You guys have helped me in the past when I had a stumper (for me) of a
question, and it was much appreciated. I've got another one. Before Ghost
comes along to remind everyone, I will say up front that I have no formal
training in the field. I'm a home studio guy who has been in and out of bar
bands for 20+ years.

So, from a practical standpoint I have no problems understanding or dealing
with feedback. What I'm getting stuck on is more technical in nature.
Feedback as I understand it happens when a frequency enters a loop and
becomes an oscillator. The output from the speakers hits the source of the
sound in-phase and the gain is above unity so that with each pass it becomes
progressively louder. I could live the rest of my life thinking this way and
always be able to get my SR system up and running smoothly.

I can best explain what I don't understand (how's that for an oxymoron) by
giving you a sample situation. Suppose I place my mic so it is pointing
directly at my speaker at some particular distance. In order to potentially
feed back the frequency must be in phase, which means the delay has to be a
multiple of the period of oscillation, right?

So, now factor in gain. Here's where I start getting lost. If the source is
already in phase, then why doesn't feedback happen at any gain level? I
know, I know, the gain has to be at or above unity. But it's that mechanism
which is my sticking point.

If I have sound source in my room putting out a steady frequency Y, that
frequency either will or will not be in phase. Suppose that it is. Suppose
my mic picks it up and amplifies it. The energy from the source combines
with the energy from the speaker which *should* be the start of oscillation.
But obviously it's not.

I need to add gain to induce feedback. To stop the feedback I can bring the
gain back down.

What I don't get is that if the amp adds just a teeny tiny bit of energy
in-phase (my initial gain setting) and *that* gets picked up by the mic, so
the next round trip it doubles, eventually the damn thing should end up in a
squeal. It can't be the sensitivity of my mic because if it's not sensitive
enough to pick up my source in the first place, the source will never get
amplified.

My current theory is that the energy from the speaker dissipates before it
ever has a chance to hook up with the source frequency. I'm guessing that
sound waves detoriate and possibly deform over distance/time. When I
increase the gain, I am in effect making sound waves that "last longer" and
thus have enough umph to combine with the existing source to be additive.

I figure it can't be the gain of my amp because if I leave the amp at the
same gain level and simply move mic close so that it is still in phase for
that frequency, eventually I will get close enough for it to start feeding
back.

I figure at this point my theory is right and somebody with more expertiese
will confirm. Or maybe I am misunderstanding something at a low level. Or
perhaps there's another factor of which I am unawares. I've read a number of
sites and none of them seem to address this nor do any of my books.

Any assistance will be greatly appreciated. I know this is esoteric and the
answer probably won't make me any better at setting up the band's PA system,
but it's been gnawing at me for several days.

--
Jim Carr
http://www.azwebpages.com/bass
Information for Bass Players

"Jim Carr" wrote in message
news:IUiFf.347096$0l5.281904@dukeread06...
.. In order to potentially
feed back the frequency must be in phase, which means the delay has to be
a
multiple of the period of oscillation, right?

Right - which is why feedback usually occurs at one frequency

If the source is
already in phase, then why doesn't feedback happen at any gain level?

If the gain of the entire loop is less than unity, then the second round
trip has less amplitude than the first, the 3rd is even smaller, etc - until
the sounds eventually dies out.

If I have sound source in my room putting out a steady frequency Y, that
frequency either will or will not be in phase.

No external sound source is required to initiate feedback.You'll get
feedback in a completely quiet environment. Its the self-noise of the system
(mic, amplifiers, etc.) which initiates feedback.

Suppose
my mic picks it up and amplifies it. The energy from the source combines
with the energy from the speaker which *should* be the start of
oscillation.
But obviously it's not.

Again, its got nothing to do with the "source" energy combining with the
speaker energy - no source is required.

Just because the sound is amplified doesn't mean it will reach the mic at an
amplitude greater than the during the previous round trip - because every
time the sound makes a round trip, not only does it undergo amplification
(mic preamp, power amp, etc.), it also undergoes attenuation (losses in
converting the electrical signal to physical motion of the speaker cone,
attenuation in traveling through the air from speaker to mic, etc.) If the
attenuation is greater than the amplification, then gain is less than
unity - and no feedback.

My current theory is that the energy from the speaker dissipates before it
ever has a chance to hook up with the source frequency.

Well - it dissipates enought to make the total gain less than unity. But
again, its not "hooking up" with any source frequency.

I'm guessing that
sound waves detoriate and possibly deform over distance/time.

Yes sound is attenuated as it travels through the air.

When I
increase the gain, I am in effect making sound waves that "last longer"
and
thus have enough umph to combine with the existing source to be additive.

Doesn't matter how "long" the sound lasts - if the total gain is greater
than unity, then feedback occurs. And one way of making the gain greater
than unity is by cranking up the gain on your amp.

I figure it can't be the gain of my amp because if I leave the amp at the
same gain level and simply move mic close so that it is still in phase for
that frequency, eventually I will get close enough for it to start feeding
back.

This is the other way of increasing the gain in the total loop - by moving
the mic closer, you're decreasing the attenuation the sound undergoes on its
way from the speaker to the mic, because it travels a shorter distance.

Steve

"Jim Carr" wrote in message
news:IUiFf.347096$0l5.281904@dukeread06...

So, from a practical standpoint I have no problems understanding or
dealing
with feedback. What I'm getting stuck on is more technical in nature.
Feedback as I understand it happens when a frequency enters a loop and
becomes an oscillator. The output from the speakers hits the source of the
sound in-phase and the gain is above unity so that with each pass it
becomes
progressively louder. I could live the rest of my life thinking this way
and
always be able to get my SR system up and running smoothly.

