is more than 10-20. Also this is why you have to be careful when buying speakers. You might not think its worth another $100 to go from 112 to 114 decibles but thats much louder.

Re:I have to wonder......

[nexthec]

and electrolytic cap is a capacitor constructed with an oil as the dialectric, as oppsed to paper, this allows for a much greater range of use. They can be found in your PSU, on your motherboard, and probably on your harddrive. In general Capacitors store charge, however they do much more than that, the can be used with an opamp to make filters, to do intergration/derivation and a multitude of other neat things. Take a circuits class or 4 and you learn some neat stuff

Re:I can think of one idea to get even cooler

[Moofie]

No.

Heat exchangers want to maximize surface area, not mass. You are trying to present the maximum surface to the surrounding airflow, cooling by convection. More mass would simply be a thermal battery, storing heat on the processor. The thin heat sink fins actually remove heat from the processor by transferring it to the surrounding air.

Re:20 times quieter?

[Anonymous Coward]

sound is measured in terms of power intensity... W/m^2 and is converted into decibels by taking the logarithm to base 10 and multiplying by 10... i.e., 10 log_10 (X)... so 3 dB is a doubling in sound intensity (not 2 dB)

Re:deciBells

[Maniakes]

I'm not sure where this 20x quieter thing comes from, but generally. a 10dB increase in sound output is considered "twice as loud." Note that a 3dB increase is twice as much energy (well, 3.0something, but close enough). Similarly, -3dB is the "half power point."

Your physics teacher will tell you that 10 dB is a 10x difference in the intensity of a sound wave.

Your psychology teacher, or your friendly local audiophile, will tell you that a 10x difference in wave intensity is only a 2x difference in perceived loudness.

Most slashdotters, geeks that we are, payed more attention in our physics classes. As a result, I forgot about the perception issue until just now.

I am currently listening to 128 kbps mp3s on a $25 pair of headphones. Whether or not I am an audiophile is left as a exercise to the reader.

Re:I can think of one idea to get even cooler

[ColaMan]

R134A is good except that it boils at -30 deg C, and its pressure to stay liquid at room / processor temp is about 70psi. It won't short your board out if it leaks, but the frost that'll be generated if you get leak liquid R134A out will.

Silence!

[Deton8]

I'd accept a few degrees hotter silicon for the huge reliability boost of getting rid of the fans on the processor and graphics card (MTBF circa 15,000 hours in the real world contrary to their b.s. specs, divided by two since there are two of the little bastards). Your remaining fan in the PSU case needs a fan rotation alarm on it, and if unattended, some kind of thermal shut-off or redundant fan. One nice trick for quiet fans is to use one much bigger than you need and then run it at a slower speed. Another tip is to mount the disk drive and fans on Sorbothane [sorbothane.com] standoffs, and maybe stick a couple of slabs of Sorbothane on the walls of the PC case. One quibble with the article -- for best cooling, you want as small a case as possible, not as big as possible. The objective should be to maximize the velocity of the airflow over the heatsinks, and you do this by constricting the space around them. One innovative way this has been done is through the use of engineering foams like E-PAC [dmtgmbh.de] which allows the designer to create engineered air ducting which forces the airflow over the parts where it is needed. Some other people have asked why the PSU fan is necessary -- having just gone through CE and UL testing on one of my products, you can't imagine the kind of pain the test lab would make you go through if you took the PSU fan out of the PSU case. It's only a practical proposal for a major corporation with a lot of money and time to throw at it.

Re:Missing the point

[jkramar]

This is horribly offtopic, but seemingly necessary.

Logarithmic or not, it doesn't matter. You can't be x times quieter than something. Its not possible. You can be x times louder, but not quieter. Ignoring negatives, x * y is always going to be a larger than either x or y. It really drives me nuts how people misuse that term. You'd think a geek site would be capable of using mathematics terms correctly.

That is incorrect. x*y is not necessarily larger than either x or y, because either can be between 0 and 1. Furthermore, "x times quieter" is equivalent to "1/x times louder", usually. Expressions such as these can hardly be considered mathematics terms.

Now, why did I reply to that?

