the breakup behavior of the various woofer/midrange drivers), so at least you're getting the best spatial imaging that the Taminos can deliver. If however one Tamino is oriented toward you even a fraction of a degree differently than another Tamino, then its destructive interference notch will null out a different frequency of the music, and the spatial imaging will naturally suffer.
Note also that this situation bodes ill for listeners who are not seated in the ideal central hot seat, since they cannot possibly be aligned at the same horizontal angle to all the surrounding Taminos, so they will hear the destructive interference notch problem nulling out the music at quite different frequencies from each of the surrounding Taminos. For them, the Tamino will probably have poorer spatial imaging performance than competing loudspeakers that do not have this problem of a destructive interference notch that varies so sensitively with the slightest change in horizontal alignment.
The imperfect matching of the non-identical drivers, and the imperfection of the woofer/midrange driver's contribution to the crossover due to its breakup in that spectral region, create significant sonic problems (especially this last destructive interference problem) only because the steep phase rotation of the particular crossover design chosen by Verity makes this loudspeaker system especially sensitive to these imperfections. In other words, it is only because the Tamino's crossover changes phase so rapidly within a small frequency span in the crossover region that we get the problematic destructive interference from the necessarily mismatched and imperfect driver pair. Another different crossover design, with less total phase rotation and especially with less steep phase rotation, would not have such a sensitive vulnerability.
-- Testing Inverted Polarity for Tweeter
In fact, as part of our investigative research we decided to test this thesis. To test this, we simply deliberately inverted the polarity of the tweeter connection on a pair of Taminos. This inverted tweeter connection cut in half the total amount of phase rotation through the crossover region, from 360 degrees down to 180 degrees, and also cut in half the steepness of the phase rotation slope, so that there was only half as much phase rotation per frequency increment. This should reduce all the above midrange coloration problems. And sure enough, it did. Voices especially became much more natural, with more realistic flesh and blood presence, less tonal balance recession, and less of that peculiar ghostly phasiness. However, with the tweeter connected in inverted polarity, the upper midrange and treble spectral regions, which are above the crossover region and are totally within the tweeter's purview, became notably softer and sweeter, with the indirectness characteristic of inverted absolute polarity in general. This soft sweetness from the tweeter was not too objectionable, and in fact could be beneficially euphonic in combination with typical solid state electronics, which tend to sound artificially hard and glaring in the upper frequencies, and which can therefore sonically benefit from some softening and sweetening by the tweeter. However, this inverted polarity soft sweetness in the upper midrange and trebles is still an inaccurate coloration, and is definitely not as accurate in those regions as the correct polarity tweeter connection chosen by Verity.
Thus, the cruel choice for a third order crossover, dictated by the laws of physics, comes down to choosing between getting the midrange colorations described above, which then gives you accurate upper midrange and treble regions -- versus getting less of the midrange colorations described above, which then gives you softened and sweetened upper midrange and treble regions. If the Tamino had dual sets of terminals for bi-wiring, you could hear the difference for yourself, and see which set of sonic tradeoffs you prefer, by simply connecting the cable to the tweeter terminal in inverted polarity (technical aside: optimizing this inverted connection for actual use might also require some changes to parameters of the electrical crossover network). But since the Tamino does not offer bi-wiring, you can't do this experiment. As it stands, then, the Tamino, with its tweeter connected in correct polarity, gives you stunningly accurate upper midrange and treble regions, but a full dose of all the midrange colorations discussed above. And our research experiment here, where we deliberately tried the Tamino with the tweeter inverted, serves as further proof that the Tamino's midrange colorations which we heard are indeed primarily caused by that large, steep phase rotation that the laws of physics dictate when you employ a third order crossover and then follow the purist route of connecting both drivers in the same correct polarity.
