Reference

Postscript August 2020

As a band plays a lazy jazz-blues in the corner, and everyone has gotten refreshed with eine kaltes Bier, the professor's central-European audience are warned that a small spectral leakage across multi-channel ICs is developed into a far greater and wider kind of audioland X-contamination, as outlined below.

Another threat is the Ring of 3, an op-amp "state variable" topology used in posh equalizers and crossovers. As with the innards of NFB amplifier stages, while low distortion on the outside lures use, voltage waveforms inside the ring aren't pretty. They're triangular, so nasty harmonics predominate, so any leakage is a no-no. Yet as the R-o-3 uses four op-amps, engineers and accountants will demand quad(ruple) ICs, ensuring that the R-o-3 perverts music. One can see why live-show engineers who use 1/3rd octave graphic equalizers, have preferred the sonics of instruments (like EMO) that employ inductors.

As for ICs, Texas Instruments, the world's longest-lived transistor maker, is also one of the analog IC makers left in the US. Great companies with initials BB, LTC & AD have been subsumed. Decades back now, Walt Jung came up with splitting op-amp stages, separating the output part. This cuts out the thermal distortion and harmonic contamination, but doubles a load of costs.

Symmetry—A Brutalist Architecture
It's routine for audio since the 1920s, thro' to 2020, to employ circuits that are variously termed bridged, balanced, differential, push-pull, and symmetrical.

The signal is split: It flows through opposing halves, that are presumed to be exact complements. This arises all over, from vinyl disc front-ends, thro' high-power amplifiers' driver and output stages. These arrangements can also be nested, most often in output stages. One symmetrically driven output is pitted against another, speaker in the middle. The second drive is inverted, so voltage and current swings are doubled, and power delivery is potentially quadrupled. Oft used where serious oomph is sought for bass and sub. With analog amplifiers, such bridging, where two channels become one, is often assignable. Across 30 years, sonic opinions of bridging with otherwise fine stereo amplifiers, have been lackluster.

Power-heads can even bridge a bridged amplifier—an Amcron specialty. (Amcron used to be called Crown and Crown amplifiers were last deemed "high-end" in the mid 1970s, and this source of industrial muscle dissolved some years ago. "Go figure.")

There are also balanced input stages. Sat in just part of the path, the noise/EMI rejection may be more important. Then balanced outputs, that aren't needed in homes (footnote 1) and when active, are miniature bridged amplifiers. In a common, low and nasty circuit that's became standard, positive feedback (zut Alors!) is even added to risk sub-stable operation, resonances and abnormal harmonics. Marlowe frowns, and slips the catch on his Colt.

Meantime, symmetrical discrete circuitry was popularized in the USA in the early 1970s (footnote 2) and a decade later promoted to drive Lateral MOSFETs by Erno Borbely. It's neat on paper. The same sort of mirror-image topology/architecture still appears across US products, and unimaginative copies from the orient.

A study of national circuits is informative. In other places, symmetrical circuits are rarely met. They appear profligate to engineers, and eyebrow-raising to French audio factory accountants. Quelle Horreur!

Same goes for the circuits inside US-designed op-amps—most are un-symmetric. Symmetrical appears smart—a dangerous, hubris-attracting adjective. Mistaken ideas couldn't be clearer now.

First, symmetrical circuits need parts of opposite gender. For example, PNP and NPN bipolar transistors, or N- and P-channel FETs. But, these genders are in no way exact opposites, any more than with humans. There are complicated differences. For example, PNP parts depend on slightly differently-behaving "holes", not electrons, to transact business (footnote 3).

Some astute makers (footnote 4) have mastered ways to inter-weave NPN and PNP bipolar transistors and symmetrize them in triples. But however neat, symmetry remains inherently damaging to musicality. There are no such pairings for tubes. Sounds good. But in the '30s, designers still managed to create a form of bridge: push-pull.

Reasons to be careful: With symmetrical circuits, any approach towards overload (clipping), will generate mainly odd harmonics. Just what the circuit does not reject. "Voll Petard und Hoist!" exclaims the German-speaking Swiss host, mangling Shakespeare.

Secondly when symmetrical and push-pull halves are rejoined, even (2,4,6) harmonics made within, are cancelled. Overall, odd harmonics are built-up and evens diminished. This is even brazenly stated in textbooks.

