This point needs to developed further. There were many times in the history of audio where a new technology sounded wonderful to the first group of listeners, then a few years later more and more people start noticing unpleasant artifacts, and finally discovered what caused the artifacts.
A quick history: the Williamson amplifier in 1947 swept away other circuits for nearly ten years, then stability problems were gradually discovered and resolved with better, more linear, and more stable vacuum-tube circuits.
The first transistor amplifiers of the late Sixties were dreadful circuits with quasi-complementary output stages with inherently high levels of Class AB switching artifacts, poor feedback stability and overall reliability, and serious problems with slew-rate distortion. It took more than ten years for better, more stable, and higher-speed circuits to appear.
The first pro-level digital tape recorder from Sony, the 1630, used the appallingly bad 741 opamps (which are completely unsuited for any kind of audio), with no dither, no jitter reduction, and ran at 44.1/16 PCM. From a modern perspective, the worst digital imaginable. Yet it was the industry standard for submitting a master to the pressing plant for more than five years. (Even the sainted J. Gordon Holt gave the Crown DC300 and Phase Linear 700 amplifiers a top-rank Class A rating, alongside the Marantz Model 9. I subscribed to the original "Stereophile" magazine back in the early Seventies, and remember reading the reviews.)
I should note that all of these technologies were hailed as the "best sound ever" at the time of introduction, with a few naysayers holding back ... and were eventually proven right when measuring technology discovered the problems. It took several years for digital engineers to admit that dither was a necessary part of digital audio, or that jitter reduction was worthwhile. It took more than ten years for transistor-amp designers to admit that slew rates of more than a few volts/microsecond might be a good idea. It took more than five years for the wretched, high-distortion quasi-complementary output circuit to be replaced by full-complementary.
I think sigma-delta converters and the latest Class D amplifiers fall in the same category. It's a new sound, and a lot of opinion-makers and reviewers like it. But I suspect it's the same thing all over again; the troubles of analog are once again shuffled around into a new place, where they won't be discovered right away.
Like Thorsten, I feel the real drivers behind the widespread adoption of sigma-delta converters and Class D amplification is the holy trinity of lower cost, better specs, and ease-of-application. For most audio engineers, those three parameters are the most important aspects of a new design.
Last edited:
The detailed response is appreciated and you addressed one of my concerns about most horn systems (I have heard)- the direct "wall of sound" effect.
I suspect that with class d, board layout and component selection play much bigger roles than with more traditional amplifiers. This leads to a wide divergence in their SQ. Whatever the reason, Lynn is right that eventually there flaws will be laid out naked before the world. Off course, there is also the "this is new, therefore it is bad" and vice-versa school of reviewing. Regards
I'd like to point the readers of diyAudio to an article by J. Oppenheim and M. Magnasco in Physical Review Letters, Human hearing beats the Fourier uncertainty principle.
The more experienced listeners beat the Fourier uncertainty principle by a factor of 13(!), which falls well outside the predicted capabilities of the standard hearing models. I strongly recommend this article, which has sound samples so you can hear for yourself what the authors are discussing. The modified samples of the "Casablanca" soundtrack are interesting: destroying the time information makes them completely unintelligible, while destroying spectral information retains intelligibility surprisingly well.
This research undermines the standard models of hearing that have been the basis of digital and lossy-digital designs for several decades, as well as key assumptions about loudspeaker design. I can certainly see implications for noise-shaping algorithms used in sigma-delta and DSD converters, the basic assumptions used in the design of lossy-compression algorithms like MP3, AAC, Dolby Digital, etc., and loudspeaker design, in particular, the effect of horn and front-panel-edge diffraction, and internal reflections inside loudspeaker cabinets.
Remember the comments I made a few pages ago about 1/4" changes in the front-to-back location of the horn being surprisingly audible? Well, here you are.
The more experienced listeners beat the Fourier uncertainty principle by a factor of 13(!), which falls well outside the predicted capabilities of the standard hearing models. I strongly recommend this article, which has sound samples so you can hear for yourself what the authors are discussing. The modified samples of the "Casablanca" soundtrack are interesting: destroying the time information makes them completely unintelligible, while destroying spectral information retains intelligibility surprisingly well.
