On 1/23/05 10:35 AM, in article , "John
Matheson" wrote:

Now take a magnesium cone loudspeaker - it's a great "hash" generator. Many
people mistake the hash for "detail" and conclude that some recordings that
are made unlistennable on these speakers (strident female vocals, saxophone,
complex choral works, etc.) are bad. Well it ain't true - they're just bad
speakers and incapable of producing natural sounding music as far as I am
concerned.

So you're saying Joe D'Appolito, Vance Dickason, Siegfried Linkwitz, Jack
Hidley at NHT, and so many others have no ability to discern a good driver?
All this time we all thought the magnesium cones' orders of magnitude
reduction of harmonic and IM components were responsible for the clean sound
they heard from those drivers. But according to you, they're all wrong.
They're just hearing hash. By your count these guys must just be plain deaf.

Regarding the driver used as a midrange drive in the Orions, you
can see Seas's own data he http://www.seas.no/excel_line/excel/E0022.pdf
It shows cone break-up modes (i.e. resonances) more than 20dB higher in
level than the passband response!!!!

In my world, and that of every speaker designer I know, you reference the
peak magnitude to the average sensitivity of the driver's pass band, not to
a response dip. Given that the data sheet clearly places the average SPL at
about 88dB (normalized to 2 pi radiation), and the peak at a bit over 100dB,
that would give you a 12-13dB peak, not the 20dB!!!!!! you so dramatically
quote. And that well defined peak is easily suppressed using a simple notch
filter, leaving a very clean pass band to work with. That's the whole idea
behind the design.

I've measured these speakers and if you
want to be truly horrified, you should see what they do to a step waveform,
which by the way is a good way of understanding what they will do when
excited by transients.

Forgive me if I don't quite grasp this. You put a step waveform into a
transducer that is, by definition, bandwidth limited with a defined out of
band resonance. Then, you're "horrified" to see that the resultant waveform
is not a step and has the out of band resonance superimposed because your
step signal excited it? Just exactly what have you learned from this
exercise that you could not have just as easily figured out by simply
examining the frequency response curve on the data sheet?

Now, since you're so easily horrified by tests, you might drop dead when you
see a laser holography scan of virtually any soft or composite cone (paper,
poly, aerogel, glass fiber, kevlar,etc). What you see is utter chaos-one
part of the cone moving in while the other moves out-sometimes in violent
fashion and constantly changing with frequency. Compare that with a scan of
a magnesium cone below breakup where all you see is pistonic motion.
You can hypothesize all you want that all of this mechanical chaos somehow
is rendered audibly insignificant because of "self damping", and that the
resultant output emerges undistorted. But it doesn't take exotic
measurements to prove this isn't true.
Prior to introducing the Excel Magnesium cone drivers we introduced an 8"
woofer with a paper cone.
This is a highly regarded Kurt Mueller paper cone, chosen for low distortion
and excellent damping properties, and used by many high end driver
manufacturers. You can see a data sheet at:
http://seas.no/excel_line/excel/E004.PDF. The only acoustic variable between
this and the W22EX001 is the cone. The motor and suspension are virtually
the same between the two. Compare the distortion graphs between the paper
and magnesium versions and look at you'll see how much higher the second and
third order in band products are with the paper cone. And, trust me, this
distortion is not confined to lower order harmonics. A simple sine sweep of
this paper driver clearly reveals audible resonances in the pass band while
a sweep of the magnesium cone unit under the same conditions sounds much
cleaner. The paper cone drive would probably look better in your step
response test. But that's more an indication of the uselessness of such a
test than the quality of the driver itself.
Would it be nice not to have the break-up mode? Sure, but it is not the
fatal flaw you make it out to be, nor is it responsible for some kind of
"false detail" as you claim. There are plenty of diverging opinions about
what constitutes "the best driver" This is the reason why we and many other
manufacturers make them available with different materials. If there was one
perfect way to do it, there would be very few drivers in our line. As it
stands today, we have hundreds.

Even though the crossover attempts to remove this part of the speaker's
response, the hash these modes generate is blatantly obvious in music
reproduced by them - once you are able to identify it. (Linkwitz's crossover
attempts to suppress the first ring mode by 50 dB or so but they are still
starting from 20dB up. At 50 - 20 = 30dB below the music, the ring modes are
clearly audible in the Orions and a serious enough defect for give the
speaker a "fail" in my opinion.)

The "ring mode" is at 5kHz. The midrange crossover network attenuates output
at 5Khz near 50dB on its own. You seem not to take into account the active
notch filter centered at 5kHz that adds an additional 16dB attenuation over
and above the crossover's 4th order slope. Taken together, this adds up to
approx 65dB signal attenuation from the pass band level. If you take the
corrected calculation of the driver's peak magnitude, you end up with the
level of that peak suppressed at least 20 dB (100x) more than your 30dB
claim. I verified this in conversation with Dr. Linkwitz.

The Peerless XLS bass drivers are another driver that I can not like, though
I don't know exactly why in a technical sense. It has been put to me by
another loudspeaker design engineer that the hysteresis in the roll
surrounds is probably the cause of their "sound".

I doubt it. We have done extensive Klippel analysis of these drivers and the
hysteresis issue is no better or worse than any other 10" long throw bass
driver we have tested. These drivers use high quality natural rubber
surrounds that keep hysteresis to a minimum. Besides, hysteresis effects are
most pronounced at small excursions. Given that the Orion woofers are
equalized for dipole radiation, their excursion is far greater for a given
spl than would be the case in an enclosure application. If anything ,this
would minimize the hysteresis effect. Linkwitz also wisely mounts the
drivers in "push pull" (one pointing in, the other pointing out) and this
serves to balance out some of the even order products generated by the
modulation of the surround roll.

Unfortunately the Thiele Small alignment model
and it's derivatives are just that - models - that have fundamental
limitations. They can't predict the real world sound of systems - yet I have
known many speaker designers that put more faith in the modelling than their
own ears!

Well isn't that what you are doing here? You say the midrange can't be good,
not just because you don't like the sound, but because you see an out of
band breakup mode on the data sheet. Then you turn around and say you don't
like the bass, but you admit to having no technical reason for doing so. It
just seems to me that your dislike of this speaker is based as much on
preconceived notions as it is upon what you are hearing. And I strongly
suspect that one influences the other.

John Stone
SEAS USA