I have some vented speakers with an F3 of about 40 Hz in an EBS
alignment. I have severe room interaction problems, clearly audible
and confirmed by LSPCad: a VERY deep valley around 60 Hz. I can't
solve it without really inconvenient furniture rearrangements, so I'm
going to add a subwoofer (closed, Q=0.7, F3=37Hz) actively crossed
over to the main speakers with a 4th order LR filter. I will have to
use a xover frequency of about 100 Hz, above which the main speakers
behave reasonably well where they are placed. Luckily I have found a
very convenient spot for the sub where its response will be very flat
and it's at exactly the same distance from the listening position as
the main speakers.

Just for fun I checked what would happened if I closed the vents in
the main speakers and what I get is a Q of about 0.5, although the
rolloff starts slightly above 100 Hz so there would be a small dip
(less than 1 dB anyway) in the overall frequency response.

I understand the closed main speakers would have a much better
transient response than if I leave them vented, but I'm not sure about
their contribution to the overall system transient response. Since in
any case I'm filtering them electrically with a 4th order slope, does
it really matter how they are aligned below the xover frequency? In
other words, will the overall transient response depend on the box
alignment of the main speakers, or only on the overall system
response, crossover and sub included?

Many thanks in advance,

Carlos

Carlos,

I have severe room interaction problems

No matter what speakers you use and in what room you put them, there will
always be problems unless the room is properly treated. However, large rooms
generally have fewer low frequency problems than small rooms. How large is
your room?

--Ethan

Carlos,

I have severe room interaction problems

No matter what speakers you use and in what room you put them, there will
always be problems unless the room is properly treated. However, large rooms
generally have fewer low frequency problems than small rooms. How large is
your room?

--Ethan

"Ethan Winer" ethanw at ethanwiner dot com wrote in message ...
Carlos,

I have severe room interaction problems

No matter what speakers you use and in what room you put them, there will
always be problems unless the room is properly treated. However, large rooms
generally have fewer low frequency problems than small rooms. How large is
your room?

--Ethan

It's large but squareish (4.6 x 5.5 x 2.5 m), maybe that's the
problem. I have physically tried two possible arrangements and then
simulated in LspCAD all the other "reasonable" ones (i.e. the ones
that will allow me to keep all my furniture), and there's always a big
valley at the listening position, so I think the best solution is the
sub. At its intended spot it will blend very well with the room gain
and will give me a very flat and extended response. Besides, I love
building stuff, so I've got the perfect excuse.

Carlos

"Ethan Winer" ethanw at ethanwiner dot com wrote in message ...
Carlos,

I have severe room interaction problems

No matter what speakers you use and in what room you put them, there will
always be problems unless the room is properly treated. However, large rooms
generally have fewer low frequency problems than small rooms. How large is
your room?

--Ethan

It's large but squareish (4.6 x 5.5 x 2.5 m), maybe that's the
problem. I have physically tried two possible arrangements and then
simulated in LspCAD all the other "reasonable" ones (i.e. the ones
that will allow me to keep all my furniture), and there's always a big
valley at the listening position, so I think the best solution is the
sub. At its intended spot it will blend very well with the room gain
and will give me a very flat and extended response. Besides, I love
building stuff, so I've got the perfect excuse.

Carlos

Carlos,

It's large but squareish (4.6 x 5.5 x 2.5 m)

As far as room acoustics is concerned that's a small room, and so it suffers
from the same problems of skewed low frequency response that are common to
all small rooms.

there's always a big valley at the listening position, so I think the best
solution is the sub.

A big dip in the bass range in a room that size is ALWAYS caused by acoutic
interference due to reflections off the walls, floor, and ceiling. It might
be fun to build a sub, but it's the wrong approach. (Unless your main
loudspeakers really are deficient in the bass range.) Not only will a
subwoofer not solve the real problem, it will likely make things even worse.
However, it will give you more "thump and boom" if that's all you care
about.

Have a look at my Acoustics FAQ, second in the list on my Articles page:

www.ethanwiner.com/articles.html

It's not what you asked for, but I believe it's what you need.

--Ethan

Carlos,

It's large but squareish (4.6 x 5.5 x 2.5 m)

As far as room acoustics is concerned that's a small room, and so it suffers
from the same problems of skewed low frequency response that are common to
all small rooms.

there's always a big valley at the listening position, so I think the best
solution is the sub.

A big dip in the bass range in a room that size is ALWAYS caused by acoutic
interference due to reflections off the walls, floor, and ceiling. It might
be fun to build a sub, but it's the wrong approach. (Unless your main
loudspeakers really are deficient in the bass range.) Not only will a
subwoofer not solve the real problem, it will likely make things even worse.
However, it will give you more "thump and boom" if that's all you care
about.

Have a look at my Acoustics FAQ, second in the list on my Articles page:

www.ethanwiner.com/articles.html

It's not what you asked for, but I believe it's what you need.

--Ethan

"Ethan Winer" ethanw at ethanwiner dot com wrote in message ...
Carlos,

It's large but squareish (4.6 x 5.5 x 2.5 m)

As far as room acoustics is concerned that's a small room, and so it suffers
from the same problems of skewed low frequency response that are common to
all small rooms.

Well, I didn't make myself clear: it's largER than the one I had
before, where I didn't have problems like this. I know, I was just
lucky then.

there's always a big valley at the listening position, so I think the best
solution is the sub.

A big dip in the bass range in a room that size is ALWAYS caused by acoutic
interference due to reflections off the walls, floor, and ceiling.

Of course. That's exactly what LspCAD models and, from the looks of
it, quite accurately. But those reflections depend also on where the
loudspeaker is and where the listening position is, so a possible
solution is to move the source of the frequency band that gives
problems to a different location, provided you can't locate spatially
sounds in that band. I have made experiments in the past and, below
100 Hz, I can't.

It might
be fun to build a sub, but it's the wrong approach. (Unless your main
loudspeakers really are deficient in the bass range.) Not only will a
subwoofer not solve the real problem, it will likely make things even worse.
However, it will give you more "thump and boom" if that's all you care
about.

Well, my main speakers are actually really good in the bass range, but
their bass range gets seriously screwed up by the room acoustics below
100 Hz, and, as I said, I can't move them and/or the listening
position around to improve that; however, the subwoofer, placed in a
certain spot, results in a very flat response (no "thump and boom")
below 100 Hz at the current listening position, so if I design and
build it properly, which I know how to do, and also design and build
properly a crossover that removes the problem band from the main
speakers and feeds it to the sub, which I also know how to do, I will
solve the problem.

I'm not saying other approaches, like changing the room acoustics,
won't work (though I doubt a very deep valley in the frequency
response like the one I have can be solved without very seriously and
disruptively altering the room), I'm just saying there's no reason why
the sub must be "the wrong approach" and "make things worse".

Just to clarify: I'm not adding a boom-box to my system to get more
"thump". I'm adding a very fine Scan-Speak 25W/8565 bass driver in a
closed box with maximally flat alignment adequately crossed over
(Active 4th-order Linkwitz-Riley) to the main speakers and with its
location carefully chosen to have the flattest possible FR. Frankly I
don't see what could be wrong with that.