As circuitry inside op-amps and many discrete signal stages outside of high-end fashion circles, isn't obsessively symmetrical, harmonics are permitted to be a mix of even and odd.

"So circuit symmetry is a rather localized viewpoint phenomena" summarized the professor.

...As Long as it's Groovin'
No less immense yet ignored, is that odd harmonics (footnote 5) inherently generate nasty intermodulation (IM) products. Even at low levels, such additions to the noise floor aren't stochastic. In the grunge is a chorus of ugliness built from inharmonic intervals. Unsurprisingly it masks detail. Whenever there's less IM, lo, new details are heard. And when there are fewer odd harmonics, stage-by-stage, sonics are bound to be less harsh, brittle and fatiguing.

Over half-a-century ago, effects of IM were categorized as "acoustic roughness" (footnote 6). Tests exist (footnote 7) to show the IM products' degree of "infill." The fact no one appears to be using such, demonstrates the low state of audio test.

Late Realizations
Alice thought this was very peculiar and counterintuitive. Throughout nature, symmetry was the diamond standard in beauty. Though with human faces, almost-symmetry being even more alluring. In electronics, symmetry is no less elegant. The counterintuitive exception is handling music signals. So there are fundamental conflicts, that listeners, not authorities, have uncovered. "And on Hammond B3" said the compére in a lazy southern drawl, "the great-test in show-biz'nezz toooo day." With a jolt, Alice realized it was The Toad (footnote 8) disguised with a turban, sunglasses, and mask. All would be safe, now the B3 pedals were being thrashed "à la Dennerlein." Some weasely men in suits at the back began to look uncomfortable.

Some types of balancing can be used with a high benefit ratio, such as dual polarity supplies. Red flags are balanced and bridged outputs; and symmetrical stages relying on "complementary" transistors. A simple corollary is cruder electronics—the worlds of Zen and SET—which generates larger harmonics, and some are odd, but mainly all low-order, and higher ones taper swiftly, as in the nature. This isn't to say SE is all. Far from. But the merit is clear. Hi-Fi is after all, for music. Corollary: Electronics Engineering is in the dock for low musical intelligence, and future analog topology designers must meld linearity with natural musical harmonic structures. Otherwise, Toads, with vast popular support, are known to vigorously evict weasels.—Ben Duncan

---

Footnote 1: For over 30 years, hi-fi-land has shown some mental difficulty grasping that any unbalanced or floating source interfaces just fine with a true balanced input. Balancing at the sender is only relevant to large systems, long cables, many channels, and extensive toxic/noisy environments such as TV studios, venues, stadia, and hotels. Stay safe, stay away, keep your outputs unbalanced.

Footnote 2: In about 1997,Walt Jung kindly mailed me Xerox-copies of DIY amplifier projects with high symmetry, appearing in early 1970s Audio—a magazine that was not made available nor advertised in England (unlike The Audio Amateur). These advanced and elegant-looking topologies took over a decade to percolate to UK and Europe. If they ever did. These circuits depended on a complementary transistor illusion that transistor makers sold, but didn't fix.

Footnote 3: In the late 1960s, I was working in a research lab and used a massive electromagnet to measure charge mobility—electrons or the absence of electrons, called "holes"—in semiconductors.—John Atkinson

Footnote 4: Bryston in Canada—see the author's High Performance Audio Power Amplifiers (1996, Newnes), section 4.3.4, Triple sub-topology. Many other topologies (including symmetrical ones) are also shown and analyzed.

Footnote 5: Cyril Bateman, Electronics World, series on ultra-low distortion measurement of components and circuits, 2002-3.

Footnote 6: P. Tharma (Mullard Application Labs), Transistor Audio Amplifiers, Iliffe, London 1971, Section 9.4, Acoustic Roughness (around 1967). Wigan's work on harmonic weighting (BBC, 1960) is also cited.

Footnote 7: Belcher, Alan, "A New Distortion Measurement Using Comb Filtering," Wireless World, May 1978. Also BBC engineering report, Dec 1976.

Footnote 8: The anthropomorphic Mr. Toad (as vast landowner, stern industrial magnate, and extreme hi-fi fanatic) has appeared in the author's articles since 1984, and in wider literature about English society for over a century, a more recent example being The Wind in the Pylons of 1993.