Digital Active Speaker Thread..... - Page 6  

First, Keith Howard was not talking about crossover filters, and JA and Kal both had a hard time identifying the ringing filters on his blind test. I think it would reasonable to think that ringing artifacts are fairly subtle at best after reading that article.

Second, the Apogee comparison is not appropriate given that they were using analog crossovers, which have lots of artifacts as you increase steepness, and they were trying to combine two very different radiators: a dipolar ribbon and an omni woofer. The NHT design blends two similar drivers.

Also, is your argument about acausality or steep filters? The Stage was done with analog filters, and therefore cannot have acausal filters.

Third, it is not clear at all that the aspect of the sound of the NHTs you dislike is due to the ringing.

--Andre

Interesting that Dennis Murphy posted those comments just a month ago in response to Alimentall's gushing over DEQX as the next coming.

Duh.

Reverence?

Just more Alimentall-twisted interpretations. I think Dennis' words speak for themselves. No one here needs your twisting, John.

It is interesting that he said:

"There is a theoretical advantage, but the real world advantage just hasn't been substantiated yet." And this is from someone you say has reverence for it.

Hyperbole, oversimplification will destroy any serious discussions. I have posted my point of view and sincerely believe that the Digital Route to crossovers and music reproduction in general is the future... I refuse the point of view that any design using a Digital implementation is Automatically superior to ANY product utilizing more traditional approaches. I also refuse the weak argument that if it measured better and even it sounds crappy then it remains superior and that those who did find it lacking do not have good hearing... Very similar to those derided "Audiophile" arguments. I welcome balanced and constructive arguments not hyperbole and simplistic rationalization ... I am however beginning to see the contrary here...We shall see where that leads...

P.S. It is very easy to say the Xd measured better but better than WHAT? Can anyone point me to measurements of very large speakers such as the Wilson X-2? I have not seen any. I have heard the X-2 and my ears tells me very few speakers are in its league.. Very few.. Those who have heard it can attest to that..

When is the cost of an analog to digital converter and DSP crossovers going to be cheaper than the cost of analog crossover components. Don't forget, even if you use external amplification, you are still going to need a power cord going to the speaker to power the ADC and DSP making installation more difficult. I certainty think digital active speakers are terrific - I own three of them myself - but it it is never going to succeed on a cost basis when a handful of analog parts cost so little.

Have you ever actually listened to the NHT Xd and the Wilson Maxx2 in the same room at the same time. If not, then you can to continue to jump up and down and say measurements, measurements, measurements, but you are in no position to say the Xd is a better sounding speaker than the Maxx2.

John,

That's why I said the "original" Apogee in my previous post.

There's a guy named "Graz" in Australia who has been making replacement ribbons
for Apogee speakers that has acquired the rights to the Apogee name a couple years
ago. The "Definitive" and the "Synergy" are his speakers.

Neither the "Definitive" nor the "Synergy" is a product of the design efforts of original
Apogee founders Jason Bloom and Leo Spiegel. The original Apogee company
was acquired by a Canadian company in the '90s; and they essentially let the company
die:

http://www.apogeespeakers.com/company_history.htm

Jason Bloom unfortunately died after falling off a ladder several years ago.

I achieved very good results with the full-range Apogee ribbons using the active crossover and 4 Krell monoblock amplifiers. There were some problems in the lower bass with that model (the Apogee Grand, the $80K flagship that followed, incorporated a dynamic subwoofer with the 3-way ribbon design) and ribbon excursion, however.

I'm sad about Jason, as we were good acquaintances.

Lee

Andre,

NO - Leo Spiegel was NOT attempting to combine a dipole and an omni. This interview
was on the design of the Apogee Stage - which is a ribbon tweeter and ribbon woofer.
[Later on Apogee mated the two-way Stage with a omni woofer and called it the
"MiniGrand". But the speaker in question in the interview is the Stage which has
two dipole ribbon drivers.]

My argument is twofold. One is with steep filters in general because of the discontinuity
in radiation pattern. The filter in that case can be either analog or digital.

The other is the pre- and post-ringing that one can get with a digital filter.
The ringing of the Xd is something that I heard when we demoed the Xd at OB's house.

In comparison to the resident Lamm-driven Wilson X-2s; transients - in particular the
crash of cymbals was "smeared" and "overhung" on the Xd in direct comparison to a
very crisp "cymbal-like" sound from the resident Wilsons.

The Wilsons with OB's fine front-end electronics gave a very convincing illusion of a
"you are there" sound; while the Xd's seemed to call attention to the fact that they were
a reproduction.

John,

It doesn't matter whether the drivers are "pistonic" - besides that is only an approximation
anyway - true drivers are really not pistonic.

The relevant physics here is "antenna' theory - and loudspeaker drivers are "antennas".

Suppose we have a speaker like the Xd satellite - with a 5" cone and a 1" dome - or
whatever the Xd has. Suppose "fc" is the crossover frequency and the crossover is
very steep - so the range in which the two drivers overlap is very narrow, if not zero.

A frequencies just below fc; the 5" cone is doing the radiating - and the radiation pattern
is that of whatever wavelength corresponds to frequency fc being radiated with a 5"
antenna. The size of the antenna vis-a-vis the size of the wavelength determines the
radiation pattern.