This research undermines the standard models of hearing that have been the basis of digital and lossy-digital designs for several decades, as well as key assumptions about loudspeaker design. I can certainly see implications for noise-shaping algorithms used in sigma-delta and DSD converters, the basic assumptions used in the design of lossy-compression algorithms like MP3, AAC, Dolby Digital, etc., and loudspeaker design, in particular, the effect of horn and front-panel-edge diffraction, and internal reflections inside loudspeaker cabinets.
Remember the comments I made a few pages ago about 1/4" changes in the front-to-back location of the horn being surprisingly audible? Well, here you are.
Last edited:
Lynn,
Though I don't always agree with some of your design assumption this is one that I have also come to the same conclusions long ago. Others seem to discount the phase shift over frequency as being audible and wonder why an accurate sound reproduction can not be achieved. I believe that phase is just as important as frequency response being ruler flat. It amazes me that others discount phase shift as being audible if it is a constant change over the full frequency range. Group delay is just ignored as unimportant to the reproduction system. I do not agree that it is only an issue at the crossover point, I think this has much to do with the loss of accuracy of a reproduced sound field. There are many aspects of design that are looked at this way and that is something that I am actively working to overcome.
Though I don't always agree with some of your design assumption this is one that I have also come to the same conclusions long ago. Others seem to discount the phase shift over frequency as being audible and wonder why an accurate sound reproduction can not be achieved. I believe that phase is just as important as frequency response being ruler flat. It amazes me that others discount phase shift as being audible if it is a constant change over the full frequency range. Group delay is just ignored as unimportant to the reproduction system. I do not agree that it is only an issue at the crossover point, I think this has much to do with the loss of accuracy of a reproduced sound field. There are many aspects of design that are looked at this way and that is something that I am actively working to overcome.
Listening to FR drivers introduced this to me. IF you have any amount of time on them and them switch to traditional multi ways, things become very apparent, in some cases.
To answer your first question - the Ariel is described in detail by Lynn in the Ariel page at Nutshell HiFi.
To answer your second question you need to know how big your room is and how loud you music choices!
It was designed with low power amplifiers in mind and is more than happy at "normal" levels with under 10W. Lynn himself states "The perennial question ... which amplifier? For most readers, it's simple; just get a classic old EL84/6BQ5 push-pull amplifier, and there you go. There are plenty of old Fifties-vintage Eico's, Dyna's, and Scott 20-watt amplifiers still kicking around, and restoration can be done any competent guitar-amp tech. ". Click through for more
Last edited:
The midbass driver for the Ariel has gone out of production. Actually, it's gone in and out of production several times, but Madisound assures me that it really, truly, is out of production now. Since the Ariel was designed in 1993, it's probably not surprising the drivers have gone away after some twenty years.
Back when I designed it, I was hoping there could be a Super Ariel that was similar, but with a bit more efficient drivers, maybe even a version with a 6.5" or 7" midbass. That never happened. Instead, Vifa, Scan-Speak and the others went in the direction of less efficiency and rougher response - not exactly what I expected. My guess is the designer of the Vifa P13 driver and the Scan-Speak D9000 left the company and started Skaaning on his own, along with his son.
If I were to do a revised version of the Ariel with modern drivers, I'd probably look at the Skaaning drivers first, even though they are pretty expensive these days. I would not use Chinese-made drivers that are copies of Scandinavian drivers.
If anyone is interesting in doing their own Super Ariel, you need to find drivers that have extremely smooth in-band and out-of-band response curves. At the time the Ariel was designed, the Vifa midbass and Scan-Speak dome tweeter were the smoothest in the world, which made the suitable for the project. Smooth response in the 2~8 kHz region was extremely important for the midbass driver: most modern midbass drivers, particularly those with rigid Kevlar, carbon-fiber, or other exotic cones have severe breakups in the 4~8 kHz region, requiring notch-filtering or high-order rolloffs in the crossover, which defeats the simplicity of the Ariel project.
The other characteristics that make the Ariel sound like it does are the asymmetric large radii on the left and right cabinet edges, Baltic Birch for the interior construction, the transmission-line vent on the floor, and external crossover.
The key design feature of the crossover is precise phase-tracking of the midbass and tweeter, which is better than 5 degrees, as measured by temporarily connecting the drivers out-of-phase and inspecting the depth of the resulting null (25 to 30 dB).