I can best explain what I don't understand (how's that for an oxymoron) by
giving you a sample situation. Suppose I place my mic so it is pointing
directly at my speaker at some particular distance. In order to
potentially
feed back the frequency must be in phase, which means the delay has to be
a
multiple of the period of oscillation, right?

The phasing part of acoustical feedback is one of those things that just
happens. The most significant of feedback is the other thing you
mentioned - that loop gain is greater than one. In fact, if you look at the
equations that are used to predict maximum available gain before feedback -
they assume that somehow the phasing issue will work itself out. Out in the
real world, it does.

If a SR system system has loop gain of greater than one over a modest range
of frequencies, things will almost always work themselves out so that there
is feedback. After all what does it take to alter the phase of an
acoustical signal? Usually just a slight chance in path length or frequency.

So, now factor in gain.

Gain is amost all that ever matters. Here's an experiment - make up or
obtain a polarity inverter cable or adaptor. Hosa makes an XLR polarity
inverter adaptor that can often be gotten for under $10. Set up your sustem
with one open mic and one speaker so that it feeds back. Now, plug in the
polarity inverter into the mic cable and see what happens. Usually the
system will still feed back, just at a slightly different frequency.
Polarity is the most extreme case of a change in phase, right?

On Sun, 5 Feb. 2006 01:47:36 -0700, "Jim Carr"
wrote:

So, now factor in gain. Here's where I start getting lost. If the source is
already in phase, then why doesn't feedback happen at any gain level? I
know, I know, the gain has to be at or above unity. But it's that mechanism
which is my sticking point.

First of all, stop thinking phase. At any distance certain
frequencies will be in phase and others out of phase, but there's
always some frequencies in phase at any distance.

If I have sound source in my room putting out a steady frequency Y, that
frequency either will or will not be in phase. Suppose that it is. Suppose
my mic picks it up and amplifies it. The energy from the source combines
with the energy from the speaker which *should* be the start of oscillation.
But obviously it's not.

Think of a short burst of sound source that cuts off and then is heard
a split second later (not exactly the way real life works) but say you
could make a sound and then completely stop it before it hits the mic
a second time. If it is louder when it hits the mic the second time,
it feeds back. If not it doesn't

What I don't get is that if the amp adds just a teeny tiny bit of energy
in-phase (my initial gain setting) and *that* gets picked up by the mic, so
the next round trip it doubles, eventually the damn thing should end up in a
squeal. It can't be the sensitivity of my mic because if it's not sensitive
enough to pick up my source in the first place, the source will never get
amplified.

No, it has to add more than a teeny bit. It has to be louder than the
original itself. the speaker always adds something to the volume at
the mic but it doesn't;feed back until it is adding more than the
source material.

My current theory is that the energy from the speaker dissipates before it
ever has a chance to hook up with the source frequency. I'm guessing that
sound waves detoriate and possibly deform over distance/time. When I
increase the gain, I am in effect making sound waves that "last longer" and
thus have enough umph to combine with the existing source to be additive.

Yes, that's what it is. Sound levels are determined by the square of
the distance to the mic. Two sounds are equal volume, but one is
twice as far from the mic, The sound twice as far away is heard to be
only 1/4 as loud, not 1/2 as loud. A sound 10 times father away is
only 100/th as loud at the mic! So when you are talking in a normal
voice one inch from a mic and the voice coming out of the speaker is
10 or 20 feet away the sound from the speaker will still be very much
quieter at mic than the original. If the speaker is more than 100
times father away than your mouth (say one inch and nearly 10 feet
which is a common example) the volume is theoretically attenuated
100 X 100 = 10,000 times less by the distance. So the speaker can be
quite a bit louder than your lips before it adds significantly to the
sound from your voice. The closer you get to the speaker, the less
advantage you have.

Of course in the real world room acoustics warp this and make certain
frequencies jump out..


I figure it can't be the gain of my amp because if I leave the amp at the
same gain level and simply move mic close so that it is still in phase for
that frequency, eventually I will get close enough for it to start feeding
back.

Distance squared. Distance is all important in sound levels. that's
why they measure speaker SPL at one watt one meter. You must specify
the volume and distance.

I figure at this point my theory is right and somebody with more expertiese
will confirm. Or maybe I am misunderstanding something at a low level. Or
perhaps there's another factor of which I am unawares. I've read a number of
sites and none of them seem to address this nor do any of my books.

I don't find anything inaccurate with you way of thinking. I hope
knowing about the square of the distance relationship helps you
understand things better.

Julian

Jim Carr wrote:

I can best explain what I don't understand (how's that for an oxymoron) by
giving you a sample situation. Suppose I place my mic so it is pointing
directly at my speaker at some particular distance. In order to potentially
feed back the frequency must be in phase, which means the delay has to be a
multiple of the period of oscillation, right?

Think of it in terms of delay. There is a certain time delay between the
mike and the speaker. The system feeds back at a frequency where that
delay is a half-period of the wave.

If you change the distance, the frequency the system feeds back at changes.

So, now factor in gain. Here's where I start getting lost. If the source is
already in phase, then why doesn't feedback happen at any gain level? I
know, I know, the gain has to be at or above unity. But it's that mechanism
which is my sticking point.

Because if you place your microphone carefully, it will be _less_ than unity.
The whole key to feedback rejection is to get as much as possible of the
instrument into the mike (so the gain can be turned down), and get as little
as possible sound from the speakers into the mike.

