starting to think about pump sound measurement

[bobkoure]

This may or may not belong in this forum as I'm probably not going to end up with results that can be graphed or absolute dB readings - but this review at cooling-masters (auto-translated to English by Google) got me to thinking that comparative sound recordings might actually be useful (and are something I can produce without access to an anechoic chamber).

I'm thinking that the exclude-external-noise box idea in that review is good for a start, but that I'd, at least
- make it larger internally
- make it a bit better at excluding external noise (so probably a couple of layers of foam with "limp wall" between.
- try to reflect as little pump noise as possible (i.e. try for a "near field" recording).
- try to diffuse what sound I do reflect - try to avoid reflective sound re-enforcement.
- control vibration transmission with something like rubber band suspension (like the silent-pc folks are doing with their hard drives).

I really don't know much about microphones (but have a friend who owns/operates a recording studio and he may have answers for me) and know even less about using a PC as a recorder (which, lacking a digital recorder is probably what I'll do).
It seems to me that there at least used to be PC sound recording software that did a differential between two audio channels - the idea being that you attach a microphone to one side of a stereo audio card, leave the other with nothing, which then has whatever (electrical) noise being generated in the PC - software does the subtraction of noise from the microphone-attached side. Anyone remember something like this? What was it called?

I'm just at the "ideas" stage with this, so would really appreciate any input. BillA, if you're still lurking, please email/pm me as I know you have some "hands on" experience trying to get decent sound measurements. Feel free to call me an idiot for trying to sidestep the whole "get dB measurement without an anechoic chamber" issue

Meanwhile, I guess it's time for me to re-read the F. Alton Everest sitting in my bookcase and see what I've forgotten. If anyone's interested, I'd suggest starting out with the Master Handbook.

Credit for the original idea goes to David D. at cooling-masters.

[TerraMex]

just keep in mind that pumps in a full loop (wb(s)+rad+tubes) behave differently than on an open or low resistance loop.
testing a pump should be with a standard loop, imho.
by standard i mean one waterblock and a radiator besides the pump, at least.
gopher it!.

[Roscal]

Sound recordings go to no restriction at full restriction to have the full noise range because rotation speed changes with DC pumps, pinch-clamp on the tube isn't on the pic.

[bobkoure]

Originally Posted by Roscal

..full noise range because rotation speed changes

I suspect there is more to the noise output / restriction relationship than just rotation speed - but, in a way, it doesn't really matter, so long as you get the full range (well, not really zero-head - not actually clear how to get an actual zero-head reading if you think about it).
Yes, I know David did this (restrict via pinch) - it's in the article text. No reason why he couldn't have just used his fingers to pinch the tubing rather than use a pinch valve.
TerraMex - I'm not sure there's such a thing as a "standard" loop - and if there is, whether it'd be the same from year to year. IMHO David did this right.
PS: I sent David an email thanking him for his article and pointing him at this thread in case he was interested... I tried to do it in French, and we may hear from him after he's done laughing at it - I'm pretty bad and Systran didn't seem to help a lot

[TerraMex]

Well ... i meant one waterblock, one radiator, pump and about a meter , or meter and half , of tubing. A more or less "standard" loop. As the waterblock, any that has an acceleration nozzle. Heatercore rad ?.
At least, i'd like to see it, gives , IMO, better perception on noise levels than a fullscale testing.
You're simply not going to have a loop with so little restriction that the pump is close to full head, neither one close to zero head. But hey, i'm just saying.

A ball valve is another way to go. Pinch clamp? isn't that a little too ghetto?
Or a motorized valve. I bet you can find a few within "the decent price range" on Ebay.

Please, no french literature.
"je ne pense pas que c'est un bon idée"
things might get a little ... er, murky.

[Les]

Originally Posted by bobkoure

PS: I sent David an email thanking him for his article and pointing him at this thread in case he was interested... I tried to do it in French, and we may hear from him after he's done laughing at it - I'm pretty bad and Systran didn't seem to help a lot

Roscal is David.

[Happy Hopping]

I am never a big fan of puting a pump, aka, an EM current device, inside a computer, for obvious reason. I hate to see it dangle outside the computer, but it make me feels better.

[bobkoure]

Originally Posted by Les

Roscal is David.

I got an email back that he's not only 'way ahead of me here but working on something that sounds very good.
I'm assuming that it being email and not a post means I shouldn't quote him directly here - but I'd suggest keeping an eye on cooling-masters. (Roscal - please let me know if I've gotten my assumption wrong).

There's still the issue of calibrating PC speakers, which I can only think of a couple of ways to deal with:
- If there's enough "headroom" in the recordings, include a 50dB white noise snippet - most inexpensive sound meters go this low, so you could literally calibrate to this (like you do with a home multi-channel audio system)
- Include a sound "everyone knows" - which is what Roscal has done in the Eheim 1046 and 1048. I'm afraid these pumps aren't as ubiquitous in North America, so maybe something like a Panaflo L1A - although that might be problematic in a sealed box (wind noise)
- Include a sound that goes with a cheap electronic device - like those audio chips that sometimes come in novelty greeting cards. Probably 0.25 or less in built - and very mail-able.

