2way speakers

I agree with @shadorne that most two-ways have a radiation pattern discontinuity in the crossover region. It can be viewed as an off-axis dip at the top end of the woofer, or an off-axis emphasis at the bottom end of the tweeter.

Let’s assume we have a 6" woofer and a 1" dome tweeter, crossover at 2.5 kHz. What happens is, the woofer’s radiation pattern is getting narrow as it approaches the crossover region, and is roughly 90 degrees wide (-6 dB limits) at around 2.5 kHz. But down that low, the 1" dome tweeter’s pattern is so wide that it would be virtually omnidirectional except that the baffle itself limits the tweeter’s dispersion to 180 degrees.

So if this speaker is essentially "flat" on-axis, its off-axis response has a lot of extra energy in the octave above the crossover, where the tweeter’s pattern is 180 degrees or so. The 2-4 kHz region just happens to be where the ear is the most sensitive. So that extra off-axis energy at the bottom end of the tweeter’s range can make the speaker fatiguing to listen to for long periods of time. This is partially because the extra energy is right smack in that most sensitive region, and partially because a significant spectral discrepancy between the first-arrival sound and the reflections can be detrimental anyway. Let me explain the latter; feel free to skip the next three paragraphs because they get kinda nerdy:

The ear derives directional cues from the first-arrival sound but ignores directional cues from reflections (repetitions of that first-arrival sound). So even in a reflective room, you can close your eyes and point right straight at me when I speak from across the room. The reflections still affect loudness and timbre, among other things, but their influence on sound source localization is minimized.

The way the ear/brain system does this is pretty cool: It puts a copy of each new sound into a short-term memory, and compares all incoming sounds with all of the sounds in that short term memory. If the spectrum is a match, then it’s classified as a reflection. If not, then it’s a new sound. This suppression of directional cues in the reflected sound is called the "precedence effect".

So, what happens if the spectrum of a reflection is off somewhat? In that case, the ear/brain system literally has to use more computing power to correctly classify the sound as a reflection. This extra usage causes the CPU to overheat (figuratively speaking), and the result can actually be a headache!

Imo there are several possible ways to minimize listening fatigue in the design of a loudspeaker. One is, to put a dip in the on-axis response at the bottom end of the tweeter’s range. This simultaneously reduces the excess off-axis energy in that region, and a happy medium can usually be found. It doesn’t totally solve the problem, but at least the speaker’s in-room response isn’t extra-loud in the 2-4 kHz region where the ear is most sensitive.

Another solution is to widen the pattern where it would otherwise be narrow, at the top end of the woofer’s range in our example. The (sadly discontinued) Cliffhanger Bulldog does this. It uses a low-bass-optimized 6" woofer with a 2" dome mid, and a quasi-ribbon tweeter for the top end. The different drivers were close enough in width (in the horizontal plane) that we end up with an off-axis response that tracks the on-axis response quite well. (Yes I know it’s not a two-way - but is neatly solves a problem that many two-ways have).

A third solution would be, to make the tweeter’s pattern narrower at the bottom end of its range by using a waveguide-style horn. Amphion would be an example. Imo this may be the preferred solution, as I subscribe to the school of thought that says early reflections are generally undesirable.

Obviously the radiation pattern in the crossover region is not the only consideration when choosing stand-mount speakers, but imo it’s worth being aware of, and it may explain things that we can hear but which do not show up in an on-axis frequency response measurement. And unlike many other issues, a radiation pattern problem cannot be fixed with EQ.

Best of luck in your quest.

Duke

dealer/manufacturer

"Dr Floyd Toole research conclusively demonstrated that these two factors [even frequency response and even dispersion] are however extremely important."

@shadorne , Yesss!

Toole found five factors to be reliable predictors of subjective preference:

1.  The first-arrival sound’s frequency response at the listening position;

2.  The spatially-averaged response across a window plus or minus 30 degrees horizontal and plus or minus 10 degrees vertical;

3.  The spatially-averaged response of the early reflections from the four walls, ceiling, and floor;

4.  The sound power, or the sum total of the speaker’s acoustic energy radiation; and

5.  The directivity index, which looks at the difference between the first-arrival sound’s frequency response and the sound power (summed omnidirectional response).

By making these measurements, Toole could predict with a high degree of reliability how several competing speakers would rank in a controlled blind listening test.   Notice that 4 of the 5 have to do with the radiation pattern.

Obviously nobody here is saying that these are the only things that matter, but they may be what matters the most. 

Duke