Sign in with
Sign up | Sign in
Your question

Crossover design question

Last response: in Home Audio
Share
Anonymous
a b \ Driver
August 24, 2004 9:51:03 AM

Archived from groups: rec.audio.tech (More info?)

I'm designing a two-way loudspeaker system using Dynaudio drivers. I
modeled the impedance and acoustic frequency responses of the drivers
using CALSOD from measurements that I've taken. I designed the
crossover for each driver in order to get 4th order L-R alignments. So
far so good. My question is, since drivers aren't perfect, should I
optimize the final -system- response for flattest frequency response
or should I try to get as close as possible to the theoretical L-R
alignments? If I try to get flattest system response then each driver
+ crossover section deviates quite a bit from the theoretical L-R
alignments. I may get very flat frequency response but I lose
advantages of the L-R alignments.

Which is the better compromise for getting best subjective results?

Thanks in advance,

Michel
Anonymous
a b \ Driver
August 24, 2004 5:19:33 PM

Archived from groups: rec.audio.tech (More info?)

"Deafboy" <cocomymy@hotmail.com> wrote in message
news:ad33a607.0408240451.aedd210@posting.google.com...
: I'm designing a two-way loudspeaker system using Dynaudio drivers. I
: modeled the impedance and acoustic frequency responses of the drivers
: using CALSOD from measurements that I've taken. I designed the
: crossover for each driver in order to get 4th order L-R alignments. So
: far so good. My question is, since drivers aren't perfect, should I
: optimize the final -system- response for flattest frequency response
: or should I try to get as close as possible to the theoretical L-R
: alignments? If I try to get flattest system response then each driver
: + crossover section deviates quite a bit from the theoretical L-R
: alignments. I may get very flat frequency response but I lose
: advantages of the L-R alignments.
:
: Which is the better compromise for getting best subjective results?

After you are done with all your calculations and computer models and measurements, change the
CO however necessary to make the speaker sound good. Theory, measurement and design calcs only
assume the result will sound good and there is no way to judge sound quality from such.

From my experience it takes a certian type of person many years of fiddling with this stuff
to make a speaker that sounds good and is true to the source.

Philip Abbate
www.philsaudio.com
:
: Thanks in advance,
:
: Michel
Anonymous
a b \ Driver
August 24, 2004 10:34:45 PM

Archived from groups: rec.audio.tech (More info?)

"Deafboy" <cocomymy@hotmail.com> wrote in message
news:ad33a607.0408240451.aedd210@posting.google.com...
> I'm designing a two-way loudspeaker system using Dynaudio drivers. I
> modeled the impedance and acoustic frequency responses of the drivers
> using CALSOD from measurements that I've taken. I designed the
> crossover for each driver in order to get 4th order L-R alignments. So
> far so good. My question is, since drivers aren't perfect, should I
> optimize the final -system- response for flattest frequency response
> or should I try to get as close as possible to the theoretical L-R
> alignments? If I try to get flattest system response then each driver
> + crossover section deviates quite a bit from the theoretical L-R
> alignments. I may get very flat frequency response but I lose
> advantages of the L-R alignments.
>
Approaching it the way you are doing gives a good methodical basis to reply
on. The real question is would optimizing the combined frequency response in
a given situation which is generally a good thing, lead to increases in
other contaminants which are bad, and where would the tradeoff
considerations take you?

Minimal changes in transfer function are not inheirantly noticeable and thus
optimizing FR is of benefit. If you find yourself departing greatly from the
LKR template, then it might be good to know what is happening with the
entire complex signal. CALSOD doesn't compute groupdelay or time domain
functions (Step function or Impulse response) which can help to visualize
this. Otherwise look at phase curve in comparison to LKR ideal curve.
Remember the choice of primary crossover point itself is of importance as
well.

Lastly listen and be patient

> Which is the better compromise for getting best subjective results?
>
> Thanks in advance,
>
> Michel
Related resources
Anonymous
a b \ Driver
August 24, 2004 10:37:45 PM

Archived from groups: rec.audio.tech (More info?)

"Deafboy" <cocomymy@hotmail.com> wrote in message
news:ad33a607.0408240451.aedd210@posting.google.com...
> I'm designing a two-way loudspeaker system using Dynaudio drivers. I
> modeled the impedance and acoustic frequency responses of the drivers
> using CALSOD from measurements that I've taken. I designed the
> crossover for each driver in order to get 4th order L-R alignments. So
> far so good. My question is, since drivers aren't perfect, should I
> optimize the final -system- response for flattest frequency response
> or should I try to get as close as possible to the theoretical L-R
> alignments? If I try to get flattest system response then each driver
> + crossover section deviates quite a bit from the theoretical L-R
> alignments. I may get very flat frequency response but I lose
> advantages of the L-R alignments.
>
> Which is the better compromise for getting best subjective results?
>
> Thanks in advance,
>
> Michel

Approaching it the way you are doing gives a good methodical basis to reply
on. The real question is would optimizing the combined frequency response in
a given situation which is generally a good thing, lead to increases in
other contaminants which are bad, and where would the tradeoff
considerations take you?

Minimal changes in transfer function are not inheirantly noticeable and thus
optimizing FR is of benefit. If you find yourself departing greatly from the
LKR template, then it might be good to know what is happening with the
entire complex signal. CALSOD doesn't compute groupdelay or time domain
functions (Step function or Impulse response) which can help to visualize
this. Otherwise look at phase curve in comparison to LKR ideal curve.
Remember the choice of primary crossover point itself is of importance as
well.