Carlos

"Ethan Winer" ethanw at ethanwiner dot com wrote in message ...
Carlos,

It's large but squareish (4.6 x 5.5 x 2.5 m)

As far as room acoustics is concerned that's a small room, and so it suffers
from the same problems of skewed low frequency response that are common to
all small rooms.

Well, I didn't make myself clear: it's largER than the one I had
before, where I didn't have problems like this. I know, I was just
lucky then.

there's always a big valley at the listening position, so I think the best
solution is the sub.

A big dip in the bass range in a room that size is ALWAYS caused by acoutic
interference due to reflections off the walls, floor, and ceiling.

Of course. That's exactly what LspCAD models and, from the looks of
it, quite accurately. But those reflections depend also on where the
loudspeaker is and where the listening position is, so a possible
solution is to move the source of the frequency band that gives
problems to a different location, provided you can't locate spatially
sounds in that band. I have made experiments in the past and, below
100 Hz, I can't.

It might
be fun to build a sub, but it's the wrong approach. (Unless your main
loudspeakers really are deficient in the bass range.) Not only will a
subwoofer not solve the real problem, it will likely make things even worse.
However, it will give you more "thump and boom" if that's all you care
about.

Well, my main speakers are actually really good in the bass range, but
their bass range gets seriously screwed up by the room acoustics below
100 Hz, and, as I said, I can't move them and/or the listening
position around to improve that; however, the subwoofer, placed in a
certain spot, results in a very flat response (no "thump and boom")
below 100 Hz at the current listening position, so if I design and
build it properly, which I know how to do, and also design and build
properly a crossover that removes the problem band from the main
speakers and feeds it to the sub, which I also know how to do, I will
solve the problem.

I'm not saying other approaches, like changing the room acoustics,
won't work (though I doubt a very deep valley in the frequency
response like the one I have can be solved without very seriously and
disruptively altering the room), I'm just saying there's no reason why
the sub must be "the wrong approach" and "make things worse".

Just to clarify: I'm not adding a boom-box to my system to get more
"thump". I'm adding a very fine Scan-Speak 25W/8565 bass driver in a
closed box with maximally flat alignment adequately crossed over
(Active 4th-order Linkwitz-Riley) to the main speakers and with its
location carefully chosen to have the flattest possible FR. Frankly I
don't see what could be wrong with that.

Carlos

Carlos,

it's largER than the one I had before, where I didn't have problems like
this.

What's much more likely is you had many peaks and nulls at the listening
position, but they were merely at more pleasing / flattering frequencies.

those reflections depend also on where the loudspeaker is and where the
listening position is

Absolutely. But in a small room there are NO locations that have a flat
response. Again, some places may sound better than others, but NONE of them
are accurate.

my main speakers are actually really good in the bass range, but their
bass range gets seriously screwed up by the room acoustics below 100 Hz,
and, as I said, I can't move them and/or the listening position

If you understand that it's a room problem, why do you resist the idea of
fixing the room?

You are welcome to keep trying to fix an acoustic problem with gear, but in
my opinion it really is the wrong approach.

--Ethan

Carlos,

it's largER than the one I had before, where I didn't have problems like
this.

What's much more likely is you had many peaks and nulls at the listening
position, but they were merely at more pleasing / flattering frequencies.

those reflections depend also on where the loudspeaker is and where the
listening position is

Absolutely. But in a small room there are NO locations that have a flat
response. Again, some places may sound better than others, but NONE of them
are accurate.

my main speakers are actually really good in the bass range, but their
bass range gets seriously screwed up by the room acoustics below 100 Hz,
and, as I said, I can't move them and/or the listening position

If you understand that it's a room problem, why do you resist the idea of
fixing the room?

You are welcome to keep trying to fix an acoustic problem with gear, but in
my opinion it really is the wrong approach.

--Ethan

"Ethan Winer" ethanw at ethanwiner dot com wrote in message ...
Carlos,

it's largER than the one I had before, where I didn't have problems like
this.

What's much more likely is you had many peaks and nulls at the listening
position, but they were merely at more pleasing / flattering frequencies.

Sure. What I said I didn't have is "problems like this", "this" being
the problem of having such a deficiency in the frequency response that
I can clearly hear it and prompts me to try to find a solution. This
is not just a narrow null, it's a very wide valley (my acoustic
perception confirmed by simulation) that makes the system sound as if
it had no bass at all.

those reflections depend also on where the loudspeaker is and where the
listening position is

Absolutely. But in a small room there are NO locations that have a flat
response. Again, some places may sound better than others, but NONE of them
are accurate.

If that was true, the ONLY way of solving the problem would be to get
a larger room. Fortunately it isn't: In the simulation of my room I
can show you a location that has a fairly good response in the band of
interest, of course not absolutely flat, but flat enough to solve my
problem. If you want I can e-mail you the plots.

my main speakers are actually really good in the bass range, but their
bass range gets seriously screwed up by the room acoustics below 100 Hz,
and, as I said, I can't move them and/or the listening position

If you understand that it's a room problem, why do you resist the idea of
fixing the room?

C'mon, Ethan, don't play with words: I, and I'm sure you too,
understand it's a room-loudspeaker interaction problem, so there are
two variables I can play with, and I have decided to play with the
second one and, in the process, I have actually found a solution,
despite which you keep saying:

You are welcome to keep trying to fix an acoustic problem with gear, but in
my opinion it really is the wrong approach.

but not answering my question: WHY is it the wrong approach if it
solves the problem?

--Ethan

"Ethan Winer" ethanw at ethanwiner dot com wrote in message ...
Carlos,

it's largER than the one I had before, where I didn't have problems like
this.

What's much more likely is you had many peaks and nulls at the listening
position, but they were merely at more pleasing / flattering frequencies.

Sure. What I said I didn't have is "problems like this", "this" being
the problem of having such a deficiency in the frequency response that
I can clearly hear it and prompts me to try to find a solution. This
is not just a narrow null, it's a very wide valley (my acoustic
perception confirmed by simulation) that makes the system sound as if
it had no bass at all.

those reflections depend also on where the loudspeaker is and where the
listening position is

Absolutely. But in a small room there are NO locations that have a flat
response. Again, some places may sound better than others, but NONE of them
are accurate.

If that was true, the ONLY way of solving the problem would be to get
a larger room. Fortunately it isn't: In the simulation of my room I
can show you a location that has a fairly good response in the band of
interest, of course not absolutely flat, but flat enough to solve my
problem. If you want I can e-mail you the plots.

my main speakers are actually really good in the bass range, but their
bass range gets seriously screwed up by the room acoustics below 100 Hz,
and, as I said, I can't move them and/or the listening position

If you understand that it's a room problem, why do you resist the idea of
fixing the room?

C'mon, Ethan, don't play with words: I, and I'm sure you too,
understand it's a room-loudspeaker interaction problem, so there are
two variables I can play with, and I have decided to play with the
second one and, in the process, I have actually found a solution,
despite which you keep saying:

You are welcome to keep trying to fix an acoustic problem with gear, but in
my opinion it really is the wrong approach.

but not answering my question: WHY is it the wrong approach if it
solves the problem?

