Cancelling Out CPU Fan Noise

Percy_Blakeney writes "After realizing how noisy his computer was, a professor at BYU has created a new CPU fan that uses small microphones and speakers to cancel out its own noise. It isn't perfected yet -- it only nixes the whine, not the whoosh -- but it looks like it could be promising, especially given the professor's background: making jet engines quieter."

I should have patented it...

[zeux]

I though about it a long time ago because I know we are using the same kind of technology in the airports.

Near the landing strips you can sometimes find some "sound reflectors" which just reflect the sound wave they receive from the planes. The sound is then cancelled by itself.

I saw it once in an airport in France and it works really well and costs next to nothing. AFAIK there's no sound wave modification in that system but I'm not sure (maybe the surface of the reflectors is made in a certain shape to change the sound wave a little).

But in this case it's different because the "box" must produce the counter sound wave. It's not just reflection, there is sound generation here. It means that the microphone and the speakers must be very precise or you just end up with more sound.

But if this guy can do it with 20 bucks it means that it's much easier than I though.

Wow

[michaelhood]

Active Noise Cancellation stuff is a really cool technology. I wonder if this could be applied to cars and other "larger louder" things in the future.

Re:No need to worry.

[hampton]

Installing a Zalman HSF is exactly what I did. Highly recommended. Or (and), you can just buy a fan controller for the money (or both) to really quiet down your system.

Where this would be really useful is for the whine of hard drives. It would be far better than the current system of enclosing it in some casing thus making it run even hotter.

Mods...

[c0dedude]

Mod points to a poster who can point me to a download site for this or something like it, want to try it myself, will put a little speaker by the fan. Or is this not the way it works? Would a computer be too slow to pull something like this off sucessfully?

Re:now for the hard drives

[WaterTroll]

My western digital hard drive is completely silent unless you put your ear nearby. However, I recall that I could hear a distinct difference in volume when I was using linux or windows 98. it was much audible from a couple feet away while running linux, but completely silent under windows 98. in fact, the first time i ever actually could hear my hard drive clearly (this is all reading/writing data, not the spinning platters) was when i first installed mandrake and used the disk partitioner.

Re:I should have patented it...

[Martin Blank]

I have a question about active cancellation, as I've heard of it being used in other places. Does the cancellation of a noise of a given frequency have any potentially harmful effects that become less obvious through cancellation? For example, does a high-pitch tone that could cause hearing loss over time become more dangerous now that there are two high-pitch tones (albeit directly off-phase) now sounding, or is the cancellation that complete?

I could use this

[crass751]

My laptop is so loud that if the fan(s) isn't/aren't running my roommate asks me if I turned it off. This thing generally has two fans running at times, and when it's really working hard, a third kicks in. My four year old desktop machine is much quieter than this thing.

Stupid HP. Had to go sticking a desktop chip in a laptop. Oh well, it still runs circles around my roommate's silent Centrino-based machine.

Absorbing technique instead

[fembots]

In most offices, they don't use noise generators (ie Gossip Support Group) to cancel out talking noises, instead they put in a lot of plants, cubicles, which act to absorb most of the noises.

If the noise is pointing at your directly, then you probably need a cancelling method. If it is a general-direction noise, it should be absorbed rather than trying to cancel it (where you need to find it in the first place).

Re:Wow

[krosk]

Most car noise these days is not from the engine running. Technology these days allows new cars to run incredibly quiet. Probably 99% of the noise you hear while traveling down the highway is road noise. Noise produced from your rubber tires against pavement. I saw a news flash a couple years ago about a new type of pavement that dramatically reduces that noise, but it's too expensive to be widely used.

Re:No need to worry.

[Anonymous Coward]

It gets harder and harder the higher in frequency you go because of the decrease in wavelength. Seagate drives are pretty quiet. Also, use 5400RPM drives for data storage, 7200 to run your system. 5400 drives are cooler and fail less often.

