what kind of speakers do you have? In general (this is how I setup my crossovers) I look at the minimum frequency response of the midrange speakers and use twice that for the crossover since the low end rolloff is generally the +-3 db part (advertised as 60hz-20khz +-3 db generally means that its falling 6 db from 'normal'
http://www.mackie.com/products/hr824...HRfreqresp.gif
if you look at that image of the response of the HR824's (freq response of 39-22.5khz +-1.5 db) anything above 50 hz appears to be +-.5 db.
What I am trying to get at is on midrange drivers the nonlinearities are generally in the low frequency range below the second octave you can produce. The formula I use assuming i had HR824's would be 39*2(one octave up)=78hz for the crossover. Sorry to make such a lengthy response. Please bear in mind that you shouldnt excede the max frequency response of the sub; I would not go above 150 hz for a crossover point.
Another thing to consider is the efficency of each driver and the amplifiers powering them. (this is easiest to calculate if you dont have active speakers. Some of the info such as the gain of the amp will probably need to be obtained from the manufacturer of the amplifier by contacting them (most companies dont list such figures))
How I matched my speakers and sub:
1.efficency of sub: 86.3 db/w/m (JL audio 13w7)
2.gain of sub amp: 33.1 db (Crown K2)
3.efficency of mids: 92 db/w/m (Klipsch Synergy SB-1)
4. gain of mid amp: 31.5 db (Crown K1)
33.1-31.5=1.6 db (so if the sub and mids had the same efficency, the sub would play 1.6 db louder they dont though SO...)
86.3+1.6=87.9
92-87.9=4.1db
therefore I reduce the output on the mids by 4.1 db in my dsp to "match" the sub and midrange speakers. It works quite well
Granted I should actually measure the response with a db meter but i dont have one at the moment and this has been sufficent so far.
