I don't know a lot about it, and the weighting or importance of frequency vs. audibility. I was shown this plot as a distinction between two different types of components. The second order harmonics are lower on the second component. But the third, and IM are much higher.
It is a complex signal of three different tones, two of them down in level from the first.
Attachments
An audio system is a device that reproduces sound, usually separated into a number of boxes. If you combine them into a single box, does that change the "equation"? I have a kitchen radio, and it squeaks and squawks at times, and the station fades in and out -- the most likely culprit is the tiny full range speaker then? A somewhat out there example, but pointing out that the last step in the chain is, well, the last step in the chain ...
So you want to play semantic games again? Or do you just think speakers are supposed to function as FM tuners?
In any case that has nothing to do with what was under discussion, namely if there is a "signature" to types of speaker drivers. Nor with what is the most important part of the sound reproduction.
If the fades and squeaks or squawks are part of the output the speaker is being fed, then it should be reproducing them. That is the job of a speaker. Not to function as a radio tuner.
The speaker is the primary limiting factor in audio reproduction, because the physical act of creating sound waves is vastly more difficult and has many more limitations, than simply moving bits around or amplifying a signal limited to a pretty narrow range (ie. that of human hearing).
Would you expand on this. I am thinking that the IM is a result of the speaker causing random farting. Not being able to recreate the original signal. Such as a loose bolt in a cabinet. Generally producing sound not related to the original. And harmonic distortion is very much related to the original.
In the loose bolt example I think of the IM as being stimulated by the signal, and then to a lesser extent the harmonics. I am not saying you are wrong, I am telling you how I think of it currently. What am I missing?
HD and IMD are just different views of the same issue - nonlinearity. It has been shown that there is no correlation between HD or IMD and perception. Thats not to say that nonlinearity is not audible, just that IMD and HD don;t tell us anything about it that is relavent to audibility.
This isn't IMD its simply uncorrelated noise.
IMD is completly correlated to the input, just as HD is.
Earl points to this already. IMD and HD have the same cause, that's non-linearity. If there's no non-linearity, there is no HD. If there's no HD, there's no IMD. It's just that simple.
If you measure low HD, you likely will measure low IMD.
I haven't followed the whole thread, and I don't know anything about the loose bolt example, but a loose bolt is no IMD but noise*. Audible, but no IMD.
*@Earl: isn't it correlated? The bolt, or wire, or anything, is vibrating with the excitation frequency. I'm not sure, but that's what I would call correlated.
"Signature" is just another word for distortion: linear is less important than non-linear because it can be compensated for with sufficient DSP -- which is presumably what was done with those tests mentioned earlier. So then weightings need to be applied to the distortion types - some are obnoxiously irritating, others are barely noticeable
Yes, the job of the speaker is to reproduce what's being fed to it. And if what's being fed to it is tainted then the speaker can't do much to fix that ...
"Vastly more difficult" ...?? It's as easy as pie, about as straightforward an action as you can think of; a decent speaker driver can do distortion at 0.1%, of relatively benign nature; solid engineering will give a good result every time ...
Of course, once you combine a conventional audio amplifier with a speaker then this scenario collapses; the normal amplifier is too poorly engineered to do the job properly ...
If its vibrating at the same frequency then it is correlated and then its a resonance. But in general a loose bolt will not vibrate at the same frequency as the excitation, but tends to bounce around erratically and that is uncorrelated. Everything in between is also possible as well - it could be half correlated and half uncorrelated. But the correlated part in this case would usually be linear and hence would not generate any IMD. It could be a nonlinear weak coupling and generate IMD, HD, a correlated resonance, and uncorrelated noise. All things are possible, but generally a loose part will just be mostly uncorrelated noise.
If I wanted to do it "right" first time round I would get drivers shown to have low intrinsic distortion, mount them in an enclosure that's as solid as a bank safe, drive each individually with an adjacent, well sorted out amp, with each amp fed from a high performance DAC, latter fed in turn with DSP corrected spectrum tailored for specific driver, with slopes as sharp as anyone could want. FR as flat as you want, no added distortion; job done, end of story ...
Of course, no-one really wants a system that works correctly, so such an item would die instantly in the market ..
Of course, no-one really wants a system that works correctly, so such an item would die instantly in the market ..
I assume you haven't picked up an issue of JAES since the 1970s or so...
Yes. Those "number of boxes" are the mains and subs. The other boxes should make any sound at all. At least, I have a low tolerance for ones that do (transformer hum, fan noise, vinyl snap-crackle-pop etc.). I suppose may others have a higher tolerance for that sort of stuff.
Correct. I was never in the game, so pick up what's relevant from outside the "circle" ...
If you can give me a precise reference, to a specific article in a particular issue that nailed the current thinking, once and for all, I would appreciate it ...
I will note that in relevant material that I've come across, the aspect of the electronics driving the system is treated as trivial -- phrases like "sufficiently competent" or "low distortion" are chucked in, which of course completely elucidates the situation ...
I don't understand what you mean by this. Levels of IMD and HD don't tell us how audible nonlinearity is?
I'm starting to suspect someone is running a turing test on us with Fas42 and the random sentence generator is breaking down.
Curses, foiled again!
When will they learn...they should make it an oblate spheroid
Curses, foiled again!
Earl, OK, I think I understand.
That's what I mentioned above: you have to apply weighting to those numbers, otherwise you just can tell which device is more linear than the other.
Take two drivers, one produces 0.1% F2 at 1 kHz/90 dB, the other one 0.5% F2. If you compare both, the latter one looks horrible. If you apply proper weighting then you realise that both distortion levels are not audible. So why take the better, probably much more expensive driver? No question that we WILL take the better one, because we are all completely insane
It's the same with IMD, just much more complicated. Which tone modulates another tone at which order of intermodulation? Which signal is best for testing, bass sweep, voice sweep, 2-tone, dense/sparse multitone with high/low bandwidth? Very difficult.
That's what I mentioned above: you have to apply weighting to those numbers, otherwise you just can tell which device is more linear than the other.
Take two drivers, one produces 0.1% F2 at 1 kHz/90 dB, the other one 0.5% F2. If you compare both, the latter one looks horrible. If you apply proper weighting then you realise that both distortion levels are not audible. So why take the better, probably much more expensive driver? No question that we WILL take the better one, because we are all completely insane
It's the same with IMD, just much more complicated. Which tone modulates another tone at which order of intermodulation? Which signal is best for testing, bass sweep, voice sweep, 2-tone, dense/sparse multitone with high/low bandwidth? Very difficult.
I can vouch for the fact that Frank is a real person. (But maybe I'm just another voice from the same Turing machine, with a different login ID, as far as you know.)
Indeed, it does.
Smart people treat things that are trivial as trivial. Marketers and the ignorant treat things that are trivial as being of great import.
The former group have a reason to massively exaggerate the trivial: it helps them push whatever they're selling.
The latter are just sad.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.