What you want is a squirrel cage fan blowing into a duct. The end of the duct is covered by a perforated solid plate. It needs to have multiple holes in the plate of different sizes. Each hole acts as a specific air motion noise band pass filter. With the right combination you can approximate pink noise or just about any curve you want to do.
Any decent port program should allow you do do a first pass at the hole design pattern. You will have to verify it with actual measurements, as your fan velocity will be influenced by the load.
Any decent port program should allow you do do a first pass at the hole design pattern. You will have to verify it with actual measurements, as your fan velocity will be influenced by the load.
Same with room modes . . . you can hear them with noise, but they are *much* more obvious with a sweep. Regardless what the theory says different tests expose different things . . . which is, in part, why I want a noise source in addition to an impulse source.
That might work, but I haven't seen an example of it. The McIntosh device (B,B&N device?) does it entirely based on turbulence at the fan . . . which has a clear "simplicity" advantage that I like. If I can find the right fan
Noise doesn't have a given frequency and being random, the phase will also be random compared to something resonant at a single frequency like a bell or room resonance.
As for a sine, if the sine is sweeping the frequency is changing, so it will go in and out of phase with the resonant object or mode. For a slow sweep it will spend significant time in phase, with energy building up in the resonant device. As the frequency changes, it will go out of phase, and the resonance will decay. We're not talking about a given frequency here, except for the resonant frequency of the bell or room.
As for a sine, if the sine is sweeping the frequency is changing, so it will go in and out of phase with the resonant object or mode. For a slow sweep it will spend significant time in phase, with energy building up in the resonant device. As the frequency changes, it will go out of phase, and the resonance will decay. We're not talking about a given frequency here, except for the resonant frequency of the bell or room.
That might work, but I haven't seen an example of it. The McIntosh device (B,B&N device?) does it entirely based on turbulence at the fan . . . which has a clear "simplicity" advantage that I like. If I can find the right fan
Look at the cage vent spacing.
I have a special synthesizer that lets me play as many pure tones as I want simultaneously, each with identical sound power. I start out with a nice 1 kHz tone. The reason I picked that frequency is I have a bell in the room with a resonant frequency that is 1 kHz. Predictably, my tone makes the bell ring.
Then I play another tone, say 2345 Hz, in addition to the 1 k tone. Does that make the bell stop ringing?
I punch in 5 more tones, randomly chosen, between 100 Hz and 5 kHz. Do any of them make the bell stop ringing? How about 25 more tones?
The more tones I add, the more the sound in the room is going to seem like white noise. But it's still just pure tones added together, right? At what point does it actually become noise? How many tones do I need to add to the mix before the bell stops being excited?
Then I play another tone, say 2345 Hz, in addition to the 1 k tone. Does that make the bell stop ringing?
I punch in 5 more tones, randomly chosen, between 100 Hz and 5 kHz. Do any of them make the bell stop ringing? How about 25 more tones?
The more tones I add, the more the sound in the room is going to seem like white noise. But it's still just pure tones added together, right? At what point does it actually become noise? How many tones do I need to add to the mix before the bell stops being excited?
Yes, I'm sure that both blade density and blade contour (profile) make a difference . . . but what, and which favors "broadband" noise? No doubt I'll be grabbing blowers off scrap piles for the foreseable future
. . . finding a specific part at Grainger or McMaster so others can reproduce the "source" would be nice . . .
So what is it then?
Let me retract previous statements I made about reference noise sources no longer being produced or used. I've never had this type of test come up, but this measurement technique is useful for determining sound power of equipment in situ in reverberant spaces. Here is the relevant standard. Whether this is useful for measuring a speaker's sound power, I don't know, but I was wrong in saying these devices are no longer available.
Here are some spectra from a manufacturer:
An externally hosted image should be here but it was not working when we last tested it.
It's even less "flat" than I would have anticipated. But that makes sense, since it has to be a small enough device that you can carry it around with you. What's important, above all else, is knowing the per-band sound power of the device. This is where the difficulty of a home-brewed noise source presents itself, as accurate sound power measurements are difficult without a proper reverb or anechoic chamber.
One thing maybe not immediately intuitive is that a high Q resonance is excited by *all* excitation frequencies - it's just excited more by those nearer its resonance.
And there is a noise that is mathematically equivalent (convolves) to an impulse, it's called an "MLS". Pink noise doesn't work the same.
Thanks,
Chris
And there is a noise that is mathematically equivalent (convolves) to an impulse, it's called an "MLS". Pink noise doesn't work the same.
Thanks,
Chris
"noise is random unwanted perturbation to a wanted signal"
"Electronic noise exists in all circuits and devices as a result of thermal noise, also referred to as Johnson Noise. It is caused by random variations in current or voltage"
The operant word in both cases is "random" . . . which is not a property of continuous tones.
The whine of a jet engine or the hum of a refrigerator is called "noise" in some circumstances, and at least parts of it may be deconstructed to a sum of continuous tones. That is not the kind of "noise" we are talking about here . . .
So what determines what kind of function can be broken down into sines and cosines? In other words, which parts of the jet engine whine are eligible for Fourier transform and which are ineligible?
If it's not possible to break down acoustic noise into pure tone components, what is a sound level meter showing me when I look at the RTA display?
"These sources offer extremely flat response over a
broad frequency range and are ideal for use in such
applications as sound power, sound absorption, and
sound insulation measurements.
http://www.larsondavis.com/docs/REF600_500Datasheet.pdf
Well doesn't that look sort of like what they used at McIntosh . . .
It certainly does. Hence my post explaining that I was wrong in saying such devices are no longer manufactured.
Is the reason the bell does not ring (or windows rattle, etc) with pink noise or MLS because there is simply less power at the resonant frequency? That's what I always figured.
With sweep, there is the total power at a given frequency while the sweep passes thru it. In a noise signal that power is divided up all thru the band, meaning that at any one frequency it's just a fraction of total power.
With sweep, there is the total power at a given frequency while the sweep passes thru it. In a noise signal that power is divided up all thru the band, meaning that at any one frequency it's just a fraction of total power.
A statistical average. If there are "pure tone" components to the noise a spectrum analyzer (using FFT) will separate them out and display them as separate peaks in the frequency domain . . .
A statistical average of what? How does it know how much energy to put into each bin?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.