Crossing-over to tweeters - a difficult choice
That could well be in the designer's mind. But, as with most technical issues involving loudspeakers it is 'one step forwards and 1.2 steps backwards': really not advised.
Originally Posted by EricW
A tweeter is like a car with an F1 racing suspension: great when the surface is mirror smooth, but hopeless on a bumpy road. The reason is that unlike the bass/midrange cone which has a generous (rubber) surround which permits the cone to move in and out several mm (the equivalent of shock absorbers on a bumpy road) the tweeter has no real surround. It is virtually a rigid diaphragm that perhaps at best has a tiny fraction of a mm of permitted forwards/backwards movement - perhaps only the thickness of a sheet of paper. That's absolutely fine when the crossover is nice and high (say, > 3kHz) because there is virtually no energy in music at those frequencies (see another thread). But we know that the movement of any diaphragm is directly linked to the frequency that it is reproducing, and a small change in frequency can give a big change in necessary amplitude. So, in the real world, every tweeter has a naturally optimum crossover frequency which is non-intuitive: you have to find it by hours/weeks of experimentation and listening.
What about distortion? That, like amplitude, is directly correlated with the amount of forwards/backwards movement. So again, a small lowering of crossover frequency results in a potentially large increase in distortion. How does a tweeter that is being asked to operate below its naturally optimum crossover frequency actually sound? - i.e. a unit designed for 3.5kHz being operated at a ridiculously low 1.75kHz (one octave below?). Easy to demonstrate that ....
Loading the player ...
Clip A8: sound of tweeter at different crossover frequencies
In the above clip ...
0:00 - 07:00 normal speech, full bandwidth
07:00 - 18:00 - speech with bass/midrange driver disconnected - we hear the tweeter via 3.5kHz crossover (i.e. remove biwire links on Harbeth speaker)
18.00 > - speech with bass/midrange driver disconnected; same tweeter but crossed-over at 1.8kHz (never used by Harbeth)
As you can hear, when a tweeter is pushed too low it barks. There is no solution to this other than raising the crossover frequency to the tweeter's 'natural' crossover point. If, as you astutely point out, the designer is trying to mask the coloration at the top end of the bass/mid driver by handing that frequency band over to the tweeter rather early, he's traded one material coloration problem for perhaps an even more serious mechanical coloration problem: nothing is gained. This new coloration may be somewhat covered by music (esp. pop music which is multi-processed and devoid of reality) but on speech - example above - it is very obvious anywhere in the listening room. The human voice box doesn't sound like a strained mechanical resonator.
The only escape from this is to solve the bass/midrange driver's material coloration issue by developing a better material and then working that unit right up to the point (around 3kHz) where the tweeter can comfortably take over without straining. Oh, and it goes without saying that the fragile tweeter's life expectancy is directly linked to how hard it is working; the lower crossover frequency will inevitably shorten its life. These are the penalties of using bog-standard, mass produced, outsourced woofers purchased for $3 in so-called high fidelity loudspeakers.
Hope this makes sense.
Alan A. Shaw
Harbeth Audio UK