Because mikes are directional and speakers are too (although somewhat less so),
you can have a system that has substantial gain for signal sources right in
front of the mike and very low gain for signals coming from the speaker into
the mike.

If I have sound source in my room putting out a steady frequency Y, that
frequency either will or will not be in phase. Suppose that it is. Suppose
my mic picks it up and amplifies it. The energy from the source combines
with the energy from the speaker which *should* be the start of oscillation.
But obviously it's not.

There is always some background noise in the room and the equipment, which
will start the system oscillating if it's possible for it to oscillate.
The key is to avoid leakage from the speaker to the mike to make it more
difficult for the system to oscillate.

If the system _can_ oscillate, it will.

I need to add gain to induce feedback. To stop the feedback I can bring the
gain back down.

Right, because you are increasing the gain above unity. To stop the feedback,
you reduce the acoustic gain of the system below unity.

My current theory is that the energy from the speaker dissipates before it
ever has a chance to hook up with the source frequency. I'm guessing that
sound waves detoriate and possibly deform over distance/time. When I
increase the gain, I am in effect making sound waves that "last longer" and
thus have enough umph to combine with the existing source to be additive.

Yes, sound waves drop off more or less according to the inverse square law.
Microphones and speakers are directional, too. You want to think of the
total _acoustic_ gain of the system and not just the electrical gain.
--scott

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

To add to the explanations, remember that "in phase" covers a lot of
ground. If you're a little bit out of phase at a particular frequency
(but not fully 180 degrees out) there will be some loss of gain, but
you'll still have a signal fed back, so it takes more than unity gain
to make up for it. And you'll be completely in phase at some other
frequency.

In an acoustic situation, you have room modes that reinforce certain
frequencies, so that reduces the amount of electrical gain that you
need in order to create feedback. If the microphone is positioned at an
acoustic peak, you'll have some extra gain at that frequency (and, yes,
this is related to phase).

It's really easier to understand feedback when it's all electronic
because you can actually measure the gain and you don't have
reflections that change things when the physical setup changes.

I can best explain what I don't understand (how's that for an oxymoron)
by
giving you a sample situation. Suppose I place my mic so it is pointing
directly at my speaker at some particular distance. In order to
potentially
feed back the frequency must be in phase, which means the delay has to
be amultiple of the period of oscillation, right?

Correct. What's confusing you (I think) is that you're forgetting that
feedback will occur only at those frequencies.

If there were no sound at all in the room -- from any source -- there could
be no feedback. But there's usually enough noise (though it might have to be
above a certain level) to initiate feedback at those frequencies.

The output of an oscillating system is not broadband noise -- it occurs at
and near specific frequencies where it _can_ occur.

"William Sommerwerck" wrote in message
...
I can best explain what I don't understand (how's that for an oxymoron)
by
giving you a sample situation. Suppose I place my mic so it is pointing
directly at my speaker at some particular distance. In order to
potentially
feed back the frequency must be in phase, which means the delay has to
be amultiple of the period of oscillation, right?

Correct. What's confusing you (I think) is that you're forgetting that
feedback will occur only at those frequencies.

If there were no sound at all in the room -- from any source -- there
could
be no feedback. But there's usually enough noise (though it might have to
be
above a certain level) to initiate feedback at those frequencies.

The output of an oscillating system is not broadband noise -- it occurs at
and near specific frequencies where it _can_ occur.


To expand on what Mike Rivers said, you also need to consider the
room/environment as part of the system. take your system and put it in an
open field and it may not feed back. Put it in a small highly reflective
room and it will probably go nuts. Every room is an enclosure, every
enclosure will have at least one frequency prone to resonance. Just like
blowing into a bottle...

Mikey
Nova Music Productions

"William Sommerwerck" wrote in message
...

If there were no sound at all in the room -- from any source -- there
could
be no feedback.

Agreed.

But there's usually enough noise (though it might have to be
above a certain level) to initiate feedback at those frequencies.

If you presume that the system is linear (and just about any modern
equipment is linear enough so that this is a good assumption), then there is
no minimum threshold below which a system that is prone towards feedback
will *not* be adequately excited.

All providing a larger stimulus does is hurry the process along. Of course
if you're trying to eliminate feedback and time is significant, than
hurrying the process along can be a good idea.

The output of an oscillating system is not broadband noise -- it occurs at
and near specific frequencies where it _can_ occur.

Feedback in a sound system is usually a pure tone, at least until something
overloads.

It is possible to have a system oscillating at two different frequencies at
the same time, but then both are pure tones.

If you presume that the system is linear (and just about any modern
equipment is linear enough so that this is a good assumption), then there
is
no minimum threshold below which a system that is prone towards feedback
will *not* be adequately excited.

I'm not sure about that. Can you say "loop gain"?

"William Sommerwerck" wrote in message
news
If you presume that the system is linear (and just about any modern
equipment is linear enough so that this is a good assumption), then there
is
no minimum threshold below which a system that is prone towards feedback
will *not* be adequately excited.

I'm not sure about that. Can you say "loop gain"?

Loop gain 1 in a linear system will lead to feedback, even only if the
only stimulus is residual noise. There is no minimum stimulus, below which
feedback will not take place.

On Sun, 5 Feb 2006 20:15:13 -0500, "Arny Krueger"
wrote:

Loop gain 1 in a linear system will lead to feedback, even only if the
only stimulus is residual noise. There is no minimum stimulus, below which
feedback will not take place.

And *this* is the answer to the OP's question. Issues
of any kind (amplitude, phase, etc.) about signal are red hearings.