Silent iMacs and Monster Heat Sinks

[Aqua OS X]

The slot loading "original" iMac (aka the gum drop) had(has) no fans, quite hardware, and vents on the top of the case. I don't quite know if Apple's eMacs and LCD iMacs have a similar set up, however I'd bet that they have fans.

Nevertheless, Apple still strives to build fairly quiet boxes when ever possible. I mean hey, look at the fan to heat sink ratio in this box: http://www.apple.com/hardware/gallery/pmg4_august2 002_480.html Honestly, I've never seen a bigger heat sink within a consumer PC. One could fry 10 strips of bacon on that beast.

Re:Fanless PC PSU - THAT would be news

[shepd]

This [snookles.com] site has a link to a 300 W fanless ATX PSU. And a lot of other really good links too.

Re:can't use flash drives

[terrencefw]

forget rotating disk drives. Get a mobo with RAID and a bunch of totally solid state flash hard drives [sandisk.com]. they're electrically identical to laptop hard drives, so a $25 adaptor will allow them to be used in place of any old IDE hard drive in your RAID, but you'll have to change the "I" in the acronym from "Inexpensive" to "Independent", if you know what I mean.

On the other hand if you think about the performance you'd get from the right kind of RAID where the individual "disks" have specs like these [sandisk.com]... suddenly everything else seems small.

Yeah... but flash drives have a limited number of write operations. They're find for digital cameras and the like, where they'll only get written on a few thousand times, but once you tried to run a full OS on them, you'd reach their limit and your data would start to disappear. They're great though on things like Linux/BSD router boxes where you can have the OS on a read-only disk.

Re:I wonder.....

[xercist]

Actually, dB is a logarithmic scale, so 0dB is not actually *zero* like you assume, as 10^0 == 1. You could have something that was at -1dB which would be quieter than that. On this scale, "complete silence" really doesn't exist. It all depends on how sensitive your measuring device is.

Re:20 times quieter?

[nathanh]

Decibels is a logarithmic scale.

Yes.

So 10db is 10x (power), 20db is 100x and so on.

Yes.

This might seem like they are understating the case, but in general 10db is considered to sound like "a doubling in volume" so a difference of 40db is roughly 2 to the 4th or 16x.

No. 3dB is a doubling in volume. The relevant formula is dB = 10 log(P/P0). You got the 10x and 20x right so I'm guessing you just typoed.

More hardcore than hardcoreware

[juventasone]

The guys at Silent PC Review [silentpcreview.com] would scoff at the "hardcoreness" of hardcoreware.net when it comes to silencing PCs. After being on their mailing list for a year, I can tell you that they're waaay ahead of these guys in every aspect of PC silencing, many of which I've implemented myself.

Why no sealed case?

[PsyQ]

I've been using an acoustically sealed case from Noise Control [noisecontrol.de]. I really can't hear my PC anymore.

I also use one of Noise Control's modified Enermax PSUs and a Silverado CPU cooler. That's all I did to my PC to make it quiet, everything else is stock. A quiet case seems to be the most logical (and least expensive) first step if you ask me. If you can still hear any of your components after you've put them behind 2 cm of noise blocking fluffy stuff, you can start replacing noisy those one by one until the noise stops.

Noise Control now has their own fan control circuitry and new modified PSUs come with it built-in. Also, they have hard drive cages that catch vibrations before they reach your case. With all of that equipment it should be easy to quiet any PC.

Re:20 times quieter?

[Seahawk]

mod this one down plz!!!

+10db is twice as loud to the human ear!

So the 65 db down to 40 db should be something like 5-6 times less loud to the human ear!

Its the energy that is 10 times as big when you go up 10db!

Example:
A 10W stereo plays x db
A similar stereo with 100W plays x + 10 db
A similar stereo with 1000W plays x + 20 db

Re:I can think of one idea to get even cooler

[Rolo Tomasi]

There's one more aspect to that. When you have very thin, long fins, like the Zalman heatsinks do, the thermal conductivity is too low for the heat to be transfered all the way to the tips. From what I can see on the photos, these heatsinks are not as efficient as they could be. IMHO the optimal heatsink would have a tree-like structure, whith thick 'stems' conducting the heat to many thin, but reasonably short 'leaves'. The Arkua [arkua.com.tw] coolers seem to somewhat follow this approach.

