home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Shareware Overload
/
ShartewareOverload.cdr
/
games
/
alpha.zip
/
XC.SLI
< prev
next >
Wrap
Text File
|
1985-03-31
|
20KB
|
1,040 lines
.W:30
.L:30
.r:d
.j:3
.s:3
.f: ...Crossovers... Slide $$$
Loudspeaker Crossovers
What They Should Do
and How to Get Them
To Do It
0. Abstract.
Considerable recent technological
change in crossover design.
Lots of "Old fashioned" crossovers
still used.
Discussion of "New" technology
seems to be in order.
I. Tasks for the crossover
network.
Crossover networks must:
1. Route signals to different
drivers.
2. Maintain desirable phase
relationships between
acoustical outputs of
individual drivers.
3. Compensate for differing
driver efficiencies.
Requirement 1: Route signals in
different frequency
ranges.
Loudspeaker system quality factors:
(as perceived by user)
* Wide frequency response range
* Smooth response
* Low distortion
* Good dynamic range
* Low price
* Small size
No single driver system known has
all these characteristics.
Multiple driver systems come closer
to the ideal.
Unfortunately they introduce the
need for crossover networks.
Primary advantages of routing
signals:
* Avoid wasting amplifier power.
* Avoid disrupting driver
operation with out-of-band
signals.
* Controlling the physical size of
the source.
Requirement 2: Maintain desired
phase relationships
between acoustical
outputs.
Reasons for maintaining phase
relationships:
One is obvious,
but sonically irrelevant.
One not so obvious,
but is important to sound
quality.
Obvious, but irrelevant:
Control the over-all phase
response of the
loudspeaker system.
Reasons:
No known controlled listening test
has ever shown any reason to be
interested in system phase
response above 1 kHz.
Tests below 1 kHz not very
conclusive.
Less-obvious and very relevant:
Control system directivity through
crossover points.
When:
* Two acoustical signals
radiated into a common
space,
-and-
* One signal has lagging phase
with respect to the other,
-then-
* Joint radiation pattern is
tilted toward the lagging
driver.
-and-
* There are dips in the
on-axis frequency response
vis-a-vis power response.
Primary advantages of controlling
phase:
* Control coverage pattern at
crossover frequency.
* Maintain desirable relationship
between on-axis response and
power response.
Ideally:
* Acoustic output of all drivers
in-phase at all frequencies.
Requirement 3. Compensate for
differing driver
efficiencies.
Why needed:
It is unlikely that drivers will
have same specified efficiency.
Sample variations within batches
of drivers.
B. Things that are nice to have
done:
* Compensate for driver
deficiencies.
* Compensate for enclosure
deficiencies.
* Facilitate use of cost-effective
amplifiers.
* Reduce "Up-front" design work.
* Reduce "Cut-and-try" elsewhere
in the system.
Accommodation 1: Compensate for
driver
deficiencies.
* Drivers have limited response
range.
* May not have "Naturally" flat
response.
* Often easier to compensate
individual drivers than entire
system.
Accommodation 2: Compensate for
enclosure
deficiencies.
* Often there are time delays/phase
shifts due to driver positioning.
* Repositioning drivers can cause
diffraction and reflection
problems.
* Possible to electrically
"reposition" drivers.
Accommodation 3: Facilitate
use of
cost-effective
amplifiers.
Power amplifier costs a non-linear
function of power rating.
Cost/availability "Barrier" at
about 400 watts per
channel (8 ohms).
Inexpensive One Chip amplifiers
limited to the 20-30 watt
per channel range.
Reactive drive requirements of
drivers, crossovers.
Accommodation 4: Reduce "Up-front"
design work.
Loudspeaker system design requires
knowledge of a wide range of
technologies.
* Mechanical Engineering:
Static analysis
Dynamic Analysis
* Electrical Engineering:
DC analysis
AC analysis
Transient analysis
Steady-state analysis
Crossovers, as components, can
reduce the span of technology
required of the system designer.
Accommodation 5: Reduce
"Cut-and-try"
elsewhere in
system.
Some crossovers make over-all
system performance less
sensitive to parameters that
are hard to control.
Crossovers can include electrical
adjustment capabilities that
simulate mechanical adjustments.
Benefit: electrical adjustments are
faster & easier than mechanical
ones.
II. Alternative crossover designs.
Historically, many alternative
designs
Only recently have full
implications of certain choices
has been understood.
Alternative A: 6 dB/octave filters.
Most common.
Simplest to construct.
May be achieved with no
additional electrical
components.
Wide ranges of driver overlap.
90 degree phase shift in electrical
drive at crossover point.
90 degrees or more additional phase
shift in drivers in crossover
region.
Drivers usually connected
electrically out of phase.
Hard to achieve passively due to
driver impedance variations.
Alternative B: 12 dB/octave
Butterworth filters.
"Maximally-flat filters"
12 dB/octave Q=1.414 roll-offs
inherent in some drivers.
Often take twice as many electrical
components.