Re:Keep everything quiet

[nmoog]

Dont get me wrong, I love earplugs - I live in a warehouse filled with dudes who love noise music. But Ive tried those bose headphones, and they are amazing. Earplugs DECREASE the noise, the headphones ERASE the noise. That makes a world of difference, especially on airplanes and the like (or offices with heaps of computers) where machine noise can drive you mad.

When you first put those headphones on it freaks you out a bit, because its like being in an anechoic chamber - the noises you use to judge distances (the rooms natural reverb) are cancelled. Its wierd.

diy?

[frankmu]

can you take apart the cheap "noise cancelling" headsets and do it yourself? i would imagine all the parts and circuitry are there. lets see if my wife will miss her airplane headset....

Specific noise cancellation.. bad.

[Anubis333]

I would much rather purchase something that can cancel out noise from any source than from just a cpu. It would be a lot better to mount something between me and my computer. After all the CPU isnt the only noisy thing, I have noisy case fans, noisy 10K HDs, and noisy RAID array's. Not to mention, the "anti-noise gun's" on the market are mobile, meaning at any given time I can turn it to face my girlfriend.

Re:Wow

[afidel]

But in this case it's absolutly retarded. For only $17 MSRP you can get the Arctic Cooling Silencer64 which is designed to handle any AMD Opteron/Athlon64 currently shipped and produces only 20dB of noise (essentially silent). It achieves this through a large, slow fan which also has the advantage of being more reliable =)

Re:Negating Sound? Its like new cars....

[Skater]

For what it's worth - in both of my cars (a Chevy Impala and a Mercury Cougar, both late-model), there's an "auto-volume adjustment". It works off your speed - the faster you go, the louder the radio gets. They are adjustable (one has 3 levels, the other has 7), or the feature can be turned off entirely.

It's a great feature - I find myself messing with the volume control a LOT less. It also doesn't require the technology and expense of sound-cancellation. :)

--RJ

Re:Like the noise cancelling headphones?

[Reverberant]

I thought the noise cancelling headphones worked because they were right against your ears.

Well yes and no. Noise cancelling headsets are particularly effective because the ANC speaker and microphone are right next to your ears. This helps insure that the signal that arrives at your year can be sampled and inverted so that the sum cancels at the ear. It would be much harder to accomplish this with speakers and/or mics located away from the ear.

However, remote mics/speakers may work if the noise source is highly directional, like a waveguide (I suspect that's happening here). If you can effectively cancel the sound at the orifice, you'll probably achieve a significant reduction in transmitted sound, no matter the location of the receiver.

Re:Wow

[Cornelius the Great]

"Active Noise Cancellation stuff is a really cool technology. I wonder if this could be applied to cars and other "larger louder" things in the future."

The 2003 Dodge Viper SRT/10 has side exhaust with active noise cancellation. It can be switched on/off to allow for "stealth mode" or a throaty sound- though I can't imagine why... I love the sound of a V-10.

I've used the headphones that do this...

[Phil John]

...as a pilot friend of mine has a pair (with mic boom as well). When you are flying a Cessna it's hard as hell to hear air traffic control, so these really help.

One really important use of these will be in ultra-quiet studio computers. Of course, its not to make sure the fan noise doesn't get recorded as its not a real recording studio if there isn't a separate recording booth/room (the studio I use in london from time to time is two rooms built within one large one, resting on a buttload of industrial springs, but I digress.

When you are listening to playback, making sure the singer was in tune, mixing the track, or whatever, you don't want ANY extraneous noise from fans. There is already a market for ultra ultra quiet pc's for this kind of application and advances like this can only help further the art.

Re:I should have patented it...

[Cosmic_Hippo]

For instance, you can buy noise-cancelling headphones, but the cancellation only works for your own ears, which gets the sound in just right right phase; to the people around you, there will be a perceptible noise coming from your headphones! Conservation of energy says you can't just destroy the energy of those sound waves. Most likely you're just sending extra-strength sounds waves somewhere else.

