"Phil Allison" wrote in message
...

"Porky"

"Phil Allison"


** Wrong direction Porker - yours is only about 85.

Plus you are autistic.

Learning to sing yet ???

Wrong, Phool, my IQ is a bit over 140 to a bit over 150, depending on
the
IQ
test used. Modest being that I am, I use the lower figure.
Yes, I am artistic, I am a musician and I paint on occasion, and I just
love
your southern drawl.


** The word was "'autistic" - which proves my point.


Ahhh, "autistic": "A psychiatric disorder of childhood characterized by
marked deficits in communication and social interaction, preoccupation with
fantasy, language impairment, and abnormal behavior, such as repetitive acts
and excessive attachment to certain objects. It is usually associated with
intellectual impairment."
Yes, it describes your behavior here quite well. I think you hit the nail
on the head with your self-diagnosis!

"Porky" writes:

"Jim Carr" wrote in message
news:1gVUc.10087$yh.1675@fed1read05...
"Arny Krueger" wrote in message
...

That's because the train is not moving back and forth in front of us, in
a
periodic sine-shaped pattern.

You OBVIOUSLY have not seen the trains in America!

Yeah, we have AMtrack, not FMtrack....

Ba-domp - bom! play horn notes here
--
% Randy Yates % "Ticket to the moon, flight leaves here today
%% Fuquay-Varina, NC % from Satellite 2"
%%% 919-577-9882 % 'Ticket To The Moon'
%%%% % *Time*, Electric Light Orchestra
http://home.earthlink.net/~yatescr

Mike, you are just feeding this troll and it doesn't
compliment you. I suggest you plonk him as I and about
everyone else has.


Bob
--

"Things should be described as simply as possible, but no
simpler."

A. Einstein

Porky wrote:

"Jim Carr" wrote in message
news:1gVUc.10087$yh.1675@fed1read05...

"Arny Krueger" wrote in message
...


That's because the train is not moving back and forth in front of us, in

a

periodic sine-shaped pattern.

You OBVIOUSLY have not seen the trains in America!


Yeah, we have AMtrack, not FMtrack....

Ouch! :-)


Bob
--

"Things should be described as simply as possible, but no
simpler."

A. Einstein

Porky wrote:

If you mean that a whistle riding on a moving train, or a speaker
swinging back and forth (or spinning round and round like in a Leslie
speaker system) in a repeating oscillation cycle will produce Doppler shift,
but a stationary speaker reproducing a complex waveform containing a mixed
LF and HF tone (or any multiple combination of tones, as would be the case
in a complex musical waveform) won't produce Doppler shift, then, by golly,
I think you're right!

I even finally agree that in many cases it has the
properties that can be called Doppler distortion, so it is a
real phenomenon in qualified conditions but not for the
reasons usually given and not from the systems for which it
has been claimed.


Bob
--

"Things should be described as simply as possible, but no
simpler."

A. Einstein

"Bob Cain" wrote in message
...


Mike, you are just feeding this troll and it doesn't
compliment you. I suggest you plonk him as I and about
everyone else has.


I have to reluctantly agree with you, but it is fun to one-up him every
time he throws an insult. However, this guy is so easy, it isn't even a
decent mental exercise, so I'll just send him to ignore-hell.

"Bob Cain" wrote in message
...


Porky wrote:

"Jim Carr" wrote in message
news:1gVUc.10087$yh.1675@fed1read05...

"Arny Krueger" wrote in message
...


That's because the train is not moving back and forth in front of us,
in

a

periodic sine-shaped pattern.

You OBVIOUSLY have not seen the trains in America!


Yeah, we have AMtrack, not FMtrack....

Ouch! :-)


Actually, I have to give Jim credit for that one, it isn't often that one
gets handed a straight line that tempting...

"Porky"


Mike, you are just feeding this troll and it doesn't
compliment you. I suggest you plonk him as I and about
everyone else has.


I have to reluctantly agree with you, but it is fun to one-up him every
time he throws an insult.


** Only in your wet dreams - you pathetic fat arsed imbecile.

You are Bob are the most blatant of trolls.

Posting such utter ****e as you two have in a pubic forum should
punishable at law.







............. Phil

"Bob Cain" wrote in message
...


Porky wrote:

If you mean that a whistle riding on a moving train, or a speaker
swinging back and forth (or spinning round and round like in a Leslie
speaker system) in a repeating oscillation cycle will produce Doppler
shift,
but a stationary speaker reproducing a complex waveform containing a
mixed
LF and HF tone (or any multiple combination of tones, as would be the
case
in a complex musical waveform) won't produce Doppler shift, then, by
golly,
I think you're right!

I even finally agree that in many cases it has the
properties that can be called Doppler distortion, so it is a
real phenomenon in qualified conditions but not for the
reasons usually given and not from the systems for which it
has been claimed.


