Trinary crossovers??? Looking for more information...

[simon5]

Hi, I stumbled on this website: http://www.orchid-precision-audio.co.uk/page6.html

They claim perfect phase, time, power linearity anywhere in the audio band because of this crossover design. So, they have a almost perfect square wave response.

More explanation here: http://www.1388.com/articles/tech_lwo/page2/page2.html

I want to know if it's ******** or not, because I don't see these crossovers used even on the EgglestonWorks Ivy or on the Wilson Audio X-1 Grand SLAMM.

If there's more information about this way of designing a crossover, I would like you to point me at the information, even Google doesn't like the term...

Thank you very much,
Simon

[richie00boy]

Sounds like they build one normal low-pass/high-pass crossover, then another narrow band-pass filter where the overlap between the normal low-high region is and feed this to an additional driver.

[dwk123]

Sounds like a 'filler driver' concept. I'm slightly sketchy on the formalities,but the basic idea:
generate a 2-way xover for the woofer and tweeter so that they are in phase at the xover. I *think* it's typical to use 2nd order acoustic slopes, but 4th might work as well. Now invert one.This creates a deep null, but due to the differing delay in the xover legs, they are actually 'time aligned'. Now, 'fill in' the null with a third driver using first-order acoustic slopes centered on the xover freq used in the original 2-way.

Done properly, this is in fact a linear-phase xover and will pass a square wave.

John Kreskovsky who usually hangs out at the Madisound board (but does post here occasionally) has (or used to have) a paper on his site covering this.

[Mr Evil]

It could be an additional subtractive crossover section which outputs the difference of (lowpass + highpass) - input.

[catapult]

JohnK's filler driver article and spreadsheet:
http://www.geocities.com/kreskovs/GenFiller.html

[johnnyx]

nuuk put a very interesting design on decibel dungeon here
The crossover sums dead flat in both magnitude and phasel, I built, measured and simulated one. The only problem I had was that if the high - pass filter has a -12dB slope, the low - pass filter has a -6dB slope. You need very good drive units to use it.

If a low - pass filter was similarly treated, and the outputs of the summing opamps combined then maybe this is the kind of thing they mean. I don't know how they do it passively though.

[catapult]

It's pretty clear they're using the filler driver concept. They say it takes 3 drivers to make a "2 way" and 5 drivers to make a "3 way."

[richie00boy]

Originally posted by Mr Evil
It could be an additional subtractive crossover section which outputs the difference of (lowpass + highpass) - input.

That was my initial thought, but I don't think you can do that passively and the info gave me no clue the design was active.

[simon5]

Thanks alot everyone, now I would like to know why not much people use this design?

[AndrewJ]

This xover concept was originally described by Bang and Olufsen back in the late '70's. It was called the "filler driver".
Basically if you look at the summation of a 2nd order low pass and high pass filter, the magnitude response sums to unity, but not the phase response. The result is an all-pass response.
The difference between the all pass response and a unity response is a 1st order bandpass response. This is the filler driver, a driver/filter whose response is centered at the xover frequency and which rolls off at 6dB/oct either side.
B&O sold many models using this principle.
The drawbacks are that the filler driver must cover a wide frequency range with low colouration. Thus it must be of high quality. It also runs in parallel with the main drivers so you have to watch impedance. Also, off axis response, particularly in the vertical direction becomes very untidy.
In the end, I believe that the potential benefits of fixing the phase were overwhelmed by the drawbacks, and the reality never lived up to the potential.
This is one of the aspects of correcting phase response that I believe is too often overlooked. Nearly all practical implemantations of "linear Phase" results in very poor off axis response. The phase is only correct at one point in space, yet the sound we hear from a loudspeaker in a room is a combination of many things, the direct sound, the early reflections and the late reflections. Only the direct response, along one narrow axis, is phase linear.

Andrew