I own a set of noise cancelling headphones and there is no perceptible noise being emmitted from them to the outside world. The noise cancellation circuitry takes the incoming signal and inverts it to cancel out the original incoming sound wave. Conservation of energy doesn't really apply. You aren't really destroying the energy of the sound waves, just cancelling it. Energy is spent on both the positive and negative signal. Although I just minored in acoustics so I'm probably missing something. Any professional opinion is appreciated.

Re:I should have patented it...

[bcrowell]

Any professional opinion is appreciated.
Well, I do have a PhD in physics, but I should bow to your actual experience with the device :-)

My guess is that the sound coming into your ears is only a tiny amount of power, so reradiating that power in all directions doesn't make any amount of sound that would be perceptible to someone a significant distance away. Your eardrums only have a surface area of a few square mm, the the amount of energy impinging on them is normally only a gazillionth of a watt. Your ears are amazingly sensitive devices.

Conservation of energy doesn't really apply.
I was obviously wrong in my prediction about an audible noise for people not wearing the headphones, but I'll bet both my testicles that it's not because conservation of energy is violated. You'd get the Nobel Prize if you found a violation of conservation of energy.

Re:now for the hard drives

[dfj225]

" The noisest part of all my computers i the hard drive, not the CPU fan."

I have one of the new design Alienware towers, and by far the most noise comes from the case fans. I have 2 seagate hard drives, and I can hear them somewhat, but the case fans are much louder. Although the fans themselves don't make that much noise, but the amount of air that they throw around does. If it is quite in the room, the whoosing sound can be quite loud.

Re:now for the hard drives

[Trejkaz]

The graphics fan is the noisiest component in my case. Those Gigacube Radeon 9600 XT Extremes are damn loud.

I also have a 30-35dBA fan which doesn't help with the noise. To help, I have an Sonata case, and when the case is on, the only thing I can hear is the clicks from the hard drive, and only when it's being used (during loading. Get more RAM if you are using swap!)

Normal operation is silent, or at least silent with respect to the other, louder computers that are not in Sonata cases. :-)

And of course right now the damn thing is quieter than it's ever been while parts are back for repairs. :-(

Re:Keep everything quiet

[extra the woos]

Heh mentioning sennheiser in the same sentence as bose is like a sin, the two aren't even remotely comparable... bose is just like re-branded cheap paper speakers with a nice looking package etc heh...

Re:It never ends though

[shadowbearer]

Damn funny post, irokitt.

One day a fan on your computer fails, and you never noticed, because you couldn't hear the squealing of the tortured bearings...

The interesting thing is with modern mobos (which control fan speed based on how hard the cpu is working/generating heat) you can actually tell to a point what's going on. I can set a compile going, go sit on the couch and read a book, and tell when it finishes (the cpu fan noise goes down and the hard drive noise goes up briefly - I run Gentoo where it writes the files at the end of the compile/emerge).

SB

Re:I should have patented it...

[Cosmic_Hippo]

Heh, I hope I wasn't implying that conservation of energy was being violated. I just though that destruction of energy and cancellation of the signal were different in this case. I think I worded it wrong. No testicles need to be wagered :-)
The experience I've had with the equipment in class showed that the noise cancellation circuitry recorded the original sound wave, inverted it and fed it back into the speaker. The combination of positive and negative voltage basically told the speaker to output zero signal for that particular frequency. Nothing is destroyed, it's more like an electronic tug-of-war. It makes listening to music a lot more enjoyable, however it works.
Thanks for the reply.

Re:Me too!

[gosh_d]

You're forgetting that sound, being simply a pressure wave, travels a _lot_ slower than an electrical signal (6 orders of magnitude). If the mic is placed closer to the fan than the speakers, the speakers can have plenty of time to invert the signal and replay it. The distance is precisely chosen such that the speakers produce their noise simultaneously with the passing noise of the fan, even though it originated farther away. Delay's not a problem--no predicting needed.