I would go so far as to say that anything generating sound and in motion
relative to the listener will generate Doppler shift, but in the special
case of a speaker reproducing a complex waveform consisting of more than one
pure tone, the velocity of the source relative to the listener is
effectively zero, and therefore no Doppler shift will be generated, because:
1) there is one complex waveform driving the speaker and producing the
energy necessary to generate the sound, not some higher frequency or
frequencies "riding" on some lower frequency.
2) the effective sound source is not the speaker cone, but some point or
plane which does not move relative to the listener. The moving speaker cone
just provides the mechanical energy which is transformed into acoustical
energy by it's interaction with the surrounding air. Each instantaneous
point in the complex motion of the cone is in direct co-relation to the
corresponding instantaneous point of compression or rarefaction in the
complex waveform and the notion that the source has motion relative to the
listener is an illusion. The sound wave can be pictured as standing still
with each bit coming into existance as the cone passes through that space.
The sound wave might be represented as:
-- - - -- - - --
with the space between the dots representing compression and rarefaction of
the wave, and the cone's motion producing the sound can be represented as:
]-- - - -- - - --
]- - - -- - - --
] - - -- - - --
]- - - -- - - --
]-- - - -- - - --
]- - - -- - - --
] - - -- - - --
]- - - -- - - --
Note that while the cone is moving, the apparent source is not moving, and
this applies no matter how complex the waveform is. This can actually be
measured with a sensitive pressure meter and graphed, showing that this is
really what is happening.
3) the effective sum of all velocities is zero relative to the listener
4) due to the nature of how a speaker produces sound, the listener is
effectively "riding on the train"
5) In order to produce Doppler shift, there must be a sound source in motion
relative to the listener and in the case of a stationary speaker there is a
complex sound source, but no motion relative to the listener
6) If a speaker did generate Doppler distortion, simply turning it so that
the listener was looking at the edge of the speaker would eliminate the
motion toward and away from the listener (the cone would move side to side,
but would stay at a constant distance from the listener), and thus would
eliminate Doppler distortion. I don't believe that this is the case in the
real world.
7) the empirical measurements I made indicated that there was no audible or
measurable Doppler shift when the Doppler equations predicted that there
would be audible and measurable shift, therefore the Doppler equations do
not apply to this special case.
8) I just have a gut feeling about it
9) none of the above
or
10) any of the above
Personally, I prefer number two.

Porky wrote:


I would go so far as to say that anything generating sound and in motion
relative to the listener will generate Doppler shift, but in the special
case of a speaker reproducing a complex waveform consisting of more than one
pure tone, the velocity of the source relative to the listener is
effectively zero, and therefore no Doppler shift will be generated, because:

Tell me if this is equivalent or not: There is Doppler type
mixing between two frequencies if and only if the pressure
in the far field due to them is a different function of the
velocity of the piston. Where the transfer function is flat
in the Fourier sense, nothing mixes. That is what it all
boils down to in the end. That is not at all the same as
the standard argument because it won't happen in a tube and
the standard argument says it will.

Also, a single tone cannot produce Doppler distortion so I
am definitely wrong about that. Yikes! I see why. The
definition I've been bandying about for a linear system is
actually the definiton of a linear, time invariant system.
The system we are considering must be time variant in terms
of the impedences involved. This is getting wierd.

Can this yield what I've been asking for, a general
expression for far field pressure as a function of piston
velocity that includes the Doppler distortion? I'm not sure
yet, but I'll be thinking about it. It is straightforward
for any two frequencies, however, and is left as an exercise
for the student. :-)


Bob
--

"Things should be described as simply as possible, but no
simpler."

A. Einstein

"Porky" wrote in message


The experiment I suggested will give the results I gave, and if it is
right at under the circumstances I suggested, it should be right
under all circumstances with the same conditions, right? In other
words, if it applies with a LF of .1 Hz or 1 Hz, it will still apply
at LF's 20Hz or 50Hz, is that not correct?

right. However, its a lot harder to properly measure doppler when the LF
tone has a very low frequency. To measure it with a FFT you must use a FFT
size that covers at least one cycle, and hopefully several cycles of the
process. If the LF tone is 0.1 Hz, this means an absolute minimum of 10
seconds of data, and ideally 30 or more. At 44,100 Hz sampling, this would
be a FFT composed of a minimum of 441,000 samples, and preferably several
million samples.

Consider the original example - the LF tone was 50 Hz. It had an 882 sample
period. Note how much overkill there was when analyzed using a 65k sample
FFT, or as I used a one million point FFT.

"Jim Carr" wrote in message
news:1gVUc.10087$yh.1675@fed1read05
"Arny Krueger" wrote in message
...

That's because the train is not moving back and forth in front of
us, in a periodic sine-shaped pattern.

You OBVIOUSLY have not seen the trains in America!

LOL!

There's a crossing I frequent north of here that must have a lot of trains
doing switching, that seem to specialize in repetitive, back-and-forth
movements.

"Arny Krueger" wrote in message
...
"Porky" wrote in message


The experiment I suggested will give the results I gave, and if it is
right at under the circumstances I suggested, it should be right
under all circumstances with the same conditions, right? In other
words, if it applies with a LF of .1 Hz or 1 Hz, it will still apply
at LF's 20Hz or 50Hz, is that not correct?

right. However, its a lot harder to properly measure doppler when the LF
tone has a very low frequency. To measure it with a FFT you must use a FFT
size that covers at least one cycle, and hopefully several cycles of the
process. If the LF tone is 0.1 Hz, this means an absolute minimum of 10
seconds of data, and ideally 30 or more. At 44,100 Hz sampling, this would
be a FFT composed of a minimum of 441,000 samples, and preferably several
million samples.

Consider the original example - the LF tone was 50 Hz. It had an 882
sample
period. Note how much overkill there was when analyzed using a 65k sample
FFT, or as I used a one million point FFT.