--Ethan

Carlos wrote:

Sure. What I said I didn't have is "problems like this", "this" being
the problem of having such a deficiency in the frequency response that
I can clearly hear it and prompts me to try to find a solution.

If you are short of bass, then as first resort try moving the listening
position closer to a room boundary and/or move the loudspeakers closer
to a room boundary, preferably a corner.

but not answering my question: WHY is it the wrong approach if it
solves the problem?

It is the wrong approach because it doesn't solve the problem - it
creates a problem in the opposite direction, i. e. it clips your
poweramp or destroys your loudspeakers or both - worst case that is. You
could be 10 dB short of bass and that is the difference between "100
watts being enough" or "1000 watts being enough". Also you create
another problem when you equlize: the tonal overall balance may get
right, but the wavefront gets anemic so it doesn't sound right and feel
right anyway.

Moving the listening position away from the mega sonic valley you have
placed in in by mishap costs nothing and preserves equipment headroom.
You may have then too much or too peaky bass, eq may be the simplest
solution, there are also bass absorbers that one can place in corners.
Sometimes putting the loudspeakers in - or nearer - the corner will be
all you need to linearise the response so that a normal tone control can
lighten it up.

Electronic approaches, yes - they may be a part of a solution so that
one can find a compromise that is also livable, after all it is a living
room, but no: they are hardly ever the only adequate solution, dip
filters may occasionally so be, but boosting hardly ever is (with normal
system designs).

--Ethan


Kind regards

Peter Larsen

--
************************************************** ***********
* My site is at: http://www.muyiovatki.dk *
************************************************** ***********

Carlos wrote:

Sure. What I said I didn't have is "problems like this", "this" being
the problem of having such a deficiency in the frequency response that
I can clearly hear it and prompts me to try to find a solution.

If you are short of bass, then as first resort try moving the listening
position closer to a room boundary and/or move the loudspeakers closer
to a room boundary, preferably a corner.

but not answering my question: WHY is it the wrong approach if it
solves the problem?

It is the wrong approach because it doesn't solve the problem - it
creates a problem in the opposite direction, i. e. it clips your
poweramp or destroys your loudspeakers or both - worst case that is. You
could be 10 dB short of bass and that is the difference between "100
watts being enough" or "1000 watts being enough". Also you create
another problem when you equlize: the tonal overall balance may get
right, but the wavefront gets anemic so it doesn't sound right and feel
right anyway.

Moving the listening position away from the mega sonic valley you have
placed in in by mishap costs nothing and preserves equipment headroom.
You may have then too much or too peaky bass, eq may be the simplest
solution, there are also bass absorbers that one can place in corners.
Sometimes putting the loudspeakers in - or nearer - the corner will be
all you need to linearise the response so that a normal tone control can
lighten it up.

Electronic approaches, yes - they may be a part of a solution so that
one can find a compromise that is also livable, after all it is a living
room, but no: they are hardly ever the only adequate solution, dip
filters may occasionally so be, but boosting hardly ever is (with normal
system designs).

--Ethan


Kind regards

Peter Larsen

--
************************************************** ***********
* My site is at: http://www.muyiovatki.dk *
************************************************** ***********

Peter Larsen wrote in message ...

[SNIP]

Sorry Peter, but I never spoke about equalisation and I did mention I
had simulated all PRACTICAL (I have some furniture, you know)
loudspeaker / listening position combinations (including, but of
course, moving the speakers closer to the back wall and shifting the
listening position) and none solved the problem. What DOES solve the
problem is a subwoofer placed in a spot that gives a very flat
response at the current listening position. Please re-read the whole
thread.

In fact, I never requested advice on that problem, I just stated that
I had found a way to solve it in a particular way and asked about the
influence on the transient response of my system of the alignment of
the main speakers once their bass has been filtered out by the
crossover. As it often happens in chaotic cyberspace:

- I was taken for what I'm not: a bass freak wanting a lot of boom and
thump in his system and with no idea of the whys and hows of room -
loudspeaker interaction problems;

- Although, just to put things in context, I happened to explain that
I HAD FOUND A SOLUTION to a particular problem in that area, I was
repeatedly told I actually hadn't because my approach is wrong (no
further explanation as to why that is so) and given an alternative
approach to finding it;

- The thread lost completely its initial topic and NOBODY ANSWERED MY
QUESTION.

Fortunately, as I saw how things were going, I did some more searching
myself and found the answer from Mr. Pierce (who else?) in an old
thread on the same topic.

Having said that, I do thank you and Ethan because I know you were
just trying to help.

Case closed, I hope.

All the best,

Carlos

Peter Larsen wrote in message ...

[SNIP]

Sorry Peter, but I never spoke about equalisation and I did mention I
had simulated all PRACTICAL (I have some furniture, you know)
loudspeaker / listening position combinations (including, but of
course, moving the speakers closer to the back wall and shifting the
listening position) and none solved the problem. What DOES solve the
problem is a subwoofer placed in a spot that gives a very flat
response at the current listening position. Please re-read the whole
thread.

In fact, I never requested advice on that problem, I just stated that
I had found a way to solve it in a particular way and asked about the
influence on the transient response of my system of the alignment of
the main speakers once their bass has been filtered out by the
crossover. As it often happens in chaotic cyberspace:

- I was taken for what I'm not: a bass freak wanting a lot of boom and
thump in his system and with no idea of the whys and hows of room -
loudspeaker interaction problems;

- Although, just to put things in context, I happened to explain that
I HAD FOUND A SOLUTION to a particular problem in that area, I was
repeatedly told I actually hadn't because my approach is wrong (no
further explanation as to why that is so) and given an alternative
approach to finding it;

- The thread lost completely its initial topic and NOBODY ANSWERED MY
QUESTION.

Fortunately, as I saw how things were going, I did some more searching
myself and found the answer from Mr. Pierce (who else?) in an old
thread on the same topic.

Having said that, I do thank you and Ethan because I know you were
just trying to help.

Case closed, I hope.

All the best,

Carlos

Carlos wrote:

- I was taken for what I'm not: a bass freak wanting a lot of boom

No.

- The thread lost completely its initial topic and NOBODY ANSWERED MY
QUESTION.

Erm yes, such can happen. I didn't actually read it from the very
beginning.

Fortunately, as I saw how things were going, I did some more searching
myself and found the answer from Mr. Pierce (who else?) in an old
thread on the same topic.

Having said that, I do thank you and Ethan because I know you were
just trying to help.

[bowing]

Case closed, I hope.

Certainly, deploying a subwoofer where it provides linear response at
the listening position as you say you did is the same type suggestion as
mine of moving the main loudspeakers closer to the room boundary.

All the best,

Same to you, thanks for tidying the thread up.

Carlos


Kind regards

Peter Larsen

--
************************************************** ***********
* My site is at: http://www.muyiovatki.dk *
************************************************** ***********

Carlos wrote:

- I was taken for what I'm not: a bass freak wanting a lot of boom

No.

