Quote Originally Posted by Mike Anderson
...
Can someone tell me what is the precise definition of "crossover frequency"? I understand what it means generally, but what is it exactly?
...
Do the signals from the two sources always sum linearly? If so, then when the sub and the satellites have two different slopes, it's impossible to pick a crossover frequency that gives a flat response, correct?
There are various crossover "alignments" which others can better explain than I can. However roughly speaking the crossover is where the signal is -3 decibels. The slopes are generally defined in terms of decibels per octave of attenuation; these a discussed in terms of "orders" of crossover:
  • 1st order: 6dB/octave
  • 2nd order: 12dB/octave
  • 3rd order: 18dB/octave
  • 4th order: 24dB/octave
  • and so on ...
No, the two sources don't necessarily sum linearly. It depends on the phase shift that is induced by the crossover. Active crossovers, which we have mostly been talking about, are designed to avoid the phase shift, (or so I understand -- I'm no engineer).

But all passive crossovers, (like the one inside the MMG), induce phase shift. Again, it relates to the "order". Each order induces a 90 degree difference in phase between the low and high pass signals. Hence a 2nd order crossover induces a 180 degree phase difference between the two -- which leads to complete cancellation at the crossover point In this case the remedy is to invert the polarity of one of the drivers