Time coherence - how important and what speakers?

Time coherence is difficult to achieve because the acoustical center for a moving coil driver is not a fixed physical point, but actually varies somewhat with frequency. Source - The Loudspeaker Design Cookbook, sixth edition, page 113.

Phase coherence in a multi-driver system requires first order acoustical slopes for both the low-pass and high-pass driver. The acoustical slope is the sum of the inherent driver response and the crossover response. Note that time and/or phase coherence can only occur in a single plane with a multi-driver system.

Human hearing characteristics aren't linear, but are full of thresholds and masking effects. For example, after .68 milliseconds the directional cues from a repetition of the original signal (an echo) are suppressed, and this suppression (masking effect) lasts for about 40 milliseconds. Now, if we knew precisely the thresholds that apply to time and phase coherence, we could take them into account in loudspeaker design. Alas, the published research is inconsistent on the audibility thresholds of phase and timing errors.

Mr. Joseph wrote: "I've found that other things such as the dispersion characteristics, smoothness of power response, and clean decay corellate more powerfully with imaging and driver integration."

As an amateur speakerbuilder for about 20 years (who tried it all, even my own version of an ultra-high-slope crossover), and now as a dealer, I have to agree with Mr. Joseph's emphasis on radiation pattern and decay characteristics. I personally would trade off phase coherency in exchange for good radiation and decay characteristics.

I once built a speaker with true a first-order acoustic crossover (which is a helluva lot harder than it sounds). Yes it imaged quite well from the sweet spot, but the timbre wasn't very good. Recombining the same drivers (a Jordan module and an Audax woofer) with a higher-slope crossover at a much lower frequency sounded much better overall.

Ya'll remember the legendary Dahlquist DQ-10? Well, Jon Dahlquist found that he could either align the leading edge of the waveform, or the trailing edge, but not both. After listening tests, he chose to align the trailing edge, as he found that getting the decay right was more beneficial than getting the initial attack right.

Just for the record, I don't sell Jeff Joseph's speakers; he's a competitor. And a damn fine one at that.

First off, the participation of Jeff Joseph and Roy Johnson have made this thread among the best I've ever encountered here.

Like the guy said to Einstien after one of his classes, "Before I heard your lecture, I was confused. Having heard your lecture I'm still confused, but on a much higher level."

I thank you both, Roy and Jeff, for taking the time to educate us. I hope you will feel welcome here whenever you see something that calls for your participation.

From the body of evidence presented, one might conclude that the ideal would be a single-driver full range loudspeaker whose radiation pattern remains uniform up and down the frequency range. Just so happens this has already been done. Sound Lab's big full-range electrostats have the innate coherence of a single driver, and their unique (and brilliant) faceted-curved geometry maintains coverage over a 90 degree angle front & back, from the dipole-pattern bass all the way up through the high treble. Of course the Sound Labs have compromises of their own (low efficiency, high cost, somewhat limited maximum SPL, large size, and demanding room placement). But to the best of my knowledge they are the only truly full-range loudspeaker to incorporate the otherwise mutually exclusive properties of time/phase coherence and uniform radiation pattern. Non-coincident multi-driver systems can't do both; there are no coincident (concentric tweet systems) that use first order crossovers; and "full-range?" moving coil driver speakers (like Lowthers) aren't truly full range. And besides, the radiation patterns of all these cone drivers change significantly over their operating range.

I suppose a concentric-tweeter driver with a first order crossover and a cardioid radiation characteristic in the lower octaves could also achieve both time/phase coherence and uniform radiation with respect to frequency. But I don't think that's been done yet.

Disclaimer - in case anybody doesn't know or hasn't figured it out, I'm a Sound Lab dealer.

Soix -

Thanks for pointing out my oversight! I stand corrected.

Theil actually has at least two concentric drivers, one of which is a mid/tweet module (requiring a separate woofer), and one of which is a midwoof/tweet module, requiring no separate woofer.

The radiation pattern would still vary considerably with respect to frequency, but output from the concentric units would be free of the off-axis anomalies that otherwise occur with non-coincident drivers using first-order crossovers.