One of the problems with FFT analysis that we've all overlooked is that we
aren't really dealing with analog waveforms in our simulations, and we can
get erroneous results when using high FFT numbers because we start playing
in the digital "cracks", so to speak, Having done a bit of experimentation,
I've found that I get the most consistant results across the whole range be
using an FFT number of 16K or 32K, higher rates give false results,
especially at higher frequency HF tones. Alternatively, if your equipment
will handle it, try creating the wave models at 24 or 32/96, or even 32/192,
you'll see a considerable difference in your resluts, especailly at higher
HF tones and FFT numbers, and your results will be more consistant across
the entire range of LF and HF tones.

"Bob Cain" wrote in message
...


Porky wrote:


I would go so far as to say that anything generating sound and in
motion
relative to the listener will generate Doppler shift, but in the special
case of a speaker reproducing a complex waveform consisting of more than
one
pure tone, the velocity of the source relative to the listener is
effectively zero, and therefore no Doppler shift will be generated,
because:

Tell me if this is equivalent or not: There is Doppler type
mixing between two frequencies if and only if the pressure
in the far field due to them is a different function of the
velocity of the piston. Where the transfer function is flat
in the Fourier sense, nothing mixes. That is what it all
boils down to in the end. That is not at all the same as
the standard argument because it won't happen in a tube and
the standard argument says it will.

Correct, for the simple reason that a speaker reproducing a complex
waveform does so coherently, however, if you add another force vector such
as moving the speaker, Doppler shift will occur in the entire reproduced
sound, not just the HF component, this is how a Leslie speaker system works.

Also, a single tone cannot produce Doppler distortion so I
am definitely wrong about that. Yikes! I see why. The
definition I've been bandying about for a linear system is
actually the definiton of a linear, time invariant system.
The system we are considering must be time variant in terms
of the impedences involved. This is getting wierd.

Can this yield what I've been asking for, a general
expression for far field pressure as a function of piston
velocity that includes the Doppler distortion? I'm not sure
yet, but I'll be thinking about it. It is straightforward
for any two frequencies, however, and is left as an exercise
for the student. :-)

It's just as straightforward for a complex musical waveform, you don't hear
any Doppler shift unless you start moving the speaker.
Here's another way to look at it, there is no Doppler shift in a speaker
because the speaker is not the source of the sound, the air it compresses
and rarifies is the actual sound source, and since moving air can't cause
Doppler distortion (try having a friend whistle a steady tone in a variable
gusty wind if you doubt it), there is no Doppler distortion introduced by a
speaker.

"Porky"

Here's another way to look at it, there is no Doppler shift in a
speaker
because the speaker is not the source of the sound, the air it compresses
and rarifies is the actual sound source, and since moving air can't cause
Doppler distortion (try having a friend whistle a steady tone in a
variable
gusty wind if you doubt it), there is no Doppler distortion introduced by
a
speaker.



** Hey - Porky boy.

If pig ignorant, asinine, hee- haw stupidity was an Olympic event - YOU
would be the world record holder.

Dumbness is YOUR forte - you excel at it in every way.

You and Bob Cain should be the USA team for the " Synchronised Imbeciles "
event.

Just like synchronised diving - but he pool is empty !!!!!!!!!!!





............... Phil

high frequency with a low frequency and applies that complex electrical
waveform to the speaker voice voil. The result is NOT a high frequency tone
riding on a low frequency tone, it's a single complex waveform containing
elements of both tones, and thus there is no Doppler distortion.

That doesn't make sense. To generate a frequency, the speaker has to
move back and forth at a certain rate. The higher the frequency, the
faster the rate at which it moves. Surely if we want to hear both
frequencies at once, the speaker has to vibrate at both speeds at once ?

If you look at a low frequency sine wave with an amplitude against time
graph, and SUM a much higher frequency much lower amplitude wave to it,
surely you'll see the original low frequency sine wave, but the line
itself instead of being a smooth sine wave, will be oscillating at the
high frequency.

This is what I imagine would happen anyway, I'm not near anything I can
test this with right now. Just looking at what the line is doing will
then surely tell you what the speaker is doing ? Surely it'll be
following the wave ?

So, the speaker would be slowly moving back and forth, following the
amplitude of the low frequency signal, but as it moves back a forth,
it'll be oscillating a small amount back and forth at its current
position in the low frequency wave because that is what the input signal
is doing.

IMHO at least ....

Here is an example;

Our keyboard player once made his keyboard output such a low frequency
signal, that you could watch the speaker slowly move back a forth quiet
far, perhaps once every second. Are you telling me that, if I "mixed" a
high frequency with that signal, the speaker would be no longer moving
back and forth slowly about once per second ?

--
Mark Simonetti.
Freelance Software Engineer.

"Porky" wrote in message

"Arny Krueger" wrote in message
...
"Porky" wrote in message


The experiment I suggested will give the results I gave, and if it
is right at under the circumstances I suggested, it should be right
under all circumstances with the same conditions, right? In other
words, if it applies with a LF of .1 Hz or 1 Hz, it will still apply
at LF's 20Hz or 50Hz, is that not correct?

right. However, its a lot harder to properly measure doppler when
the LF tone has a very low frequency. To measure it with a FFT you
must use a FFT size that covers at least one cycle, and hopefully
several cycles of the process. If the LF tone is 0.1 Hz, this means
an absolute minimum of 10 seconds of data, and ideally 30 or more.
At 44,100 Hz sampling, this would be a FFT composed of a minimum of
441,000 samples, and preferably several million samples.

Consider the original example - the LF tone was 50 Hz. It had an 882
sample
period. Note how much overkill there was when analyzed using a 65k
sample FFT, or as I used a one million point FFT.

