Saturday 19 September 2020

As We Hear It: Thoughts from an audiophile friend on assembling his sound system.


Every so often, I have the pleasure of having friends and readers add their thoughts to these pages. The other day my buddy whom I visited at the end of 2019 (a.k.a. "linnrd" online) sent me this discussion piece. Back in December when I visited, I invited him to take the opportunity to write about his equipment and how he came to this interesting collection of gear. Well, he delivered...

--------------------

A Complex Path to Simple Sound


Audio systems are generally an assemblage of products from different manufacturers. While significant R&D resources go into designing the various pieces that make up a system, it is left to the consumer to deal with the "R&D" of system design from "soup to nuts", as it were. The popular audio press is more oriented to singing the praises of individual components instead of educating and helping the reader actually assemble a decent combination of the aforementioned. Perhaps this a result of whom they view as customer and who as the product. A graveyard visit allows one to see much touted wonders such as Shakti Stones and the green CD marker (Ed: also see Harley's extensive investigation in Stereophile in 2004 amounting to nothing of course), to name just two, with the rotting corpse of MQA lying in wait for a plot (in both senses of the word!). 

Unfortunately, the self-proclaimed experts actively ignore existing knowledge and science and make it extremely difficult to fix this fundamental problem of the spread of misinformation by also subscribing to junk science. In other words, they are really messing up the S:N ratio of information transmission. As Bill Whitlock from Jensen says, reiterated in a recent discussion: "It often feels like an uphill battle to educate in a world of marketing deception and self-appointed experts."

System design may be viewed as a process that involves decision-making and judgement in order to achieve desired outcomes. I have had the luxury of being able to indulge in my proclivities of having several unique disparate systems, a clutch of interesting components that are historic (some of which are timeless design, whereas a couple are definitely dated). The setup that Archimago featured (see "Visiting an audiophile friend" in December 2019) is one of those systems, and this write up is a very belated account of why and how it was put together, whether it had achieved the goals that it was set out to fulfill, and some of my biases and peccadilloes so as to place things in perspective. I hope that it will be an interesting read and, possibly, educational for some. 


The why...

The speakers are a pair of Avantgarde UNOs with active woofers, with extremely high sensitivity and controlled dispersion that don’t require prodigious amounts of power to reach high sound levels, and are relatively immune to many vagaries of in-room interactions. The spherical horns provide the acoustic coupling giving rise to the high sensitivity and energy density at the listening position. The lower frequencies are further enhanced by a REL Britannia B1 subwoofer that I had sitting around. This combination maintains a high crest factor and power bandwidth which results in uniform output at a wider range of sound levels in the room.

The Nelson Pass designed First Watt SIT-2 amplifier is an interesting amplifier because it is a single-ended amplifier with the audio signal passing through a single device — a static induction transistor ("SIT") — without any feedback. It is the simplest imaginable amplification circuit; the input signal arrives directly at the gate, and leaves the drain in a direct path to the speaker. The transistor is biased by a constant current FET, and is therefore immune to variations in the electrical mains.

The extremely wide bandwidth of the SIT (-0.8dB @200kHz, and -3dB @500kHz!) along with its very rapid switching time of 250ns results in a near perfect reproduction of a 20kHz square wave with the absence of overshoot or ringing. The flat frequency and phase response, along with distortion harmonics that are consistent in amplitude and phase, and a constant damping factor further add to the appeal of the minimalism. The price one pays for all of the above is electrical efficiency (10%) and distortion (1% @2W and 4% @10W, both into 8Ω).

I find the theoretical concept, the practical application and the design aesthetic elegant, unique, and, (having on hand speakers operable within the amplifier’s design parameters and limits) irresistible. The ability to reach 109dB levels at 2.5% distortion from the amp (mostly 2nd order and entirely uniform in magnitude across the audio frequency band) makes for a very functional system. Before one gets alarmed by the 2.5% distortion in the amplifier, it is worth remembering that:
a) The grossest distortions in the audio production/reproduction chain arise at the transducers at either end of the chain, and

b) 109dB peaks in home audio reproduction are extremely short in duration even at enthusiastic listening levels — at least, they should be if one values one’s long-term hearing!
The input impedance is 10kΩ where the source sees the gate directly, which makes it particularly well suited to mate with the Manley Jumbo Shrimp — a pre-amp which has an output impedance of 50Ω, allowing it to play well with any power amplifier. For those who like to talk about different interconnects, the low output impedance pretty much makes the signal immune to any weird cables following it which, IMHO, is what all preamplifiers should strive for. The circuit is extremely simple with a short signal path, modest feedback, a wide bandwidth, and ease of drive. At the risk of repeating myself, once again the appeal is the simplicity and, in this case, the ability to play well with any of the power amplifiers I have kicking about. With two outputs, it can comfortably feed a pair of power amplifiers with input impedance as low as 10kΩ!  

The EAR 324 phono preamp is a very long-lived unchanged design from deParavicini who excels in designing transformer-based circuits. The EAR has a unique circuit topology in that it uses a transformer coupled gain stage, making it an electrically efficient circuit design. It is possible to replace the EAR with a digital solution, and carry out an accurate RIAA compensation in the digital domain, but the EAR’s minimal fuss plug-and-play, instantly accessible flexibility, and transparency to sources at hand provide a lot of inertia to overcome. Thoughts of substituting it with a more complicated, less flexible digital replacement have been just that — thoughts. As you may realize, I hold no allegiances to either analog or digital, and I view these domains merely as suitably useful tools to enable me to listen to my choice of music. I have no qualms listening to a vinyl rip of a recording for which there is no suitable digital version [insert obligatory loudness war gripe here]. 

The how...

The overall goal was to end up with a system that was unconstrained and consistent in dynamics from low to high listening levels with the components operating linearly and predictably, well within their design parameters. The system is located in a multipurpose room and had to be unobtrusive (for a given value of unobtrusive), not audibly overwhelming at any seated location, and play well with and within the room. 

The power requirements of the amplifier were well within its linear range since the SIT-2 was only audibly providing the frequencies at and above those reproduced by the crossoverless Avantgarde UNO midrange with its acoustical roll-off at the bottom of its bandwidth. There is always a flicker of perverse pleasure when recalling the unique fortune of listening to signals between 140Hz and 3.5kHz going directly from the preamp into the gate of the SIT and then directly to the transducer without the presence of anything else in the signal path. It is about as close as one can get to a straight wire with gain. The bass duties down to about 18Hz are handled by twin 10" drivers connected in parallel in a sealed enclosure driven by a 200W amplifier operating well within its linear range. In parallel, there is the REL subwoofer with its 12" driver (powered by a 500W Class A/B amplifier) acoustically coupled to successive chambers bringing to the party frequencies down to 12Hz (@-3dB) at suitable SPLs. 

Physical placement of the main speakers was somewhat constrained by the limits of acceptance of a 5’ contraption in the living space, but fortunately coincided with near optimal placement. ["A surprisingly accurate first approximation for speaker placement is to put them where they do the most visual damage to a room; that's probably where they'll sound their best" - Markus Sauer, from an otherwise forgettable article in Stereophile.] The location of the listening position, as dictated by domestic requirements, also fortuitously turned out to be most advantageous for a smooth response.

The speakers were toed-in such that they were crossing over slightly in front of the listening position. The added benefit to this is that the apparent position of a mono signal does not laterally shift when one is sitting in the adjacent seats; one is more on axis with the far speaker and off axis to the near. This is convenient when one is lounging on the couch with one's head on the armrest. (Ed: the "Time/Intensity Trading" that can particularly benefit narrow dispersion speakers.) Rake angle adjustment, as expected, allowed for dialing in the coherence of the sound from the midrange and tweeter horns, and following the instructions in the speaker manual yields acceptable results (for a given value of acceptable). NOTE: The use of DSP allows one to skip this step. The graphs of the system behaviour in the frequency domain illustrate this. 

Getting to the bottom of things...

When Archimago was visiting, while we were measuring at the listening position, I adjusted the settings of the B1 subwoofer to minimize any destructive interference between it and the woofers on the Avantgardes. The levels were set relatively high, which gave observers the initial impression that I am a bit of a bass-head. However, there is a method in my madness. A careful look at the response at the listening position reveals that the troughs of the dips are level with output from 300Hz upwards:


The idea is that, with ample power to spare for the woofers and subwoofers, I can afford to throw away all that LF gain when digital correction filters are applied to the system without losing any of the system dynamics from 300Hz on up. This is evident from the graph of the frequency response after DSP is implemented.

