Back in 1994, I moderated a discussion of high end amplifier designers:
Amplifiers have always been the products that have made icons of audio designers and engineers. Companies like Krell, Threshold, Audio Research, and VTL have, in a way, helped to define this industry. For a number of reasons, amplifiers are easy to view as being the most significant link in the electronic aspect of the audio chain. The Linnies tell us that the source, be it LP or CD, is the most significant qualitative link of the audio chain. Many meter-dulled objectivists would have you save money on electronics while spending freely on speakers.
What do I think?
To be truthful, I don’t know. My basic belief is that amplifiers are the most critical (note avoidance of the word important) device in the audio chain. The amplifier is literally and figuratively caught in the middle. It cannot reproduce more than the upstream components pass on to it, nor can it control or drive the loudspeaker to a level that is beyond the latter’s electromechanical potential. Further, it is the only device in the chain that has to deal with a potentially wild and varying electrical load, in other words, the typical audiophile speaker.
So, there, in the dark, sits the lonely amplifier designer. All the while asking himself, “What kind of load will my baby have to drive?,” “What if bipolars really were the way to go?,” “Will this single-ended craze last?” TAO is all about observations. The observations in my reviews and articles are purposefully limited to descriptions of the musical potential of the audio gear I evaluate. And while my technical competence may actually be somewhat greater in reality than I let on in print, I still believe that mine should be largely a non-technical presence.
Still, technical issues are of great interest to me particularly as they concern amplifier design and philosophy. So, I sat down and did some pondering. What questions can I come up with that will tend to reveal the thought process that goes into a successful amplifier design? What’s more, who will answer them?
The respondents to my questions are some of the best audio minds I know. Some of their names will be well known to you. Others may be new to you.
Kevin Halverson / Muse Electronics
John Kovacich / Pointsource Audio
Eric Lauchli / Coda Technologies
Paul McGowan / Genesis Technologies
Nelson Pass / Pass Laboratories
At the outset, I’d like to thank all of these guys for their willingness to take part in this humble survey, and for simply taking the time to respond to my questions.
The Round Table will be divided over three issues with each designer answering the same two questions in each installment.
Here are the questions for this issue:
Question 1: In amplifier design, science never really separates itself from art. To what degree are your design decisions based upon what you can hear versus what you can measure?
Question 2: As a follow-up to the first question, if you could only make three measurements of a design’s electrical performance what would they be and why?
Kevin Halverson As for the question about which skill (art or science) is most important in the design of hi-fi amplifiers, I would have to answer science. This is not to say that art does not play a strong role in my designs. If, for example, I find during listening that an aspect of what I’m hearing is either better or worse, no amount of artistic knowledge would do anything to assist me in understanding the phenomenon. Similarly, no amount of science will do anything to enhance my listening skills. If, by your question, you are suggesting that there is a fundamental conflict between what one hears and what one measures, I would reply that there is not, provided that one knows where and most importantly how to look (measure).
John Kovacich Since it’s quite easy to create audio equipment that has good specifications, there is no reason not to at least try and design amps and other products that look good on paper. Unfortunately, specifications and what you hear don’t always go together. In other words, a product can spec-out very well and still sound poor. Of course, we have to ask ourselves, does the product truly sound bad or is it simply revealing problems in the rest of the system that a lesser amp would obscure? For instance, some people believe that tube amplifiers sound superior to solid state because they are better in design, and their poor specs don’t mean anything. Other people feel that tube amps sound better because they create a pleasant sound. The key word being, create. Mainstream designers generally design for specs and then assume that their products will sound good.
Audiophile designers generally design, then listen to the product with little concern for specs. Now, we have to ask ourselves two questions: Can they really hear any difference? Since none of these listening sessions are done in a double blind fashion, they are, from a purely scientific point of view, invalid. I have heard many differences between designs but could never reliably and consistently pick the superior design when denied the right to know which circuit I am listening to. Secondly, should the amp be transparent? Or, should it rather reflect the designer’s personal opinion of what the sound should be like? And, if the amp reflects the designer’s point of view, that point of view may or may not be consistent with your own.