One of the problems with FFT analysis that we've all overlooked is
that we aren't really dealing with analog waveforms in our
simulations, and we can get erroneous results when using high FFT
numbers because we start playing in the digital "cracks", so to
speak,

I'd like to see a believable fuirther explantion of that.

FFT and I are going on our 42nd year, and we've been pretty good friends the
whole time.

Having done a bit of experimentation, I've found that I get
the most consistent results across the whole range be using an FFT
number of 16K or 32K, higher rates give false results, especially at
higher frequency HF tones.

I'd like to see a believable further explanation of that.

Alternatively, if your equipment will
handle it, try creating the wave models at 24 or 32/96, or even
32/192, you'll see a considerable difference in your results,
especially at higher HF tones and FFT numbers, and your results will
be more consistent across the entire range of LF and HF tones.

I'd also like to see a believable further explanation of that.

"Mark Simonetti" wrote in message

high frequency with a low frequency and applies that complex
electrical waveform to the speaker voice voil. The result is NOT a
high frequency tone riding on a low frequency tone, it's a single
complex waveform containing elements of both tones, and thus there
is no Doppler distortion.

That doesn't make sense. To generate a frequency, the speaker has to
move back and forth at a certain rate. The higher the frequency, the
faster the rate at which it moves. Surely if we want to hear both
frequencies at once, the speaker has to vibrate at both speeds at
once ?

The speaker is one entity, so it has only one speed at a time. In your case
the speaker's speed is the sum of the two speeds. You later talk like you
believe this. So why not say it up front?

If you look at a low frequency sine wave with an amplitude against
time graph, and SUM a much higher frequency much lower amplitude wave
to it, surely you'll see the original low frequency sine wave, but
the line itself instead of being a smooth sine wave, will be
oscillating at the high frequency.

Agreed

This is what I imagine would happen anyway, I'm not near anything I
can test this with right now. Just looking at what the line is doing
will then surely tell you what the speaker is doing ? Surely it'll be
following the wave ?

Pretty much.

So, the speaker would be slowly moving back and forth, following the
amplitude of the low frequency signal, but as it moves back a forth,
it'll be oscillating a small amount back and forth at its current
position in the low frequency wave because that is what the input
signal is doing.

Ironcially, the velocity due to the lower frequency may be the greater of
the two velocities that are summed together.

IMHO at least ....

Here is an example;

Our keyboard player once made his keyboard output such a low frequency
signal, that you could watch the speaker slowly move back a forth
quiet far, perhaps once every second. Are you telling me that, if I
"mixed" a high frequency with that signal, the speaker would be no
longer moving back and forth slowly about once per second ?

Of course it will. And its instanteious velocity is added to the
instantaneous velocity due to the upper note.

Arny Krueger wrote:
"Porky" wrote in message

Having done a bit of experimentation, I've found that I get
the most consistent results across the whole range be using an FFT
number of 16K or 32K, higher rates give false results, especially at
higher frequency HF tones.

I'd like to see a believable further explanation of that.

I bet it's an implementation issue with whatever Arny is using, probably
due to rounding.

Alternatively, if your equipment will
handle it, try creating the wave models at 24 or 32/96, or even
32/192, you'll see a considerable difference in your results,
especially at higher HF tones and FFT numbers, and your results will
be more consistent across the entire range of LF and HF tones.

I'd also like to see a believable further explanation of that.

This definitely sounds like a numeric precision issue.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Of course it will. And its instanteious velocity is added to the
instantaneous velocity due to the upper note.

Okay, but it seems like people are saying that this isn't the case, and
is the reason doppler distortion would NOT occur, which seems plain
bizzare, IMHO. Surely, as the lower frequency moves the cone back and
forth, which at the same time is vibrating to create the higher
frequency, that is IDENTICAL to moving a single vibrating source back
and forth, like the train analogy (except the train doesn't move back
and forth unless the driver is very confused). Therefore, the doppler
effect surely DOES occur.

It just seems really obvious, so I must be missing the whole point of
this, I'm no physics scientist !

--
Mark Simonetti.
Freelance Software Engineer.

One of the problems with FFT analysis that we've all overlooked is
that we aren't really dealing with analog waveforms in our
simulations, and we can get erroneous results when using high FFT
numbers because we start playing in the digital "cracks", so to
speak,

I have to agree with Arny on this. (I used to do FFT and waterfall measurements
when I reviewed for Stereophile.) Higher sampling rates are almost always
better, other than their effect on measuring LF response. Regardless, the higher
the rate, the _fewer_ the cracks.

"Mark Simonetti"

Surely, as the lower frequency moves the cone back and
forth, which at the same time is vibrating to create the higher
frequency, that is IDENTICAL to moving a single vibrating source back
and forth, like the train analogy (except the train doesn't move back
and forth unless the driver is very confused). Therefore, the doppler
effect surely DOES occur.

It just seems really obvious, so I must be missing the whole point of
this, I'm no physics scientist !



** The matter is intuitive to many - but forever obscure to those with
no mental capacity to imagine the situation in their heads. It pretty much
divides up between the science types ( using mental, physical models ) and
the arts subject types ( using only grammar and phrase matching).

The fact that cones have **vastly greater** excursions at low frequencies
than at high ones - even for the same SPL - is at the heart of the matter
and clearly bamboozles as well.

The fact that those large low frequency excursions have the greatest
velocity also confounds the easily confoundable.





.............. Phil

"Mark Simonetti" wrote in message

Of course it will. And its instanteious velocity is added to the
instantaneous velocity due to the upper note.