I had an interesting idea based on this (I don't know if it's actually used). Fixing the distance and delay may not be accurate enough to match the signals completely, so you could have a second mic which listens _after_ the cancellation for beats. Superposition of the two similar waves produces the "beats" that musicians use to tune an instrument. By observing the frequency of the beats, the microprocessor could adjust the delay to more perfectly cancel the noise.

Not a new concept ...

[scdeimos]

... just a new application.

Before CPU's came along, this sort of thing used to be done with BBD (Bucket Bridge Delay) circuits, replaying the sampled sound 180 degrees out of phase. Of course, this only worked with single-frequency tones and the BBD had to be clocked at just the right correct frequency. Cancelling white noise (ie: fan whoosh) is a somewhat more difficult problem.

A number of "professional" aircraft pilot communications headsets have had active-cancelling (as in the article) built into earpieces (as opposed to the microphones) for several years, so as to reduce engine noise and pilot stress.

Car buffs here might even remember that VW had a Concept Car in the nineties which had an (I think) Bose-powered active-cancelling system in the cabin, the purpose being to cancel road noise and engine bay noise so you could replace it with sound samples of your favourite sports cars: Ferrari's, Porsche's, etc. Not sure it ever took off, though. ;)

Re:Nope

[opec]

This is a really neat aspect of sound waves. I saw an experiment once where two speakers were playing at each other, they tuned it until each completely negated each other and there was silence, though each was pumping out the tone. It applies to musicianship as well. Good marching bands with great intonation control can play really loud and sound good, while not so good bands will be limited in their maximum volume.

Re:now for the hard drives

[Endive4Ever]

They do in my collection of aging monster hardware.

You can't touch anything else with the reliability and stability of that system without spending four figures today.

And Pentium Pro systems are cool. A good PPro server is like a diesel truck. I can't afford to collect diesel trucks but I can afford to keep around some nice hardware, and even use it.

Re:now for the hard drives

[Kris_J]

Get a slower hard drive. 5400rpm drives are nearly silent, draw less power and are much cooler (thus you can turn down other fans). If you're concerned that this will affect your system's performance, you can compensate to some extent with lots of RAM.

Re:I should have patented it...

[FallLine]

Well I'm no PhD in physics, but I (did stay at the holiday inn express last night^H^H^H^H^H^H^H^H^H) do own a QuietComfort2 headphone set. I would agree that there is no discernible noise coming from outside the earphones. I suspect the explanation though is relatively very simple (or perhaps I'm just naive). The earphone cup is already quite muffled. The noise cancelling circuity only needs to "cancel out" a tiny amount of noise, that which would reach your ear by leaking through the insulation. Whatever noise is created by the cancellation speaker is quiet relative to outside background noise and will be further muffled by the headset before someone sitting next to you can hear it. In other words, whatever background noise you'd hear without the headphones is almost certainly (in real world applications) going to be orders of magnitude louder to a neighbooring passenger than the noise coming through your earphones.

Make fans with predictable noise?

[noidentity]

I wonder if the cooling fans themselves could be designed to make more predictable noise, such that noise cancellation could be done without a mic, and synchronized to the fan via the rotation sensors.

Re:Me too!

[martensitic]

It's true -- most active noise cancellation schema have both a source transducer and a feedback transducer. For instance, most active systems for reducing ductborne HVAC noise (a prime candidate, since in-duct fan noise is transmitted via a controlled, conduit, and the noise comprises fairly constant, discrete frequency characteristics) include a mic downstream of the cancellation "speaker". Feedback from this mic is used along with the source signal to improve the performance of the system as a whole.

Mod up!

[Ayanami Rei]

We're currently doing experiments monitoring this at work. We want to isolate engine vs. tire vs. transmission noise across various makes and models of cars during ramp up, idle, and braking. It's a fun project involving lots of wireless and embedded tech, with audio, sig proc, and linux thrown in there too to make it interesting.

