Home Speaker Asylum

General speaker questions for audio and home theater.

[ Follow Ups ] Thread: [ Display  All  Email ] [ Speaker Asylum ]
[ Alert Moderator ]

-just a strange attack

Hi David-
It's not about "my way or the highway". Campy made explicit charges that everything I had said about loudspeaker design was ficticious- that "only he knew the truth..." (I asked him to technically elaborate and to show me where my claims were mistaken).

To clarify some mis-information:
Parallel 1st-order and QSO (which are SERIES 1st-order) filters are the only known ways to most closely approach time coherence throughout the crossover range- known since filters were first designed. Filter responses are controlled by physics, and these are the only filters to do a specified (minimum-phase) job.

The "constant phase shift" you mention, David, is not quite an accurate description of what is going on with the filter circuit:

At the crossover point, a parallel 1st-order crossover, by definition, delivers a signal with a +45 degree shift (an eighth wave-period) to one driver and a signal with a -45 degree shift to the other. That makes for a 90 degrees phase difference between the high and low signals at the crossover point- the voltage outputs are said to be "in quadrature".

As you move away from the crossover frequency- higher or lower- that phase difference remains at 90 degrees- which is a "constant phase difference", not a "constant phase shift". This means that the high and low signals are NOT changing in their relative phase at other frequencies- their relative phase change is zero.

With every other filter except QSO, they do change (their relative phase change is non-zero), which means time delays are injected- different ones at each frequency. And this means the transient response and amplitude response, and timbre and image are warped. It means the wave envelope is changed. B&O's papers in the AES journals, among others, show the math and the square-wave responses of all the commonly used filters. It's also in any filter-theory book for electrical engineers. This link also has related math:
http://ccrma-www.stanford.edu/~jos/filters/Group_Delay.html


To come close to passing on to the ear a first-order circuit's "no relative phase shift" output, we need the loudspeaker drivers to have extended bandwidths, rigid cones, linear motors and suspensions, that effective Zobel circuits are put on any moving-coil drivers, that the acoustic time-of-travel to the ear be the same for each driver, and that the cabinetry not color the sound. There are other factors, but those are the most important ones, or you simply cannot preserve nearly as much of the original signal's "wave envelope". And note that I am addressing direct-radiator technology here, not horns, which do have distinct advantages if done correctly.

"Aren't there other ways to manipulate drivers with various crossover topologies and physical placements for "time alignment?"
No David, not for direct radiator designs. One can confuse phase-coherence at the crossover point with having an overall minimum-phase behaviour (quite common). Bear with me here:
Minimum-phase system behaviour is what results in "time coherent" response. Miminum-phase behaviour is a linear-systems term, meaning what goes in comes out, in some linear ratio, undistorted. It would mean a perfect transducer. If you have a minimum-phase system, you have time coherent behaviour. Which is not the same as a "phase coherent" behaviour. That is a term used for a good-looking "in-phase" output from woofer and tweeter when both are producing one pure, un-ending sine wave at the crossover frequency, without regard to what happened at the beginning or end of that (common-to-both-drivers) single tone.
"Time alignment" itself can be used as just a reference to describe what happens in a small part of the tone spectrum (around the crossover point) or to a larger chunk of the audible bandwidth, like for a three-way design.

The problem achieving a completely minimum-phase speaker system is that the drivers have their own mechanical time-delays (phase shifts) at their frequency extremes. And no crossover can work backwards in time to "speed up" the signal, to "make up" for those mechanical delays. All one can do via higher-order circuits is to introduce enough extra time delay at the crossover point, on top of what the drivers already might have, to make the hi and lo signals at the crossover-point have a total phase DIFFERENCE of 360 degrees. Which is just "phase coherence". Again, phase coherence does not indicate anything about the drivers starting and stopping together, just that they are in-phase when reproducing the same continuous tone at the crossover frequency.

"For example, placing a tweeter below a mid-woofer, and using a third order parallel filter will result an upward polar tilt of 15 degrees (approximately) of the tweeter's response. If the two drivers are dead-on vertically aligned (and driver separation is properly calculated), the resulting response will at least nearly approximate "time-alignment" of the drivers."
The result is phase coherent at the crossover point, but not nearly as time coherent as a properly-done 1st-order design. There are other AES papers that plot the time delays of those circuits- which are different at each frequency. I have them here if you need a title. That work was done in the late 70's or very early eighties. Again, it is peer-reviewed, which means no errors were seen by other researchers in the math or the interpretation of the measured data.

"Didn't the Spica TC-50 have a fourth order Bessel filter on the woofer, a first order filter with the tweeter, and a sloped baffle to achieve "time alignment?""
I think those were the filters, yes. They were used in conjunction with the woofer's HF mechanical phase shift (its natural high-end rollof), and the tweeter's LF mechanical phase shift (its natural low-end rolloff). Spicas achieved only phase coherence at the crossover point. The wide-band output was not time coherent- revealed in an impulse-response measurement.

I hope this does clarify some of what confuses people about speaker design. Please do not take this as an attack on you or others! Hopefully you can see that this is also not about "my way is the only way". It's about following the laws of physics to their logical conclusions. And of course, as I re-read this, I probably could have explained things better. What I am proud of is that we have reduced phase shift from 200Hz to > 8kHz to levels unknown to dynamic speakers, and this really means a lot to the music. Why just 200Hz to 8kHz? Because at the very lowest and highest frequencies, we still have the time delays caused by the drivers' mechanical rolloffs. And to truly eliminate those, one needs drivers with bandwidths extending from DC to light. To extend the bandwidths, one could use Dr. Hill's Plasmatronic tweeter, or at least a ribbon tweeter flat past 100kHz. Or an electrostatic woofer for the lowest bass if it weren't for the size and low efficiency. But our little Europa speaker is relatively inexpensive too, which rules those out.

Campy says I could not have possibly designed products with minimal phase shift over this wide range, having broad and uniform dispersion, with low distortion and high power handling. Independent tests show I did. Our owners, dealers and reviewers hear the results. So he doesn't believe it- fine! If he wants to call me dishonest, he should take the initative and show how so, and not just make pronouncements, don't you think?

Best regards,
Roy

This post is made possible by the generous support of people like you and our sponsors:
  VH Audio  

Follow Ups Full Thread
Follow Ups
You can not post to an archived thread.

[ Top Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2025, Audio Asylum, all rights reserved ]
[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]