Before early 1970's, virtually all speaker manufacturers in the world positioned their speakers on front baffles following a logic of design rather than acoustics. The arrangement was to put the tweeters at the top and midranges and woofers at the bottom… mainly a matter of size. Then someone decided it was better to arrange the speakers in phase, aligning the acoustic centers by making them equidistant from the listening position. After considerable acoustic field research, ESB reached some conclusions about sound reproduction and fidelity. These conclusions have since been adopted by many other speaker manufacturers. Our conclusions were:

1) The equidistance of the centers of the speaker drivers from the listening point is not critical for a woofer, and have minimal effect at the coupling of a midrange and tweeter. Therefore, it makes no sense to position speaker drivers on different planes. (In fact, “phase aligned” loudspeakers soon disappeared).
2) The centers of the speaker drivers that reproduce the mid-high frequencies must be aligned on a vertical line.
3) The location of speakers that reproduce the mid-high frequencies must be asymmetric as it relates to the center line of the speaker enclosure.
4) The speaker drivers that reproduce the mid-high frequencies should be as close as possible, with reference to the frequency at which they intersect.

Thus, was born what became known as UMA (Unit Medium High) or Mid/Tweeter, mounted adjacent to each other on the same flange. UMA distinguished virtually all future design and production at ESB.

 

 

Phase and the Research

(Article published in the 1975 ESB catalog, shown above, page 16 and 17)

​

The most relevant and oft discussed topic in the field of loudspeaker design these days is "phase". Speaker builders around the world produce models with diffusers and speakers drivers arranged on different planes. These designs are an attempt to align the acoustic centers of the drivers to allow perfect phase response. They have built number of very sophisticated tools that can detect phase differences of only a few degrees between the waves from the various drivers in the speaker system.
 

At the same time, conflicting views on the sensitivity, or lack thereof, of the human ear to detect delays of a few milliseconds, corresponding to phase differences of several hundred degrees, called the whole issue into question.

But consider, for a moment, that the human ear is at least as sensitive to phase as the best tools. Suppose the recording of a musical piece is carried out impeccably (unlikely given the presence of microphones in various locations during the recording). Suppose also that the mixer, equalizer, disk or tape, the recorder, and the amplifier do not induce any phase distortion, and the signal reproduced by the loudspeaker maintains the same phase relationships as the original signal.  

Still, the listener remains the concern. He must be in the position recommended by the loudspeaker manufacturer in relation to the position of each driver on the front panel of the speaker. If the listener is out of position, phase lags will be far greater than those that might be introduced by the recording/playback systems. And all this makes sense only if the high playback quality of the “perfect” speaker is actually appreciated by the human ear in normal use with musical program material.

Among today’s engineers, rather than phase shifts, they prefer to speak of signal delay; the difference in time it takes signals from each driver to reach the ears of the listener. This does not change the nature of the problem; it is simply looking at it from a different angle. Practical controlled tests show in this case that the human ear is unable to perceive quality differences in the signal when these delays are contained within a few tenths of a millisecond for the medium-high frequencies and up to a millisecond for the lower frequencies. 

Evaluated in these terms, and making more practical considerations for actual and theoretical listening, the "phase" remains a topic whose importance is yet to be demonstrated, and not the one that is assumed in many manufacturers’ advertising pages.

 

About ESB, engineers have said: it is preferable to have the speakers perfectly in phase with the original recording, but the price of the required constructive solutions are not justified when other factors can have a greater effect on the quality of reproduction.
A far more important consideration is to have the speakers’ crossover frequencies not only on the axis of the speaker, but if possible, on as broad a projection as possible. The crossover frequencies simultaneously operate on two speakers. If at the listening point, their signals arrive in opposing phase, they will interfere with one another, creating a hole in the frequency response that theoretically matches the annulment total. It can be 10dB or more. This is easily noticeable because the width of this hole is generally equal to 2 or 3 third-octave bands. For signals to maintain a proper phase relationship, it is necessary that the speaker distances (at the acoustic centers) from the listening position differ as little as possible. The maximum difference is preferably not greater than one third of the wavelength; in this case, the hole is only 3dB. Let's take an example. Suppose we have two speakers whose frequency bands are crossed at around 5000 Hz. The wavelength of this frequency is equal to 6.8 cm. We will have a correct response in amplitude in space in the front of the speakers when the difference of the distances of the two acoustic centers from the listening point is less than 2 cm.
Clearly, the closer the acoustic centers are and the more distant the listening point, the greater the linear response listening area will be.