-- Flattening On Axis Response
Finally, there's yet another factor which contributes to the Tamino's midrange coloration, specifically to its tonal balance recession here. The crossover includes a 6 dB per octave filter to equalize the midrange output from the woofer/midrange driver. This filter begins rolling off this driver's response above 1200 Hz, and reaches its 3 dB down point at 3750 Hz (which is close enough to the nominal 3500 Hz system crossover point that this filter acts as one pole of the 3 pole third order crossover filter network, for the woofer/midrange driver, the other two poles being obtained from this driver's natural mechanical breakup and rolloff behavior). The goal of including this filter, which slopes downward above 1200 Hz, is again an example of praiseworthy purist design thinking. The design goal is to equalize the on-axis frequency response of the woofer/midrange driver, which evidently begins gently rising above 1200 Hz, and thereby produce a flat measured on axis frequency response from this driver.
That's a valid design choice from an abstract purist viewpoint, but it might not be the best pragmatic choice for this particular loudspeaker system. Given the Tamino's other problems with midrange coloration, many of which (as discussed above) contribute to making the Tamino's midrange tonal balance already sound recessed, it might not be the best pragmatic move to introduce a filter that rolls off the midrange even more, regardless of what the purist academic abstraction of an on axis frequency response measurement says. Furthermore, when we also take into account the manner in which human hearing works, we might realize a flatter, less colored midrange tonal balance from the Tamino, as actually perceived by a human listener, by deliberately letting its on-axis response naturally rise gently in the midrange.
Say what? How can a rising on axis response produce a flatter, more neutral perceived tonal balance? This story starts with the fact that most drivers begin to progressively narrow their radiation pattern for the higher frequencies of their spectral output capability. Typically this narrowing produces a gently rising on-axis frequency response measurement at these higher frequencies. But, because the driver's radiation pattern is progressively narrowing at these higher frequencies (which is the midrange for the Tamino's woofer/midrange driver), it spreads these higher frequencies (the midrange here) less widely into your room, so there is less of this upper frequency (midrange) energy bouncing off your room's side walls and reaching your ears.
Now let's discuss how human hearing works. Our ear/brains judge the temporal performance of a loudspeaker (its attack speed, coherence, etc.) by evaluating the direct first arrival sound we receive, along the direct path between each loudspeaker and our ear. But our ear/brains judge the tonal balance of a loudspeaker in a completely different way. We judge tonal balance (warm vs. lean, bright vs. dull, forward vs. recessed midrange) by evaluating the total power we receive from the room, at various frequencies of the musical spectrum. The reverberant energy we receive from the room walls around us, at various frequencies, plays a key role in the total power we receive at various frequencies. The sound energy we receive directly from the on-axis response of each loudspeaker plays a minor, indeed insignificant, role in the total power we receive.
That's why acoustic treatments of a room (or concert hall) play such a large part in the tonal balance we hear from our system (or live music). Think about it. If the reverberant energy from your room all around you were not important in judging tonal balance, then it would not make any difference to your perception of your system's tonal balance if you put room décor items all over the place that absorbed primarily high frequencies (e.g. stuffed furniture, carpet, drapes, etc.). But of course it does make a huge difference, which proves that our hearing perception of our system's or our loudspeakers' tonal balance does indeed depend heavily on the tonal balance of the total reverberant energy bouncing around the listening room.
Now, if a given loudspeaker driver (say any woofer/midrange driver) begins to progressively narrow its radiated field at its upper frequencies (say here the midrange), then naturally it will radiate progressively less of these upper frequencies at the room's side walls -- whereas it will radiate its lower frequencies in full measure at the room's side walls, because its radiated field is wide instead of narrow at these lower frequencies. The obvious result will be that the total reverberant power we receive, from the reverberant energy bouncing off the side walls, will be progressively less at these upper frequencies (here the midrange). Consequently, because we humans judge tonal balance based on total reverberant power received at various frequencies, we will hear this driver's tonal balance as progressively declining in its upper frequencies (here the midrange), so we will hear it as having a colored tonal balance, and in this case we will legitimately judge this driver as having a recessed midrange. The fact that most woofer/midrange drivers narrow their radiated field in the midrange, thereby radiating less midrange energy to your room's side walls, causes a tonal balance recession in the midrange, as surely as if you had installed special acoustic absorbers along your side walls that soaked up and absorbed only the midrange region of the musical spectrum.