Last edited:
The LTO is just about finished; I've given it a fair amount of subjective fine-tuning, and the crossover, surprisingly, is close to the "book" values for the woofer and AH425/Radian745Neo. That's mostly because the autoformer/transformer used for attenuating the horn by 12 dB isolates the impedance variations from the crossover, the GPA 416-16 Alnico woofer uses Zobel compensation to correct for VC inductance, and both drivers are very flat in their passbands.
However ... it is a much more expensive loudspeaker than the Ariel, and a lot bigger too. Pretty much the inevitable consequence of wanting more efficiency and more headroom.
Re: the Revelator line. Hmm. The published curves of all the 5.5" Revelator drivers are much rougher than the measurements I've made with MLSSA on the Vifa 5.5", and the efficiency is 2 dB lower, as well. The Revelators I've heard in commercial speakers did not seem acceptable to me; rough response, grainy sound, and I fail to see how a slit cone improves performance.
I hate to say this, but I think Scan-Speak has lost its way. For decades, they made drivers that were the smoothest in the industry, and were very musical. I don't understand what Scan-Speak is trying to do now: they're no longer very flat, don't have outstanding impulse response, are low efficiency compared to other vendors, and are expensive. Why?
Frankly, I was very disappointed when Scan-Speak came out with the Revelator line; the illusion of "more detail" that reviewers mention is actually an artifact of HF breakups from the cone, and is measurable in the time domain. For some reason, reviewers that use high-power Class AB amplifiers consistently mis-identify driver breakups as "more detailed", when really it's just more grainy and edgy-sounding.
I'm sorry if this sounds arrogant: it's really not meant that way. I design loudspeakers and amplifiers to satisfy my own tastes ... and that taste has now diverged pretty far from the mainstream high-end industry.
However ... it is a much more expensive loudspeaker than the Ariel, and a lot bigger too. Pretty much the inevitable consequence of wanting more efficiency and more headroom.
Re: the Revelator line. Hmm. The published curves of all the 5.5" Revelator drivers are much rougher than the measurements I've made with MLSSA on the Vifa 5.5", and the efficiency is 2 dB lower, as well. The Revelators I've heard in commercial speakers did not seem acceptable to me; rough response, grainy sound, and I fail to see how a slit cone improves performance.
I hate to say this, but I think Scan-Speak has lost its way. For decades, they made drivers that were the smoothest in the industry, and were very musical. I don't understand what Scan-Speak is trying to do now: they're no longer very flat, don't have outstanding impulse response, are low efficiency compared to other vendors, and are expensive. Why?
Frankly, I was very disappointed when Scan-Speak came out with the Revelator line; the illusion of "more detail" that reviewers mention is actually an artifact of HF breakups from the cone, and is measurable in the time domain. For some reason, reviewers that use high-power Class AB amplifiers consistently mis-identify driver breakups as "more detailed", when really it's just more grainy and edgy-sounding.
I'm sorry if this sounds arrogant: it's really not meant that way. I design loudspeakers and amplifiers to satisfy my own tastes ... and that taste has now diverged pretty far from the mainstream high-end industry.
Last edited:
Lynn, I dont think you sound arrogant, I completely follow you. The developement of the Scan Speak drivers are much like where most drivers have gone, to make much bas and high soundpressure in small boxes. An it seems lowest distortin in the pass band has become more important than the out of band response curves.
I think you entered the age which is beyond the target of mainstream high-end industry. I don't know to whom the hell they design their speakers now. Must be the hip-hop generation and old pricks who will fight to last breath(until family drags them by force to nursing home) to stay current , take whatever pill and drink the blood of virgins if necessary 
And even if one finds decent speaker which doesn't make one run for the shelter the price makes you chuckle ..
Anyway, I look forward to see final incarnation of your speaker . With your design experience I'd love to see something ..ekh more evolved in a shape similar to Living Voice Vox Olympian but I understand all the limiting factors . Rgrds, L
And even if one finds decent speaker which doesn't make one run for the shelter the price makes you chuckle ..Anyway, I look forward to see final incarnation of your speaker . With your design experience I'd love to see something ..ekh more evolved in a shape similar to Living Voice Vox Olympian but I understand all the limiting factors . Rgrds, L
Last edited: