Wednesday, October 28, 2009

Memory Distortion: The Bane of Solid-State Amplifiers?

Some audio purist still claim that solid-state amplification is not truly hi-fi in comparison to it’s thermionic / vacuum tube counterpart, is there a reason behind this?


By: Ringo Bones


During the start of the 1970s, the solid-state transistor-based audio power amplifier started to make the price of high-powered hi-fi amplifiers – i.e. over 50-watts – much more affordable. But many hardcore audiophiles complained that transistor-based audio power amplifier didn’t sound as musical as their thermionic / vacuum tube-based counterparts. Even newer MOSFET devices – whose characteristic curves resembles that of a pentode tube – still didn’t sound quite as musical in comparison to their thermionic brethren during their rollout near the end of the 1970s. But is there a reason – hopefully a scientifically verifiable one – that explains why solid-state amplification (transistors and MOSFETS) don’t sound as good as their vacuum tube counterparts?

Back in the summer of 1997, a French amplifier manufacturer – Lavardin Technologies – announced that they have discovered why solid-state amplification didn’t sound as musical as their vacuum tube-based competition. They called the phenomena “Memory Distortion” which Laverdin Technologies describes it during 1997 as “the greatest discovery in analogue audio design in the previous twenty years”. Memory distortion, Lavardin Technologies says, is responsible for the shrillness and mechanical-sounding artifacts identified in solid-state amplifiers. Unfortunately in the intervening years, my “richer” audio-buddies can only listen to Lavardin Technologies’ amplifiers in hi-fi shows because they are so prohibitively expensive when compared to vacuum tube-based amplifiers of similar power output and features. But they swear that Lavardin Technologies' low-powered integrated amps do sound like they use vacuum tubes as power output devices.

Fast-forward in 2009 when one of my audio-buddies managed to purchase one used – although still very costly – one of those Lavardin Technologies integrated amplifier. It is the 30-watt Lavardin IS Reference which sells for almost 4,000 US dollars when brand new. He got one for a shade under 2,000 US dollars, and say’s its all worth it because the single-pair of transistors used in this integrated amp will last for thousands of years when properly used. And they do sound like vacuum tube amps – vacuum tube amps that could drive speakers with tricky impedance curves. Albeit only within their “meager” 30-watt rating. But despite of the obvious overpricing in electronics engineering terms, why do these amps sound so good?

Given our sample no longer has the company’s warranty and my audio-buddy was generous enough – albeit up to a point – to allow our local hi-fi association a peak inside the innards of Lavardin’s famed integrated amps. A look inside might make every “mainstream” electronic engineers accuse of Lavardin Technologies of recto-cranial inversion. Those mainstream folks usually accuse everyone of encasing the circuits of their designs in some kind of black goop in the name of copyright protection, as suffering from recto-cranial inversion. But these has been proven – probably since the 1980s – that it could improve the sound quality by controlling spurious vibrations from affecting the sensitive circuit layout.

Lavardin did divulge the reason why tubes sound better than transistors, which they used to their advantage in making their solid-state amplifiers sound as good as tubes. It was probably the consensus view of quantum physicists who looked into the differences in operation of vacuum tubes and solid-state devices during the 1990s. According to their findings – though it has been noted in every post World War II vacuum tube-based electronic textbook in existence – which start at the basic fundamental differences between vacuum tubes and solid-state devices.

In a vacuum tube any particular electron – i.e. strictly speaking the electron’s wave function – travels through free space, influenced only by the electric fields caused by the various electrodes in the electron’s wave function’s path within the confines of the tube. The control grid’s field hold’s back a proportional number of electrons from the total number of electrons emitted by the cathode, in which a change in grid voltage change’s it’s field and thereby varies the total number of electrons reaching the anode and hence the resulting anode current.

The velocity of an electron by the time it reaches the anode after being accelerated by the anode’s field is truly mind-boggling. As an example, a tube with a typical anode voltage of 450-volts, the electrons will hit the anode at approximately 28 million miles per hour or about 4% the speed of light – which is around 670 million miles per hour in vacuum. Thus the reason for the vacuum tube’s somewhat high-temperature operation. The electrons which make it past grid are the same ones which – a tiny fraction of a second later – appear at the anode and becomes the signal that drives the load.

