You might use FFT to approximate a filter, but generally you would still use filters to break out your bands.
I think so.
Assuming that by "random noise" you mean pink noise, you're mistaken. Pink noise will excite the bell continuously while a sine sweep will only excite the bell for the moment the sweep is in the bell's resonant frequency.
Pink noise is equal power in all octaves (or 1/3-octaves, or 1/n-octaves). That means it has power in whatever frequency excites the bell. So the bell will be excited.
Why then the "perfect" this and "perfect" that in your comments?
Perhaps it doesn't . . . that was the question. But a squirel cage on a motor need not be expensive, and there might be alternate ways that you don't know about as well. Things might exist that "[you] don't see" . . .
"Walking through" your process I see your problem . . . it is different from (and doesn't address) what I'm asking.
I'm not, as the original post should have made clear, looking for "looks similar". "Looks" are not the issue, or the desired function. Ultimately, I suppose, addressing the flawed power response of most DIY speakers, and the flawed absorbtion profile of most listening rooms, is. Discussing a "constant power noise source" (and all the implications and issues surrounding it) may lead people to think about those issues in useful ways that "impulse response" almost certainly does not.
Yes. But whether (or not) the results are identical in practice is a different question, as each method imposes its own "flavor" on the data. "Noise" will excite the bell, but since it is of random phase I suspect it will not excite it in the same way as a sine wave sitting on the resonant frequency.
I really am trying to be helpful here. I don't understand why you can't just quote what I said. This is silly so let's just go ahead and nip it in the bud now. Here's everywhere in this thread I used the letters p-e-r-f-e-c-t together.
...and that's it. So where are all the "perfect" this and "perfect" that that you're referring to? Nowhere. So knock it off.
You're missing the point of a detail that doesn't even matter that much. Pedantic, much?
Decay time is decay time. What difference does it make what the nature of the source is for measurement?
You don't have to use an impulse source if you don't want. I've explained multiple times already that it's common practice to use a pink noise build-up and cut off to measure room response. You don't have to go anywhere near a balloon if you don't want to. Maybe they scare you due to a childhood event involving a clown? It doesn't matter. Use pink noise and you won't have to worry about it.
OK, original question...
This is what you asked for:
And it doesn't exist, as far as anyone who has contributed to this thread knows. People had a few ideas but you dismissed them all.
I didn't get to this thread until post #27. It seemed appropriate to indicate, based on my professional experience as someone who does these kinds of tests for a living, that there's no point in you seeking out this magical device. If you really want such a device, don't let me stop you from looking, but I can't imagine you'll ever find it. And here's why: There's no reason for anybody to make one. Not that I can think of.
If you can't find one, but you still really, really want one, you can build/measure your own. Pretty much any fan will do, as long as you can get it to spin at a known speed. Shouldn't be a problem, since electrical grid frequency is reasonably constant.
Then measure it. You'll have to go through the process of characterizing the room you're using through RT tests, but fear not, you'll only have to do it just this once. Measure the room you're in, then measure the fan. Back-calculate the sound power in each band and you'll have your device.
You will not find a device that has even sound power across the spectrum. I will try to save you time and tell you not to look for it, we'll see if you listen. But you don't need it to be even across the spectrum. I don't know why you think you need that, but you don't. As long as levels in adjacent bands are within ~10 dB of each other, they will behave independently.
Thank you for that . . .
Since you imply "none" then why the objection to using a known noise source?
Well, perhaps not a "perfect" one, but how about one "as good as" the one they had at McIntosh? Do you have any reference to the performance (how "good" or "not good") of that one?
[if=""]
That was the point of the original question. In the absense of any help, thanks at least for the encouragement. But, as I commented to Pano, I would rather skip the "project" (of making something that "doesn't exist", even if it once did) and get straight to the use . . .[/if]
His objection seem perfectly clear to me. It isn't needed with today's hardware and software and will be more trouble than it's worth.
Would be fun if you found one, but that doesn't look likely.
