[JN] Linear and minimum phase crossovers

Phil toob-man at sbcglobal.net
Sat Jul 2 18:26:47 CDT 2011

Okay, found the AudioXpress issues: In June and July 2008 G. R. Koonce
had a two-part article on linear phase crossovers. There was something
of a "letters follow-up" in February 2009, with responses by Koonce to
two letters. The second response is quite useful to those of us with a
poor knowledge of crossovers, as it reviews the amplitude and phase
characteristics of 1st, 2nd, 3rd, and 4th order two-way crossovers, and
includes graphs of the amplitude and phase characteristics for all 4 of
them. Sweet.

The 1st order is interesting in that you can reverse the output of one
driver and you still get flat amplitude response, but with the phase
gradually shifting 180 degrees (takes about 3 decades!). They mentioned
in the article that most people actually didn't hear any significant
problems, indicating that gradual phase shifts really aren't that bad.

The 2nd and 4th order require one of the drivers to be reversed, i.e.,
for a pulse, one driver pushes while the other pulls, in order to get
flat amplitude response, as Bill said; otherwise you get a sharp notch
in both the amplitude and the phase. For standard crossovers, therefore,
you can get all the drivers moving forward for a pulse only with 1st and
3rd order crossovers (or 5th, 7th, etc.).

However, there is a non-standard 2nd order crossover that allows all the
drivers to move forward together in response to a pulse, while having
phase that is constant within 7 degrees, and amplitude that is flat to
within 2 dB (total, not +/-). The design is from George Augspurger, who
helps people design and build recording studios, and has of course had
quite a few articles in both Speaker Builder and JAES (including the
ones on transmission lines). Koonce gave the information about it in his
first response/follow-up (with permission from George). It can be used
with 2 and 3-ways, but for 3-ways, it works best if the two crossover
are "not too close" (whatever that means).

The high and low pass sections are all 2nd order filters with a Q of
0.5. In a 2-way speaker, for example, with a crossover of 1 KHz, the
low-pass is designed for 2 KHz, while the high-pass is designed for 500
Hz. That sounds crazy, but in practice, if you look at the region from
330 Hz to 1 KHz for the tweeter, and from 1 KHz to 3.3 KHz for the
woofer, you find that both drivers see almost EXACTLY the same thing
that they would see with a 1st order 1 KHz crossover, to within 1 dB or
less. The difference is that with a 1st order, the tweeter response is
-10 dB at 330 Hz, -20 dB at 100 Hz, and -40 dB at 10 Hz, whereas with
the modified 2nd order it's -10 dB at 330 Hz, -34 dB at 100 Hz, and -74
dB at 10 Hz, with similar results for the woofer, of course.

In THEORY, it should make any speaker that works, or barely works, with
a 1st order crossover, work a lot better, and with little or no sonic
penalty, but I have not heard of anyone who's actually tried it. My
understanding is that Augspurger has been using it in at least some of
his studios.


More information about the Sound mailing list