 | |  | |  | |  | |  | |  |
Introducing the ICTA Loudspeaker
By Music and Design TM....
By Music and Design TM....
John k....
crossovers, however, at the
crossover point the individual HP
and LP responses are greater than
0dB indicating that for a flat
summation cancellation is required.
This means that the phase
difference at the crossover point
must be between 120 and 180
degrees.(At 120 degrees the
summed response would be equal
to the amplitude of the individual
responses and at 180 degrees the
sum would have a null). Such a
phase difference between the HP
and LP sections would render poor
polar response for non-coincident
drivers. Figure 2 shows this result
when the driver separation is 1/3 of
a wave length at the crossover
crossover point the individual HP
and LP responses are greater than
0dB indicating that for a flat
summation cancellation is required.
This means that the phase
difference at the crossover point
must be between 120 and 180
degrees.(At 120 degrees the
summed response would be equal
to the amplitude of the individual
responses and at 180 degrees the
sum would have a null). Such a
phase difference between the HP
and LP sections would render poor
polar response for non-coincident
drivers. Figure 2 shows this result
when the driver separation is 1/3 of
a wave length at the crossover
frequency. The response
resembles that of a 1st order
crossover with the main lobe tilted
down. However, the off axis
response peaks at a level
exceeding 7 dB higher than the on
axis response. Clearly this is
unacceptable. But what of the
case where the driver separation
is only a small fraction of a wave
length. For example assume a
midrange and woofer with 18 and
28 cm basket diameters (typical
numbers for a 6 1/2" midrange
and a 10" woofer). The minimum
separation would be 23 cm. This
corresponds to about 1/10 a wave
length at 150 Hz, the intended
crossover point of the ICTA.
Figure 3 shows polar response for
this case. The improvement is
apparent, but the radiation pattern
is still undesirable.
Lastly, since the ICTA will be built
as a WMTMW system, as shown in
Figure 4, we can examine the
polar response for such a system.
The woofers are separated by 74
cm and the midrange driver by 28
cm. Here we see that the potential
polar response exhibits great
uniformity over the front
hemisphere.
As a final preview of the ICTA
development a brief examination
of the square wave and impulse
response is presented in Figures
6 and 7. The square wave
response is for a 100 Hz square
wave. The fundamental frequency
is below the W/M crossover point
with harmonics above it. The pulse
response is for a 0.1 msec wide
pulse. This reflects characteristics
of the M/T crossover which is
handled by a TP, 2nd order
crossover. In both figures the
response is shown on axis and at
15 degrees off axis. Further
discussion of this type of
crossover is presented here.
resembles that of a 1st order
crossover with the main lobe tilted
down. However, the off axis
response peaks at a level
exceeding 7 dB higher than the on
axis response. Clearly this is
unacceptable. But what of the
case where the driver separation
is only a small fraction of a wave
length. For example assume a
midrange and woofer with 18 and
28 cm basket diameters (typical
numbers for a 6 1/2" midrange
and a 10" woofer). The minimum
separation would be 23 cm. This
corresponds to about 1/10 a wave
length at 150 Hz, the intended
crossover point of the ICTA.
Figure 3 shows polar response for
this case. The improvement is
apparent, but the radiation pattern
is still undesirable.
Lastly, since the ICTA will be built
as a WMTMW system, as shown in
Figure 4, we can examine the
polar response for such a system.
The woofers are separated by 74
cm and the midrange driver by 28
cm. Here we see that the potential
polar response exhibits great
uniformity over the front
hemisphere.
As a final preview of the ICTA
development a brief examination
of the square wave and impulse
response is presented in Figures
6 and 7. The square wave
response is for a 100 Hz square
wave. The fundamental frequency
is below the W/M crossover point
with harmonics above it. The pulse
response is for a 0.1 msec wide
pulse. This reflects characteristics
of the M/T crossover which is
handled by a TP, 2nd order
crossover. In both figures the
response is shown on axis and at
15 degrees off axis. Further
discussion of this type of
crossover is presented here.
