Something to think about...

[tonym]

I don't have the time to experiment much lately, but there are a lot of bright young folks out there that are students with abbudanza curiosity (and some time on their hands) that might have the time to do some experimentation on REALLY reducing the noise generated by a PC, and then pehaps MARKET their discovery for all PC users to utilize.

Consider that active noise cancellation (ANC) has been understood since the 50's and implemented since the mid-70's early 80's (anyone who's seved on a flight deck knows what an ANC microphone and headset are!). Besides military applications, ANC is routinely used to lower ambient noise in music studios and in anechoic chanbers, and it is used to greatly reduce exhaust/muffler noise in luxury cars.

Frankly, I think ANC can be very effectively used to greatly lower the fan/cooling noise in PCs. I'm guessing the reason it's not implemented to production PCs is cost...manufacturers trade a little noise versus the extra cost of eliminating it.

Essentially ANC for a PC would work by sampling the ambient environment (at the air exhaust port of the PC) with a microphone, amplifying this signal appropriately and feeding-back this signal anitphase (180 degrees out of phase) which acts to subtract it from the ambient (with a loudspeaker -- probably a 4-6" midrange speaker). The result is a tangiable noise cancellation. Think of the implementatio as a pipe that the exhaust air passes through into which a speaker is introduced (in the side) to provide the antiphase noise cancellation.

It's not quite as easy as I describe it...there are stability and loop dynamic characteristics for the feedback system and the amplifier, but it is very much possible, and the only thing standing between the description and successful implementation is some elbow-grease experimentation.

I even think that digital signal processing (DSP) technology can be applied this problem to add "personalizing" so the ANC could be used with any PC configuration and the system would auto-tweak itself (automatic gain control AGC!!).

Anyway, I think that a problem like this is right up the alley of some of the inquiring minds out there. I throw out the idea to you to use as you will...I'm not looking for ANYTHING from this. Just to perhaps be able to BUY a finished product for my PC some day!

I you're interested in the pursuing this...I'd be glad to give you further insights if you wish. Just drop me an e-mail!

Tony

[EastSt]

Tell you what TONYM...
Since you obviously are so knowledgeable
on the subject.......
And have so much free time...
I volunteer you to do it
and report back to us.

[tonym]

Hey EastSt,

I was looking for someone that had a measurable IQ to reply. Constructive comments please, otherwise www.moron.com. (http://www.moron.com.)


Tony

[psyklone]

tonym ... interesting you would bring this up, i'm working on a program that essentially does the same thing ...

i'm working on writing a piece of software that attenuates DOS attacks by slowing down an incoming request and replicates it, sending back out to the originating host. the downside so far is that it requires running a separate server for the program (operates on a system that runs as a router ... doesn't require a real beefy system) and that the majority of it so far is theory and i'm working out some kinks. another downside is that so far it doesn't operate with connectionless protocols like UDP at this point. but anyway ...

a couple of problems with implementing this (ANC) in a running any type of anti-phase noise cancellation inside a pc is this: a standard system (off the shelf, so to speak) is going to have only one system fan (aside from the power supply fan) and that won't generate much noise at all. the systems that are the noisiest are the ones that are overclocked or need extra cooling for one reason or another. that being said, to introduce any type of acoustic signal amplification circuit into the pc would be counterproductive because one of the most efficient ways to amplify a signal and have it reproduced cleanly is through a push-pull circuit .... and these generate plenty of heat. as a result, the owner of the machine would probably need to add an additional fan or two inside the system to offset the heat generated by the ANC implementation.

anyone else have any feedback?

cheers,

psyklone

<edit> oh yeah ... and the shape of nearly all computer cases would cause a problem with standing waves produced by any type of acoustic signal regeneration, even on a small scale so chances are it would also require some type of noise dampening lining on the inside of the case anyway. i'm not saying it can't be done, because i'm a firm believer that nothing's impossible, i'm just saying it may not be able to be done by conventional methods.</edit>

[This message has been edited by psyklone (edited 05-02-2001).]

[barbados]

I'm an electronics engineer by training and I think the ideas have merit.

Noise cancelling is difficult with short-burst, varying amplitute noise, varying frequency noise, but should be achievable with the constant drone of a computer fan.

Let us know your progress and GOOD LUCK!

[dcc1977]

You know, I never thought about that. And your base idea, from what I understand of sonics is correct.

Considering I'm heading back to the University of Missouri-Rolla for a masters in computer engineering to further my studies with microcontrollers and FPGAs... I think I may be able to come up with something during my free time.

I mean mic's and speakers are cheep, the problem is I don't know how much of an effect speaker qauality would be in it.

Thanks for the idea regardless, I'll try to try it out sometime.

[tonym]

psyklone,

One could use a class D switching amplifier. These have efficiencies 90+%. They're used in high-fidelity systems (Bose!!) and have a wide dynamic range and very low distortion if switching frequency is just high enough.

I've actually done this task (the ANC/noise reduction thing) for another source several years ago...it was military and complicated! But the result was a 10kHz or so bandwidth with a typical attenuation of 18db across the board (and greater in certain specified ranges!).

I was sitting here reading the numerous "quieter PC" posts and listening to the drone of my computers and I just thought that someone a little younger and more ambitious than me would maybe like something to try ANC rather than the same old "get a quieter fan and muffle the case..." approach!

Go for it dcc1977 -- this is a perfect application for adaptive signal processing using uC's (PIC seies is a good example) and/or DSP's. Since most of the noise-power bandwidth is below 15kHz, this fits in quite well with an inexpensive, parts-available electronics solution.

BTW, you already know it but U Missouri/Rolla is a real good school. They turn out a lot of good Reliability and EMC engineers!!


Have fun...

Tony