Horn Question

[whitebroncoii]

Mustang Marvin and Altec Best: Please re-read bfish's response. He describes an excellent method for aligning a two-way loudspeaker. The acoustic center (source of the sound) on a woofer is not the voice-coil or the edge of the frame, and the acoustic center of a HF horn/driver is not the voice-coil or the edge of the horn. bfish's method is described in old Altec literature for theater system set-up . . .

Another way to align a two-way loudspeaker is to measure the time difference using an analyzer. A passive system is then physically aligned (fore/aft position of the source). You can adjust the alignment in a by-amp system using digital delay. Delay the closer acoustic center to the distant acoustic center. These methods require special equipment (expensive), highlighting why bfish has described an excellent method for most of us to use.

An AES white paper that I assisted on offers this;
The closer the acoustic center is to aligned, the more phase damage, until they are perfectly aligned and in-phase. The converse is that the farther out of alignment the less damage occurs. Aligning the mouth of the horn with the front edge of the woofer can sound better than lining up the voice-coils!

[bfish]

Thanks G!

I tried to be simplistic, that's usually pretty easy for a simpleton like me...

[Steve Schell]

As I understand this situation (may be a bit faulty here and there) the "acoustic center" or virtual source of sound can be considered to be at the plane of the voice coil on a high frequency compression driver due to the very light moving mass. On a woofer with many times the moving mass the acoustic center is somewhat behind the plane of the voice coil due to inertia slowing the response time. This helps somewhat on a coaxial driver like a 604, at least partially compensating for the distance between low and high frequency voice coils.

Bfish's method of reversing polarity of one driver is a good one, as it creates a cancellation at certain points in space due to one driver moving out while the other is moving in. At points in space where the distance between acoustic centers of the two drivers are equal the cancellation is most complete and the null is greatest. Assuming vertical alignment of high and low frequency drivers, one can visualize how the locations of greatest cancellation would form a straight line (or beam) whose angle vertically would vary as one driver is moved closer or farther from the listening position while the other is held constant. Find that point where the beam coincides with the listening position and the alignment is optimal.

There is more to this Humpty Dumpty job of putting the audio signal back together, at least when it comes to creating a playback which is most transparent to the source, preserving localization (imaging) information in a stereo recording. Second order passive crossovers require the high and low frequency drivers to be connected in opposite polarity. This preserves even power response of the system through the crossover region, though imaging flies out the window. Audio waveforms are highly transient and asymmetric, and a system with drivers connected in opposite polarity cannot reconstruct such waveforms. First and third order networks permit the drivers to be connected in the same polarity. First order networks can sound terrific though they offer very little protection from low frequencies for the high frequency driver. Third order networks offer great protection, though at the cost of a high parts count and possible sonic effects due to the signal having to fight its way through all those parts. No free lunch!

[tomt]

Thanks G!

I tried to be simplistic, that's usually pretty easy for a simpleton like me...

doesn't the "quote" go something like this?

“Any fool can make things complicated, but it requires a genius to make things simple.”

[bfish]

Guilty.

[whitebroncoii]

As I understand this situation (may be a bit faulty here and there) the "acoustic center" or virtual source of sound can be considered to be at the plane of the voice coil on a high frequency compression driver due to the very light moving mass. On a woofer with many times the moving mass the acoustic center is somewhat behind the plane of the voice coil due to inertia slowing the response time. This helps somewhat on a coaxial driver like a 604, at least partially compensating for the distance between low and high frequency voice coils.

For several years we have been measuring loudspeakers and components for design programs like "Ease". These measurements have shown that the acoustic center of a horn is a couple of inches up the throat. The exact distance depends on the horn and throat. In addition, the acoustic center of a constant directivity horn changes slightly with frequency. This is not a real concern when working with one frequency at the crossover point. Your description of the woofer is interesting. Not exactly what we have measured, but interesting. It is true that woofers will lag behind HF drivers due to their mass. The larger the woofer, the more they lag. This also means that smaller woofers have little lag and not really a consideration.

Bfish's method of reversing polarity of one driver is a good one, as it creates a cancellation at certain points in space due to one driver moving out while the other is moving in. At points in space where the distance between acoustic centers of the two drivers are equal the cancellation is most complete and the null is greatest. Assuming vertical alignment of high and low frequency drivers, one can visualize how the locations of greatest cancellation would form a straight line (or beam) whose angle vertically would vary as one driver is moved closer or farther from the listening position while the other is held constant. Find that point where the beam coincides with the listening position and the alignment is optimal.

Alignment of a loudspeaker is about aligning the components. Although the listening position is critical for stereo imaging, once the loudspeaker components are aligned, they are aligned at 2, 20 or 200 feet away. Home hi-fi folks are encouraged to align at the listening position.

There is more to this Humpty Dumpty job of putting the audio signal back together, at least when it comes to creating a playback which is most transparent to the source, preserving localization (imaging) information in a stereo recording. Second order passive crossovers require the high and low frequency drivers to be connected in opposite polarity. This preserves even power response of the system through the crossover region, though imaging flies out the window. Audio waveforms are highly transient and asymmetric, and a system with drivers connected in opposite polarity cannot reconstruct such waveforms. First and third order networks permit the drivers to be connected in the same polarity. First order networks can sound terrific though they offer very little protection from low frequencies for the high frequency driver. Third order networks offer great protection, though at the cost of a high parts count and possible sonic effects due to the signal having to fight its way through all those parts. No free lunch!

Where to start here . . . Yes, audio waves can be reconstructed with the components connected with opposite polarity. Please try and review "polarity" versus "phase". These are not the same. Connecting the drivers with the same polarity does not necessarily put them in-phase. In-phase, or aligned, means the signal generated by the two components reaches the listener at the same down point in playback time.

Most Altec loudspeakers with passive crossovers have the HF polarity inverted to put them in-phase. This correction is needed because there is a phase shift introduced by the crossover (all filters cause a phase shift and a passive crossover is a filter). Add this phase shift to non-aligned components and inverting the polarity often puts the system closer to in-phase.

Not sure what power response has to do with phase and alignment. Certainly as more and more power is applied to the components they will begin to react in a non-linear fashion (lagging behind). This problem is not automatically the crossover's fault.

It is a common belief in some circles that crossovers are a bad thing. This concept leads some folks to running their systems with bare minimum high-pass filters on the tweeter and letting the components overlap. Starting from this concept can get you to this statement; ". . . possible sonic effects due to the signal having to fight its way through all those parts. No free lunch!" I suggest that the comb-filtering caused by the overlap is much more damaging then the crossover.

[Steve Schell]

Whitebroncoii I offered my thoughts and opinions with the caveat in the first sentence. I would be happy to discuss these concepts further with you, though hopefully without the overlay of condescension and schooling. I assure you that I fully understand the differences between polarity and phase, and I chose my words carefully.

First, could you explain how high and low frequency drivers connected in opposite polarity (i.e. one moves in and the other moves out on a positive signal) might properly reconstruct an asymmetric waveform (i.e. a snare drum hit) at the listening position? If there is a way to do this I am all ears. I am of the opinion that most audio signals are not simply periodic but rather highly transient.

[whitebroncoii]

Whitebroncoii I offered my thoughts and opinions with the caveat in the first sentence. I would be happy to discuss these concepts further with you, though hopefully without the overlay of condescension and schooling. I assure you that I fully understand the differences between polarity and phase, and I chose my words carefully.

First, could you explain how high and low frequency drivers connected in opposite polarity (i.e. one moves in and the other moves out on a positive signal) might properly reconstruct an asymmetric waveform (i.e. a snare drum hit) at the listening position? If there is a way to do this I am all ears. I am of the opinion that most audio signals are not simply periodic but rather highly transient.

I also tried to choose my words carefully, not knowing who the audience might be. Can a HF and a LF component reproduce a signal at crossover with one of them reversed in polarity without horrible things happening? Yes, when one component is half a wave length away physically or electronically. Am I going to be able to convince you? Probably not.