- The thread lost completely its initial topic and NOBODY ANSWERED MY
QUESTION.

Erm yes, such can happen. I didn't actually read it from the very
beginning.

Fortunately, as I saw how things were going, I did some more searching
myself and found the answer from Mr. Pierce (who else?) in an old
thread on the same topic.

Having said that, I do thank you and Ethan because I know you were
just trying to help.

[bowing]

Case closed, I hope.

Certainly, deploying a subwoofer where it provides linear response at
the listening position as you say you did is the same type suggestion as
mine of moving the main loudspeakers closer to the room boundary.

All the best,

Same to you, thanks for tidying the thread up.

Carlos


Kind regards

Peter Larsen

--
************************************************** ***********
* My site is at: http://www.muyiovatki.dk *
************************************************** ***********

Carlos,

This is not just a narrow null, it's a very wide valley (my acoustic
perception confirmed by simulation) that makes the system sound as if it had
no bass at all.

I realize this! The cause of the problem is acoustic interference as waves
bounce of the walls, floor, and ceiling, and combine with each other and the
direct sound from the loudspeakers. Yes, you can fool around with
loudspeaker placement an in attempt to shift the problem frequencies and
where they occur. But since the cause is acoustics based and so, in my
opinion, should be the solution.

I can show you a location that has a fairly good response in the band of
interest

How would you do that? You need to meaure in 1 Hz increments to truly see
what's going on. Standard 1/3 octave tests average far too wide a range to
be useful. Even 1/12 octaves is too coarse to get an accurate reading.

I'm sure you too, understand it's a room-loudspeaker interaction problem

This is exactly my point. It is NOT so much an interaction as simple
acoustic interference, no matter how badly you want it to be interaction.

WHY is it the wrong approach if it solves the problem?

Because it doesn't solve the problem! If you could solve the problem we
wouldn't be having this discussion!

--Ethan

Carlos,

This is not just a narrow null, it's a very wide valley (my acoustic
perception confirmed by simulation) that makes the system sound as if it had
no bass at all.

I realize this! The cause of the problem is acoustic interference as waves
bounce of the walls, floor, and ceiling, and combine with each other and the
direct sound from the loudspeakers. Yes, you can fool around with
loudspeaker placement an in attempt to shift the problem frequencies and
where they occur. But since the cause is acoustics based and so, in my
opinion, should be the solution.

I can show you a location that has a fairly good response in the band of
interest

How would you do that? You need to meaure in 1 Hz increments to truly see
what's going on. Standard 1/3 octave tests average far too wide a range to
be useful. Even 1/12 octaves is too coarse to get an accurate reading.

I'm sure you too, understand it's a room-loudspeaker interaction problem

This is exactly my point. It is NOT so much an interaction as simple
acoustic interference, no matter how badly you want it to be interaction.

WHY is it the wrong approach if it solves the problem?

Because it doesn't solve the problem! If you could solve the problem we
wouldn't be having this discussion!

--Ethan

"Ethan Winer" ethanw at ethanwiner dot com wrote in message ...
I can show you a location that has a fairly good response in the band of
interest

How would you do that? You need to meaure in 1 Hz increments to truly see
what's going on. Standard 1/3 octave tests average far too wide a range to
be useful. Even 1/12 octaves is too coarse to get an accurate reading.

Really, now, Ethan. Tell us, how wide is 1/12 octave at, say, 20 Hz?

I'm sure you too, understand it's a room-loudspeaker interaction problem

This is exactly my point. It is NOT so much an interaction as simple
acoustic interference, no matter how badly you want it to be interaction.

Mr. Winer, would you care not to, as in a previous discussion,
argue semantics but rather deal with facts?

WHY is it the wrong approach if it solves the problem?

Because it doesn't solve the problem! If you could solve the problem we
wouldn't be having this discussion!

No, it's because Ethan's only tool is a hammer, and thus ALL
problems look like nails to him.

"Ethan Winer" ethanw at ethanwiner dot com wrote in message ...
I can show you a location that has a fairly good response in the band of
interest

How would you do that? You need to meaure in 1 Hz increments to truly see
what's going on. Standard 1/3 octave tests average far too wide a range to
be useful. Even 1/12 octaves is too coarse to get an accurate reading.

Really, now, Ethan. Tell us, how wide is 1/12 octave at, say, 20 Hz?

I'm sure you too, understand it's a room-loudspeaker interaction problem

This is exactly my point. It is NOT so much an interaction as simple
acoustic interference, no matter how badly you want it to be interaction.

Mr. Winer, would you care not to, as in a previous discussion,
argue semantics but rather deal with facts?

WHY is it the wrong approach if it solves the problem?

Because it doesn't solve the problem! If you could solve the problem we
wouldn't be having this discussion!

No, it's because Ethan's only tool is a hammer, and thus ALL
problems look like nails to him.

(Dick Pierce) wrote in message . com...
"Ethan Winer" ethanw at ethanwiner dot com wrote in message ...

Because it doesn't solve the problem! If you could solve the problem we
wouldn't be having this discussion!

No, it's because Ethan's only tool is a hammer, and thus ALL
problems look like nails to him.

If you had actually READ my original post, Ethan, we wouldn't be
having this discussion. The problem I *mentioned* (which had nothing
to do with the question I *asked*) may actually have been a nail. But
I made it clear it was already well nailed in. I used a different
hammer, but hey, IT WORKED (despite your stubborn denial).

Going back to my original question, Dick, I have understood that the
overall system transient response will depend on the overall frequency
response (all subs and xovers included) and therefore the alignment of
the main speakers below xover frequency is irrelevant.

However, this link:

http://www.silcom.com/~aludwig/Crossover_demos.html

shows square wave responses of different xover topologies / orders and
there are substantial differences. Now, if they are well designed and
give a flat FR, i.e. their contribution to the system FR is the same,
why should they affect the overall transient response? Or am I wrong
in assuming that the square wave response is determined by the
transient response? What worries me is that the 4th order
Linkwitz-Riley, which is what I was planning to use, doesn't look very
good.

Many thanks in advance,

Carlos

(Dick Pierce) wrote in message . com...
"Ethan Winer" ethanw at ethanwiner dot com wrote in message ...

Because it doesn't solve the problem! If you could solve the problem we
wouldn't be having this discussion!

No, it's because Ethan's only tool is a hammer, and thus ALL
problems look like nails to him.

If you had actually READ my original post, Ethan, we wouldn't be
having this discussion. The problem I *mentioned* (which had nothing
to do with the question I *asked*) may actually have been a nail. But
I made it clear it was already well nailed in. I used a different
hammer, but hey, IT WORKED (despite your stubborn denial).

Going back to my original question, Dick, I have understood that the
overall system transient response will depend on the overall frequency
response (all subs and xovers included) and therefore the alignment of
the main speakers below xover frequency is irrelevant.

