Ashly Electronic Amplifier none - Manual (Page 30)

crossovers are incapable of flat summed response, while 18dB per octave types
do sum flat. The two determining factors in summed response, response shape
and
phase relationship between fiIters, are frequently ignored or
mi sunderstood.
PHASE SHIFT
When audio is passed through a filter, there will be a shift in the phase of
the signal. This is true of both active and passive filters. For example, in
a
simple second-order (12dB/octave) crossover network, the low-pass output
will lag the high pass output by 180°. That means that there will be an
actual time difference in the movement of woofer and tweeter at the crossover
frequency; as the tweeter moves forward, the woofer will be moving backwards,
since the woofer will always be 1/2 wavelength behind the tweeter. If the two
drivers are of equal efficiency at the crossover frequency, their simultaneous
pushing and pulling will cancel each other out, and the result will be an
audible notch at the crossover frequency. This is even more obvious when thë
two filters are electrically summed--there will be a very deep notch in the
summed response curve where the two filters meet and cancel each other out.
Fortunately, there is an easy solution.
When two signals are exactly 180° out of phase, it is a simple matter to
invert the polarity of one output, thus eliminating the notch effect. That
explains why you'll often hear people say the the outputs of a 12dB per octave
crossover should be wired out of phase. This is one way of getting around the
phase error inherent in these crossovers. However, this would be a mistake
with
an Ashly crossover; we have already performed all necessary phase
inversions internally in our crossovers so that the user can simply wire
everything in phase without having to stop and think about which outputs to
invert and which ones not to invert. Unfortunately, this is notan agreed-
upon standard among crossover manufacturers, and so there is confusion on this
poi nt.
18dB per octave crossovers also have filter phase shifts, but unlike the
second-order crossovers, their outputs come out 90° out of phase, a difference
of 1/4 wavelength. When electrically or acoustically summed, these outputs
will not produce the notches associated with second-order crossovers. Phase
inversion and polarity swapping are not necessary.
RESPONSE SHAPE
Another source of confusion in summed amplitude response testing is response
shape of the individual filters. You'll hear it said that, even when wired
in-phase, 12dB per octave crossovers won't sum flat. It is true that 12dB per
octave Butterworth shape filters won't sum flat, and this is the source of the
unfortunate
generalization. However, if the filters are more highly damped,
then flat summing is certainly possible. This was illustrated in figure 30.
CHOICE OF SLOPE
Once it is realized that the slope of a crossover doesn't necessarily have
anything to do with summed amplitude response, we can proceed to choose a
crossover slope based on more important criteria. The choice of a particular
slope is largely subjective; some people like the "sound" of an 18dB per
octave crossover, while others might feel that its steep cutoff characteristic
29