Reply-To: "Kevin Aylward"
From: "Kevin Aylward"
Subject: Re: Bullshit wins v. science
X-Newsreader: Microsoft Outlook Express 6.00.2800.1106
Date: Sat, 9 Nov 2002 11:38:15 -0000
NNTP-Posting-Date: Sat, 09 Nov 2002 11:38:24 GMT
"N. Thornton" wrote in message
> "Kevin Aylward" wrote in message
> "N. Thornton" wrote in message
> >> The approach and industry practice you describe for measuring
> >> distortion only measure certain types of distortion. They do not
> >> _define_ distortion.
> >In practice it does. Its what is meant by using the term distortion,
> I agree that is what it is normally taken to mean.
> I think it thru beyond what is printed in the text books,
And what *advanced* books on acoustics have you read?
> because I
> discovered this issue when trying to produce a super efficient
> speaker. I went for a resonant design for best efficiency, and
> encountered a problem that made me rethink the generally accepted
> concept of distortion.
With all due respect here, there is no realistic chance that you will
have "discovered" something or come up with something never seen before.
> Resonance causes a non flat freq response, which can be corrected, but
> it also causes something else.
>Resonance does make what you hear
> different to the signal analogue, and that dfference is conceptually
> distortion. I accept we probably won't agree on that. More later...
There is no problem at all in considering aberrations other than THD/IMD
as "distortion", *provided* one specifically points out that one is
discussing, e.g. phase, frequency, delay distortion etc. Using the word
"distortion" unqualified was the issue. Not that distortion could not
refer to other forms of aberrations.
> >For example if someone was referring to phase aberrations,
> >on would have said "phase distortion".
> Sure, granted. You're talking about accepted conventions there, I am
> going past them. I am saying accepted convention is not the whole
> picture, and although accepted, is not entirely accurate.
I think we are missing something here. We're discussing terminology for
known effects, not that effects *other* then say, THD and frequency
response are unknown to everyone but you.
> >The issue here is that you are simply not conversant with standard
> >engineering practise.
> Sure I am. I just believe there more to it now. My muddy speaker
> proved that to me.
It does not prove that what you are discussing is not known about. It
only means that you are using an incorrect name for the effect, if one
> > Lets take a really clear example to illustrate. Lets say you delay
> > your audio signal by 1 second twice, and add the 3 signals: the
> > original one, the 1 sec delayed one, and the 2 sec delayed one. Any
> > speech put through that would be unintelligible. According to the
> > _definition_ of distortion, it is indeed distorted, badly distorted.
> > But if you measure the THD of this distorting process, it is zero.
> >In fact if had taken probably any course on this in university,
> I did.
imo, it don't seem like it cos your terminology seems at odds with
> >example, it would have been pointed out that delay distortion is not
> >usually refered to as "distortion".
> Sure, I quite agree. But the example I gave still is distortion, even
> if its not what's measured and referred to as distortion in the
But you have to *qualify* what *you* mean by distortion. If the norm is
to reserve "distortion" for specific effects, you cannot *dispute* an
argument based on that understanding. Sure, no one denies that "delay
distortion" is not a valid and common term used in he field. However, if
you whant to refer to delay or frequency response distortion, you must
explicitly say so if you want people to understand what you mean.
> >THD has never implied
> >anything about what the frequency response of a speaker is.
> The delay/ time spread type of distortion caused by resonance, and
> uneven frequency response, are 2 separate things:
Ahmmm... Did you ever actually do a course in Fourier Transforms? Or,
indeed in passive networks?
Since by assumption we are not discussing non-linear amplitude
distortion, this issue can be addressed completely by conventional
linear analysis. A linear network can be described in terms of its
impulse response f(t), or by the Fourier transform g(w) of its impulse
response. g(w) is a complex transform and contains all the information
that there is to know about the time delay response and frequency
>I am talking only
> about the former, not of anything to do with unflat freq response
It is certainly true that a system can have a flat frequency response,
but still have a varying phase response with frequency. e.g.
And it is also true, that there is an asssociated group delay with this
phase response given by Tg=d(phase)/dW
> Of course resonance is prone to causing both of those. But they can be
> separated, and are 2 different effects of resonance. I'm not sure I'm
> putting this too clearly though :)
Yes, one can certainly examine the frequency response and the
phase/group delay response of any system, all characterised by g(w).
This is all routine stuff, as noted above.
> Have you ever tried putting sound through a resonant transducer and
> flattening the frequency response? When I did, I discovered something.
I think you were mistaken.
> Although the final freq response may be flat, or near flat, the sound
> is still, to the ear, very distorted. It has a 'muddy' blurred'
> character. It is harder to make out the words, everything sounds a
> mess. But the frequency response was perfectly reasonable, and the
> linearity quite good.
Ahmmm...Well I would certainly need to know the set-up in more detail.
By and large, a proper set up bass reflex system with horns,hf system
usually sounds very good. Its why they are used. Any cursory
examination, shows that a typical bass reflex system with HF horn sounds
much better then a single speaker.
If the system was highly resonant, this would mean a highly
anti-resonant filter to correct it. I would be surprised that one would
be able to match such resonance's in practice due to inherrent errors in
Certainly, flattening out the frequency response may introduce phase
errors, and these phase errors can cause frequency response
cancellations and additions when speaker outputs are finally summed at
the ear. Indeed, a good x-over system is designed so that this does not
> The cause of this, what I would call distortion, is that the sound
> reflects back and forth across the speaker surface, and thus an input
> at time t comes out at times t, t+1, t+2 etc. The example I gave above
> is a simplified version of this type of distortion.
Of course, the transfer function of any system could, in principle
involve delay terms, such as exp(-s.tau). This usually results in comb
filter effects when such delays are combined. If the delay is very large
you might well notice an echo. I would have to do the numbers, but off
hand, I'm inclined to take the view that the delays involved in a bass
reflex system are such that if you truly equalise out the frequency
response, they should be insignificant.
> What I was attempting to say originally, and maybe I was being muddy
> about it myself, is it is this type of distortion that bass reflex
> speaker cabs cause at low frequencies. And its for the same reason,
> because they are designed to be resonant at the bottom of their f
> And that is why the bass from a bass reflex cabinet always sounds
> different to that from a simple closed cabinet.
Yes they do usually. They have different frequency response!
>It is less clear. A
> bass note will have slower attack, and delayed decay.
This does not really make sense. Clarity is usually due to the higher
frequencies, which are not carried by a the bass reflex at all.
Secondly, it would seem your comparing apples and oranges. You have not
convinced me at all that you have set up a closed back system to have an
identical frequency response to that of the bass reflex.
> >Your pissing in the wind. An engineering definition of distortion is
> >the same as an English language definition of distortion. Try looking
> >the definition of an "Observable" in Quantum Mechanics, e.g.
> I agree the accepted eng. concept of it is not what I have given. What
> I am saying is that the accepted concept isn't the full story.
I doubt it. The accepted concepts are such as "Distortion", unqualified,
to mean THD/IMD, and "phase distortion", "frequency response
distortion", "delay distortion" etc to have their obvious meanings. All
of these concepts are very well understood and *fully* appreciated. The
issue here was that you were referring to the use of the word
"distortion" in such a manner, that was not the norm. Because of this,
you are apparently claiming that you have discovered something new. This
is daft. Industry simply has different words for well known effects that
you are, apparently, unaware off.
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.