However, this link:

http://www.silcom.com/~aludwig/Crossover_demos.html

shows square wave responses of different xover topologies / orders and
there are substantial differences. Now, if they are well designed and
give a flat FR, i.e. their contribution to the system FR is the same,
why should they affect the overall transient response? Or am I wrong
in assuming that the square wave response is determined by the
transient response? What worries me is that the 4th order
Linkwitz-Riley, which is what I was planning to use, doesn't look very
good.

Many thanks in advance,

Carlos

Dick,

Tell us, how wide is 1/12 octave at, say, 20 Hz?

It doesn't much matter because that's not where his problems are! The
perception of "not enough bass" is around 70-200 Hz. The graph at
www.ethanwiner.com/response.gif shows actual measurements I made in a
friend's smallish control room. Note in the upper portion that the peaks and
dips between 130 and 200 Hz are all about 1/12 octave apart. So, depending
on which 1/12 octave frequencies you measure, you could see the true
response, a flat line, or something in between.

Ethan's only tool is a hammer, and thus ALL problems look like nails to
him.

There's no need to be rude or insulting. If you disagree with something I
say, just address the issue in a professional and civil manner.

Also, I was disappointed you didn't stop by my company's booth at the AES
show and say Hi. I really wanted to meet you in person (and still do).

--Ethan

Dick,

Tell us, how wide is 1/12 octave at, say, 20 Hz?

It doesn't much matter because that's not where his problems are! The
perception of "not enough bass" is around 70-200 Hz. The graph at
www.ethanwiner.com/response.gif shows actual measurements I made in a
friend's smallish control room. Note in the upper portion that the peaks and
dips between 130 and 200 Hz are all about 1/12 octave apart. So, depending
on which 1/12 octave frequencies you measure, you could see the true
response, a flat line, or something in between.

Ethan's only tool is a hammer, and thus ALL problems look like nails to
him.

There's no need to be rude or insulting. If you disagree with something I
say, just address the issue in a professional and civil manner.

Also, I was disappointed you didn't stop by my company's booth at the AES
show and say Hi. I really wanted to meet you in person (and still do).

--Ethan

(Carlos) wrote in message


shows square wave responses of different xover topologies / orders and
there are substantial differences. Now, if they are well designed and
give a flat FR, i.e. their contribution to the system FR is the same,
why should they affect the overall transient response? Or am I wrong
in assuming that the square wave response is determined by the
transient response? What worries me is that the 4th order
Linkwitz-Riley, which is what I was planning to use, doesn't look very
good.


A crossover can have a flat response and yet can degrade the square
wave response of a signal passing through it. All 4th order crossovers
have non-linear group delay. (Non-linear group delay causes various
frequencies to take different times to pass through the filter.)

A square wave, is componsed of a fundamental and all odd harmonics. If
the harmonics are delayed longer than the fundamental, the square wave
will lose it's shape.

It seem conterintuitive that a filter can be flat, yet delay various
frequencies different amounts, but it happens. The Linkwitz-Riley is a
flat filter, but it has a non-linear group delay and thus it will have
poor transient response.

Bob Stanton

(Carlos) wrote in message


shows square wave responses of different xover topologies / orders and
there are substantial differences. Now, if they are well designed and
give a flat FR, i.e. their contribution to the system FR is the same,
why should they affect the overall transient response? Or am I wrong
in assuming that the square wave response is determined by the
transient response? What worries me is that the 4th order
Linkwitz-Riley, which is what I was planning to use, doesn't look very
good.


A crossover can have a flat response and yet can degrade the square
wave response of a signal passing through it. All 4th order crossovers
have non-linear group delay. (Non-linear group delay causes various
frequencies to take different times to pass through the filter.)

A square wave, is componsed of a fundamental and all odd harmonics. If
the harmonics are delayed longer than the fundamental, the square wave
will lose it's shape.

It seem conterintuitive that a filter can be flat, yet delay various
frequencies different amounts, but it happens. The Linkwitz-Riley is a
flat filter, but it has a non-linear group delay and thus it will have
poor transient response.

Bob Stanton

(Bob-Stanton) wrote in message . com...



... The Linkwitz-Riley is a
flat filter, but it has a non-linear group delay and thus it will have
poor transient response.


I need to make a correction here.

I used an incorrect term: "non-linear group delay". I was thinking of
group delay that changes across the band.

It would have been better if I had used the term "non-linear phase
shift".

Bob Stanton

(Bob-Stanton) wrote in message . com...



... The Linkwitz-Riley is a
flat filter, but it has a non-linear group delay and thus it will have
poor transient response.


I need to make a correction here.

I used an incorrect term: "non-linear group delay". I was thinking of
group delay that changes across the band.

It would have been better if I had used the term "non-linear phase
shift".

Bob Stanton

"Bob-Stanton" wrote in message
om
(Carlos) wrote in message


shows square wave responses of different xover topologies / orders
and there are substantial differences. Now, if they are well
designed and give a flat FR, i.e. their contribution to the system
FR is the same, why should they affect the overall transient
response? Or am I wrong in assuming that the square wave response is
determined by the transient response? What worries me is that the
4th order Linkwitz-Riley, which is what I was planning to use,
doesn't look very good.


A crossover can have a flat response and yet can degrade the square
wave response of a signal passing through it. All 4th order crossovers
have non-linear group delay. (Non-linear group delay causes various
frequencies to take different times to pass through the filter.)

A square wave, is componsed of a fundamental and all odd harmonics. If
the harmonics are delayed longer than the fundamental, the square wave
will lose it's shape.

It seem conterintuitive that a filter can be flat, yet delay various
frequencies different amounts, but it happens. The Linkwitz-Riley is a
flat filter, but it has a non-linear group delay and thus it will have
poor transient response.

Not a matter of just "will" but simple fact. The only way to rationalize the
transient response of a Linkwitz-Riley filter is to argue that absence of
phase shift and perfect transient response aren't absolute necessities. The
good news is that it's hard to do experiments that show that phase shift and
perfect transient response are important, if other issues are hadled well
like frequency response on, and off-axis.

Listening tests involving Linkwitz-Riley filters that trash phase and
transient response:

http://www.pcabx.com/technical/LR-300-3K/index.htm

"Bob-Stanton" wrote in message
om
(Carlos) wrote in message


shows square wave responses of different xover topologies / orders
and there are substantial differences. Now, if they are well
designed and give a flat FR, i.e. their contribution to the system
FR is the same, why should they affect the overall transient
response? Or am I wrong in assuming that the square wave response is
determined by the transient response? What worries me is that the
4th order Linkwitz-Riley, which is what I was planning to use,
doesn't look very good.


A crossover can have a flat response and yet can degrade the square
wave response of a signal passing through it. All 4th order crossovers
have non-linear group delay. (Non-linear group delay causes various
frequencies to take different times to pass through the filter.)

A square wave, is componsed of a fundamental and all odd harmonics. If
the harmonics are delayed longer than the fundamental, the square wave
will lose it's shape.

It seem conterintuitive that a filter can be flat, yet delay various
frequencies different amounts, but it happens. The Linkwitz-Riley is a
flat filter, but it has a non-linear group delay and thus it will have
poor transient response.