Re:20 times quieter?

[Maniakes]

+10db is twice as loud to the human ear!

You're right, but I'm not wrong. For a full explanation [slashdot.org]

Re:I can't do that

[kolbeinn]

I have an AMD 1800+ and justbought a Zalman CNPS5100-Cu CPU fan, Zalman [zalman.co.kr].
Now my only complaint is how loud my hard drives are.

Re:20 times quieter?

[Anonymous Coward]

I think he was referring to the fact that there is the human perception of sound which has to be considered.

I think the statement that 10db is roughly perceived to be a doubling in volume is not far off.

My silent PC

[chocolatei]

My PC is underclocked with an overspec'ed PSU and hence there are zero fans. With 1G of cache memory the harddisk never spins up. The loudest thing is the monitor - it clicks when changing res. and hisses slightly, sort of like a very distant stream. I don't know what that is in dB but then it is quieter than ambient noise.

Re:20 times quieter?

[packeteer]

HAHA its so true. I just read another post i was goign to reply to but saw you so let me go off here. The wattage of a stereo doesn't matter. Its does matter in that its part of the equation but not the total measurement. I have seen peizo electric and other speakers that run off fractions of a watt that can put out over a hundred decibles. I hate all this crap about "i have a 500w stereo, oh yah well mines 600w". Some speakers can make it up to 130dB but will start to distort at 110dB, where as some wont distort up to 115dB but can only go to 125dB. Which one would you rather have? Im tired of people doing stupid things with speakers. Running 1kw systems in their drunks using $5 cables that have way over acceptable limits of capacitance. And dont even start on capacitors they sell for subs in your trunk. Thats just total bullshit. When the sub runs out of power its because of impedance on the line most likely from bad connections (solder em if you really want good sound) or just plain bad wires (10 gauge wont cut it). Personally im going to stick with my mono radio output and take the $1000 i saved and do something better.

Could be quieter..

[Fweeky]

The Barracuda V [storagereview.com] is actually somewhat quieter [storagereview.com] than the Barracuda IV [storagereview.com] they used.

Use the extra leeway to add a few fans; don't forget, if everything's running close to their design limits now, it'll probably get hairy if you have a hot summer.

Plus it's really a good idea to keep components like HD's fairly cool. Let them fry and you risk reducing the service life of the drive and increasing the chances of data loss. You at least want reliable storage, right?

Also, you should be careful with that huge-ass Zalman cooler. They're very heavy, and will happily tear off the socket if you happen to move the machine anywhere. The full Cu version is about 200g heavier than AMD's maximum recommended weight.

Contain PC in a box

[TheVidiot]

What about putting a noisy PC inside some sound-absorbing case (with good air flow). Wouldn't that be much easier? Does anyone have any experience with this model?

Thanks!

Re:silence overrated?

[Zathrus]

i dont see why people go to such extremes, for little to no improvement.

It's called the noise floor, and while it may not matter much to you in standard computing environments, it matters a lot in non-standard ones. Like if you're building a home theater PC. If your PC runs at 40 dB then forget hearing anything below that -- which can be either quiet passages in music (classical or rock - both have 'em) or downkeyed scenes in movies. Turning the volume of the system up isn't an option unless you want to constantly change the volume during a movie so you don't lose your hearing during action/climactic scenes.

Additionally any noise like a PC can interfere with subtle nuances in music or movies, which is equally bad. With a 20 dB noise floor you may be able to hear sound queues for things you can't hear with a 40 dB noise floor.

So it's not little to no improvement - it can be pretty substantial in the right environment. Personally I'm looking to build my next PC to be quieter, not because it's going to be a HTPC, but because I'm tired of having an absurdly loud PC. I'm not going to take it to extremes, but I will try to purchase quieter components.

Re:Longevity? Hardware Burnout?

[Jucius Maximus]

"Has any other /.er made a comparable system? How has it lasted?"