Now we increase the frequency slightly to get us above fc into the tweeter passband.
The frequency is not only slightly above fc - so the frequency hasn't changed much -
but it is enough to put us into the tweeter's passband.

Now we have a 1" antenna radiating a wavelenght that is only marginally different from
the one the woofer was radiating. However, the size of the antenna has changed
suddenly from 5" to 1"; and hence the radiation pattern suddenly became more
isotropic with a slight increase in frequency.

Real musical instruments don't do that - they don't change their radiation pattern in the
frequency span of just a few Hertz. It would have been better to blend the output of
the drivers so that the transition from the 5" radiation pattern to the more isotropic
pattern of the 1" driver took place over a greater span of frequencies so that it was
less noticeable.

A sharp crossover between two different sized drivers is going to give a sharp transition
in the radiation pattern as a function of frequency at the crossover. Real instruments
don't have sharp transitions in radiation pattern. The artificiality of this transition calls
attention to the fact that the reproduction is an artificial one.

Again - Mother Nature doesn't have these sharp transitions in radiation patterns - why
would one try to emulate Nature with a characteristic that Nature doesn't have?

John,

I agree with that - with the proviso that the phrase "used properly" means that steep
crosssovers are to be avoided.

John,

Unless you have actually done the experiment and put a DEQX on the X-2;
your assurances are meaningless.

If you want to make that claim to a scientist - you have to do the experiment.

Digital filtering is a powerful technique and you can do a lot with it.

However, like any technology it can be mis-used.

Steep crossovers are not a virtue in and of themselves. The ideal speaker is not one
in which one has a cadre of independent, decoupled drivers; each handling a given
passband with zero regard to what its neighbor is doing.

Ideally, the response of the system should be continuous in both frequency and in
radiation pattern. Sharp transitions are an anathema to the continuity and smoothness
of the response.

John,

I had no preconceived notion. I listened to the X-2 and I listened to the Xds.

There was just "something" about the Xds that sounded "un-natural" There was something
about what I was hearing from the Xds that said "You are listening to a reproduction."

The Wilson X-2s were saying "You are listening to the real thing"; and although I knew
I wasn't listening to the real thing - the Wilsons were VERY convincing.

So I listened very critically!! I wanted to know what about the sound from the Xds
was subconsciously tipping me off that they were artificial.

One of the things that I identified was the sound of cymbals. The Wilson X-2 produced
the correct transient sounds that a cymbal makes. You hear the initial "click" of the
drumsticks hitting the metal. You hear a "ringing" as the sound waves in the cymbal
propagate outward from the hit site; but have not yet "discovered the boundary of the
cymbal" You hear the "shimmering" as the multiple reflections from the edge of the
cymbal are all interacting. The Wilson gives you a cymbal that really sounds like
a cymbal.

The Xds just sort of went "shhh" - and totally missed all this microdynamics that made
the cymbal sound so life-like. The DEQX digital processor in the XdA unit is doing the
temporal integration demanded by the digital approximation of the filter by summing
together the output of the various "taps" along the unit's delay chain.

That summation and averaging is what smears the transients that I heard so crisply
on the Wilsons. I could hear the "smeariing" and also understand WHY the digital
processing was giving me the smearing.

The frequency and temporal responses of a filter are not independent. [ They are
Fourier Transforms of each other ] If you are not judicious in your use of the DSP
in the frequency domain [ like demanding steepness in the filter ], you can very well
screw up the temporal response.

John,

The problem is that you can mess up the temporal response when you over-constrain
the frequency response.

As an analogy; you can build a car that has great gas mileage - if you build it out of
balsa wood - so that it has terrible crash survivability.

You can also build a tank that is great in crashes - but gets horrible gas mileage.

Good engineering design is that art of compromise - and the optimal design rarely, if
ever lies at the limits.

You are showing me the mileage figures over and over again for the balsa wood car;
and totally MISSING the point that my objection is to the poor crash performance.

The Xd graphs just show that it was designed to get good frequency response - but
the Xd is a "one trick pony". It does one thing well - but not the pantheon of multiple
characteristics that I expect in a great speaker.

The Xd is like the amps and receivers of the '70s that had tons of negative feedback
so that they would measure well on a THD benchmark when fed a 1 kHz sinewave.
Yes - those amps really "locked in" on that 1 kHz sinewave and got great THD metrics.

But music is not a 1 kHz sinewave - and those amps were terrible on music.

You can show me the THD metrics of those amps all day - but that is not going to
convince me that they were good amps. It just showed that the engineers designed
the product to the test; so they would measure well, and thus make good ad copy.

The Xd evidently was designed with the same philosophy in mind.

Then you need to be perfectly clear that when you say things like "easily outperforms" you really just mean measures better if you actually measured them or would theoretically measure better if you did not measure them. As you admitted, while you might eliminate a speaker on the basis of poor measurements, you would not buy a speaker based on measurements alone. Therefore all the pie in the sky about DEQX doesn't mean much unless you are actually comparing two speakers in a listening room. You may say one "performs better" but unless it sounds better, all the measurements in the world won't do you a bit of good. If you say that you are keeping your arguments about objective performance, then limit them to instances when actual measurements exist and are readily available. Otherwise, they are not objective arguments, they are merely theoretical arguments or opinion.