Not a matter of just "will" but simple fact. The only way to rationalize the
transient response of a Linkwitz-Riley filter is to argue that absence of
phase shift and perfect transient response aren't absolute necessities. The
good news is that it's hard to do experiments that show that phase shift and
perfect transient response are important, if other issues are hadled well
like frequency response on, and off-axis.

Listening tests involving Linkwitz-Riley filters that trash phase and
transient response:

http://www.pcabx.com/technical/LR-300-3K/index.htm

"Ethan Winer" ethanw at ethanwiner dot com wrote in message ...
Dick,

Tell us, how wide is 1/12 octave at, say, 20 Hz?

It doesn't much matter because that's not where his problems are! The
perception of "not enough bass" is around 70-200 Hz. The graph at
www.ethanwiner.com/response.gif shows actual measurements I made in a
friend's smallish control room. Note in the upper portion that the peaks and
dips between 130 and 200 Hz are all about 1/12 octave apart. So, depending
on which 1/12 octave frequencies you measure, you could see the true
response, a flat line, or something in between.

If those are the resuults you getm then you're not doing your filtering properly.

Ethan's only tool is a hammer, and thus ALL problems look like nails to
him.

There's no need to be rude or insulting. If you disagree with something I
say, just address the issue in a professional and civil manner.

Not insulting at all: your penchant to treat all problems as "room"
problems because you seem to be in the business of solving room
problems, whether it works or not, has made you COMPLETELY miss the
point of the poster's question, and lead you down a path which has
NOTHING to do with answering the poster's question.

"To someone that has a hammer, all problems look like nails"

is hardly a rude insult: it is, in fact, a common analogy used in
a wide variety of instances, including professional. If you are
insulted by it, perhaps you might want to consider expanding your
toolbox.

Also, I was disappointed you didn't stop by my company's booth at the AES
show and say Hi.

Do you fear, sir, that I have snubbed you?

If so, I did it you by spending the entire time of AES convention
225 miles away working on several clients projects.

"Ethan Winer" ethanw at ethanwiner dot com wrote in message ...
Dick,

Tell us, how wide is 1/12 octave at, say, 20 Hz?

It doesn't much matter because that's not where his problems are! The
perception of "not enough bass" is around 70-200 Hz. The graph at
www.ethanwiner.com/response.gif shows actual measurements I made in a
friend's smallish control room. Note in the upper portion that the peaks and
dips between 130 and 200 Hz are all about 1/12 octave apart. So, depending
on which 1/12 octave frequencies you measure, you could see the true
response, a flat line, or something in between.

If those are the resuults you getm then you're not doing your filtering properly.

Ethan's only tool is a hammer, and thus ALL problems look like nails to
him.

There's no need to be rude or insulting. If you disagree with something I
say, just address the issue in a professional and civil manner.

Not insulting at all: your penchant to treat all problems as "room"
problems because you seem to be in the business of solving room
problems, whether it works or not, has made you COMPLETELY miss the
point of the poster's question, and lead you down a path which has
NOTHING to do with answering the poster's question.

"To someone that has a hammer, all problems look like nails"

is hardly a rude insult: it is, in fact, a common analogy used in
a wide variety of instances, including professional. If you are
insulted by it, perhaps you might want to consider expanding your
toolbox.

Also, I was disappointed you didn't stop by my company's booth at the AES
show and say Hi.

Do you fear, sir, that I have snubbed you?

If so, I did it you by spending the entire time of AES convention
225 miles away working on several clients projects.

(Carlos) wrote in message . com...
(Dick Pierce) wrote in message . com...
"Ethan Winer" ethanw at ethanwiner dot com wrote in message ...

Because it doesn't solve the problem! If you could solve the problem we
wouldn't be having this discussion!

No, it's because Ethan's only tool is a hammer, and thus ALL
problems look like nails to him.

If you had actually READ my original post, Ethan, we wouldn't be
having this discussion. The problem I *mentioned* (which had nothing
to do with the question I *asked*) may actually have been a nail. But
I made it clear it was already well nailed in. I used a different
hammer, but hey, IT WORKED (despite your stubborn denial).

Going back to my original question, Dick, I have understood that the
overall system transient response will depend on the overall frequency
response (all subs and xovers included) and therefore the alignment of
the main speakers below xover frequency is irrelevant.

However, this link:

http://www.silcom.com/~aludwig/Crossover_demos.html

shows square wave responses of different xover topologies / orders and
there are substantial differences. Now, if they are well designed and
give a flat FR, i.e. their contribution to the system FR is the same,
why should they affect the overall transient response?

Because you and many others make the mistake of assuming that there
is an exact and unique link between frequency response and transient
response. This is ONLY true if the system has what is called "minimum
phase" behavior. Indeed, the very definition of "minimum phase" is that
the phase response is a unique transform of the frequency response
of the system.

In other words, if you measured the frequency response of a system,
and, mathematically, derived the phase response of the system from
that response (the process involves the Hilbert transform), you'd
get one phase response. But if you measured the actual phase response
of the system and go something different, you'd be looking at a non-
minimum phase system.

Let's look at one VERY simple example: a simple linear, non-dispersive
delay, something as simple as a 10 foot piece of air through which
sound is traveling. If we look at the frequency response of that
delay, it's flat, just plain flat (or falt to well within what we are
looking at here). Now, the Hilbert transform of that flat response is
a flat, 0 phase response across the entire band. Yet if we MEASURE
the phase, we will find the phase increasing monotonically with
frequency. At 1 kHz, for example, the phase will be 3600 degrees,
while at 2 kHz, it's 7200 degress (approximately). Since the phase
response is NOT the same as the Hilbert phase response, the system has
non-minimum-phase behavior.

Now, change the distance over which the sound travels: the resulting
frequency response is the same, but the pahse response is different,
again, non-minimum-phase behavior.

These are seemingly silly and trivial examples, but they serve to
precisely illustrated the point: there is not necessarily unique
relation between frequency and phase response UNLESS the system
is known to be minimum phase.

Now, before we co on to another example, let's just make sure that
EVERYONE understand that non-minimum-phase behavior IS NOT NECESSARILY
A BAD THING: these two illustrations above show that clearly and
unambiguoulsy: the non-minimum phase behavior of a simple acoustical
delay is clearly not determinental to the "sound" of the system, per se.

Now, let's look at a case more closer to your point. Consider a
typical two-way speaker system with a symmetrical 2nd-order crossover
network. We'll assume that, within their bands, the drivers behave
ideally, and the net acoustical response of each driver is "ideal,"
i.e., it has the desired 2nd order low-pass or high-pass response at
the right frequency, and that there is no effective delays involved.

If we were to simply sum the in-phase response of these two drivers,
we would get a non-flat frequency response at the crossover point
due to the fact that in the woofer, you have -90 degrees of phase
rotation at the crossover, and in the tweeter, you have +90 degrees,
and the result is that the two are at equal amplitude and 180 degrees
out of phase with one another, and the cancel at the crossover.