Last May or June I invested in some good cooling equipment and reduced the sound level of my system by about 20 dB. Frankly my solution is not as elegant or inexpensive as this guy [slashdot.org] but at least the machine is more portable and you don't need access to a wood shop.

I replaced the noisy-as-hell stock heatsink (coller master or whatnot) on my CPU with an Alpha 8045 and artic silver 3 thermal compound and then put a vantec stealth fan* on top of it. The great thing is that this HS is 80mmx80mm so you can use an 80mm fan as opposed to the standard 60mm fan. This means that it's possible to get greater airflow with lower speeds and thus lower noise.

I also replaced my other case fans with Vantec Stealths and got some light machine oil and lubed up all the fans' bearings to eliminate any tiny fan noises.

At an overall cost of CAD$120 or so, my box actually runs cooler than before and I don't get headaches when working.

My one mistake was usnig a slot loading DVD drive because the slot lets noise from inside the machine get out easily.

*Yes, I know that Panaflos are cheaper and only a tiny bit less well performing.

Where are you measuring from? dB or dBA?

[TibbonZero]

Well, this is really gonna depend on what you are measuring and where you are measuring it from.
To say that a Jet Aircraft is 140dB is meaningless. 140dB at almost any frequency would hurt alot, and probably cause a bit of damage after a few minutes. I know that there are jets taking off from Logan right now but, for some reason I am not losing my hearing. It all depends where you measure it from. There isn't a standard distance.

So quiet PC could be measuring from farther away (and because of the inverse square law, it would get 10 dB quieter pretty fast).

Another thing that nothing here is mentioning is dB @ a freqency @ a distance, or if it's dBA @ a distance. Your ears wouldn't be able to hear 50dB @ 40hz, let alone 30dB @ 80 hz. You could hear 30dB at 2000hz though. dBA is a weighting of multiple bands, and is another beast altogether.

Another thing is if you are measuring the sound right beside the power supply, you are screwing your measurements anyway. It's acting as a Plane Source if r So basically, move the mic a few feet away, not right up on the thing.

I am personally wondering if they used a good measurement system, or just a radioshack thing...

Explanation of log

[Jucius Maximus]

Here, once and for all, is a log scale mathematics explanation. I'm still a student so I remember something about it:

Sound level is how loud a sound is to human ears. It can be measured in dB and an increase of 10 dB sounds ten times louder to human ears.
Sound intensity measures the energy of the sound, often in W/m^2. (Watts per metre squared.) If you multiply the sound intensity by the face area of your eardrum, you'll get the number of Joules per second (W = J/s) that your ear is perceiving. This scale is linear with human hearing perception, so double the intensity means it sounds twice is loud.

The Equation:
B = 10log(I/Io)

B = sound level in dB
I = sound intensity in W/m^2
Io = sound floor of human hearing, Io = 1x10^-12 W/m^2

So, doing the math, 40dB = 1.0 x 10^-8 W/m^2.

And 31 dB = 1.26x10^-9 dB

So therefore, 40 dB is 7.94 times more intense, and therefore 7.94 times louder to human ears.

(7.94 = 1.0 x 10^-8 / 1.26x10^-9 )

Note: the previous poster's comment about one being 2,512 times quieter than the other was for different values, and this information does not override that person's (correct) calculation.

Thank you, and have a nice day :-)

Re:Not as loud, but its still a space heater

[Natalie's Hot Grits]

"Apple has been making quiet computers since before the Mac came out. That would be before Gateway, Dell and Compaq were even in existance (not to mention Pentiums)."

Before the pentium, x86 PC's didn't have fans in them. If they did, it was 1 very slow fan in the PSU. IIRC, most of the pre-Mac apple PSU's also contained fans in them. If you like, I will go through the apple parts database and find all the models that contain fans. But I'm not at work now, so it will have to be later. So your point is irrelevant. Back then, cooling and noise was a non issue, and we didn't have 300-400W going through our computers. Fans were slow, small, and didn't push much air, thus noise was not an issue.

Yes, my first hand computer knowledge starts around the intel 386, that doesn't mean I don't know anything about old PC's (Personal Computers, I don't mean x86 here). My 386 had 1 fan; it was in the PSU. So it was just as quiet as the apples.