Thanks, as always,

Chris Hornbeck

Jim Carr wrote:

So, now factor in gain. Here's where I start getting lost. If the source is
already in phase, then why doesn't feedback happen at any gain level? I
know, I know, the gain has to be at or above unity. But it's that mechanism
which is my sticking point.

Let's assume a system where the gain from the mic to the speaker due to
the electronics is A and that between the speaker and the mic as the
sound traverses the path is B.

So, when in phase the input from the source, i, and output of the system
at the speaker, o, is given by

o = A*(i + B*o),

right? (i + B*o) is what reaches the mic from the source plus what
reaches it fed back from the speaker.

rearranging algebraicly, we find that

o = [A/(1 - A*B)]*i

The A*B term in the denominator is what is called the loop gain.

Consider what happens if you start with your electronic gain A=0 and
gradually turn it up. The output amplitude starts at zero and goes up,
amplifying i, as A is increased and remains finite until A*B=1 (loop
gain is 1), at which the denominator of the overall gain is zero and the
total gain, in brackets, is infinity. As you approach that point the
gain gets pretty high and you hear that as the ringing (because of
delay) that tells you that you are approaching "feedback."

When A*B=1 is reached, the amplification becomes infinite and the system
goes unstable. Instability in such a system causes oscillation that you
hear as "feedback".

I figure it can't be the gain of my amp because if I leave the amp at the
same gain level and simply move mic close so that it is still in phase for
that frequency, eventually I will get close enough for it to start feeding
back.

As you get closer B gets bigger which in terms of loop gain has the same
effect as making A bigger.


Bob
--

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

A. Einstein

Arny Krueger wrote:

If you presume that the system is linear (and just about any modern
equipment is linear enough so that this is a good assumption),

As soon as a room enters the equation, we no longer have a linear
system. And it's somewhat difficult to eliminate the acoustic
environment in live sound.

--
ha

On Mon, 06 Feb 2006 04:15:42 GMT, (hank alrich)
wrote:

Arny Krueger wrote:

If you presume that the system is linear (and just about any modern
equipment is linear enough so that this is a good assumption),

As soon as a room enters the equation, we no longer have a linear
system. And it's somewhat difficult to eliminate the acoustic
environment in live sound.

True that (am I hip or what?). But I think that Arny's
use of the term "linear" is different. "Linear" in this
sense means that if you turn something up one notch,
everything downstream goes up one notch. And the vice
is in the verse.

I remember getting excited about a month or so ago
about the idea that oscillation could be considered
chaotic, but I was wrong; it's linear.

Still interesting to think about though...

Thanks, as always,

Chris Hornbeck

"Chris Hornbeck" wrote in message
...
On Sun, 5 Feb 2006 20:15:13 -0500, "Arny Krueger"
wrote:

Loop gain 1 in a linear system will lead to feedback, even only if the
only stimulus is residual noise. There is no minimum stimulus, below
which
feedback will not take place.

And *this* is the answer to the OP's question. Issues
of any kind (amplitude, phase, etc.) about signal are red hearings.

Red "hearings"? Is that a pun or a fortunate typo?

I disagree that phase is a red herring. Set up a simple mic, amp and
speaker. Put the gain near ringing, then drop it back just a hair. Now put a
phaser in-line. In my tests I hear the system ring then go "quiet" at a rate
equal to how phast my phaser is going.

I also quote this source:
http://www.svconline.com/mag/avinsta...c_enhancement/

When the direct and reflected sound sum in phase, amplitude at that
frequency increases. Likewise, when the direct and reflected sound sum out
of phase, amplitude at that frequency decreases. Thus, the transfer function
between the loudspeaker and the mic has many peaks and valleys as a function
of frequency due to interference between the numerous reflections in the
sound path. If the electronic gain of the system is continually increased,
the system will begin to oscillate at the frequency with the highest
statistical gain or the path of least acoustic impedance.

"Bob Cain" wrote in message
...

So, when in phase the input from the source, i, and output of the system
at the speaker, o, is given by

o = A*(i + B*o),

snip

Thanks, Bob. I understand what you're driving at, but it's in my
visualization that I am getting stuck. I'll try again to explain where
exactly I am going wrong. Maybe I'm not adding decibels right. :-)

Take the simple mic-amp-speaker loop. I have a widget generating a steady
1kHz tone. I'm in an open field wth the speaker facing away from the mic. I
can crank this bad boy up and not worry about feedback.

Okay, so now I turn the speaker to face the mic. Everything is just right so
that speaker output is in phase with my widget, which is between the speaker
and the mic.

So, using Matrix-style bullet time we see that wave #1 leaves my widget at
60dB and enters the system at 50dB (SPL). A certain number of cycles occur
(9) before wave #1 leaves the speaker and hits my widget. When it hits my
widget it is also at 60dB. So wave #10 leaves the widget at 63dB, right?

Therefore, when wave #10 hits the mic, it will be greater than 50db. Thus
wave #20 (next round trip) will get a bee's dick more energy added to it
perhaps pushing it to 63.1dB. Keep doing this and we get feedback. Since we
know the delay and the gain per round trip, we can predict how quickly the
gain will accellerate.


But what if my speaker is just sending 10dB to the 60dB signal at my widget.
If you tell me that I really only end up with 60dB, not 60.0001dB (or
whatever), then everything makes perfect sense to me. Otherwise I don't
understand how an in-phase signal is not additive every round trip.