Meanwhile, the direct on-axis sound we receive from a loudspeaker plays a very minor, indeed virtually insignificant, role in its perceived tonal balance. To understand why, try visualizing the following thought experiment. Suppose you sit 6 feet away from each loudspeaker. To keep things simple, imagine a huge beach ball, 12 feet in diameter, with its center at the loudspeaker location. This beach ball represents the total radiated power that the loudspeaker puts out in all directions (which it does indeed do at low frequencies). Now, imagine bringing one of your ears up to the surface of this giant 12 foot beach ball, and laying your ear against the surface of the ball. Compare the miniscule area of your ear canal opening to the huge surface area of this giant 12 foot beach ball. That teeny fraction, about .000006, represents how much of the loudspeaker's total radiated power your ear can intercept from the direct on-axis path from loudspeaker to you. The entire remainder of the beach ball's huge surface area represents reverberant power that you hear only after it bounces off your room walls, after it has been directed there by the loudspeaker's hopefully wide radiated field.
So let's consider a typical woofer/midrange driver, whose radiated field narrows at its upper frequencies (the midrange region) to say 1/10 the surface area of that beach ball. Its frontal dispersion at these midrange frequencies would still measure very well, apparently allowing a wide choice of listening angles. But its total radiated field into the room would narrow to just 1/10 of that at lower frequencies, so its total radiated power into the room in the midrange would decline to a mere 1/10 of what it is at lower frequencies, and hence its tonal balance as we hear it would be down a whopping minus 10 dB. That's a very unacceptable amount of tonal coloration, which we would surely hear as a tonally recessed midrange.
But, interestingly, many drivers produce a compensating error, as their radiated field narrows at their upper frequencies. Their on-axis frequency response tends to progressively rise, as their radiated field progressively narrows, at progressively higher frequencies. From a tonal balance point of view, this gradual rise in on-axis response is very beneficial, since it offsets and can approximately balance out the gradual decline in power radiated off to the sides at progressively higher frequencies (going up into the midrange in this case). Put another way, just as this driver begins progressively failing to output an adequate amount of progressively higher frequencies off to the side, in order to contribute to the overall room power, it also begins to output progressively higher amounts of progressively higher frequencies in the general direction of the listener, in order to contribute to the same overall room power.
Thus, the overall room power contributed by this driver can remain moreless constant at progressively higher frequencies toward the upper reaches of its operating range, which means that this driver's tonal balance as we humans perceive it can remain moreless flat and neutral. But of course this compensating error can only produce flat and neutral perceived tonal balance if we allow it to, by allowing the on-axis response of this driver to gradually rise as it naturally wants to do. If on the other hand we intervene and flatten down this driver's on-axis response, via an added equalizing filter, then this compensating error of rising on-axis response won't occur, and won't be able to work its compensating magic of delivering neutral overall tonal balance - so we would instead hear the driver's tonal balance declining and getting recessed in its upper frequencies (here the midrange), by virtue of its steadily narrowing radiated field and hence its steadily declining total radiated power into the room.
Why on earth would we as a loudspeaker designer want to flatten down this driver's naturally rising response, when it can give us the benefit of flatter, more neutral tonal balance? Two reasons.
First, as conscientious purists, we might want to proudly show that our loudspeaker system's measured on-axis frequency response looks flat, rather than evincing a progressive rise in the midrange, which then apparently goes down in level when the tweeter cuts in above the crossover frequency (which by the way would in fact be entirely appropriate, since the smaller diameter tweeter cuts in with a much wider radiated field angle, thereby once again bringing up the total room power, in spite of its apparently lower on-axis response output).