Eric Lauchli Without a good technical understanding and adherence to reasonable measurement standards, a truly fine amplifier is unlikely to result. Having said this, number chasing can be and often is sonically disastrous. I must admit to using a considerable degree of intuition while designing, though this mental process is difficult to fully explain. An elusive but important concept seems to be design elegance. This is a characteristic of designs solutions which are both simple and powerful, while yielding overall circuits that can be said to be much more than the sum of their individual parts.
Paul McGowan This is awfully tough for an old goat like me to answer because so many of my decisions, both technical and sonic, are based on my years of experience in both realms. Where the specific knowledge comes from (for a given situation), meters or sonic experience, is not always clear.
As an example, my sonic experience tells me not to use a lot of global feedback because I know that it will typically make the amps sound bright and hard. There are also very good technical arguments against excessive feedback as well (TIM, SID). However, it would be the sonic experience that would lead me to pursue a low feedback approach.
Still, measuring is a tool that I use in virtually every design that I undertake. I measure so that I am sure not to stray too far in any one direction. I make sure that the THD hasn’t strayed too far under load. I look for ringing on the square wave under different conditions. I keep an eye on the damping factor (to make sure that there’s enough). I measure frequency response to make sure that there are no anomalies. I look to be sure that the noise floor is low enough, etc. Still, I rarely use my instruments as part of my basic design decisions.
For me, designing an amplifier starts out as a philosophical exercise and develops into a product when I can come up with enough simple and elegant solutions to the problems inherent in the chosen philosophy. Philosophically, most of my designs come out of a desire to correct a fundamental question or problem. The things that interest me are the most basic problems of amps that may entail a bit of new thinking. I get excited when I attack fundamental problems.
Nelson Pass I care so little about subjective versus objective arguments that this is all I offer: Listening is a measurement, and in high end audio is the most important measurement. Not all the art is in what you hear, and not all the science is in what you measure.
Kevin Halverson As for simply picking three measurements, I must admit that I would not allow a product to go to market if I were allowed only three measurements. Each and every product should be evaluated based upon several factors including: intended application, design expectations, and most importantly a typically elusive path whereby prior measurements lead to new discoveries. No two products will ever undergo the exact same regiment of tests. Rather, each product, having an individual character, will require a different approach in order for the designer to feel confident that all aspects of the product’s performance meet the intent. I might also add that I strongly endorse the use of blind listening tests. While these can be very humbling experiences, I always come away with a greater sense of understanding and validation.
John Kovacich I would make all the standard measurements, then I would make some special ones that I feel reflect the real world a little better. First, I would use a square wave, not a sine wave, at the input and I would use an actual loudspeaker or an equivalent circuit for the load. I would measure the level of each harmonic at the input and then at the output (their amplitude as well as their phase). This test would show how the amp does under complex signal as well as complex load conditions. Then I would use an actual piece of music as the source and a loudspeaker for the load. The test would consist of measuring the input and the output using a simple null test. This test would show how the amp is doing under the most realistic of conditions.
Eric Lauchli The direct, simple and inherently linear nature of our designs make ordinary THD measurements surprisingly useful, particularly at high frequencies and with a distortion waveform displayed for careful examination. High frequency square wave testing, especially into reactive loads, can reveal much about an amplifier’s stability and composure under transient conditions. A plot of output impedance against frequency and power level can help predict if an amp will remain stable and linear into any load.
Paul McGowan Frequency response, square wave performance, distortion. Frequency response: It has been my experience that as long as there are little to no restrictions in frequency response between 2Hz and 50kHz that the ear will not detect any anomalies. Square wave performance: This, to me, has proven to be extremely valuable. I can tell all kinds of things from an amplifier’s square wave performance. Ringing of the square wave is a key to a number of mistakes that an amp can make especially when a capacitive load is added to the output. A spike on the square wave’s leading edge can also spell big sonic trouble if not addressed. Typically, this relates to some feedback type problems. Also, the actual quality of the square wave is important in how gently it maintains its shape when the frequency rises, etc. Distortion: I use a spectrum analyzer to view distortion products. If the distortion rises over 0.1%, I get concerned. Another factor is the harmonic structure of the components. Odd order harmonics do sound worse than even order products.