I guess the point is, how would you use it?
Say you have the exact same fan as used by macintosh, what next?
What kind of microphone do you plan to use with this setup, and what kind of measurement instrument? What are you even measuring?
How can you measure RT60 shorter than the fan takes to slow down? I think a loudspeaker source is going to be much more useful for real room measurements because it stops faster than a fan. Directivity doesn't matter much in a reverberant test room, but it does in a real room, and I doubt that fan is effectively an omnidirectional point source at anything but very low frequencies.
Deward,
take a look at Room Equalization Wizard and Holm Impulse.
Two free tools that are extremely powerful and easy to use.
Speaker workshop can also do impulse measurements.
You could even do measurements with Audacity - record a balloon pop, or a hammer blow on a piece of stone, or a squirrel cage fan
I have SoundEasy and would not recommend it for a beginner, it is really a bewildering kluge of tools with no discernable workflow - it works if you essentially understand how to write a program to do measurements yourself...
take a look at Room Equalization Wizard and Holm Impulse.
Two free tools that are extremely powerful and easy to use.
Speaker workshop can also do impulse measurements.
You could even do measurements with Audacity - record a balloon pop, or a hammer blow on a piece of stone, or a squirrel cage fan
I have SoundEasy and would not recommend it for a beginner, it is really a bewildering kluge of tools with no discernable workflow - it works if you essentially understand how to write a program to do measurements yourself...
The RT60 measurement is different than the known sound power measurement. You use the sources in a different way.
For the RT60 measurement, you excite the room with your source (whether impulse or pink noise), remove the source (automatic for an impulse, switch off the source for pink noise), and watch the SPL drop in each band with respect to time. Your room response is a function of the time it takes the SPL to die down by 60 dB in each band.
For the McIntosh known sound power measurement, you simply leave the noise source on and measure the resulting steady state SPL. Your room response is a function of the known sound power and the measured SPL. It's an easier measurement to take, but you have to know the sound power of your source, and it's not likely to be as accurate. I also think this would require a very reverberant room to have a chance of working correctly, whereas RT60 can be measured in live rooms and dead rooms.
I don't, but in the 7 years I've been doing mechanical noise analysis I've never seen a fan with a flat Lw spectrum. It's possible they tweaked it somehow to get a more specific Lw spectrum, but that would take a lot of effort and that unit is almost certainly 1 of a kind.
I certainly agree with that
. . . I've had it for years and still find many aspects confusing or difficult (and if I don't use it for a couple months it's re-learn all over, 'cause *nothing* makes the kind of sense that "sticks with me". OmniMic is much more direct and easy, and since it uses an "external" sound source (on CD) it doesn't require setting up any kind of "jig" or "loopback" at all. Just plug the mic into the laptop and go. Kind of lends itself to the "external noise source" method Yes . . .
Yes . . . although that presumes an impulse loud enough to be 60dB (plus headroom) above the existing noise floor. If you switch to RT30 you lose considerable accuracy.
Yes . . . the "easier measurement to take" part is one of its (potential) charms. It is also correspondingly easy to compare to a loudspeaker (driven by pink noise) in the same environment to get a direct measure (or at least reasonable sense) of the loudspeaker's power response. The simplicity is one of the technique's advantages
I don't find RT60 all that easy in small dead rooms . . . resolution is poor . . . and as with the "noise source" method it can be position dependent (in fact I can see no reason the two methods should be any different in that respect). A "very reverberant" room (like the McIntosh room) would reduce position dependence, I would expect, and be better for measuring loudspeakers.
There we go with the implied "perfect" again
. . . there is significant import to how far from "flat". Within a dB or two is probably as good as any such measurement is going to get anyway . . . and that's easy to accomplish over a more limited range (say three or four octaves), even with a small cone driver. "Flat" power from 100-1000 Hz. is easy with $20 in parts . . . it's extending to 10,000 that gets a bit difficult. I'm sure that's why they took the "'cage" out of any housing and ran it in "free air" where the turbulence can follow its own rules (which, as I recall, remain fairly uniform over a wide range of Reynolds numbers).But the ability to actually listen to a noise source in direct comparison to a loudspeaker, rather than just look at lines or numbers on paper, is a considerable advantage. My experience is that we hear and think about hearing in quite different ways.