The common technique, then is to reverse the phase of the tweeter,
so it's +90 is converted into +90-180 or -90 degrees, and now it's
in phase with the woofer and, voila, we have flat frequency response
through the corssover and beyond.

HOWEVER, we notice that the TOTAL phase of the system (ignoring the
acoustical delay) undergoes a 180 degree rotation starting from
0 below the corssover to -180 degrees above it. Now, that is NOT
the phase predicted by the Hilbert transform of the frequency response.
The frequency response is flat: it's Hilbert transform results in a
phase response that's also flat, but the speaker's ACTUAL phase shows
this 180 degree rotation: the resulting speaker has nopn-minimum-phase
behavior, different than that of simple delay.

Let's look at it another way. The complex response of the system can
can be examined in pieces and in total. The complex transfer function
of the woofer, with its 2nd order low-pass filter, is:

fw = 1 / (s^2 + ds + 1)

where s is the complex frequency variable and d is a damping coefficient.

while the high-pass filter on the tweeter has the transfer function of

ft = s^2 / (s^2 + ds + 1)

So, sum them together, as we get:

fwt = (s^2 + 1) / (s^2 + ds + 1)

Which is CLEARLY not unity. Perfect summation would have to be:

(s^2 + ds + 1) / (s^2 + ds + 1)

so CLEARLY there is a portion missing:

ds / (s^2 + ds + 1)

Now, what does phase have to do with transient response? Simple:
phase is merely one orthogonal view of the complex transfer function
of the system. Phase is essentially a frequency-dependent time variable.
Fourier shows us that, for example, one can transform the impulse
response of the system into the complex transfer function for the
system via the Fourier transform. THat complex transfer function is
in the form of the real and imaginary responses of the system. What
we call the "frequency response" is really the modulus or the vector
sum of the real and imaginary parts of the response, i.e.:

modulus = sqrt(real^2 + imag^2)

while the phase is essentially the resulting angle of the modulus
vector on the complex s-plane, or more directly:

phase is arctan(imag/real)

So, what this is telling us is that the frequency response ALONE
(or the phase response ALONE) does NOT tell us what the transient
response of the system will be. It also tells us that for a given
frequency response, without the contraining knowledge of the
phase response, there are an infinite number of possible transient
responses that will fit that frequency response.

Some simple concepts to help in answering your next question:

Transient response is the behovior of the system in the amplitude
vs time domain.

Frequency response is the behavior of the system in the modulus
of complex amplitude vs frequency domain.

Or am I wrong
in assuming that the square wave response is determined by the
transient response?

Yes, completely, since you are talking, in both cases, about the
behavior of amplitude vs time. It is NOT possible to have the same
square wave response and a different transient response within the
limits of the behavior of the signals at hand.

What worries me is that the 4th order
Linkwitz-Riley, which is what I was planning to use, doesn't look very
good.

"doesn't look very good" needs some definitions and constraining,
since we've now leapt from the realm of objective definitions and
examinations of physical system behavior to value judgements.

A Linkwitz-Rilet system is simply another system with non-minimum-
phase behavior. It can get you flat frequency response but will not
necessarily, therefore, give you perfect transient response.

As to whether it's "very good" or not requires someone, you, in this
case, to make a judgement call: which evil to you consider most
tolerable. With the information you provided, it's not possible for
me to make a recommendation that carries any weight. My guess is
that IF the 4th LR is implemented properly, the concerns you have for
the transient response issues are overly emphasised.

One final note: "minimum phase" has a very specific technical definition,
summarized above. You'll see a similar term, "linear phase" which also
has a quite specific and unambiguous definition: it means a phase
which increase in direct proportion to frequency and is characteristic
of simple linear delay. There are a lot of other buzz words that have
the word "phase" in them, most of which, in the realm of speakers, are
marketing gobbledygook.

(Carlos) wrote in message . com...
(Dick Pierce) wrote in message . com...
"Ethan Winer" ethanw at ethanwiner dot com wrote in message ...

Because it doesn't solve the problem! If you could solve the problem we
wouldn't be having this discussion!

No, it's because Ethan's only tool is a hammer, and thus ALL
problems look like nails to him.

If you had actually READ my original post, Ethan, we wouldn't be
having this discussion. The problem I *mentioned* (which had nothing
to do with the question I *asked*) may actually have been a nail. But
I made it clear it was already well nailed in. I used a different
hammer, but hey, IT WORKED (despite your stubborn denial).

Going back to my original question, Dick, I have understood that the
overall system transient response will depend on the overall frequency
response (all subs and xovers included) and therefore the alignment of
the main speakers below xover frequency is irrelevant.

However, this link:

http://www.silcom.com/~aludwig/Crossover_demos.html

shows square wave responses of different xover topologies / orders and
there are substantial differences. Now, if they are well designed and
give a flat FR, i.e. their contribution to the system FR is the same,
why should they affect the overall transient response?

Because you and many others make the mistake of assuming that there
is an exact and unique link between frequency response and transient
response. This is ONLY true if the system has what is called "minimum
phase" behavior. Indeed, the very definition of "minimum phase" is that
the phase response is a unique transform of the frequency response
of the system.

In other words, if you measured the frequency response of a system,
and, mathematically, derived the phase response of the system from
that response (the process involves the Hilbert transform), you'd
get one phase response. But if you measured the actual phase response
of the system and go something different, you'd be looking at a non-
minimum phase system.

Let's look at one VERY simple example: a simple linear, non-dispersive
delay, something as simple as a 10 foot piece of air through which
sound is traveling. If we look at the frequency response of that
delay, it's flat, just plain flat (or falt to well within what we are
looking at here). Now, the Hilbert transform of that flat response is
a flat, 0 phase response across the entire band. Yet if we MEASURE
the phase, we will find the phase increasing monotonically with
frequency. At 1 kHz, for example, the phase will be 3600 degrees,
while at 2 kHz, it's 7200 degress (approximately). Since the phase
response is NOT the same as the Hilbert phase response, the system has
non-minimum-phase behavior.

Now, change the distance over which the sound travels: the resulting
frequency response is the same, but the pahse response is different,
again, non-minimum-phase behavior.

These are seemingly silly and trivial examples, but they serve to
precisely illustrated the point: there is not necessarily unique
relation between frequency and phase response UNLESS the system
is known to be minimum phase.

Now, before we co on to another example, let's just make sure that
EVERYONE understand that non-minimum-phase behavior IS NOT NECESSARILY
A BAD THING: these two illustrations above show that clearly and
unambiguoulsy: the non-minimum phase behavior of a simple acoustical
delay is clearly not determinental to the "sound" of the system, per se.

Now, let's look at a case more closer to your point. Consider a
typical two-way speaker system with a symmetrical 2nd-order crossover
network. We'll assume that, within their bands, the drivers behave
ideally, and the net acoustical response of each driver is "ideal,"
i.e., it has the desired 2nd order low-pass or high-pass response at
the right frequency, and that there is no effective delays involved.

If we were to simply sum the in-phase response of these two drivers,
we would get a non-flat frequency response at the crossover point
due to the fact that in the woofer, you have -90 degrees of phase
rotation at the crossover, and in the tweeter, you have +90 degrees,
and the result is that the two are at equal amplitude and 180 degrees
out of phase with one another, and the cancel at the crossover.

