Chapter 1 Part 2

A Tiny History of
High-Fidelity,
High-End, & Ultra-Fi

by Lynn Olson
lynno@enjoythemusic.com 

 

The Demise of the American Hi-Fi Industry

 In the latter part of the 1960’s, the High Fidelity industry continued to grow, but the pace of innovation slowed down after the conversion to stereo sound. More ominously, the transistor revolution proved the undoing of the domestic Hi-Fi industry. (This is why the Hi-Fi line in the Markets chart breaks off around 1968 ... sadly, the dreams of many Americans came unraveled in those bitter years, with the assassination of both Kennedys and Martin Luther King.)

 To return to the story of audio, the early transistor amplifiers used the notorious quasi-complementary output scheme, since matched complementary output transistors were not then available. In addition, the dangers of exceeding the Safe Operating Area were poorly understood, so the first generation of transistor amps weren’t even remotely “solid-state reliable.” Instead, they failed at the drop of a hat, and were so harsh with Class AB crossover distortion and Transient Intermodulation Distortion (TIM) that an entire generation of “East Coast Sound” speakers became duller and duller to compensate.

 The return rates eventually became so bad that Scott, Fisher, Sherwood, and many other well-known names were driven out of business, while the hallowed Marantz name was sold to the Tushinsky brothers who owned the US distribution rights for Sony. I've talked to folks who owned hi-fi shops during this time, and some early transistor amps and receivers had a failure rate of more than 50% when you first turned them on!

 This was the market opening for Pioneer, Kenwood (Trio in the rest of the world), and Sansui. The Japanese had extensive experience with transistors in cost-sensitive consumer products (the US experience was limited to the military and instrumentation sectors), and they made products that were fairly reliable, looked expensive, had good reviews in the mass-market magazines, and best of all, had generous profit margins for the retailer!

 The Japanese, unlike the American manufacturers, were quite willing to absorb financial losses for 5 to 10 years in order to gain a controlling market share. For the neighborhood Hi-Fi salon, it was, as they say in Sicily, “an offer you cannot refuse.” 

During this grim period, slow progress was made in the development of output transistors, allowing the use of a matched PNP/NPN complementary (or push-pull) output stage with direct-coupling. This removed the worst of the bias reliability problems of the quasi-complementary output stage, but problems with high-frequency instability and thermal runaway still plagued transistor amplifiers, keeping the long-stated goal of “solid-state reliability” as empty as the contemporary digital  promise of “perfect sound forever.”

 Things weren't any better in the movie theatres. By now, widespread color television ownership had cut deeply into the movie-going audience, and most theatre owners were not interested in maintaining costly 70mm projection systems with large curved screens. The theatres had been divested from the movie studios by court order, and the new independent owners wanted to maximize profits as quickly as possible. That meant bigger popcorn concessions, faster audience turnaround (no more double features where you stay all day if you wanted), smaller screens, a much smaller (and non-unionized) theatre staff, and a technological regression to 35mm film and mono optical sound. It took more than a decade for high-fidelity stereo sound to return to the new, smaller theatres, thanks to the runaway success of the Star Wars movies. (Without Star Wars, it’s quite likely the theatre owners would have ignored Dolby Labs and stuck to good ol’ mono.)

 The tape-recorder market was in a state of flux. The introduction of small-signal transistors was a more positive development than in power amplifiers, since high power is not an issue in tape-decks, and the circuit complexity is high ... a situation tailor-made for transistors. Japanese tape recorders, some of high quality (TEAC and Sony), and some not, steadily drove the US-made machines like Ampex, Magnecord, and Viking off the consumer market.

 In an unlikely alliance, Nakamichi, Dolby Labs, and Advent co-operated in transforming the Philips Compact Cassette, originally intended as a low-fi portable dictation format, into a almost-hi-fi tape format. In less than 4 years, the new cassette format drove the unreliable 4 and 8-track endless loop systems off the market. The non-technical public had an easy-to-use, hard-to-break stereo tape recorder for the first time. For better or worse, the Compact Cassette is now the dominant medium for world sales of pre-recorded music, being dirt-cheap, good-enough for the mass market, and easy to pirate. (The latter a major advantage for nations that are not signatories to international copyright law.)