Last edited:
Not difficult at all. I do these measurements in places with background levels of 40+ dBA sometimes.
What are you talking about? Peak Lp is the same in both a live room and a dead room. Down 60 dB is the same in both a live room and a dead room. The only difference is how long in between. A low RT is 0.4 seconds, are you saying your computer isn't fast enough to divide 0.4 seconds into meaningfully small pieces?
It needs to virtually eliminate position dependence to be useful. If the room is sufficiently reverberant and diffuse (no parallel walls!) as long as you stay out of the near field of the source you're OK.
FFS! Enough with that stupid BS!. I said repeatedly that you don't need it to be flat. I said you need to have your bands within 10 dB of each other, how could you interpret that as saying it needs to be perfectly flat?There we go with the implied "perfect" again
The flat spectrum requirement comes from you. I don't understand why you think it needs to be that way. You've never explained it. Any ideas on what is close enough to flat are yours alone, because that requirement exists only in your universe.
You asked for my opinion on the McIntosh source and I gave you a response in your terms. Other than the shape of the spectrum, what other performance terms did you expect me to comment on?
I told you you can use almost any fan, as long as you know its sound power. There's no "good" or "bad" in my explanation. There's only "known" or "not known." In your world flat is important, so that's what I commented on.
Stop beating up things I never said. It's ridiculous. Why do you make it so hard for people to help you?
OK FANTASTIC.
So... how does that help you measure a speaker's Lw? Isn't that what this is all about?
There we go with the implied "perfect" again
Let me refresh your memory . . .
And I asked "how far from flat". Your tantrum obscured your answer (if there was one).
Thanks. Sorry all your "experience" couldn't translate into any useful information. I remain hopeful that more DIers will become aware of the importance of loudspeaker power response, despite the fact that few currently have any way to measure it. The advent and availability of simple tools for measuring on-axis response has led to significant improvement in DIY loudspeakers . . . the next step remains to be taken . . .
Never tried it, but I suspect a good impulse source would be a pellet gun with no pellet. Not a spring piston type, which can't be fired "dry", but something like the Daisy where one pump pressurizes a chamber. They make a nice sharp pop when fired, and quite loud.
Forgive my fascination with the difference between measurement types, but I remain unconvinced they are equivalent. A pink noise source, if it's truly random, should put as much constructive energy into a resonance (bell or whatever room resonance), as it does destructive energy. I don't see how it can ring the bell. A slow sine sweep OTOH will go through a region where a large amount of in-phase energy can transfer to the bell, causing it to ring. It will then take just as long for the out-of-phase energy to stop it. The measurement results on resonances should be different than using a noise signal.
Other than small errors and the frequency contribution of the sweep (you can't sweep without changing the waveform!), a sine sweep is near zero bandwidth. Noise and FFT isn't going to be that narrow, much less 1/3 octave, so each should emphasize different aspects of the room behavior.
Forgive my fascination with the difference between measurement types, but I remain unconvinced they are equivalent. A pink noise source, if it's truly random, should put as much constructive energy into a resonance (bell or whatever room resonance), as it does destructive energy. I don't see how it can ring the bell. A slow sine sweep OTOH will go through a region where a large amount of in-phase energy can transfer to the bell, causing it to ring. It will then take just as long for the out-of-phase energy to stop it. The measurement results on resonances should be different than using a noise signal.
Other than small errors and the frequency contribution of the sweep (you can't sweep without changing the waveform!), a sine sweep is near zero bandwidth. Noise and FFT isn't going to be that narrow, much less 1/3 octave, so each should emphasize different aspects of the room behavior.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.