October 1, 2007
Midrange-Tweeter crossover:
The development of the midrange tweeter crossover is based on the simple subtractive approach. However,
since the driver spacing is typically a significant fraction of the wave length at the crossover point variation of the
inter-driver phase difference with vertical position will affect the lobing characteristics of the response. This leads
to several objectives: 1) The overshoot in the response of either the midrange or the tweeter in the vicinity of the
crossover frequency should be minimized to prevent off axis peaks in the response; 2) The polar response
should eb as uniform as possible; and 3) A minimum of a 3rd order high pass acoustic response is desired to
protect the tweeter from over excursion at low frequency and to reduce potential distortion.
Before proceeding a word about inter-driver phase variation with position is in order. This variation with vertical
position is only a function of the geometry of the driver layout on the baffle. Once the driver positions are
specified this variation is fixed. Thus, what remains for consideration is the affect of this variation on the off axis
response. For a two-way MT speaker a Linkwitz/Riley crossover tends to be the least sensitive because the
inter-driver phase is (or should be) zero on axis. The result is the well known symmetric lobing pattern with the
maximum response on axis. With crossover that do not have zero inter-driver phase at the crossover point, such
as Butterworth crossover, the polar response yields a null on one as we move off axis in one direction and a
peak as we move off axis in the other direction. The magnitude of this peak depends on the inter-driver phase
on axis. For Butterworth crossover where the inter-driver phase is ideally 90 degrees on axis, the off axis peak
will be +3dB. For certain types of transient perfect crossover where the on axis inter-diver phase is ideally 120
degrees the peak is +6dB. In order to reduce or eliminate the peak and MTM format may be implemented. The
well know D'Appolito configuration is useful for Butterworth type crossover. In the ICTA and MTM format is
implemented with a transient perfect crossover to minimize irregularity in the vertical off axis response. While the
polar response should be as uniform as possible it is particularly import to eliminate significant off axis response
peaks.
The desired crossover point for the ITCA is in the 1.5k to 2k hz range. The development presented here
assumes a 2kHz crossover. If a suitable result can be obtained for 2k Hz, the crossover can be scales to 1.5k Hz
without concern that the desirable polar response will degenerate. A simple subtractive crossover with minimal
over shoot can be obtained using a 2nd order Linkwitz/Riley high pass section. The resulting low pass section
will have a small overshoot of approximately 1.25 dB 1.5 octaves below the crossover frequency. The high pass
and low pass responses are shown below. The problems here are that the low pass response rolls off 1st order
and the high pass response rolls off at only 2nd order.
Midrange-Tweeter crossover:
The development of the midrange tweeter crossover is based on the simple subtractive approach. However,
since the driver spacing is typically a significant fraction of the wave length at the crossover point variation of the
inter-driver phase difference with vertical position will affect the lobing characteristics of the response. This leads
to several objectives: 1) The overshoot in the response of either the midrange or the tweeter in the vicinity of the
crossover frequency should be minimized to prevent off axis peaks in the response; 2) The polar response
should eb as uniform as possible; and 3) A minimum of a 3rd order high pass acoustic response is desired to
protect the tweeter from over excursion at low frequency and to reduce potential distortion.
Before proceeding a word about inter-driver phase variation with position is in order. This variation with vertical
position is only a function of the geometry of the driver layout on the baffle. Once the driver positions are
specified this variation is fixed. Thus, what remains for consideration is the affect of this variation on the off axis
response. For a two-way MT speaker a Linkwitz/Riley crossover tends to be the least sensitive because the
inter-driver phase is (or should be) zero on axis. The result is the well known symmetric lobing pattern with the
maximum response on axis. With crossover that do not have zero inter-driver phase at the crossover point, such
as Butterworth crossover, the polar response yields a null on one as we move off axis in one direction and a
peak as we move off axis in the other direction. The magnitude of this peak depends on the inter-driver phase
on axis. For Butterworth crossover where the inter-driver phase is ideally 90 degrees on axis, the off axis peak
will be +3dB. For certain types of transient perfect crossover where the on axis inter-diver phase is ideally 120
degrees the peak is +6dB. In order to reduce or eliminate the peak and MTM format may be implemented. The
well know D'Appolito configuration is useful for Butterworth type crossover. In the ICTA and MTM format is
implemented with a transient perfect crossover to minimize irregularity in the vertical off axis response. While the
polar response should be as uniform as possible it is particularly import to eliminate significant off axis response
peaks.