 

The Rise from the Ashes 

As the 1960’s fizzled out into the spreading gray twilight of Asian mass-fi, a tiny speck of light emerged from an unlikely corner of the industry: J. Gordon Holt's typewritten zine, the Stereophile. This was a very different magazine than the current slick 4-color product. Every pint-sized issue had a grainy black-and-white picture on the cover, and J. Gordon's zany sense of humor was evident in the writing, the goofy cartoons, and the funky photos.

 This was just about the only place where you’d see serious discussion of the sonic merits of the JansZen 130 electrostatic tweeter, the Paoli Mark III, the KLH Nines, the Fulton FMI-80, and a brand-new vacuum-tube preamp from a small company in Minneapolis called Audio Research. This was something else ... a different set of values was in place here. If you looked hard, you might even find a funky little hole-in-the-wall dealer that actually carried this stuff. In the formative years of high-end, “image,” “style,” and “fashion” counted for nothing; the designer-jean, style-over-substance, marketing-uber-alles zeitgeist was still 5 to 10 years in the future.

 J. Gordon Holt played a pivotal role in unearthing the interesting things when the American Hi-Fi business had to all appearances entered a state of terminal collapse. Sure enough, here and there, in little workshops all around the land, folks were building strange new products ... vacuum-tube electronics, electrostatic speakers, making direct-to-disc recordings, all kinds of offbeat items. They made their way to the door of the early Stereophile, and the market kept on growing. Slightly less funky dealers started picking them up. More companies appeared. In a few years, a new industry, risen from the ashes of the old, had a name: the High End.

 I walked into this story in 1970 by subscribing to Ed Dell’s The Audio Amateur and Gordon Holt’s Stereophile during my sophomore year at Claremont Men’s College. After I left college, my introduction to the real world of audio was the not-so-gentle art of high-pressuring customers at Pacific Stereo. They’d get ten ... fifteen ... sometimes even twenty feet into the store before a gang of commission-hungry salesmen would jump on them. I was one of the those salesmen, and what kept me motivated was a large bar chart posted in the stockroom that compared the week-by-week commissions of each salesman. If you were the lowest-commission salesman for four weeks in a row, that would be your last paycheck at Pacific Stereo. The average salesman lasted 3 months. (Still the industry average 30 years later.)

 I knew I couldn’t keep this up forever. It was time to do something ... anything! ... to get out of retail sales. At the time, I was fascinated with quad sound, and spent much of my modest paycheck on assorted decoder boxes from Sony (SQ), Sansui (QS), and JVC (CD-4). After living with these gadgets for a few months, I could hear them manipulate and steer the signal around the room. After a while, you just got annoyed and went back to listening in stereo. I started to understand why my customers weren’t too impressed with quad sound.

 During a rare 3-day weekend, I fell into a deep sleep, and images of a point of light traveling around a transparent sphere came to mind as I awoke. I was familiar with the 3-D Scheiber notation for quadraphonic encoding, and it dawned on me I had been given the solution for a better way of decoding the complex CBS SQ matrix (the market leader at the time). I quickly wrote down the dream sequence of images using Scheiber notation. A few days later, I took a good hard look at what I’d written down and realized that it would really work. Not only that, but it would work very smoothly, be elegant, and just possibly, might even be patentable. This was my ticket out of Los Angeles!

 I spent the next several weeks of my spare time in the Patent Room of the Los Angeles Public Library. To my surprise, nobody else had thought of it. (In 1973, the quadraphonic subclass was easy to research, with fewer than 10-15 inventors filing.) I filed a Document of Disclosure with the Department of Trademarks and Patents, and sent copies of the proposed invention to Advent and Audionics. Well, Advent never sent me a reply, but Audionics was interested.

 To make a long story short, I left Pacific Stereo and Los Angeles behind, moved up to Portland, Oregon, and built the Shadow Vector prototype for Audionics. After we got the prototype up and running, it was time to go on the road, first to New York, then on to the BBC, EMI, KEF, and our agent in Milan. No nibbles, but it was a lot of fun, and Shadow Vector was one of the best decoders of the day. It was the only one designed to keep a smooth reverberent field as it gently enhanced the directivity of the dominant sound. Seeing the raised eyebrows at EMI as our Shadow Vector outdid the most advanced CBS Labs prototype made it all worthwhile.

 After we returned to Portland, we witnessed the CBS SQ system fight for its life against JVC and MITI, with Sansui QS system gradually squeezed out. The buying public was getting burned out on the whole oversold quadraphonic thing ... while the truly advanced decoders, the ones that should have been on the market right from the beginning, languished in the laboratories. Strange to think the very technology (dynamic matrix-steering) ignored by the major companies back then now forms the basis of the Dolby Surround system.

 When I received the Shadow Vector patent 3 years later, quad was history. I was drafted into doing loudspeaker design, partly as a result of all the tips I’d gotten when we visited the BBC and KEF. In a company as small as Audionics, there were two real EE engineers and one speaker designer (me). Over time, I ended up designing a full line of speakers for Audionics. Some were fairly good, some weren’t. I was learning as I went along, and our parts suppliers tended to be the vendor-of-the-month that hadn’t yet discovered Audionics’ credit record. Once they found out, it was time for us to go looking again ... that’s why many audiophile vendors use different parts from month to month. There’s a good reason for that!

 As a young speaker designer I had a ringside seat watching and participating in the profound changes in speaker design techniques in the middle Seventies. The most significant was the re-discovery of Neville Theile’s landmark paper first published in Australia in 1963 (and promptly forgotten). Theile, a lead engineer for Australia’s color TV project, had gone on to fully analyze both closed and vented box loudspeakers as 2nd and 4th-order high-pass filters.

No more cut-n-try approximations, no hypercomplex theoretical math filled with mistaken assumptions, just straightforward Butterworth and Chebychev filter design ... and none of this outdated “M-derived filter” stuff, either. Theile’s paper also provided precise methods of measuring fundamental properties of the driver such as Fs (resonance frequency), Qt (damping), and Vas (compliance). With a scope, an oscillator, a voltmeter, a frequency counter, a test box, and a hand calculator, you could accurately design a closed or vented box system and get results within a fraction of a dB of the prediction ... a genuine breakthrough in low-frequency design that removed all of the cut-n-try guesswork of previous decades.

 What’s a little sad is that Theile’s paper was ignored for nearly a decade simply because it was published in a little-known Australian journal. It took Robert Ashley of the Audio Engineering Society to pick up Theile’s work and also that of Richard Small, who published a very comprehensive summation and extension of Theile’s work in his doctoral thesis (as well as the modern near-field method of measuring loudspeakers). All of this material appeared in the AES Journal in 1973, and it took the speaker-designing world by storm. Within a matter of months, Theile-Small became the accepted method of designing, prototyping, and measuring closed-box, vented-box, and passive-radiator speakers all over the world.

 Small simplified the system so powerfully that all it took was one of the new scientific calculators (the slide rule was beginning to fade away) and a set of nomograms to design accurate bass response. After personal computers were introduced in the early 1980’s, the T/S equations became an integral part of commercially available software for designing loudspeakers. 

Over at KEF in England, Laurie Fincham was extending the analytic techniques pioneered by Theile and Small to the more difficult problem of crossover design. Using the best available HP minicomputer of the day, he was able to acoustically measure the driver using FFT techniques, measure its impedance characteristics using the new T/S techniques, set up a prototype “target function” for the desired crossover filter, and let the computer optimize all of the possible values of crossover elements. In effect, the computer goes through thousands of potential crossovers and picks the closest approximation to the desired response curve. 

Although nearly everyone adopted Theile-Small techniques for bass design (except the transmission-line holdouts), it took ten years longer for computer-based crossover optimization to be widely adopted, due to the very high cost of minicomputers, and the even higher cost of hiring experienced FORTRAN programmers to run the things. KEF, Celestion, and Bowers & Wilkins were pretty much alone in using this technique until the advent of low-cost, powerful PC’s with off-the-shelf speaker-design software. Today, the designer clicks the “optimize crossover” function after measuring the driver, choosing the desired crossover topology, and selecting a set of starting values. After the PC models the crossover and shows you the results, you can build the physical crossover, measure the speaker system, and sure enough, it’ll be within a small fraction of a dB of the software model.

 The systems-modeling approach perfected in the early 1970’s extended to driver design, an even more intractable realm. The BBC was seeking a cone material that would provide exact pair-matching as well as permitting the design of a highly consistent and repeatable monitor speaker. Bextrene, an acetate plastic derived from wood products, first saw use in the KEF B110 driver, which had a starring role in the legendary LS3/5a compact monitor.

 (While I was at Audionics, we applied for permission to officially license the LS3/5a design from the BBC. After 10 long months of British silence, the BBC sent us a very gracious letter that boiled down to “no way.” If Audionics had been as clever as our competitors, we would have copied the LS3/5a, and then pretended we designed it all by ourselves.)

 In the late 1970’s, the BBC perfected polypropylene cones, which had the significant advantage of not requiring a treatment with doping material, as well as higher efficiency and much flatter response. By a process I still don’t understand today, the BBC patents were circumvented in less than 3 years, and everybody and their brother started making polypropylene-cone drivers. Even mass-fi rack stereos use polypropylene drivers these days, which tells me that they must be even cheaper than paper to make. However, the BBC was very much on the mark in not using poly drivers any larger than 8 inches; the latest BBC monitor (the successor to the LS3/5a) uses a Dynaudio 5.5" driver with a poly cone, which I feel is the just about the right size for getting the best sound from polypropylene.

 Moving on to electronics, the power amps of the late Sixties and early Seventies blew up a lot and sounded pretty nasty. The engineers of the early 1970’s were still wrestling with problems like maintaining adequate phase margin with real loudspeaker loads, Nyquist feedback stability criteria, Safe Operating Area for the driver and output stage, and little things like that. Audionics’ first amp, the PZ-3, fit right into this picture: loads of feedback, and very low THD distortion measurements. (0.03%, get it?) It measured just fine, but it wasn’t too stable in the real world, with an alarming fondness for shorting out driver transistors, smoking bias resistors, and shooting flames out of the cooling vents (in anticipation of the much larger solid-state melt-down at Three Mile Island).

 I remember many days when more of these dogs came back for repair than we shipped out. Some of the amps had circuit boards scorched beyond recognition, and top plates discolored by lines of dark-gray soot. We’d replace the circuit board, repaint the top cover, and ship ‘em right back out again. Needless to say, the PZ-3 was not a big money-maker for Audionics. The only consolation was knowing that all the rest of the high-powered transistor amps were just as bad.

 In the mid-Seventies, along came Matti Otala and the discovery of TIM (slewing) distortion. Our Number One engineer (the conservative old-timer who designed the PZ-3) was utterly horrified by Otala's first AES paper and said it was unscientific bunk or worse. Our young Number Two engineer took Matti seriously, let “traditional values” go by the board, and tried different approach.

 Bob Sickler let the distortion rise up to the 0.1% level, by making very large decreases in feedback (feedback dropped from 40-50 dB to 20 dB) and using the most linear complementary-symmetry topology possible. The slew rate and power bandwidth improved by a factor of 10 to 50 times. Best of all, we couldn’t toast it, even with my speaker simulator load hooked up.

 In 1976, Audionics introduced the CC-2, which was probably one of the first low-TIM amps in the US. Sure enough, it sounded much better than the PZ-3, and the failure rate in the field was well under 1%. The reason for both was probably the >200kHz power bandwidth and an excess phase margin of 60 degrees, both quite unusual at the time. Although I rarely listen to my CC-2 these days, it’s still not a bad transistor amp; by now, though, nearly all transistor amps use the same design principles as the CC-2. Matti’s paper had such a profound impact on the solid-state design community that nearly all high-end engineers got on board ... besides, it’s hard to argue with better reliability, which a high slew rate and adequate phase margin certainly provide.

 As the high end market firmed up in the mid-Seventies, Harry Pearson’s “The Abso!ute Sound” magazine made its first appearance. HP’s approach was more subjective than Holt’s, and he was attracted to more offbeat products than Holt. I wasn’t a big reader or follower of HP, since he was attracted to things that left me cold, like the Dahlquist DQ-10, the really big Infinity panels, and the ever-changing galaxy of Magneplanar and Audio Research products. Still, despite my personal feelings, I must credit HP with playing an absolutely crucial role in blowing the whistle on the truly appalling sound of the first CD’s, and in kindling the flames of the tube revival. Also, HP provided a continuity that could have been lost when J. Gordon Holt sold Stereophile to its new owners. (The modern Stereophile bears no resemblance in style or content to the sassy and contrarian earlier magazine.)

 I still remember the massive PR blitz (similar to the Windows95 onslaught) pushing the first CD’s and their players. I didn’t take the claim of “perfect sound forever” too seriously, but I really did expect that digital sound would be a significant advance for the entire industry. After all, LP’s have serious problems with end-of-side distortion, noise buildup, guessing the correct VTA adjustment, etc., and tape has its own set of troubles with scrape flutter, IM distortion, setting bias and EQ for the exact tape formulation, Dolby mistracking, etc. etc. Digital sidesteps all these problems, and has noise and distortion approaching that of an op-amp ... hundreds of times better than any LP or tape medium. In principle at least, it should sound as transparent as a good amplifier. Little did I know.

 As it turned out, the designer of the CC-2, Bob Sickler, went out and bought one of the very first players, the Sony CDP-101. His roommate was a professional musician for the Portland Symphony, so when they both sang the praises of the new medium, I was expecting my first experience with digital-in-the-home to be something like hearing a well-done mastertape, or even a live mike-feed straight from the console. After all, both them had good taste in music, and I knew that the digital process was almost distortionless compared to any other medium. It had to be good, right?

 They put on a all-digital DGG classical disc and all I heard from the TAD studio monitors was screech, screech, screech. The massed violins were far worse than any Columbia LP I’d ever heard ... they really did sound like bandsaws, and I heard outright tearing and ripping sounds in the loud passages. The quiet passages were dead silent, as I expected, but any sensation of space, of stereo depth, and of acoustic presence was totally absent. The reverberation sounded as flat as a paper moon and just as fake. I was appalled ... this was the worst sound I had heard in many years, yet I knew the rest of their hi-fi system was quite respectable.

 My friends were grinning the whole time and saying, “Isn’t it so clear! There’s no noise and clicks at all!” I was silently thinking “Is this the future of audio!?” So I simultaneously experienced both bitter disappointment and astonishment that my friends having a utterly different sonic experience than I was having.

 That whole experience really opened my eyes. I realized that people really do hear things in quite different ways ... my friends thought the Sony CDP-101 and the shiny little aluminm coasters were just terrific, and I thought they sounded awful! After that, I started taking the reviews in any magazine a lot less seriously. After all, how was I to know the reviewer was hearing things in the same way that I did?

 I left Audionics in 1979, realizing there was no upward mobility in a little company that was going to stay little. By then, the ego strokes of being introduced as a “world-famous designer” at the CES had worn thin. That's when I decided to take a break from making a living from audio and simply work for Tektronix and be done with it. As it was, my timing was good; the high-end audio business fell like a stone in the early 1980’s, thanks to competition from early adopters of VCR's and computers, coupled with a dry spell of creativity in the industry. Even Audio Research gave up on tubes for reasons I have never quite understood (rising costs and pressure from increasingly conservative dealers?)

 During my 9 years at Tektronix, I peeked in at what was stirring in the audio world from time to time. I remember a brief flurry of interest created by the Quad 405 quasi-feedforward amp, with different exotic mixed feedback schemes appearing in every issue of the Audio Engineering Society Journal. For a while there, it looked like the old-time high-feedback engineers and the new boys on the block could have it all: zero distortion, very wide power bandwidth, ultra-high slew rates, a simple output stage that didn't require any bias tweaks, and the complete elimination of crossover distortion!

 Unfortunately, once you hooked these miracle amps up to real loudspeakers, the balance equations evaporated. (Isn’t that just like speakers ... they’re always screwing up the latest “wonder amp.”) As a result, none of this intense theoretical activity resulted in any lasting sonic breakthroughs, except to once again point the finger at loudspeakers.

 Amplifier designers keep expecting loudspeaker designers to make speakers that are resistive loads, and it never happens. The very best drivers are quite reactive, including electrostats, top-rank horns, and advanced direct-radiators. Speaker designers quickly find out that yes, the load curve can be smoothed out a little, but making the whole thing look like a big resistor is counterproductive in terms of sheer complexity and cost. Like it or not, it is the responsibility of the amplifier designer to make products that work successfully with the loudspeakers we have right now.

 

Click here for Chapter 1 Part 3

 

 

Copyright© 1996 by Lynn Olson