The common technique, then is to reverse the phase of the tweeter,
so it's +90 is converted into +90-180 or -90 degrees, and now it's
in phase with the woofer and, voila, we have flat frequency response
through the corssover and beyond.

HOWEVER, we notice that the TOTAL phase of the system (ignoring the
acoustical delay) undergoes a 180 degree rotation starting from
0 below the corssover to -180 degrees above it. Now, that is NOT
the phase predicted by the Hilbert transform of the frequency response.
The frequency response is flat: it's Hilbert transform results in a
phase response that's also flat, but the speaker's ACTUAL phase shows
this 180 degree rotation: the resulting speaker has nopn-minimum-phase
behavior, different than that of simple delay.

Let's look at it another way. The complex response of the system can
can be examined in pieces and in total. The complex transfer function
of the woofer, with its 2nd order low-pass filter, is:

fw = 1 / (s^2 + ds + 1)

where s is the complex frequency variable and d is a damping coefficient.

while the high-pass filter on the tweeter has the transfer function of

ft = s^2 / (s^2 + ds + 1)

So, sum them together, as we get:

fwt = (s^2 + 1) / (s^2 + ds + 1)

Which is CLEARLY not unity. Perfect summation would have to be:

(s^2 + ds + 1) / (s^2 + ds + 1)

so CLEARLY there is a portion missing:

ds / (s^2 + ds + 1)

Now, what does phase have to do with transient response? Simple:
phase is merely one orthogonal view of the complex transfer function
of the system. Phase is essentially a frequency-dependent time variable.
Fourier shows us that, for example, one can transform the impulse
response of the system into the complex transfer function for the
system via the Fourier transform. THat complex transfer function is
in the form of the real and imaginary responses of the system. What
we call the "frequency response" is really the modulus or the vector
sum of the real and imaginary parts of the response, i.e.:

modulus = sqrt(real^2 + imag^2)

while the phase is essentially the resulting angle of the modulus
vector on the complex s-plane, or more directly:

phase is arctan(imag/real)

So, what this is telling us is that the frequency response ALONE
(or the phase response ALONE) does NOT tell us what the transient
response of the system will be. It also tells us that for a given
frequency response, without the contraining knowledge of the
phase response, there are an infinite number of possible transient
responses that will fit that frequency response.

Some simple concepts to help in answering your next question:

Transient response is the behovior of the system in the amplitude
vs time domain.

Frequency response is the behavior of the system in the modulus
of complex amplitude vs frequency domain.

Or am I wrong
in assuming that the square wave response is determined by the
transient response?

Yes, completely, since you are talking, in both cases, about the
behavior of amplitude vs time. It is NOT possible to have the same
square wave response and a different transient response within the
limits of the behavior of the signals at hand.

What worries me is that the 4th order
Linkwitz-Riley, which is what I was planning to use, doesn't look very
good.

"doesn't look very good" needs some definitions and constraining,
since we've now leapt from the realm of objective definitions and
examinations of physical system behavior to value judgements.

A Linkwitz-Rilet system is simply another system with non-minimum-
phase behavior. It can get you flat frequency response but will not
necessarily, therefore, give you perfect transient response.

As to whether it's "very good" or not requires someone, you, in this
case, to make a judgement call: which evil to you consider most
tolerable. With the information you provided, it's not possible for
me to make a recommendation that carries any weight. My guess is
that IF the 4th LR is implemented properly, the concerns you have for
the transient response issues are overly emphasised.

One final note: "minimum phase" has a very specific technical definition,
summarized above. You'll see a similar term, "linear phase" which also
has a quite specific and unambiguous definition: it means a phase
which increase in direct proportion to frequency and is characteristic
of simple linear delay. There are a lot of other buzz words that have
the word "phase" in them, most of which, in the realm of speakers, are
marketing gobbledygook.

(Dick Pierce) wrote in message . com...
(Carlos) wrote in message . com...

Thank you so much for the very detailed explanation, Dick.

What worries me is that the 4th order
Linkwitz-Riley, which is what I was planning to use, doesn't look very
good.

"doesn't look very good" needs some definitions and constraining,
since we've now leapt from the realm of objective definitions and
examinations of physical system behavior to value judgements.

You're right, in fact it was almost an aesthetic judgement, those
waveforms look ugly! Seriously, what worries me is that by solving one
problem, one that is clearly audible and annoying to me, I might be
creating another one, which may or may not be even more audible and
annoying than the first one.

I'll try to be more specific: I would like to know whether the final
system (closed sub with Q=0.707, F3 = 37 Hz, 4th order LR crossover @
100 Hz) will have noticeably worse transient response than the current
one (vented boxes with 37% larger than maximally flat box volume,
tuned 22% lower than maximally flat, F3 = 40 Hz, 19.5 dB/Oct. slope),
assuming that the filter is well implemented, the levels are matched
and the distance from listening position to main speakers and sub is
the same (+/- 10 cm) and, for the sake of simplicity, the sub and main
speakers have substantially flat frequency response within their
respective passbands.

If "noticeably worse" is still too subjective, ok, let's say
"measurably worse".

Many thanks again,

Carlos

(Dick Pierce) wrote in message . com...
(Carlos) wrote in message . com...

Thank you so much for the very detailed explanation, Dick.

What worries me is that the 4th order
Linkwitz-Riley, which is what I was planning to use, doesn't look very
good.

"doesn't look very good" needs some definitions and constraining,
since we've now leapt from the realm of objective definitions and
examinations of physical system behavior to value judgements.

You're right, in fact it was almost an aesthetic judgement, those
waveforms look ugly! Seriously, what worries me is that by solving one
problem, one that is clearly audible and annoying to me, I might be
creating another one, which may or may not be even more audible and
annoying than the first one.

I'll try to be more specific: I would like to know whether the final
system (closed sub with Q=0.707, F3 = 37 Hz, 4th order LR crossover @
100 Hz) will have noticeably worse transient response than the current
one (vented boxes with 37% larger than maximally flat box volume,
tuned 22% lower than maximally flat, F3 = 40 Hz, 19.5 dB/Oct. slope),
assuming that the filter is well implemented, the levels are matched
and the distance from listening position to main speakers and sub is
the same (+/- 10 cm) and, for the sake of simplicity, the sub and main
speakers have substantially flat frequency response within their
respective passbands.

If "noticeably worse" is still too subjective, ok, let's say
"measurably worse".

Many thanks again,

Carlos

Dick,

If those are the resuults you getm then you're not doing your filtering
properly.

I dunno - we set up a mike, played some tones, and that's what we got. What
else is there to do?

perhaps you might want to consider expanding your toolbox.

Actually, loudspeaker placement in rooms is of great interest to me too. In
fact, later today I'm going to do some tests of the response obtained with
speakers flat against a wall versus various spacings away from the wall.

Do you fear, sir, that I have snubbed you?

No, I just wanted to meet this mysterious man behind the curtain in person.

--Ethan