Anyone got any other ideas...?

[TerraMex]

well ... i would like to see some testing with something like a NoiseBuster to see if the overall system noise can actually be countered properly.

[JWFokker]

Originally Posted by Happy Hopping

I am never a big fan of puting a pump, aka, an EM current device, inside a computer, for obvious reason. I hate to see it dangle outside the computer, but it make me feels better.

Why not put it in a metal enclosure and ground the enclosure? Actually, since I was thinking about putting my MCP350 in a box to make it dead silent, I should probably integrate that. Now which metal is supposed to be better at blocking EMI? I can't remember if it's copper or aluminum. One's good for RF, the other EM. I can never keep them straight.

[TerraMex]

copper then.

btw:
http://www.parker.com/chomerics/tech...ng_methods.htm

[mwolfman]

My sugestion:
Ad a ballvalve, a flowmeter and a pressure meter.
Set up some diffrent flows (or pressures) that you want to test, and then do all the test whith that flow (or pressure)...

For test #2: Ad a plate to put the pump on (dont mount it), say a 200x200x1 steal plate, (if the pump vibrates then the case will make loads of sound even if the pump it's self is quiet)

[Marci]

Hmmm mic factors largely. Need super/hyper cardioid pickup ideally... Lots of non reflective insulation directly opposite in the cardioid pattern. Off to sides won't be picked up.

BUT

Down side - cardioid mics add bass at close proximity. Not good for nearfield.

Large diaphram = more artifical bass.
Small diaphram = less artificial bass but commonly cardioid pattern so nearfield get woofy.

BEST for job in 1mtr cubed insulated enclosure... ribbon mic. Coles 4038 / Royer Labs R-121

Downside = Expensive and VERY fragile. BLOW on ribbon wrong way and death.

Look to condensers, small diaphram. Hypercardioid. Recommend borrowing large diaphram condenser - spectral analysis... find the critical frequency range... depending on high and lo cutoff, can then check suitability of small diaphram. Spectral analysis of live source vs recorded should be done. Ensure what's getting recorded is what's getting heard. Quality of AD Converters becomes crucial... MOTU or better.

Budget db measurement these days is down to 40dBA... calibration hopeless. Proper calibration ££s. Ignore dBA unless comparative known baseline specified. Generate testtone using soundforge or similar at fixed dba as reference point (prob already suggested somewhere - skimmed thru)

[mwolfman]

I read a french pump review (there is a link somewere on the forum), that test had an mp3 recording of each pump... realy usefull for a system builder (diffrent pich gives diffrent sound absorbation materials)

[bobkoure]

That's most likely Roscal's work.
You're right about pitch and sound absorbtion materials. Actually it's sound absorbtion material depth - which you want to be at least as thick as a half-wavelength of the sound you're trying to attenuate. Higher pitches have shorter wavelengths and don't require as deep absorbtion material. Of course, what does get though of the higher frequency noise is more irritating (IMHO, of course)

[mwolfman]

If you have a high pitch sound then the sound can be stopped with just foam, but if you have low pitch sounds then you need to ad some weight (asphalt paper) combined with foam. Please correct me if I'm wrong.

[Roscal]

You're right, the lower the frequency is, the stronger/heavier the material should be (more energy to dissipate). Some sound absorption sheets are made of a sandwith of different density material to work on a wider spectrum.

[bobkoure]

Have a look at F. Alton Everest's Master Handbook of Acoustics.

Absorbtion (interfering with reflection) and energy transmission and re-radiation are two different things.
You don't get the re-radiation with higher frequencies because the resonant point of the wall is too low to really pick up much energy (adding a heavy material just moves the resonant point down).
Absorbtion only works when the sound energy is molecule velocity - it alternates between velocity and pressure over the course of one wavelength. If you've got a reflective surface, then the "velocity" part of the reflected sound energy is going to be one half wavelength out (and again one and a half wavelengths out, etc.) Something air-restrictive at that high-velocity point should be effective at converting velocity-energy to heat-energy.
Of course, you don't have only one frequency.
Look at the bright side, you're not worrying about varying speeds of sound in heated gasses and gas inertia (I did a stint 30+ years ago designing race bike exhaust systems that would pass noise regs).

[bobkoure]

Originally Posted by Roscal

Some sound absorption sheets are made of a sandwich of different density material to work on a wider spectrum.

I've had best luck with "limp wall" sandwich foams. A layer of foam (sometimes with composite or tar to go against the metal wall and lower resonance frequency), a flexible membrane, then sound absorbtive foam. A good portion of what makes it to the membrane converts to mechanical energy and is then dissipated by the "lower" foam (and what little makes it past that has to deal with the lower resonant point of the sheet steel + asphalt). The sound that reflects off the membrane is then absorbed by the foam, if the wavelength is short enough. The problem is that, to be effective, this stuff is usually 2"+ thick - which is inconvenient for the inside of a PC case.
I used to use something called "Black Hole" acoustical foam (limp wall layered) but it seems to have disappeared off the market...(?)