Driver impedance variations cause
some passive "12 dB"
designs "6dB" effective
slopes
Electrical outputs 180 degrees
out-of phase at crossover
point.
90 degrees or more additional phase
shift in drivers in crossover
region.
Drivers usually connected
electrically in phase.
Alternative C. 18 dB/octave
Butterworth filters.
Usually implemented actively.
(1 op amp)
Driver overlap respectably low.
Electrical outputs 270 degrees
out-of phase at crossover
point.
90 degrees or more additional phase
shift in drivers in crossover
region.
Drivers usually connected
electrically in phase
(some controversy).
Often achievable with 12/dB
electrical filters + driver
characteristics.
With drivers out-phase,
approximates linear phase.
Alternative D. All-pass designs:
Minimum phase versus all-pass
Minimum phase:
* Familiar LRC filters.
* Hilbert transform relationship
between phase and amplitude.
* Equalize amplitude at all
frequencies and phase will also
be equalized.
* Individual drivers are often
approximately minimum phase.
Alternative D. All-pass designs:
Minimum phase versus all-pass
(cont.)
All-pass:
* Time delays/phase shifts with
flat frequency response.
* Loudspeaker systems are usually
all-pass because of displaced
drivers.
* Loudspeakers in break-up modes
may be all-pass (displaced
sources).
All-pass crossovers are created by
cascading two Butterworth
filters.
Constant phase relationships at all
frequencies.
12 dB/Octave all pass is cascade of
two 6 dB/octave filters with no
interaction.
* Electrical outputs 180 degrees
out-of phase at all
frequencies.
* Connect drivers out-of-phase
electrically.
* Hard to achieve acoustic
12dB/octave all-pass
crossover.
All-pass crossovers created by
cascading two Butterworth
filters. (cont.)
24 dB/Octave all pass is cascade of
two 12 dB/octave filters with no
interaction.
* Electrical outputs 360 degrees
out-of phase at all
frequencies. (i.e. in phase)
* Connect drivers in-phase.
* Easier to achieve acoustic
12dB/octave all-pass crossover
since driver roll-offs commonly
12 dB/octave.
* Linkwitz of Hewlett-Packard
described mathematical
properties (1976).
* Riley described implementation
using Sallen and Key type op-amp
filters (1976).
All-pass crossovers created by
cascading two Butterworth
filters. (cont.)
Phase compensation of 3-way
and up all-pass crossovers.
* Residual phase shift of
adjacent crossover point(s).
* Second order Q=1.414 all-pass
filter compensates exactly.
* Optional for 3-way, more
necessary for 4-way and up.
* Analyzed by D'Appolito (1984)
All-pass crossovers created by
cascading two Butterworth filters.
(cont.)
Configuration of filters:
* Band-pass filters minimize
stages signal passes through.
* Tree configuration minimizes need
for phase compensation.
* Analyzed by D'Appolito (1984)
E. Delay-derived filters.
Minimize parts count via
subtraction.
Example: Subtract low-pass filter
output from input deriving
high-pass.
Problem: derived channel has gentle
slope, no more than 6 dB/octave.
Source of problem is time delay in
filter (approx 1 mSec for 1 kHz
filter).
Total subtraction impossible
without compensating time-delay.
E. Delay-derived filters (cont.)
Unexpected desirable result: Summed
response has linear phase (phase
shift equal to a time delay).
Delay-derived crossover using 5th
Order Bessel filter, and
appropriate time delay has
directional control about as
good as 4th order all-pass.
Undesired characteristic: $$$$
Advantage over 4th order all-pass:
????
F. Driver Frequency response
correction.
Goals for driver response: (either
is equally ideal)
* Flat DC to light.
* Roll-offs are 2nd order
Butterworth at crossover
frequencies.
Less-than-perfect but usually
acceptable:
* Flat an octave past crossover
point(s).
F. Driver Frequency response
correction (cont.)
Drivers with these characteristics
may be hard to obtain.
Alternatives:
* Equalize using combined outputs
of a state variable filter
(Zaustinsky exact equalization).
* Equalize using Q > 1.414 high or
low pass filter (s) (similar to
Thiel B6 alignments).
* Other forms of experimentally
derived equalization.
* Watch out for phase!
III. Performance of alternatives:
A. Passive versus Active
crossovers.
* Choose active filters when
price is no object.
* Active filters can be
cost-effective depending on
amplifier power
requirements.
* Active filters provide more
flexibility.
III. Performance of alternatives:
(cont.)
B. Butterworth versus All-Pass
crossovers.
* Choose All-pass designs
whenever flat response on and
off axis is desired.
III. Performance of alternatives:
(cont.)
C. Delay-derived versus All-Pass.
* At this time Delay-derived
crossovers are prohibitively
complex and expensive.
* With increased quality
requirements and less costly
implementation of filters and
delays, this may be the
technology of choice.
III. Performance of alternatives:
(cont.)
D. Constant voltage versus
constant power:
Ideal "Constant voltage"
crossovers have:
* Flat on-axis pressure
response.