In a solid-state device – transistors, MOSFETS, and specially including wire – the electrons have a very hard time traversing the entire length of their intended path. In the solid-state domain, electrons have to fight their way through millions of random fields caused by the atoms in the substrate material – usually at 0.001 meters per second. In which calling it a snail’s pace would be an understatement in comparison to an electron’s speed traveled though a typical vacuum tube. Furthermore, it is not the actual electrons which carry the signal, but the influence one electron on it’s neighbors. The message or signal gets carried akin to a “Chinese Whisper” albeit only with less degradation of the signal – hopefully.

This quantum-mechanical explanation of the radical difference between vacuum tubes and solid-state devices is the claim used by Lavardin in explaining how they minimized memory distortion in their solid-state amplifier designs. According to them, memory distortion has to do with the way musical signals have to slog their way through silicon – akin to being stuck in the mud. Transistors hold previous signals in memory – as in the electron’s wave function. And these “residual memories” or remnants of an electron’s previous state - maybe a few tiny fractions of a second before distort the new incoming signals. The musical signals can’t flee the silicon fast enough. But is this explanation sufficient from a scientific standpoint? After all, if “memory distortion” is about timing errors – assuming that the phenomenon is real in the first place – then why is it that there are several, albeit almost unrelated, ways of eliminating the symptoms caused by memory distortion.

James Henriot of Whest Audio also managed to do the same feat of making solid-state amplifiers more musical by eliminating “analog-domain jitter” via his Whest dap.10 processor. Which most users testify that the Whest dap.10 processor improves their already well-sorted CD playback system’s sound quality by making it sound like a big analog open-reel tape, the one often used in better recording studios. I’ve heard this only in hi-fi shows, but my impression of this product seems like it makes your typical solid-state integrated amp sound like a good vacuum tube amp.

While a Frenchman named Yves-Bernard André of YBA also manages to do the same with his solid-state integrated designs by using various techniques holistically to eliminate the symptoms that make solid-state amplifiers sound “inferior” to their vacuum tube counterparts. From using synthetic diamond powder to damp the circuit boards to the resonance control of every component used. Not to mention minimizing to the absolute minimum the inherent hystersis distortion caused by a transistor’s ferromagnetic enclosure. Even though YBA products – as with most French integrated amps - are typically priced way above a typical hi-fi enthusiast is willing to pay, Yves-Bernard André’s holistic approach to designing his solid-state based audio components seems to have removed the symptoms of what we know of as memory distortion.

LFD Mistral MOSFET-based integrated amplifiers also managed to eliminate the symptoms of memory distortion through attention in circuit layout. By orienting the resistors of their LFD Mistral integrated amps in phase on the master board. The resistors on both channels are identically oriented which they believe – and some owners of LFD Mistral integrated amps – is important to stereo imaging. So does the orientation of the wiring and the fuses. Unfortunately, this attention to detail in parts layout doesn’t lend itself well to mass production machinery used in making mobile / cellular phones and i-Pods. But the resulting product is nonetheless spectacular. LFD Mistral integrated amplifiers are often compared to single-ended triode amplifiers in terms of sound quality.

So what does this all mean? Well, it seems like the holistic tweaking techniques utilized by Yves-Bernard André and the LFD Mistral does seem to improve the sound quality of your typical solid-state audio gear – even ones using integrated circuit IC amplifiers. While the Lavardin Technologies may be on to something in explaining the phenomenon of memory distortion, James Henriot’s forays into analog-domain jitter will probably need the collaboration of other scientist with access to the very state of the art testing gear to explore further the phenomena of analog-domain jitter. Who knows that it might result in better and cheaper laptops and mobile phones ten years from now? Maybe memory distortion is just a symptom of bad circuit layout in the production of solid-state gear. Often easily solved via enclosing critical parts in a faraday cage - or by the use of exotic and boutique capacitors like Rubycon Black Gates or Philips-sourced French Blue capacitors.