From what little of the results I've looked at, it's pretty clear that tire noise is dominant during cruise.

Re:Nope

[BlueJay465]

Listening to the differences in the different types of speakers and monitors working at a Guitar Center over the holidays, I was blown away by the sound quality of some of the studio monitors in their Pro-Audio department, and then comparing them to the high-end home-theater gear at the Magnolia Hi-fi just down the street.

You will get a lot more bang for your buck with a good set of monitors while only taking fraction of the space. Check out the Event TR [event1.com] series or M-Audio [m-audio.com]'s line. Most decent music equipment stores like Guitar Center should have some demos.

Re:Nope

[canowhoopass.com]

Try it yourself some time. Take two identicle speakers and feed them both the same sound (as in one mono sound to both channels, not a single stereo source). Reverse the polairty on one speaker (plug the black plug into the red and vice versa). If you have them setup normally and listen to the sound far away, it'll simply sound defocused, as though it has no apparent centre or source. This is a good way to focus your speakers, the more defocused an out of phase sound is, the more in focus an in phase sound is. However now take them, get them right next to each other, and point them at eachother. You'll hear almost nothing. PRetty much all you hear is the sound that radiates from the cabinets.

This is one reason why I love Slashdot... you pick up all sorts of neat science project ideas for the kids. Thanks!

Rod!

Ears are sensitive indeed!

[chadjg]

My acoustics book said that if you put a person with normal hearing into a sound isolated anechoic chamber, and give them awhile to adjust, they will actually hear the blood flowing in their ear.

Point being is that it would be completely pointless for them to be any more sensitive. Quite amazing really.

Re:Nope

[Ralph Wiggam]

Seperating audiophiles and thier money is like shooting fish in a barrel. I think there is a lot of placebo effect in high end audio. You just paid 800 bucks for a few meters of cable. It was a multi-thousand dollar system before the new cables. Do you think it could sound bad now?

I'm lucky. I have very good hearing, but I'm pretty tone deaf. The medium priced stuff at Best Buy sounds just as good to me as the high end stuff at the specialty stores. There's no point for me to shell out the cash.

-B

Re:I should have patented it...

[xtal]

You'd get the Nobel Prize if you found a violation of conservation of energy.

Not mine, but.. compelling.

Casimir Effect [wikipedia.org]

Re:It never ends though

[The Famous Brett Wat]

I have (and am posting this using) a VIA EPIA-M motherboard with passive heatsinking, no hard drive, an external fanless power supply, and no case fans. (It acts as an X terminal for another system.) There are no moving parts in the system at all. Believe it or not, however, I can hear a small electrically-generated squelching sound when there is ethernet activity (when all else is quiet in the dead of night). I could try WiFi, I suppose.

A Cheaper Way

[Kombat]

My acoustics book said that if you put a person with normal hearing into a sound isolated anechoic chamber, and give them awhile to adjust, they will actually hear the blood flowing in their ear.


An even cheaper demonstration is to simply plug your ears. It works better in an area that's already quiet, but if you simply plug your ears with your fingers, you'll hear the blood flowing in your veins and arteries. That's what that low, rumbling noise is that you'll hear.

Been There, Almost Done That

[Wingsy]

As an EE I had the "opportunity" to design an active noise cancellation system for headphones to be used in aircraft. I had the perfect test bed - my own plane, an Emigh Trojan. Loudest cockpit noise in the world. Anyway, the technique seemed simple enough - place a mic outside the earpiece and apply the outside noise to the earphone speaker, phase adjusted by whatever was required to cancel the outside noise striking the eardrum (a single point). The problem was that every frequency (or small range of frequencies) needed its own amount of phase shift, which complicated matters tremendously. The phase shift needed, due to the wavelength "distance" between the mic and the eardrum, was not right on 180 degrees like one might think. The final "product" helped quite a bit but was still not something I would want to try and market (which it wasn't). If you notice in the article, NASA also gave up on designing a noise cancellation system at airports. The problem NASA faced was much more difficult than mine. The source of the noise and the eardrums of the receivers were never in one fixed location. So, not only did they have to apply a phase shift to several bands of frequencies to the noise cancellation sound generators, they also had to apply different shifts, and different amplitudes, depending on the location of the noise source and the eardrums. Yes, I can see why they passed on that one. The Professor's problem lies somewhere between the one I tackled and what NASA tried. His advantage lies in the fact that he can place his noise cancellation speakers relatively close to the noise source. That helps a lot, in that the wavefronts from the two sources can radiate outwards from a single point (almost) and be cancelled no matter where the receiver is located. I suspect one of the reasons he can only attain cancellation at the high end (the whine) is due to the poor low frequency efficiency of the speakers.