The desired crossover point for the ITCA is in the 1.5k to 2k hz range. The development presented here
assumes a 2kHz crossover. If a suitable result can be obtained for 2k Hz, the crossover can be scales to 1.5k Hz
without concern that the desirable polar response will degenerate. A simple subtractive crossover with minimal
over shoot can be obtained using a 2nd order Linkwitz/Riley high pass section. The resulting low pass section
will have a small overshoot of approximately 1.25 dB 1.5 octaves below the crossover frequency. The high pass
and low pass responses are shown below. The problems here are that the low pass response rolls off 1st order
and the high pass response rolls off at only 2nd order.
A 3rd order high pass roll off can be obtained by using a 3rd order Butterwoth or Bessel response for the high
pass section. However, this leads to excessive peaking in the low pass response. The characteristics when
using a B3 high pass response are shown below. Noted the 4 dB peak only one octave below the crossover
point and the extensive overlap between the high pass section (violet) and the low pass section (black). The
green curve is the low pass section when using the LR2 high pass. Note that the low pass response remains
1st order. Clearly this is not acceptable and another approach to obtaining a 3rd order high pass response is
required.
pass section. However, this leads to excessive peaking in the low pass response. The characteristics when
using a B3 high pass response are shown below. Noted the 4 dB peak only one octave below the crossover
point and the extensive overlap between the high pass section (violet) and the low pass section (black). The
green curve is the low pass section when using the LR2 high pass. Note that the low pass response remains
1st order. Clearly this is not acceptable and another approach to obtaining a 3rd order high pass response is
required.
This last figure shown the final result for the ICTA midrange/tweeter crossover. The black curves are
for a 2nd order LR filter as used initially and a 750 Hz B3 response. The red curve is the acoustic
target for the ICTA tweeter response. While the initial roll off is 2nd order the response steepens to
3rd order at low frequency. The high pass section is composed of a LR2 response cascaded with a
B1 with staggered poles. This is accomplished with only a little increase on the midrange peaking
below the crossover point. We also note that the low pass response now has a 4th order roll off.
This is accomplished by cascading the 1st order low pass roll off from the subtractive low pass filter
with a 3rd order Bessel response, again with corner frequency shifted well above the crossover
point. The BE3 response introduce a fairly constant delay to the midrange response which is
compensated for by physically off setting the tweeter. The response of this crossover to a 0.5 msec
wide pulse is shown to the right.
for a 2nd order LR filter as used initially and a 750 Hz B3 response. The red curve is the acoustic
target for the ICTA tweeter response. While the initial roll off is 2nd order the response steepens to
3rd order at low frequency. The high pass section is composed of a LR2 response cascaded with a
B1 with staggered poles. This is accomplished with only a little increase on the midrange peaking
below the crossover point. We also note that the low pass response now has a 4th order roll off.
This is accomplished by cascading the 1st order low pass roll off from the subtractive low pass filter
with a 3rd order Bessel response, again with corner frequency shifted well above the crossover
point. The BE3 response introduce a fairly constant delay to the midrange response which is
compensated for by physically off setting the tweeter. The response of this crossover to a 0.5 msec
wide pulse is shown to the right.
The vertical polar response using the proposed midrange/tweeter crossover is
shown to the left. It is apparent that the response is relatively uniform with
minimal peaking off axis. The off axis dips at 1k Hz are a result of the MTM
configuration and are less severe that would be observed using an LR4
crossover.
shown to the left. It is apparent that the response is relatively uniform with
minimal peaking off axis. The off axis dips at 1k Hz are a result of the MTM
configuration and are less severe that would be observed using an LR4
crossover.