The thing is, I *know* what I expect is wrong based on practical experience.
My difficulty is in figuring *what* exactly I am missing. Can you break down
my visualization and explain to me what I am missing?

Jim Carr wrote:

But what if my speaker is just sending 10dB to the 60dB signal at my widget.
If you tell me that I really only end up with 60dB, not 60.0001dB (or
whatever), then everything makes perfect sense to me. Otherwise I don't
understand how an in-phase signal is not additive every round trip.

It is additive, but if the gain roound the loop is less than 1, the
fractions added, even though they form a series going to infinity, still
add up to a finite amount.

For example, if the feedback adds exactly half to the original sound,
you get a series:
1 + 1/2 + 1/4 + 1/8 + 1/16 + 1/32....

Which tends towards adds up to 2 as the number of terms become infinite.

In practical terms this menas the signal does increase at certain
frequencies with the gain set just below feedback squeal level. This
accounts for the familiar ringing sound of a system close to feedback.

Anahata

hank alrich wrote:
Arny Krueger wrote:

If you presume that the system is linear (and just about any modern
equipment is linear enough so that this is a good assumption),

As soon as a room enters the equation, we no longer have a linear
system. And it's somewhat difficult to eliminate the acoustic
environment in live sound.

Huh? What significant non-linearity affects a room response? Not
saying you are wrong, ya unnerstan, I just can't see what would cause that.


Bob
--

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

A. Einstein

Jim Carr wrote:
"Bob Cain" wrote in message
...

So, when in phase the input from the source, i, and output of the system
at the speaker, o, is given by

o = A*(i + B*o),

snip

Thanks, Bob. I understand what you're driving at, but it's in my
visualization that I am getting stuck. I'll try again to explain where
exactly I am going wrong. Maybe I'm not adding decibels right. :-)

Anahata nailed this one.


Bob
--

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

A. Einstein

I'm not sure about that. Can you say "loop gain"?

Loop gain 1 in a linear system will lead to feedback,
even only if the only stimulus is residual noise. There
is no minimum stimulus, below which feedback will not
take place.

You can have feedback -- at least for a short time -- in a system where the
loop gain is 1, if the stimulus is strong enough. We've all seen
situations where the system is silent, then howls when someone starts to
speak.

"William Sommerwerck" wrote in message
...

I'm not sure about that. Can you say "loop gain"?

Loop gain 1 in a linear system will lead to feedback,
even only if the only stimulus is residual noise. There
is no minimum stimulus, below which feedback will not
take place.

You can have feedback -- at least for a short time -- in a system where
the
loop gain is 1, if the stimulus is strong enough. We've all seen
situations where the system is silent, then howls when someone starts to
speak.

That would be for effective gain that is just under 1.0. Basically, its a
filter that is ringing with a long settling time.

If the system gain is 1, the feedback will decrease. If the system is
otherwise quiet, the ringing will eventually stop.

When gain is close to 1.0 seeming small influences can tip things either
way. For example there are always convection currents due to differences in
air temperature in the room. When you get high enough resolution in your
measurments - real-world large-scale (i.e, room-sized) acoustical systems
are never constant, but always varying.

"hank alrich" wrote in message
. ..
Arny Krueger wrote:

If you presume that the system is linear (and just about any modern
equipment is linear enough so that this is a good assumption),

As soon as a room enters the equation, we no longer have a linear
system. And it's somewhat difficult to eliminate the acoustic
environment in live sound.

When is linear enough to ignore any nonlinearities?

It's all about quantification.

Air is linear enough from an engineering standpoint until the SPLs get
pretty high.

The "how linear is air" issue has been around at least since Paul Klipsch
tried to show that the AR-1W woofer was nonlinear due to the air in the box
being compressed.

Klipsch was ultimately proven wrong by Allison, Kloss etc.

Not that the AR-1W woofer would impress many these days, but the problem is
relatively limited Xmax, not nonlinear air compression.

However Klipsch later used a similar approach to target the Bose 901. He
played the Doppler distortion card and won.

For example, air is not linear from an engineering standpoint in the throat
of a 1" throat compression horn when the 1 meter SPL is 130 dB. But it is
linear enough at say 85 dB.

Jim Carr wrote:

I disagree that phase is a red herring. Set up a simple mic, amp and
speaker. Put the gain near ringing, then drop it back just a hair. Now put a
phaser in-line. In my tests I hear the system ring then go "quiet" at a rate
equal to how phast my phaser is going.

What do you think that demonstrates? I mean besides amplitude
nonlinearity when the phase pedal is phasing.

--
ha

Bob Cain wrote:

hank alrich wrote:
Arny Krueger wrote:

If you presume that the system is linear (and just about any modern
equipment is linear enough so that this is a good assumption),

As soon as a room enters the equation, we no longer have a linear
system. And it's somewhat difficult to eliminate the acoustic
environment in live sound.

Huh? What significant non-linearity affects a room response? Not
saying you are wrong, ya unnerstan, I just can't see what would cause that.

What are the +/- dB SPL limits of the room's response (i.e., test the
room like you'd test an electronic audio device and see what's the
result)?

Room dimensions, orientation of boundaries and surface treatments all
contribute to the respoonse characteristics of a room.

Rooms are very lumpy, IME, and a huge factor in what frequencies will
want to feedback. One seeks as linear a room as possible for a mix room,
but one rarely gets as close to perfection as an amp, preamp, or even
good measurement mic. In a show venue things are much worse.