Second, we might want our loudspeaker to have the best possible time domain performance as heard on-axis. The temporal performance of a loudspeaker (or any minimum phase system) is, generally speaking, optimized when the frequency response on-axis is flattest, all other things being equal. And, since human hearing does evaluate the temporal performance of a loudspeaker from the first arrival direct sound that we hear on-axis (not from the room reverberation energy we hear), for purposes of optimizing a loudspeaker's temporal performance it is indeed important to optimize the on-axis behavior of the loudspeaker, regardless of what this does to the overall radiated power response and hence the perceived tonal balance.
Here then is another area where the laws of physics give a loudspeaker designer a very cruel choice. Should he flatten the rising on-axis midrange response of the woofer/midrange driver, in order to give the loudspeaker the best possible temporal performance on-axis through this midrange crossover region, and also in order to able to show off a flat on-axis frequency response measurement, even though that means a colored tonal balance with a recessed midrange? Or should he instead allow the on-axis response of the woofer/midrange driver to naturally rise through the midrange, in order to give the loudspeaker a more neutral, uncolored tonal balance through the midrange crossover region, even though that means poorer temporal performance through the midrange crossover region and a less impressive looking on-axis frequency response measurement?
There is no correct answer, and there is no easy choice. It's an especially difficult quandary for designers of loudspeaker systems that employ first order (or quasi second order) crossover networks, since these systems can indeed deliver outstanding temporal performance, so it would a pity to sacrifice this, yet most listeners hear tonal balance more easily and more obviously than they do temporal performance (personally, I might try the pragmatic tactic of splitting the difference with such a system). But, for designers of loudspeakers that employ higher order crossover networks, the choice should be easier. Most such systems cannot even approach ideal temporal performance, due to the temporal imperfections imposed by higher order crossovers (unless you're a design genius like John Bau, who was able to achieve this with his Spica loudspeakers). So there's little point to trying to optimize temporal performance of these systems. For these systems, then, it probably would be wiser to provide flatter, more neutral tonal balance through the midrange crossover region, by allowing the woofer/midrange to naturally rise on-axis. This would make the loudspeaker's tonal balance actually sound more neutral and uncolored, even though it would make the on-axis frequency response measurement look less impressive (to the average consumer who did not understand the true sonic import and true sonic benefit of this non-flat on-axis curve).
According to Verity, the woofer/midrange driver of the Tamino does exhibit this gradually rising on-axis response in the midrange, which is also the general crossover region for this loudspeaker system. And Verity has made the choice to flatten down this rising on-axis response, by applying that first order electrical filter that starts flattening at 1200 Hz and then very gradually increases in slope (as all first order electrical filters do), not achieving its 3 dB down point until 3750 Hz. So this is a mild flattening, which in the interest of purist perfectionism does give the Tamino a flatter, better looking on-axis frequency response measurement. But this flattening also subtracts from the Tamino's midrange tonal balance energy, which is presumably already being reduced by the narrowing radiation pattern of the woofer/midrange driver in this upper frequency part of its spectral coverage (as just discussed, this makes the midrange tonal balance more recessed by reducing the Tamino's total power output into the room in the midrange region). Moreover, the Tamino already has further coloration problems that produce tonal balance recession in the midrange crossover region, due to the nature of the crossover, as discussed previously above. In short, the further reduction of midrange energy by this flattening filter further worsens the Tamino's midrange coloration, specifically its midrange tonal balance recession. So the inclusion of this flattening filter does not seem like a wise move.
It could be argued that this flattening filter, by flattening the on-axis response from the woofer/midrange driver in the crossover region, works to optimize the Tamino's temporal performance, which is evaluated by our hearing from the direct on-axis first arrival sound. However, the Tamino's temporal performance through the midrange crossover region is already a very imperfect mess, and beyond salvation or optimizing, due to the steep phase rotation (through complete polarity
(Continued on page 100)