This ensures that the SIT-2 and the horns are operating within their linear range and the system has a high crest factor whether one is listening at 2:00am or at prodigious (almost unbearable) levels, and everything in between depending on one's desires. The more important benefit of DSP is seen in the time domain as is evident from the step-response curves (in the original article). It is possible to do A-B comparisons on the fly by using a playlist with pairs of the same track streamed by two different server programs, one with DSP and the other without. The results are a well-behaved system that is responsive within +/-3dB to signals down to ca. 12Hz and can relay the immediacy of a live performance at concert levels as well as the the intimacy of a solo performance. 

As for overall takeaway regarding the before and after using DSP, I find that the differences are subtle yet very noticeable. The easiest way to convey my impressions is by making use of a visual analogy: if one were to imagine having blurry vision and then rubbing one's eyes to seeing the image come into sharp focus, then in the audio realm the application of the DSP has a similar effect in the system. I am at a loss to describe the "sound" of the system as it currently stands, because I am of the opinion that it has none that I can discern. To me, it ranges from raw and dirty when listening to The Clash or Ramones to saccharine and soporific when a visitor insists on playing the prerequisite Diana Krall track before deciding to bestow their approval or not.

The DSP puts the finishing touch on enabling the room and this system to be as audibly unobtrusive as I have ever experienced it, and the less I hear it the more I enjoy listening to it. Because of the thought and effort that had gone into putting the setup together, the system was eminently listenable before and I, personally, could have happily lived with it, as indeed I did. After room correction, I am unwilling to go back to listening sans correction, and that should be an indication of my endorsement. 

Do measurements matter? 

The SIT-2 has, to quote Archimago, "not exactly the typical specs a pure objectivist might be impressed by," and "distortion at 1W is 0.7% and it can push 10W at 5% THD with a 1kHz signal into 4Ω." Do these measurements matter to me? Yes... and, no. Yes, in that they inform me as to what the linear envelope of the amplifier is. No, in that they are irrelevant to me once I am sure the amplifier is operating well within its linear envelope in accordance with its design parameters. I devour the specifications on every piece of equipment I have, be it electronic or mechanical.

Amps are designed electrically to drive certain loads; if the speaker doesn’t present that load, then we are demanding the amplifier deliver higher current (more power) and potentially (haha!) operate outside its design specification; the assembled system is then set up for possible failure. The advantage of using ultra-high efficiency speakers that present a resistive load that never dips below 8Ω is the freedom to use simple, low-power amplifiers and still achieve significant SPLs. Aristotle said long ago, "if it be disgraceful in men not to be able to use the goods of life, it is peculiarly disgraceful not to be able to use them in time of leisure" (Politics Book VII) and the SIT-2 is, indeed in my opinion, one of the "goods of life", in both senses of the phrase.

Objectivist, subjectivist, or both? 

When asked which camp I would place myself in, I cop out because I find those two categories to be too restrictive, individually as well as together. I see myself as being more of a pragmatist. I prefer to use the objective data to ascertain the limits of linear operation, and ensure that the assemblage is operating within that range in all foreseeable circumstances; after that, the numbers published in the popular press don’t matter to me any more because they are irrelevant to my purposes. If, on the other hand, I find the prescribed playing field too restrictive or unsuitable for my purpose, I simply use available objective data to allow me to locate a suitable playground for my toys in question. After using published measurements to assure myself that I will not bump up against any limits in my use cases, equipment is binned into the category of "as good as it gets."

I do, however, succumb to the subjective in more insidious ways. Forethought and elegance in design, along with adherence to established scientific principles, are attractive. Historical longevity of a piece of equipment increases its appeal. The tactility of a good control switch or knob gives me way more feelings than it ought to. These are but three of many of the qualia which have no bearing on the sound reproduction and bring nothing of utility to the table, but nonetheless give me pleasure and are my undoing [Ed: Hang on man! Control knobs are the literal definition of 'utility'! :-)]. And I can live with that. 