--
ha

hank alrich wrote:
Bob Cain wrote:

hank alrich wrote:
Arny Krueger wrote:

If you presume that the system is linear (and just about any modern
equipment is linear enough so that this is a good assumption),

As soon as a room enters the equation, we no longer have a linear
system. And it's somewhat difficult to eliminate the acoustic
environment in live sound.

Huh? What significant non-linearity affects a room response? Not
saying you are wrong, ya unnerstan, I just can't see what would cause that.

What are the +/- dB SPL limits of the room's response (i.e., test the
room like you'd test an electronic audio device and see what's the
result)?

Do you mean that at some power level, objects and walls will go into the
non-linear regime in the way they reflect, absorb and re-radiate? I'm
sure that is true but not at any listening level below the threshold of
pain.

Room dimensions, orientation of boundaries and surface treatments all
contribute to the respoonse characteristics of a room.

Absolutely. This makes me think that you are using the word linear for
flat. Is that right? From that perspective what you said is clearly true.


Bob
--

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

A. Einstein

Arny Krueger wrote:

"hank alrich" wrote in message
. ..
Arny Krueger wrote:

If you presume that the system is linear (and just about any modern
equipment is linear enough so that this is a good assumption),

As soon as a room enters the equation, we no longer have a linear
system. And it's somewhat difficult to eliminate the acoustic
environment in live sound.

When is linear enough to ignore any nonlinearities?

It's all about quantification.

Air is linear enough from an engineering standpoint until the SPLs get
pretty high.

The "how linear is air" issue has been around at least since Paul Klipsch
tried to show that the AR-1W woofer was nonlinear due to the air in the box
being compressed.

Klipsch was ultimately proven wrong by Allison, Kloss etc.

Not that the AR-1W woofer would impress many these days, but the problem is
relatively limited Xmax, not nonlinear air compression.

However Klipsch later used a similar approach to target the Bose 901. He
played the Doppler distortion card and won.

For example, air is not linear from an engineering standpoint in the throat
of a 1" throat compression horn when the 1 meter SPL is 130 dB. But it is
linear enough at say 85 dB.

And all that has what to do with the supposed linearity of a _room_ the
room being an essential part of the live sound circuit? As soon as the
walls and ceiling are up, linearity haas left the building.

--
ha

"hank alrich" wrote in message
...
Jim Carr wrote:

I disagree that phase is a red herring. Set up a simple mic, amp and
speaker. Put the gain near ringing, then drop it back just a hair. Now
put a
phaser in-line. In my tests I hear the system ring then go "quiet" at a
rate
equal to how phast my phaser is going.

What do you think that demonstrates? I mean besides amplitude
nonlinearity when the phase pedal is phasing.

I think it tells me that phase and amplitude are by definition related and
not red herrings when it comes to feedback. I think that if I take a mic and
point it at a speaker and bring up the gain until I get feedback at
frequency X that I can then move the mic closer to the speaker the feedback
at frequency X will subside when it goes out of phase. I probably will get
feedback at another frequency, but that doesn't mean that phase is not a
component of feedback.

On Mon, 6 Feb 2006 22:00:19 -0700, "Jim Carr"
wrote:

I think it tells me that phase and amplitude are by definition related and
not red herrings when it comes to feedback. I think that if I take a mic and
point it at a speaker and bring up the gain until I get feedback at
frequency X that I can then move the mic closer to the speaker the feedback
at frequency X will subside when it goes out of phase. I probably will get
feedback at another frequency, but that doesn't mean that phase is not a
component of feedback.

1) Phase and amplitude are by definition unrelated.

2) Phase is not a component of feedback.

Next fallacy.

Chris Hornbeck

"Chris Hornbeck" wrote in message
...

1) Phase and amplitude are by definition unrelated.

2) Phase is not a component of feedback.

Next fallacy.

Well, *now* I get it. There's nothing like the dogmatic school of teaching!

If phase and amplitude are unrelated, then explain to me how phase
cancellation works 'cause otherwise I just don't understand.

If phase is not a part of feedback, then explain this: I place a mic at
distance X from a speaker and induce feedback by bringing up the gain. I can
then move the mic a small distance *closer* to the speaker and feedback at
that frequency disappears.

What about the techniques of using delay to reduce feedback? How do they
work if phase is not a component?

I have no doubt that feedback can be created without phase. Take a brief
source sound. Delay it 10 seconds, then run it through the system above
unity gain. Each round trip will be louder than the last with no phase
involved. But that still doesn't explain my observations above.

If you don't have the patience to explain to me where I've gone wrong, then
don't waste your time or mine by replying. I'm willing to learn if someone
is willing to teach. One source from which I have learned is
http://www.rane.com/note158.html. Maybe I am misinterpreting or they are
wrong.

hank alrich wrote:

And all that has what to do with the supposed linearity of a _room_ the
room being an essential part of the live sound circuit? As soon as the
walls and ceiling are up, linearity haas left the building.

But why, Hank?


Bob
--

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

A. Einstein

No linearity has NOT left the building. Feedback systems of all types
are always a hassle to understand because they are very non-intuitive.
The idea of a frequency starting out and going round and round the loop
until it builds up into a squeal is pretty much wrong though it does
give people a way to think.

What is really happening is you have a TOTAL system and to determine
what happens you have to describe it mathematically. You work out the
feedback equations and voila you can tell if it's stable or not. If
not, it "squeals".

Key factors are loop gain and phase. Time delay is also a factor for
system stablility. If the returned signal is out of phase, it REDUCES
the output of the system and also improves linearity. It's what goes on
in amplifiers. If it is in phase (or of the wrong delay) it can
increase the apparent gain of the system. If loop gain gets too large,
instability can occur. And let me point out that instability can take
many forms. If a system is broadband and has no frequency determining
elements a typical instability is called "motorboating" where the sytem
is rapidly driven lock to lock. The actually frequency is determined by
the slew rate of the system and forms a "relaxation" oscillator.