Okay, but it seems like people are saying that this isn't the case,
and is the reason doppler distortion would NOT occur, which seems
plain bizzare, IMHO.

So bizarre, that it is in fact fallacious.

Surely, as the lower frequency moves the cone
back and forth, which at the same time is vibrating to create the
higher frequency, that is IDENTICAL to moving a single vibrating
source back and forth, like the train analogy (except the train
doesn't move back and forth unless the driver is very confused).

Agreed.

Therefore, the doppler effect surely DOES occur.

Agreed.

It just seems really obvious, so I must be missing the whole point of
this, I'm no physics scientist !

I've got two years of physics, an undergraduate degree in engineering and
completed most of my MSE except for my thesis project (wife's pregnancy
ended that). I've also measured it quite conclusively in the lab. I've been
reading papers about it for like 30 years in the JAES and JASA. Yes, I think
that Doppler distortion exist in speakers, but no I don't think it is a
serious issue. In contrast the AM distortion in speakers is a very serious,
audible issue.

Arny Krueger wrote:

I've got two years of physics, an undergraduate degree in engineering and
completed most of my MSE except for my thesis project (wife's pregnancy
ended that). I've also measured it quite conclusively in the lab. I've been
reading papers about it for like 30 years in the JAES and JASA. Yes, I think
that Doppler distortion exist in speakers, but no I don't think it is a
serious issue. In contrast the AM distortion in speakers is a very serious,
audible issue.

This is a reasonable assessment of the situation. The thing about doppler
modulation, though, is that it's really interesting and the math is a lot
of fun. Not like typical AM distortion from amplitude nonlinearities, which
is dull, even if it's a more significant problem. So I think folks should
continue investigating doppler distortion because it's an interesting problem
even if not a terribly important one.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Phil Allison wrote:
no mental capacity to imagine the situation in their heads. It pretty much
divides up between the science types ( using mental, physical models ) and
the arts subject types ( using only grammar and phrase matching).

I often find this. When designing and writing software for instance at
work, I can see it all in my head, the mechanisms, how things interact
and such like. As soon as I try and explain it verbally to someone, I
find it difficult to transfer to language. I'm okay if I have time to
produce a document though, because then I have time to translate the
visualisation into words, and I can use diagrams.

In my original post about this, I had the same problem, I wanted to draw
diagrams showing the waves being summed, and the speaker in its
different position, etc.

--
Mark Simonetti.
Freelance Software Engineer.

"Mark Simonetti"

no mental capacity to imagine the situation in their heads. It pretty
much
divides up between the science types ( using mental, physical models )
and
the arts subject types ( using only grammar and phrase matching).

I often find this. When designing and writing software for instance at
work, I can see it all in my head, the mechanisms, how things interact
and such like. As soon as I try and explain it verbally to someone, I
find it difficult to transfer to language. I'm okay if I have time to
produce a document though, because then I have time to translate the
visualisation into words, and I can use diagrams.

In my original post about this, I had the same problem, I wanted to draw
diagrams showing the waves being summed, and the speaker in its
different position, etc.



** The lack of the facility to post sketches and diagrams on usenet is a
*real* drawback. When I need to explain stuff to non-technical folk ( and
some technical ones too) I often reach for my pen and paper !!!!

Then, on second thoughts, the sketches that might appear most often could
be kinda pornographic in nature ;-)



........... Phil

Phil Allison wrote:
Then, on second thoughts, the sketches that might appear most often could
be kinda pornographic in nature ;-)

That might not be a bad thing, I mean if it helps get the point across,
right? All in the aid of science and all that ;-)

--
Mark Simonetti.
Freelance Software Engineer.

"William Sommerwerck" wrote in message

One of the problems with FFT analysis that we've all overlooked is
that we aren't really dealing with analog waveforms in our
simulations, and we can get erroneous results when using high FFT
numbers because we start playing in the digital "cracks", so to
speak,

I have to agree with Arny on this. (I used to do FFT and waterfall
measurements when I reviewed for Stereophile.) Higher sampling rates
are almost always better, other than their effect on measuring LF
response. Regardless, the higher the rate, the _fewer_ the cracks.

Strictly speaking there are no cracks, its just that the bricks are wider.

"Scott Dorsey" wrote in message

Arny Krueger wrote:

I've got two years of physics, an undergraduate degree in
engineering and completed most of my MSE except for my thesis
project (wife's pregnancy ended that). I've also measured it quite
conclusively in the lab. I've been reading papers about it for like
30 years in the JAES and JASA. Yes, I think that Doppler distortion
exist in speakers, but no I don't think it is a serious issue. In
contrast the AM distortion in speakers is a very serious, audible
issue.

This is a reasonable assessment of the situation. The thing about
doppler modulation, though, is that it's really interesting and the
math is a lot of fun. Not like typical AM distortion from amplitude
nonlinearities, which is dull, even if it's a more significant
problem. So I think folks should continue investigating doppler
distortion because it's an interesting problem even if not a terribly
important one.

Thanks, Scott. The other thing about Doppler is that it is in some sense
irreducable, and even something that modern speaker development trends seem
to want to increase.

Some of my informants argue that in fact speakers are about as linear as
they ever will be, and that the only remaining approach is to make them
cheaper, smaller, and put their nonlinearities where they won't sound so
objectionable.

This whole discussion traces back to another discussion on another audio
groups about a month ago. My opponent in that discussion seems to have
considerably changed his position in the past month in a good way, but he
still abuses my name. So goes life!