* 3 dB loss in power response
at the crossover point.
(read as +/- 1.5 dB)
III. Performance of alternatives:
(cont.)
D. Constant voltage versus
constant power (cont):
However, the major sources of
non-uniform power response
are:
* Inter-driver phase effects.
* Driver directionality.
* Listening room acoustics.
As a practical matter, some
"Constant voltage" crossovers
have better power response
than "Constant Power" types.
III. Performance of alternatives:
(cont.)
E. Time delay alternatives:
* Use in delay-derived
crossovers.
* Useful for repositioning the
acoustic location of a
loudspeaker driver.
* Significant amounts of time
delay dictate the use of
active crossovers.
III. Performance of alternatives:
(cont.)
E. Time delay alternatives:
(cont.)
1. Digital Delay.
* Co$tly
* High quality
* Digitization can be
avoided with digital
program sources.
* Level control ganging
problems.
III. Performance of alternatives:
(cont.)
E. Time delay alternatives:
(cont.)
2. CCD delay.
* Adequate quality with
short delays.( < 20
milliseconds)
* Almost reasonable cost
III. Performance of alternatives: (cont.)
E. Time delay alternatives: (cont.)
3. Delay via all-pass filters.
* Excellent quality for very
short delays. ( < 5
milliseconds )
* Reasonable cost
IV. Loudspeaker system construction.
A. Driver selection.
1. Direct radiators:
* Low frequency cones
* High frequency domes
* Good performance at low
driver unit cost.
* Thiel parameters
important for
low-frequency design.
* Qt < 0.4 desired
* Cone area versus stroke.
* Individual port tuning.
* Midrange and tweeter
specs unreliable.
* Recommend measurement.
IV. Loudspeaker system
construction (cont.)
A. Driver selection (cont.).
2. Horn loaded drivers.
* Shunned by picky listeners.
* Never out of style in
high quality P.A.
systems.
* Regaining favor for all
applications.
* Key technology: Constant
Directivity horns.
* Require equalization.
* May be cost effective
consideraing total
system cost, including
power amplifiers.
IV. Loudspeaker system construction
(cont.)
B. Enclosure Designs can affect
sound.
Sonically important factors:
* Enclosure volume for the
low frequency driver & port
area.
* Orientation of upper driver
radiation patterns towards
listener.
* Control of diffraction
effects.
* Elimination of panel
resonances.
IV. Loudspeaker system construction
(cont.)
B. Enclosure Designs can affect
sound.
Alternatives:
1. Conventional
free-standing enclosures.
* May include wave
guides control
directivity.
* Edges of the enclosure
rounded to control
diffraction.
* "Pyramid" design with
square edges encourage
diffraction.
IV. Loudspeaker system
construction (cont.)
B. Enclosure Designs can affect
sound.
Alternatives (cont.):
2. Built-in loudspeakers
* Avoid disruption of
decor with large
boxes.
* Subwoofers can be placed
in the floor, ceiling,
or wall.
IV. Loudspeaker system
construction (cont.)
B. Enclosure Designs can affect
sound.
Alternatives (cont.):
3. Esoteric enclosure
designs.
* Concrete
* Sand between concentric
cardboard tubes, etc.
* Good suppression of
enclosure wall
resonance can be
obtained in this way.
C. Evaluation techniques are
important to any design
effort.
* Goal is subjectively
perceived "Good sound",
* Subjective perceptions
have limited reliability
w/o good experimental
controls.
* Subjective tests are
very time consuming.
* "If you measure length by
eye, you can see
discrepancies by eye"
C. Evaluation techniques are
important to any design
effort. (cont.)
1. Miniature electret
capsules with
outstanding response over
the entire range cost
just a few dollars.
Reliability enhanced
using multiple
microphones with separate
rectification prior to
summation.
C. Evaluation techniques are
important to any design
effort. (cont.)
2.Ordinary audio oscillators
can give fair results
above 500 Hz.
3. Swept oscillators and
chart recorders are fast,
reliable, and moderately
priced.
4. Pink noise sources from a
test set or CD avoid
standing waves.
5. Coherent fractional
octave sources described
by Linkwitz and
others.
6. One third octave real
time analyzer are very
effective, at a price.
C. Evaluation techniques are
important to any design
effort. (cont.)
7. Computerized techniques.
Time Energy Frequency (TEF)
test $et.
Combinations of third
octave filter banks and
personal computers.
Dedicated and personal
computer FFT analyzers
Expect cost decreases for:
* Computation
* Digital-to-analog
conversion
C. Evaluation techniques are
important to any design
effort. (cont.)
7. Computerized techniques
(cont).
All known and future audio
test instruments will be
ultimately implemented
using:
* General purpose
hardware,
* Sophisticated
software.
V. Conclusions.
A. Electronic crossovers
desirable for many reasons.
B. Best price-performance in
crossovers is obtained with
24 dB/octave acoustical
all-pass.
C. It is important to know
actual driver
characteristics.
D. Relatively sophisticated
acoustic measurements are
possible at a reasonable
cost.