"Orchid XP v3" <voi### [at] devnull> wrote in message 
news:46deec35$1@news.povray.org...
> Brian Elliott wrote:
>
>> Firstly, the crossover network divides the signal from the amplifier to 
>> the woofer and the tweeter drivers.  It must be tuned to the right 
>> frequency and shape the frequency falloff curves and phase so that the 
>> sound levels from both drivers are balanced without dips or humps or 
>> comb-filtering at the crossover band where both drivers transmit part of 
>> the signal.  It is also supposed to keep the signal phase true between 
>> the two drivers over the frequency range so they operate in a unison 
>> making the sound wavefront from both drivers combine and arrive at your 
>> ear together.  That is fekking difficult to design, because...
>>
>> Crossovers are passive networks of resistors, capacitors and inductors. 
>> Networks behave much more complicatedly than isolated components because 
>> everything interacts with everything else, not just its immediate 
>> neighbours.  Speaker drivers are *also* R-L-C networks, so crossovers 
>> must be designed with that driver's electrical properties being integral 
>> to it.
>>
>> In short, the whole thing is interbalanced, so if one resistor, capacitor 
>> or inductor changes, everything goes out of whack -- crossover frequency, 
>> frequency response, phase response, impedance response, resonances and 
>> ringing -- and the speaker sounds like crap.
>>
>> For predictability, a speaker and its crossover also rely on the 
>> amplifier's output stage being very low impedance. Signal-wise, the 
>> amplifier is near to a short-circuit, regardless of the voltage swings it 
>> generates.  A speaker is a motor and when moving, it generates back-EMF 
>> through its crossover.  It has mechanical inertia and wants to overshoot. 
>> It  also has natural resonances from driver suspension springiness, 
>> cabinet air volume, acoustic transmission line length, tuned acoustic 
>> port, etc.  That colouration is NOT part of the original signal.  If the 
>> amplifier doesn't soak up that energy (absorb the current), to damp the 
>> unwanted motion (overshoot and ringing) it will reflect back into the 
>> network and colour the sound you hear.
>
> I read about this kind of thing in a DSP book. Damn, IIR filters sound so 
> complicated! Fascinating, but... I wouldn't want to design them for a 
> living.

I'm with you on that!  Fascinating, wish I could do it, but AIUI, there's 
hardly a hope of designing it manually.  It's computerised design with an 
element of alchemy, dead cats, satanic verse and other ancient rituals 
thrown in.  Probably a reading of Pam Ayres poetry too!  Might as well find 
the diff. eqns. for a six-body gravitational orbit.

  :-)

> -- 
> http://blog.orphi.me.uk/

Post a reply to this message