Arny Krueger wrote:

Some of my informants argue that in fact speakers are about as linear as
they ever will be, and that the only remaining approach is to make them
cheaper, smaller, and put their nonlinearities where they won't sound so
objectionable.

If speakers are as linear as they ever will be, I'm giving up this whole
industry and going out to listen only to live music. If this is as good
as it gets, it's a total waste.
--scott


--
"C'est un Nagra. C'est suisse, et tres, tres precis."

"Scott Dorsey" wrote in message

Arny Krueger wrote:

Some of my informants argue that in fact speakers are about as
linear as they ever will be, and that the only remaining approach is
to make them cheaper, smaller, and put their nonlinearities where
they won't sound so objectionable.

If speakers are as linear as they ever will be, I'm giving up this
whole industry and going out to listen only to live music. If this

I'm almost with you, Scott.

"Porky" writes:

"Arny Krueger" wrote in message
...
"Porky" wrote in message


The experiment I suggested will give the results I gave, and if it is
right at under the circumstances I suggested, it should be right
under all circumstances with the same conditions, right? In other
words, if it applies with a LF of .1 Hz or 1 Hz, it will still apply
at LF's 20Hz or 50Hz, is that not correct?

right. However, its a lot harder to properly measure doppler when the LF
tone has a very low frequency. To measure it with a FFT you must use a FFT
size that covers at least one cycle, and hopefully several cycles of the
process. If the LF tone is 0.1 Hz, this means an absolute minimum of 10
seconds of data, and ideally 30 or more. At 44,100 Hz sampling, this would
be a FFT composed of a minimum of 441,000 samples, and preferably several
million samples.

Consider the original example - the LF tone was 50 Hz. It had an 882
sample
period. Note how much overkill there was when analyzed using a 65k sample
FFT, or as I used a one million point FFT.

One of the problems with FFT analysis that we've all overlooked is that we
aren't really dealing with analog waveforms in our simulations, and we can
get erroneous results when using high FFT numbers because we start playing
in the digital "cracks", so to speak,

The waveform being analog or digital makes no difference as long as sufficient
bandwidth and dynamic range has been supplied by the A/D conversion. Rather, the
problem you are ignorantly referring to is that an FFT implicitly assumes the
input is periodic. If it isn't, you can get yourself befuddled. There is also
the problem with using the FFT to estimate the spectrum of a random signal -
it can be shown that there will be variance in the frequency estimates no
matter how many points are used in the FFT (see, for example, "Signal Processing:
Discrete Spectral Analysis, Detection, and Estimation," Mischa Schwartz and
Leonard Shaw).
--
Randy Yates
Sony Ericsson Mobile Communications
Research Triangle Park, NC, USA
, 919-472-1124

Bob,

You seem to think the Doppler effect doesn't happen in speakers beacue
the air is moving with the speaker cone.

I think this is wrong. The Doppler effect happens anyway. The
Doppler effect depends only on the distnace between the Rx and Tx
changing.

Doppler happens for radio and light waves as well and there is no
ether to move or not move. The Doppler effect is a function of the
changing distance between the Rx and Tx and has nothing to do with the
propogating medium.


Mark

"Paul Guy" wrote in message

On Sat, 7 Aug 2004 07:02:36 -0400, "Arny Krueger"
wrote:

....stuff deleted........

I think that the triple tone test and modern spectrum analyzer
technology provides valuable insights into this area. I think that
I've established that when there are two upper-frequency probe
tones, FM distortion will produce sidebands with a higher amplitude
with the highest frequency tone, all other things being equal. This
finding can be, and probably should be applied to investigations
relating to both Doppler distortion and jitter.

To get an idea of the magnitude of any Doppler (FM) artifacts, you
need to know the cone velocity. It is my understanding, that in the
area where a speaker has a flat response, the velocities are fairly
consistent as frequency changes. So within this area, you should be
able to analyze and predict the Doppler effects.

I don't have much data that represents typical cone velocities at
different power levels (or SPL level, at say, 1 meter). From some of
the data shown on the Linkwitz site, he has a woofer with about 1.5
Meters/sec at 86db @1meter (that's reasonably loud). Does anyone have
typical data for other loudspeakers, especially at higher frequencies
(tweeters, midrange)?

If you can't see the motion, it must be 1/16 or less. Multiply the maximum
motion by 2 pi F to get peak velocity. Remember that peak possible cone
motion happens just below system resonance, and is less at higher
frequencies.

Using 1.5 M/s peak cone velocity, the speed of sound is about 340
M/sec, that should vary all the frequencies whose velocities were
supposed to be something else. Usually for the purpose of analysis, we
assume that all the other ones are pretty small. Anyhow, with those
numbers, you get about 0.44% change in frequency, higher or lower
depending which way the cone is moving.

Seems about right.

If the signal causing the 1.5
meter/sec was 50 Hz , and you had another signal of 4 KHz, then your
4khz note appears to be changing from 4khz to 4017 khz, then back then
to a low of 3983 khz, 50 times a second. In the frequency domain (your
ears, spectrum analyzer) things get weird.

Well, they get Besselized. ;-)

The result frequencies
(there are more than what you put in) depend on the ratio of the
change in frequency divided by the modulating frequency - this is the
modulation index - M. The total energy is unchanged, so that the
addition of extra stuff comes at the expense the main peak (unlike IM
distortion, where the modulating frequency has to add energy). If the
modulation index is less than 0.3, then there are 2 extra frequencies
(distortion), each one has an amplitude of 1/2 times M (modulation
index), or the total grunge is M . For low values of M, you get 2
extra freq., the sum and difference (just like IM distortion, but out
of phase with each other, and may sound QUITE different) . At higher M
the calculation is very complex, you can have almost all distortion
with almost no fundamental.

