Sloped baffle

@Ngjockey ... let me try to unpack what you just wrote. Let's assume we have a single dynamic cone speaker with a pass band of 35Hz to 20K Hz. Let's forget about high frequency beaming and cone breakup. Just assume this hypothetical speaker has a flat frequency response within its pass band, as measure on axis. Obviously no X-over needed here.

Now ... like all dynamic drivers, we have a voice coil, a spider, magnets, and so forth. Let's focus on your comment about the voice coil being inherently inductive. Makes sense. After all, we have a wire coil moving in a magnetic field, producing voltage and its own magnetic field. The faster it moves, presumably, the more voltage and back inductive reactance to the input signal.

Now, if a complex signal was fed into the speaker, would there be phase shifting with respect to the higher frequencies as compared to the low order fundamentals? To be more specific, say the signal was composed of a 100 Hz fundamental, plus "n" number of harmonics into the high treble. I assume this complex signal could be visually reproduced on an oscilloscope.

If the driver's output was compared to the input signal, would there be some sort of harmonic difference between input and output signals? Would the speaker's lack of inherent phase coherence be the cause of this distortion? Would this phase nonlinearity be caused by the inductance resulting from the voice coil moving in the speaker motor's magnetic field??

Let's assume the answers to my questions are -- yes?? Is there a frequency range where a speaker is phase coherent, or does phase nonlinearity increase as a function of frequency ... period??

If the answers to all of these questions are -- yes, then it seems to me using 1st order X-overs and sloped baffles is at best a rough justice engineering response to a problem that is inherent with dynamic speakers that use voice coils.

So ... where do we go from here?? Magneplaners, ESLs??

Cheers.

P.S. Bombaywalla and Al, feel free to chime in. I think I'm getting tangled up in my shoe-laces.

Bifwynne,
I would very much like Roy J to jump in here & answer your question.....
Meanwhile, have you read Roy's white paper on "Time & Phase Coherence" on his website?
http://greenmountainaudio.com/time-and-phase-coherence/
when you read this paper, skip the initial part & read this section titled "Time Coherent Speakers". You'll see the response of the individual driver & how they add up in a time coherent speaker.
Then scroll past the rest of the material & read the section titled "Where a speaker goes wrong". I *think* you might get many answers (maybe not all) to your questions. Thanks.

Thanks Bombaywalla. I read Roy's White Paper, but will re-read the sections you suggested.

Meanwhile, I just checked Stereophile's bench test report of the Maggie 3.5R and see that it is not time coherent. In fact, JA speculated that the midrange was connected in reverse polarity to the tweeter and woofer. I assume similar characteristics for the 3.7i.

Bombaywalla,

The DSP signal processing is touching your music signal in a very fundamental way in that the entire music signal has to go thru the DSP before you can hear it. Same deal with the passive x-over. But the difference is that you change the quality of the cap or the inductor or the hook-up wire & you can change the sound to your liking. It appears that it's not that simple with the DSP software - you cant go in there & change the code. Or, maybe I'm not thinking of this correctly?

I think you have it right; at least from my perspective. However, I don't have the skills to change caps or wire...so I'm basically stuck with what I get. Actually, software provides more flexibility here, in my case.

So, if I'm envisioning this correctly - computer digital out runs into the DSP software which breaks the audio signal into highs, mids, bass. You get 3 digital streams now. You feed these 3 streams into 3 identical DACs or 1 Pro quality DAC able to output 3 analog streams (one box would be better as everything sits in 1 chassis & has a better chance of being matched to the other analog stream). Then 3 analog streams into 3 power amps - you need to match these very well too: same input sensitivity, same gain, same sort of clipping algorithm, same dynamic headroom extension, same power output capability, same current source/sink capability.

Here I would say, not exactly. Inside the computer being used as audio server the DSP software runs as well. On the DSP software (eg, Acourate) you set up XO frequencies, slopes, delays, etc, and perform the driver measurements, do the adjustments, etc, perform digital room correction, and eventually get a sort of digital filter. Then you apply this through a convolver to the audio player software (eg, JRMC). Now the computer is outputting through USB several channels. Eight in my case/plan. A multichannel DAC, such as the exaSound e28 takes USB in and decodes into the 8 channels and outputs 8 analog signals. Simple 1-box solution!

Also the amps don't need to be identical. You adjust gain at the software level. Take a look at the article by Mitchco I linked before. It's an easy read and provides a nice view of his setup.

I have in the past toyed with the idea of multiamping, but always in the analog domain. It always seemed it was too cumbersome, needed too many boxes, and was creating new problems. This newer technology seems to be bridging that gap. Or maybe it's me convincing myself?

Thanks for the clarifications regarding driver time-coherency. Conceptually I understand it. My gut feeling is, though, that lack of coherency is at least one order of magnitude smaller than that introduced by passive XOs. Right? If so, most of the issue would be solved with said software/approach.

Bombaywalla, I reread Roy's White Papers. He speaks to time and phase effects caused by speaker cone mass, suspension elasticity and damping. Nothing about phase shifting (if any) that may be caused by the inductive reactance of the driver itself, namely the voice coil moving in a magnetic field and producing back EMF. Perhaps Roy will catch my Q and share some thoughts.

If the driver's inherent inductance, as a stand alone factor, causes or contributes to nonlinear phase shifting, the challenge becomes a moving target.

Any ESLs out there that don't use X-overs??