The other bits’n’bobs in the room are symptoms of the above foibles. The Omega Maxhemp speakers are unique albeit not without their flaws, which can be worked around. Implementing speaker correction via DSP on the way to an accurate crossoverless speaker is on my to-do list; no, the Maxhemp will not play loud, but will be more than adequate in a smaller room in the house. The Spendor SA1 and the descendant KEF LS50 are both historically interesting devices grounded in good experimental design and scientific principles; while there is significant improvement over the decades especially in the HF extension and step response, it is useful to remember that within the 24kHz bandwidth-limited constraints of digital media up to 16-bit/48kHz, they are comparable in the frequency domain and that any disadvantages that the SA1 has in the time domain disappear when DSP is applied. Neither of these speakers is expected to be optimally driven by the SIT-2, but that’s what the McIntosh MC275 is for, whose timeless design is reflected in the fact that used pieces fetch the same dollar as their original retail price, if not more. 

With multiple outputs on the Jumbo Shrimp pre-amp, the choice of system merely depends on which power amp is switched on. We are in an era where "accurate sound" is approachable at reasonable prices and accessible to anyone willing to devote the time to assemble a system operating within the design constraints of all component. It is also worth investing in the effort to understand how to minimize, if not eliminate, any deleterious externalities affecting the sound in the room. Hence, where some people see a couch between the speakers, I see a visually acceptable room treatment. 

The careful setup of a system is rewarded by the faithful reproduction of harmonics (timbre), the behaviour over short timeframes (dynamics), and the behaviour over a longer time period (the sense of audio space). All of the above are determined by the system, the room, and the interaction between them. Harman’s How to Listen is a useful program that aids in understanding and appreciating the rewards gleaned from the efforts and is highly recommended. Be on the lookout for a future article on a very listenable sub-$1000 system put together for a 15-year old music lover, that retains the essence of the system in this article.

I realize that the gestation period of this article has been 9 months, and there was certainly labour involved in putting this missive together — in fact, it is Labour Day as I write this. The thinking and rationale herein is hereby put up for adoption. I thank Archimago for the exchange of many ideas, for his time and efforts in providing the measurements and filters in current use, and for graciously providing a space on his corner of the Internet for my scribblings. I thank the readers for their forbearance.

---------------------------------

Nice man. As I said over E-mail, I love your artistry with the written word, historical references, and of course the thoughts put into this article! An important reminder and demonstration of what it means (IMO) to be a modern audiophile - one who is able to integrate the various domains of modern life; some science, some art, personal philosophies, personal idiosyncrasies and of course avoidance of the simplistic "objective" vs. "subjective" divide. As if things were ever so simple...

Notice that I titled this "As We Hear It", obviously with a side reference to Stereophile's "As We See It" monthly editorials. Subjective reviewers in audiophile magazines appear to almost insist that "sound quality" is the most important characteristic they're evaluating (witness the justification for many a cables for example). Sometimes it appears that there is denial that more $$$ actually went into what is seen rather than heard. Let's not pretend that differences in appearance have not overtaken actual sound quality for many an expensive device! I guess I've always found the "As We See It" title ironic yet highly accurate as reflective of why something might be adjudicated as "deserving" of big dollars especially in the "high end".

These days, with so much "good" hardware out there, it really IMO has become more interesting reading about the systems audiophiles own and their thoughts about the hobby as my friend eloquently and thoughtfully expressed above. Furthermore, it's clear that there's a ton of experience out there as well among audiophiles with fascinating stories and ideas around the hows and whys of the systems. In the past, I've had others like Allan Folz share about his system and recently 24bitbob talked about his ideas around artist appreciation and "core music". If you have stories you'd like to share, feel free to send me a note at the E-mail address above; things can get pretty slow with my response to E-mails when life is busy but I'll get to them in time! ;-)

This week, I finally had a listen to Gordon Lightfoot's Solo (2020, DR11). Mr. Lightfoot is 81 years old now and has produced an album of singer/songwriter material stripped down to (primarily) guitar and vocals. Above all else, for me the album is an authentic reminder of Lightfoot's artistry (and at times mastery!) of the folk/rock/country genre (have a listen to "Return Into Dust" for a taste). Respects to one of the greats of Canadian music.

I hope you're all having a good September so far and enjoying the music...