But when there IS a frequency determining element (essential to form a
sinusoidal oscillator) then the instability takes place at that
frequency. In the case of PA and such setups there are LOTS of
frequency determining elements. These include the mic, the speakers and
the room. In a room the boundary equations of the walls form the
frequency selective resonant system. Because lots of different
wavelengths can fit between various walls there are LOTS of resonant
frequencies. But NOTE the ROOM is NOT "non-linear" it is merely
resonant! Non linearity implies an inapropriate output response to a
given input energy. That happens in speakers, amps, mics and the like
but not in rooms. But rooms DO make wonderful frequency selecting
devices. Although there may be lots of resonant frequencies in the
system (including room) feedback will choose to occur at the frequency
with the highest loop gain. Ever notice how the frequency of a feedback
can shift as you move a mic? You are just shifting the system loop
gains so that two similar resonant peaks are shifted in loop gains and
the feedback frequency then shifts to match the higher one.

Sound guys know that cheapo mics are really prone to feedback. The
reason is that these mics have some very pronouced frequency peaks
which thence become the frequency determining element in your
room-sized oscillator. A mic/speaker combo that is very flat and smooth
in resonse forces the ROOM to be the frequency determing element and
that means you can crank higher without feedback. But there is still
more. Even if you have a perfectly flat and phased setup, you can still
get feedback because sound takes time to travel. This means delay. And
pure delay in a feedback system even of NO frequency determining
elements are present can lead to instablility!

The bottom line here is that while a lot is known about the mathematics
of feedback, a room and PA system are so complex with so many variables
that apart from some general principles, eliminating feedback in a
practical case is much more 'art" than science. And a great deal of
that cojmplexity is due to the complexities in the room. But that only
means that the room is complexly resonant not that it is "non-linear"!

In fact it is the non-linearity in the system as a whole that limits
the final amplitude of the oscillations. And typically the non-linear
element in the system that does that is the driving amplifier.

Benj

"hank alrich" wrote in message

Arny Krueger wrote:

"hank alrich" wrote in message
. ..
Arny Krueger wrote:

If you presume that the system is linear (and just
about any modern equipment is linear enough so that
this is a good assumption),

As soon as a room enters the equation, we no longer
have a linear system. And it's somewhat difficult to
eliminate the acoustic environment in live sound.

When is linear enough to ignore any nonlinearities?

It's all about quantification.

Air is linear enough from an engineering standpoint
until the SPLs get pretty high.

The "how linear is air" issue has been around at least
since Paul Klipsch tried to show that the AR-1W woofer
was nonlinear due to the air in the box being compressed.

Klipsch was ultimately proven wrong by Allison, Kloss
etc.

Not that the AR-1W woofer would impress many these days,
but the problem is relatively limited Xmax, not
nonlinear air compression.

However Klipsch later used a similar approach to target
the Bose 901. He played the Doppler distortion card and
won.

For example, air is not linear from an engineering
standpoint in the throat of a 1" throat compression horn
when the 1 meter SPL is 130 dB. But it is linear enough
at say 85 dB.

And all that has what to do with the supposed linearity
of a _room_ the room being an essential part of the live
sound circuit?

I'm wondering if there's a communications problem here. There is this old
usage or common technical words that involves phrases such as "nonlinear
frequency response" which actually means non-uniform frequency response.

Agreed that when the walls go up, the frequency response in the room becomes
very non-uniform.

However, non-uniform frequency response does not mean that the room is
nonlinear. "nonlinear frequency response" is in fact an oxymoron because
frequency response is about the linear properties of the system, just like
phase response is.

The word nonlinear used in the modern sense relates to bad things happening
in the amplitude domain. IOW if applying twice the power to a speaker at the
same frequency does not make the SPL go up by 3 dB, then the speaker is
nonlinear. If applying the same power to a speaker at a different frequency
changes the SPL, then the speaker may still be linear, but it has
non-uniform frequency reponse.

As soon as the walls and ceiling are up,
linearity haas left the building.

Take the simple mic-amp-speaker loop. I have a widget generating a
steady
1kHz tone. I'm in an open field wth the speaker facing away from the
mic. I
can crank this bad boy up and not worry about feedback.

With sufficient gain it WILL feed back, but at a frequency determined
by the response of the speaker facing away from the mic. The widget
(AKA test oscillator) is largely irrelevant to this scenario, unless
the system is otherwise devoid of frequency response peaks (which never
happens in the real world) & is just on the verge of feedback, in which
case the widget will add just enough energy to the system to push it
into oscillation at that frequency. The system will most likely settle
into a different frequency as it stabilizes at a resonance determined
by the response of the system components.

Okay, so now I turn the speaker to face the mic. Everything is just
right so
that speaker output is in phase with my widget, which is between the
speaker
and the mic.


In the real world you don't need the widget. A feedback loop is a tuned
circuit, the resonance of which is determined by the frequency response
of the microphone, the frequency response of the speaker at the
location of the microphone, the distance between the two (which impacts
the frequency response of the speaker at the mic), overall gain, plus
any additional system nonlinearities, such as distortion byproducts.
You don't need to start any given frequency with your widget, although
adding a frequency can sometimes add just enough energy in a nearly
oscillating system to force the resonant tuning to that frequency. This
is usually only the case in sound systems that have been equalized such
that any prominent response peaks have already been eliminated.
Otherwise, without the widget you simply add gain to the system & it
will resonate at the frequency which has the greatest energy, IOW the
biggest peak in the overall frequency response of the tuned circuit. I
know you wanted to know about the math behind this, & this is a
non-math explanation, but you seem to indicate an additional
unnecessary element in your example. The point is that the frequency of
oscillation is determined by the tuning of the circuit, not by the
addition of a frequency to a circuit otherwise untuned to that
frequency.