Re:I should have patented it...

[CProgrammer98]

I remember many many years ago (> 20) reading a sci-fi story about a noise cancellation device that work in exactly the way current real devices do. The plot of the story was based around "where does the energy go?" The author concluded that it would be stored in the powersource of the noise cancellation device. As I recall correctly, the fictional inventors of the device discovered this at the conclusion of the story when the device exploded - the battery ended up storing all the energy until, as Scotty would have said "She canna take any more cap'n" and the whole thing blew up spectacularly.

If anyone can remember the name of the story or even the Author, I'd be very grateful...

Re:I should have patented it...

[tiger99]

No, the cancellation can only be partial and in certain positions. The problem is simply a matter of time delay, the speed of sound is not very high so for practical dimensions there is substantial phase shift. By using a lot of speakers and very complicated processing, you can make it better, but it is still in the form of an interference pattern, and will always show peaks in certain places. You would need to make the summation of the speakers look like the original source, but in antiphase, to everyone around, to get cancellation. The wavelength of sound at 1KHz is about a foot, if you use one speaker you need to get the difference in disatnce between everyone and the speaker very much less than 1 foot so they all see cancellation. The idea is to put the maxima where they do least harm. These systems only work on regular frequencies, you can't predict random noise, and so can't cancel it unless maybe the speakers are all very much nearer people than the original source so there is time to do the computation.

A very greatly hyped and over-rated technology, which in some specific circumstances will provide a useful reduction (10dB?) in low-frequency noise, for example in the Dash8-Q400 aircraft, where propeller blade fundamental frequency noise is at 85Hz in the cruise (6 blades at 850 rpm, which is lower than most), and where people tend to sit in predictable places, it does quite well, although a fair part of the reduction is by trimming the relative phases of the two propellers (which should run in synchronism in steady flight, although this is not a safety-related function and might not always work, as it is not provided with any backup system), how that compares to the contribution from the speakers I don't know. The active noise suppression system can command the propeller controls to adjust the phasing, and indeed select which blades to synchronise, as they might be slightly unequal, of course it has only extremely limited control authority to avoid it becoming safety-critical, so it can only trim the relative angles very slowly. That is basically adjusting two noise sources so they make the least overall noise, inside the aircraft. I always had the suspicion that at certain precise positions outside (as presumaly happens with all twin-engined aircraft), the noise would be doubled, but it passed certification so it must have been acceptable. Probably much quieter than the average jet, Avro 146 excluded, anyway.

At 85Hz, the wavelength is about 12 feet, so the problem is somewhat simpler, but still very complex....

I am not a noise expert, but I can clain very intimate knowledge of the propeller sync system, called "syncrophase" in this case, being one of its main hardware designers. The propellers are synchronised at the desired angle, within about quarter of a degree, which is not bad considering there is no mechanical connection, the engine power is several thousand horsepower, and only a little pulse as each blade passes a sensor gets sent from the master controller to the slave. Oh! sorry, I forgot, can't use these words any more..... Back to the drawing board. Ground the aircraft in the interests of political correctness. Now, did it have any IDE drives on board?

On a jet, by comparison, the fundamental frequency is much greater, and the engines can't be synchronised anyway, so these systems are not worth bothering with.