10 comments:

  1. "and the behaviour over a longer time period (the sense of audio space)."

    A concept that remains elusive to me. Every valid test of audio accuracy I've found requires an instantaneous comparison (A/B). Some audiophile reviewers insist that they need to spend long periods of time with a single system to discern differences. My experience is just the opposite. Unless the difference is obvious, I'm not able to hear audio differences when the comparison is separated by "a long period of time." And if more than one variable has changed over that "long period of time" (which when considering temperature, humidity, personal physiology, HRTF vectors, etc., there are always multiple variables), then it's a non-starter.

    Though I may be missing your point here.

    ReplyDelete
    Replies
    1. Poor phrasing on my part. I was attempting to convey the appropriate in-room decay (an optimal RT60) resulting in a sound field that didn’t exaggerate (too reverberant) or truncate (too dead) tailing harmonics, allowing for a better representation of the audio space imprinted on the source medium.

      Delete
  2. I get what you're saying about assessing the overall system and operating withing the limitations of the gear, and that's certainly a benefit of measurements, but I lean on measurements to let me know if I get what I'm paying for. In "high end" audio, many times increasing prices come with decreased performance compared to more competent gear at lower prices. Measurements let me know who is doing their engineering homework and who is focused on telling a good story.

    ReplyDelete
    Replies
    1. Yeah, I agree Volt,
      That concern about expensive "high end" performing worse compared to more economically-priced products runs a very real risk. Personally, I wouldn't nit-pick on slight performance differences. For example if a good looking, expensive CD player or DAC still performs as well as a reasonably priced device, I'd just call that "high-end luxury hi-fi"; at least it's still high fidelity.

      I'm more perturbed when very expensive gear gets rolled out in the subjective press hailed as "amazing sound", gets all kinds of accolades from subjective reviewers, yet the measurements show that it isn't even capable of basic fidelity. These circumstances really challenge the audiophile to decide just what "kind" of hobbyist he/she is. Does objective "high fidelity" based on an independent "ideal" standard matter in the hardware pursuit? Or is it all subjective and personal in which case the best sounding device could literally be anything because an objective standard just doesn't exist?

      Obviously for me, I do wish that if I spend good money on a product, the performance be objectively up to par technically as a prerequisite. And for certain products like cables, power conditioners, and tweaks, that evidence at least exists that they make a difference rather than just "snake oil".

      Delete
  3. It sounds like that along your journey you reached the goal your were striving for. A very good thing. for me my spaces are not the main house spaces as my wife of 50 years would have never tolerated those speakers regardless of color or sound.

    I have much the same frequency response as you with my old AR-58's in their spots 4 feet from side walls and 3 feet from the back walls. When I have listened to other speakers it as always been mostly pleasing, always different, and generally not better without talking crazy money speakers which I will never do.

    Always interesting reads. Thanks.

    ReplyDelete

  4. "And the award for Most Understanding Wife goes to..."

    :-)

    Your system looks very cool and I bet it would be a blast to hear. Congratulations.
    (And I share a lot of your thoughts on subjective/objective in audio).

    ReplyDelete
  5. Very impressive! (But I still see the couch! ) :)

    ReplyDelete
  6. The camp I am in gear-wise, and the route to it is probably as different from linnrd's as it can possibly be on mostly all parameters, but as Jim hints, whatever makes us happy is the right choice for each of us. Nobody buys audio gear to impress friends or neighbors, at least I hope not. The main goal we all share is listening to music we love, and enjoy the endorphines it releases in our brains as we listen.

    I have to admit though, that I was kind of shocked to read about the level of distortion from the Nelson Pass amp. But when I think about it, most setups would probably display similar levels of one kind of distortion or other, measurable or not. And it shouldn't matter a bit, as long as the owner is a happy camper.

    ReplyDelete
    Replies
    1. Not surprisingly, "THD" is one of the least effective "objective" metrics used to judge perceptive audio performance. There were some blind ABX tests done years ago that showed THD under 0.5% (IIRC) was imperceptible in audio program.

      Delete
  7. ‘the rotting corpse of MQA lying in wait for a plot’ this gave me a good laugh tonight. Whenever those three letters appear, this phrase ought to ring in the ears…

    ReplyDelete