Scott Fraser

Jim Carr wrote:

"hank alrich" wrote...
Jim Carr wrote:

I disagree that phase is a red herring. Set up a simple mic, amp and
speaker. Put the gain near ringing, then drop it back just a hair. Now
put a phaser in-line. In my tests I hear the system ring then go
"quiet" at a rate equal to how phast my phaser is going.

What do you think that demonstrates? I mean besides amplitude
nonlinearity when the phase pedal is phasing.

I think it tells me that phase and amplitude are by definition related and
not red herrings when it comes to feedback.

Have you taken a look at the amplitude response of that phaser while
it's phasing? How flat is it?

I think that if I take a mic and point it at a speaker and bring up the
gain until I get feedback at frequency X that I can then move the mic
closer to the speaker the feedback at frequency X will subside when it
goes out of phase. I probably will get feedback at another frequency, but
that doesn't mean that phase is not a component of feedback.

Try it.

--
ha

On Tue, 7 Feb 2006 10:25:45 -0500, "Arny Krueger"
wrote:

I'm wondering if there's a communications problem here.

Bingo.

There is this old
usage or common technical words that involves phrases such as "nonlinear
frequency response" which actually means non-uniform frequency response.

But on Usenet, on a slow week, it passes the time. I'd
much rather do this than work on the coding for Customer
From Hell's wall keypads. Hey, it'll keep an hour.

Thanks, as always,

Chris Hornbeck

On Mon, 6 Feb 2006 23:56:39 -0700, "Jim Carr"
wrote:

1) Phase and amplitude are by definition unrelated.

2) Phase is not a component of feedback.

Next fallacy.

Well, *now* I get it. There's nothing like the dogmatic school of teaching!

If phase and amplitude are unrelated, then explain to me how phase
cancellation works 'cause otherwise I just don't understand.

Sorry; didn't mean to come across as short. You didn't
believe me the first try, or the second, but I'm game
for a third attempt.

We're still arguing over appropriate models at this
stage, so specifics like "phase", etc. are premature.

I suggest to you that an appropriate model must begin
with a model of the forward path; that is, the microphone -
electronics - speaker.

Then, separately, the model of the feedback path; that
is, the speaker - room - microphone.

For oscillation to occur, gain though the forward path
*plus* gain through the feedback path must be greater
than unity. Period.

Signal of any amplitude, phase, or religious persuasion
is not necessary or significant in feedback in linear systems.
Systems *very close* to oscillation act as if they're
non-linear, and cause a lot of confusion (in me too).

Thanks for your thoughts,

Chris Hornbeck

Chris Hornbeck wrote:

For oscillation to occur, gain though the forward path
*plus* gain through the feedback path must be greater
than unity. Period.

A pedantic picking of the nit: it's *times* not *plus* unless of course
you mean that dB forward plus dB feedback can't be greater than zero. :-)

Signal of any amplitude, phase, or religious persuasion
is not necessary or significant in feedback in linear systems.

In oscillation, or what we colloquially call audio "feedback", they are.
Except of course...

Systems *very close* to oscillation act as if they're
non-linear, and cause a lot of confusion (in me too).

Yeah, that statement causes confusion in me too. Huh?


Bob
--

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

A. Einstein

"hank alrich" wrote in message
.. .

I think it tells me that phase and amplitude are by definition related
and
not red herrings when it comes to feedback.

Have you taken a look at the amplitude response of that phaser while
it's phasing? How flat is it?

I would assume that when the original signal and the duplicated signal are
in phase, the amplitude is higher. Then as it goes out of phase different
overtones are raised/lowered.

If phase and amplitude are unrelated, then how does phase cancellation work?


I think that if I take a mic and point it at a speaker and bring up the
gain until I get feedback at frequency X that I can then move the mic
closer to the speaker the feedback at frequency X will subside when it
goes out of phase. I probably will get feedback at another frequency,
but
that doesn't mean that phase is not a component of feedback.

Try it.

I did *before* posted that. It happens as I describe.

On Tue, 07 Feb 2006 17:50:17 -0800, Bob Cain
wrote:

Systems *very close* to oscillation act as if they're
non-linear, and cause a lot of confusion (in me too).

Yeah, that statement causes confusion in me too. Huh?

This goes back to the earlier discussion about
whether oscillation was chaotic or linear. You and
Arny convinced me, by weight of conviction, that
oscillation is linear.

My lingering doubts revolve around my own poor
understanding of where the definition should lie.
Chaos, after all, was "discovered" as an artifact
while observing linear processes on an analog
computer.

Linear systems *very close* to oscillation have several
characteristics of nonlinear chaotic systems:
extreme sensitivity to initial conditions,
strongly preferred end states (attractors), and
strongly non-preferred transitional states.



IOW, can a simple system, with *no* degree of
freedom, exhibit chaos? The answer is yes. Purely
mathematical examples exist.

But, is oscillation chaotic? People whose opinions
I respect say no. Still... it's itchy.

Much thanks, as always,

Chris Hornbeck
r.a.p FAQ at www.recaudio.pro.net