Agreed.

Using the above numbers, change in frequency is about 17 Hz,
modulation freq. is 50 Hz, so M=0.34 , or about 17% for each extra
frequency. These will be at 4050 and 3950. This is NOT IM distortion.

Well, its not AM distortion. Whether FM distortion is IM is controversial. I
think that FM is IM because that's what the words seem to mean to me.

The thing to note is that theamount of distortion changes with
modulating frequency! At 10Hz modulating freq,. M is about 1.7 - that
will mess up the waveform badly. The worst case is when the
frequencies are very different. With high values of M, the note
spreads out in frequency - instead of a fundamental and two satellite
tones, there is an almost contiuous block of frequencies. With a 5 or
10 Hz modulation, instead of 4 KHz and 2 extra peaks, you get an
almost continuous band of frequencies around 4 KHz.

That agrees with experimental results. However, you can get a similar family
of tones if your modulating signal is not a pure sine wave.

The sound? High M values are VERY noticeable, usually a warbling
sound, or noticeable extra frequencies. As M decreases to about 0.3,
the original pure tone sounds indistinct in pitch, or you might just
notice extra "stuff", and as M decreases to less than 0.1, it's very
hard to tell (for me). These were done at 4 KHz, with varying
amplitudes and frequency of the modulation frequency. This was not a
really good listening test, the real Golden Ears might be better at
finding the threshold. I used 2 signal generators, one modulating the
others frequency.

One can also use the tone generator in Audition/CE or a bunch of other
software. Then, everything is rigidly phase locked.

I used a spectrum analyzer to determine M, and
adjusted the signal generators to vary M as I listened to the "tones".
My good signal generators are at work, so if you're interested, I can
compare IM and FM (Doppler) distortion with the same frequencies.


I'm
sure Arny has the equipment more readily available - and may even have
.wav files for your listening pleasure, so you can hear for yourself
what the effects are.

Slightly different context, but its all FM:

http://www.pcabx.com/technical/jitter_power/index.htm

What would be really nice, is to frequency shift
a chunk of music with different delta-freq, and different modulation
frequencies, i.e., varying M with different conditions. Multi-tone and
real music should the preferred way to check this out.

It's just a matte of twidding in the parameters with software like
Audition/CE.

The cure? Keep wide ranges of frequencies OUT of a loudspeaker,
i.e., use 2 or 3 way systems. Because the modulation index (M) is
calculated with the modulating frequency as DENOMINATOR, avoiding low
modulating frequencies reduces the distortion. That bears out in my
listening tests. As the modulating frequency increases, the less
noticeable things are. A 3 way system can have a 10 to 1 range of
frequencies for each driver, compared to almost a 1000 to 1 range for
a singe wide range speaker. That will make a big difference when you
calculate M, the modulation index.

Agreed, and since 2-way speakers are almost endemic.., and get to be 3-way
when subwoofers are added...

If speakers are as linear as they ever will be, I'm giving up this whole
industry and going out to listen only to live music. If this is as good
as it gets, it's a total waste.

If you're talking about simple harmonic and IM distortion, I'm inclined to agree
there isn't much room for improvement. But there is great room for improvement
in other areas.

On Thu, 19 Aug 2004 11:03:32 -0400, "Arny Krueger"
wrote:

......some stuff deleted.....

I'm
sure Arny has the equipment more readily available - and may even have
.wav files for your listening pleasure, so you can hear for yourself
what the effects are.

Slightly different context, but its all FM:

http://www.pcabx.com/technical/jitter_power/index.htm

What would be really nice, is to frequency shift
a chunk of music with different delta-freq, and different modulation
frequencies, i.e., varying M with different conditions. Multi-tone and
real music should the preferred way to check this out.


I tried listening to your jitter samples in a less than optimum
environment. You have some castanet samples castanets-060.wav
(unjittered) and castanets_060_jit-20FF2.wav (-20 db 60 Hz jitter). I
can barely tell them apart. To my ears, the jitter version is slightly
duller, but the difference is so tiny, I could easily be fooled. All
the other samples are far too similiar to the reference.
Your piano selections (piano1_1644.wav [unjittered] and
piano1_1644_-20FF2.wav [-20db 60 Hz jitter])are indistinguishable to
me. I noticed that they are both distorted somewhat, nowhere as nice
as your reference piano_nlref.wav file.
Either my ears are totally wrecked (not likely), but the jitter
(FM) page you have really makes the case that it is not a very big
deal. From your spectral analysis, most of the crud is very close to
the fundamentals, and as such will be largely masked. Have you
synthesized higher or lower frequency jitter components to see their
audibility?
What is the prevaling opinion about the jitter (or FM "distortion")
samples you put on your site?
From my own testing, the sidebands need to be more like -10db (or
-10 db jitter as you specify it) before they begin to be audible.
That's pretty disgusting! 30% crud! Masking theory does confirm what
my ears tell me, namely that junk very close to the fundamental is
very well masked i.e., inaudible. It interesting that conventional
spectrum analyzers have the same difficulty. The ear does have much of
the behaviour of a poor dynamic range (30db) spectrum analyzer, with
strange post processing and AGC.
Readers of this newsgroup would be well advised to read up about the
ear (especially the cochlea) to understand masking and other
mechanisms the ears uses as "garbage cleanup".

