Time alignment, important? long

"Sound wave certainly does not travel at the same speed for all frequencies, but music recording and reproduction often neglect this fact."

Are you sure about that? I'm pretty sure sound wave speed is not frequency dependent.

In time aligned speakers the drivers are offset to compensate for the fact that bass driver voice coils are deeper in the speaker than than the smaller high frequency drivers. Tilting the speaker back brings the voice coil of the tweeter to the same distance as the voice coil of the woofer.

Semi,

I'm certainly no expert here, but will offer my $.02 worth. Hopefully Roy Johnson of Green Mountain Audio will drop in on us.

First, assuming constant temperature and pressure, the speed of sound is the same regardless of the frequency being reproduced. High frequency sound does not travel faster than low frequency sound.

Now with all crossovers other than first order crossovers, there is a phase shift introduced by the filter, resulting in the low frequency driver lagging in phase relative to the high frequency driver. Take, for instance, a fourth order acoustic crossover (in which the filter topology may not look like a fourth order circuit, but the net rolloffs are fourth order). The woofer will lag 360 degrees (or 1 wavelength) behind the tweeter at the crossover frequency. Assuming a crossover frequency of 2 kHz, that 1 wavelength corresponds to a time delay of .5 milliseconds, or a path length of 6.75 inches.

Moving the tweeter back by 6.75 inches helps a bit because now the two drivers are in phase and time-aligned at the crossover frequency, but (as I understand it) that is the only frequency where we've really fixed anything, since that is the only frequency where the phase difference between the two drivers is exactly 360 degrees.

The only filter type that can preserve the time and phase relationship properly is the first-order filter, and then only when the listener is equidistant from both drivers. So in practice there is a fairly narrow vertical "window" within which time and phase correctness is preserved. Unfortunately first order filters are more sensitive to mis-alignment of the drivers, so outside of that "window" the first-arrival response will tend to have peaks and valleys that wouldn't be there with a higher-order crossover. Whether these on-axis peaks and valleys are audible or not I don't yet have enough experience to say. I'm expecting delivery of a pair of true first-order crossover loudspeakers soon, and hopefully will learn from them.

[As an aside, even-order crossovers are much more tolerant of misaligned drivers than are odd-order crossovers, as least as far as the on-axis, first-arrival response goes.]

I used to build loudspeakers as a hobby, and the best-imaging speakers I ever built used true first order crossovers. I was unable to successfully deal with the other challenges of a first-order crossover, so I went back to steeper slope filters which gave me much greater lattitide in driver selection.

Sloping the cabinet face back to align the drivers (with a first-order crossover) is a matter of simple geometry - the goal is to get the voice coils equidistant from the listener. There is no one "correct" angle - it depends on the physical characteristics of the drivers used (like how far back to the woofer's voice coil), how high off the ground they are, what their vertical separation is, and the anticipated listening distance and height.

If the speakers don't use first-order crossovers, then tipping the cabinet back so as to align the voice coils to be equidistant from the listener would not result in "time alignment", as the output from the woofer would still be delayed in time relative to the tweeter due to the inherent phase shift. That's probably why you don't see manufacturers of non-first-order crossover loudspeakers wasting their money building expensive sloped baffles.

The designers of Merlin and Dynaudio and other high quality non-first-order loudspeakers have no doubt taken the relative driver voice-coil offsets into account in designing the crossover. So tipping the speaker back to align the voice coils would more than likely be slightly detrimental, as now the condition the crossover was optimized for no longer exists. One case in which tipping the baffle back might well improve the imaging is if the front baffle has sharp edges. Tipping the baffle back would smear the diffraction out in time, hopefully making it less of a distinct sonic event and therefore keeping it from screwing up the imaging as much.

My understanding is that there are tradeoffs a-plenty in the mystical art of loudspeaker design - not that the science is mystical, but the selection of priorities (and subsequent selection of the best way to achieve them) certainly is.

I don't know if this helps any or not. I welcome correction from those who know more about this than I do.

Duke

Guess I was thinking about about the temp effect of ground vs. air which caused sound to propagate either upward or downward, depending on the temp gradient.

So with that straighten and put aside, I still have questions. If tilting the speaker back is simply to line up the voice coil of different drivers, that will be very simple and effective without moving the drivers back in vertical plane like Dunlavy which is the alternative method. I know x-over will introduce phase shift, I better since I am an MS EE graduate. A driver can potentially have a wide range of phase shift to reproduce especially woofer, but phase range decreases as we move up the freq in a 1st order x-over design. By tilting the speaker back by a few degrees, midrange driver which has less phase shift would actually either lead or lag behind on part of freq spectrum.

Take the Dynaudio Contour 3.3 measurement as an example:

http://www.stereophile.com/loudspeakerreviews/262/index6.html

Where x-over freq is at 350 Hz. Time alignment should only be done on the woofer, not midrange or tweeter. But Thiel, a big proponent of time and phase alignment, tilt the whole speaker back, which in theory can create more problem than solving it.

Who is right?

As my memory is restored, Ed Long who owned the trade mark for TA or time aligned speakers back in the late 70's had a speaker brand he enginnered that was not 1st order and it did have staggered drivers for time compensation.Hales speakers of the early and mid 90's had multi order crossovers and sloped baffle cabinets designed by Tom Thiel..Peter Snell the original, had a couple of multi order designs in the mid and late 80's that had time compensation and diffraction adjusted baffles. Peter Snell had U.S patents on baffle and room interaction designs dating back to the early 70's. The original Jon Dahlquist Dq10 was a phased array but may not have been of of 1st order design. These are examples of speakers with voice coils aligned that may not have been of true phase coherency, none the less, some were ground breaking in theory and design..Tom

I prefer single-driver speakers.

An old engineer once said,"The fewer holes you poke in your boat, the less bailing you have to do." :^)