Lastly listen, be patient, and don't be afraid to experiment
Anonymous
a b \ Driver
September 3, 2004 12:57:22 AM

Archived from groups: rec.audio.tech (More info?)

Deafboy wrote:

> I'm designing a two-way loudspeaker system using Dynaudio drivers. I
> modeled the impedance and acoustic frequency responses of the drivers
> using CALSOD from measurements that I've taken. I designed the
> crossover for each driver in order to get 4th order L-R alignments. So
> far so good. My question is, since drivers aren't perfect, should I
> optimize the final -system- response for flattest frequency response
> or should I try to get as close as possible to the theoretical L-R
> alignments? If I try to get flattest system response then each driver
> + crossover section deviates quite a bit from the theoretical L-R
> alignments. I may get very flat frequency response but I lose
> advantages of the L-R alignments.

Do you? - how is the power curve?

Anyway, I don't understand what your concern is, are you worried that
the electric output from the cross-over does not look like the required
slope-shape or are you worried that the acoustic output from each driver
does not look like the required cross-over shape?

> Which is the better compromise for getting best subjective results?

The one that takes best care of the treble unit powerhandling and
excursion concerns.

I think the compromise solution is to optimize the treble unit
cross-over components separately and then defined them as FIXED
when a fair version of a correct acoustic output is obtained.

It is probably not wise to have all cross-over components variable in
the final optimization and you are likely to be better off by letting
the low frequency unit have an imperfect slope due to overall system EQ
getting applied by CALSOD. IF the slope for the treble unit has to be
imperfect, then it should be below the target rather than above it.

It is *required* to measure the actual units in the actual box,
preferably in a location like the intended (corner, wallmount, free
standing, bookshelf or flush in wall) prior to deciding on what kind of
cross-over at what frequency that will fit the actual project.

There are very many ways in which you can nudge CALSOD to do things
differently. It may also be that the L-R target is not the wisest choice
for the design in question, look for what the design wants to do.

> Thanks in advance,

Your mileage may vary wildly.

> Michel


Kind regards

Peter Larsen

--
*******************************************
* My site is at: http://www.muyiovatki.dk *
*******************************************
Anonymous
a b \ Driver
September 3, 2004 5:21:28 PM

Archived from groups: rec.audio.tech (More info?)

"Peter Larsen" <SPAMSHIELD_plarsen@mail.tele.dk> wrote in message
news:41376D12.ABFD8464@mail.tele.dk...
: Deafboy wrote:
:
: > I'm designing a two-way loudspeaker system using Dynaudio drivers. I
: > modeled the impedance and acoustic frequency responses of the drivers
: > using CALSOD from measurements that I've taken. I designed the
: > crossover for each driver in order to get 4th order L-R alignments. So
: > far so good. My question is, since drivers aren't perfect, should I
: > optimize the final -system- response for flattest frequency response
: > or should I try to get as close as possible to the theoretical L-R
: > alignments? If I try to get flattest system response then each driver
: > + crossover section deviates quite a bit from the theoretical L-R
: > alignments. I may get very flat frequency response but I lose
: > advantages of the L-R alignments.
:
: Do you? - how is the power curve?
:
: Anyway, I don't understand what your concern is, are you worried that
: the electric output from the cross-over does not look like the required
: slope-shape or are you worried that the acoustic output from each driver
: does not look like the required cross-over shape?
:
: > Which is the better compromise for getting best subjective results?
:
: The one that takes best care of the treble unit powerhandling and
: excursion concerns.
:
: I think the compromise solution is to optimize the treble unit
: cross-over components separately and then defined them as FIXED
: when a fair version of a correct acoustic output is obtained.
:
: It is probably not wise to have all cross-over components variable in
: the final optimization and you are likely to be better off by letting
: the low frequency unit have an imperfect slope due to overall system EQ
: getting applied by CALSOD. IF the slope for the treble unit has to be
: imperfect, then it should be below the target rather than above it.
:
: It is *required* to measure the actual units in the actual box,
: preferably in a location like the intended (corner, wallmount, free
: standing, bookshelf or flush in wall) prior to deciding on what kind of
: cross-over at what frequency that will fit the actual project.
:
: There are very many ways in which you can nudge CALSOD to do things
: differently. It may also be that the L-R target is not the wisest choice
: for the design in question, look for what the design wants to do.
:
: > Thanks in advance,
:
: Your mileage may vary wildly.
:
: > Michel
:
:
: Kind regards
:
: Peter Larsen

I agree with Peters lesson above. I would like to point out that there are programs that do
the work of the Calsod optimizer, measure drivers with MLS and also allow you to HEAR what
you are modeling. All the while listening you can see the various predictions on the screen as
the program emulates playing music into your room over your drivers in your speakers. .

I sell LspCAD for this purpose. With LspCAD you have a crossover breadboard in the PC. Using a
multi channel sound card you can output low level signals which are filtered in accordance
with the computer network you enter/optimize. Connect these signals to amps and your drivers
in the box in the room and you can hear the differences you are wondering about. You can step
value for each part in the CO, You can have the optimizer step part(s) values to match a
target, be it an imported measurement or a pre defined curve like LR.

You can save up to 8 crossovers and instantly switch between them. When you finally find a
network that degrades the sound of your music with any part change you are done. Now you have
the option of building the emulated CO with passive parts, active parts (salen-key) or you can
flash it into a Behringer 2496 active DSP crossover.

Phil Abbate
LspCAD Dealer
www.philsaudio.com/lspcad.htm
!