-Paul
.................................................. .............
Paul Guy
Somewhere in the Nova Scotia fog

Porky wrote:


One of the problems with FFT analysis that we've all overlooked is that we
aren't really dealing with analog waveforms in our simulations, and we can
get erroneous results when using high FFT numbers because we start playing
in the digital "cracks", so to speak,

The FFT gives slightly misleading results due to "edge
effects" or the windows that must be applied to eliminate
them but after windowing, nothing much gets through the
cracks. If we are dealing with signals whose Fourier
components have an integral number of cycles in the length
of the transform and if the signal doesn't contain anything
in the band above half the sample rate than it's exact.


Bob
--

"Things should be described as simply as possible, but no
simpler."

A. Einstein

Mark Simonetti wrote:

Of course it will. And its instanteious velocity is added to the
instantaneous velocity due to the upper note.


Okay, but it seems like people are saying that this isn't the case, and
is the reason doppler distortion would NOT occur, which seems plain
bizzare, IMHO.

In some circumstances it does and in others it doesn't but
that isn't the reason, at any rate, for either result.


Bob
--

"Things should be described as simply as possible, but no
simpler."

A. Einstein

Bob Cain wrote:


Also, a single tone cannot produce Doppler distortion so I am definitely
wrong about that. Yikes! I see why. The definition I've been bandying
about for a linear system is actually the definiton of a linear, time
invariant system. The system we are considering must be time variant in
terms of the impedences involved. This is getting wierd.

Yikes is right. I'm not sure that thought would have
survived a nights sleep but I hit Send instead of Save at
bedtime. Anyway, that's what a raw speculation that needs
much further thought looks like when it pops outa my head
unbeckoned.

If it's nonsense I appologize for the noise. I'm trying to
understand why a system that appears linear in the case of
any pure sinusoid would produce mixing when presented with
superpositions. That defies my understanding at the moment
but I plan on fixing that. :-)


Bob
--

"Things should be described as simply as possible, but no
simpler."

A. Einstein

"Paul Guy" wrote in message
news
On Thu, 19 Aug 2004 11:03:32 -0400, "Arny Krueger"
wrote:

.....some stuff deleted.....

I'm
sure Arny has the equipment more readily available - and may even
have .wav files for your listening pleasure, so you can hear for
yourself what the effects are.

Slightly different context, but its all FM:

http://www.pcabx.com/technical/jitter_power/index.htm

What would be really nice, is to frequency shift
a chunk of music with different delta-freq, and different modulation
frequencies, i.e., varying M with different conditions. Multi-tone
and real music should the preferred way to check this out.


I tried listening to your jitter samples in a less than optimum
environment. You have some castanet samples castanets-060.wav
(unjittered) and castanets_060_jit-20FF2.wav (-20 db 60 Hz jitter). I
can barely tell them apart. To my ears, the jitter version is slightly
duller, but the difference is so tiny, I could easily be fooled. All
the other samples are far too similiar to the reference.

I agree. The actual amounts of FM distoriton seemed high, but the audible
effects seemed to be pretty innocious.

It actually was quite a bit of work to prepare those samples, and I never
returned to the situation.

Your piano selections (piano1_1644.wav [unjittered] and
piano1_1644_-20FF2.wav [-20db 60 Hz jitter])are indistinguishable to
me. I noticed that they are both distorted somewhat, nowhere as nice
as your reference piano_nlref.wav file.

Ironically, those samples were taken from a ADC/DAC vendor's site. At this
time I have gigabytes of better-sounding piano samples at my disposal.

Either my ears are totally wrecked (not likely), but the jitter
(FM) page you have really makes the case that it is not a very big
deal. From your spectral analysis, most of the crud is very close to
the fundamentals, and as such will be largely masked.

Something like that.

Have you
synthesized higher or lower frequency jitter components to see their
audibility?

I picked 60 Hz because my PCAVTech work suggested that this was a very
common, perhaps the most common jitter freqeuency.

I know know quite a bit more about the psychoacoustics of FM distortion, and
were I to revisit the topic I would shift the jitter frequency down.

In rough terms FM distortion is most audible for low modulating frequencies,
kind of plateaus from 1 to 5 Hz, and then falls off at about 6 dB/ocatave.
This is rough paraphrase of Zwicker and Fastl's comments in the matter. It
also agrees with the design of the old NAB wow and flutter weighting curve.

What is the prevaling opinion about the jitter (or FM "distortion")
samples you put on your site?

Nobody hears nuttin even though the amounts of jitter are vastly in excess
of what one sees in digital gear, even the crap.

From my own testing, the sidebands need to be more like -10db (or
-10 db jitter as you specify it) before they begin to be audible.
That's pretty disgusting! 30% crud!

That would depend on modulating frequency, of course.

Masking theory does confirm what
my ears tell me, namely that junk very close to the fundamental is
very well masked i.e., inaudible.

There is actually a separate case for low frequency modulation. Zwicker and
Fastl mention both, but in different places, as I recall.

It interesting that conventional
spectrum analyzers have the same difficulty.

1 million point FFTs don't have similar difficulties, to say the least!

The ear does have much of
the behaviour of a poor dynamic range (30db) spectrum analyzer, with
strange post processing and AGC.

Agreed.

Readers of this newsgroup would be well advised to read up about the
ear (especially the cochlea) to understand masking and other
mechanisms the ears uses as "garbage cleanup".

Agreed. I have often decried the fact that EEs & studio workers aren't
routinely taught much about psychoacoustics, or other forms of perception.