Some Audio Ramblings

The high end stores are all gone and the pursuit of two channel perfection seems a thing of the past, at least here in Rochester, NY. There was a time when we were said to have more hi-fi stores per capita than any other city in the U.S. Regardless, I still have an interest in electronics and speaker design. Unfortunately, I've noticed that the publications specializing in those areas are shadows of their former selves. For a time, science was winning, but the pendulum has shifted to the subjective. Even highly respected analog engineers (who should know better) are publishing designs aimed at solving problems that may not exist, or are failing to prove their claims with tests and measurements.

It may not be a popular view right now, but I don't believe there is any magic to good audio design. The science is well understood if one cares to do their homework, and as long as you consider all aspects of the problem, including psychoacoustics. After many years of designing analog circuits, listening to analog circuits, and reading about others adventures with analog circuits, here are some things I believe. Warning: These are my opinions. They are always subject to revision based on new information. A friend once said "There's what you know and what you think you know, and you need to know the difference". Caveat emptor.

If you can hear it, you can measure it. I have never heard a difference in sound quality between two pieces of equipment that I couldn't measure on the bench. That doesn't mean it will show up in simple frequency response or distortion measurements, but there is much more to electronic testing than that. The problem is that it takes perseverance to ferret out circuit misbehavior, and there is no simple and standard test that reveals all ills. You have to work at it and few people have the time and patience to follow every lead. The popular magazines ignore this area because the average reader wants a simple answer given in traditional terms that he or she understands.

The old rules are still correct. Components can only be compared if the levels are matched to very tight tolerance. Statistically analyzed double blind comparisons are the gold standard. Note that conducting such tests is notoriously difficult. In the end, almost all claims of sonic difference turn out to be errors in the comparison methodology, simple differences in frequency response within the audio band, or an obvious design error in the circuit.

RF throws all bets off. Most people doing audio work have audio frequency test equipment and knowledge. Unfortunately, power supply rectifiers generate RF, and the world is full of RF signals that find their way into inputs and outputs. Sometimes high gain high bandwidth circuits will oscillate at tens or hundreds of megahertz. Any of these things will play havoc with stability and operating points. I dare say that the most common cause of sound differences between amplifiers (where they do exist) is stray RF and subtle oscillations that no one looked for or realized the importance of.

Identical signals sound identical. Seems simple enough. If two amplifiers produce the same voltage vs. time signal at the load terminals they will sound the same. A differential test between amplifiers that yields a null pretty much guarantees they sound the same, assuming the same loads. Make sure to get agreement on this point before wasting too much time debating anything else!

Interconnects (cables) are completely understood. Cables can be compared using standard electrical measurements like capacitance, inductance, and resistance. There is no magic and cost has nothing to do with performance. Cables absolutely can change the sound of a system, but not because they have any inherent sonic virtue. They change the sound because they interact with the inputs and outputs based on, you guessed it, their capacitance, inductance, and resistance. Skin effect is calculable, but is probably not relevant. Velocity factor is even less so. Cables are not transmission lines at audio frequencies and talking of cable impedance as if it were an RF system makes no sense unless one is talking about amplifier stability and RF pickup. Of course it isn't in any cable makers best interest to state the simple truth, so they keep dreaming up technical sounding gobbledygook to hook the unwary.

Electrical components are completely understood. Most of the marketing claims for exotic resistors, capacitors, inductors, and connectors are nonsense. Take a capacitor as an example. Once you know the capacitance vs. voltage, dissipation factor, dielectric absorption, self resonance, aging, and temperature coefficient, you know pretty much everything about the sound quality. Excellent specifications in any or all areas are available off the shelf at reasonable prices. Good connections matter, but solder is solder. Silver solder has several advantages, but none of them are audible. Until the purveyors of expensive exotic components reveal the results of standard engineering measurements, and they prove superior to standard parts, don't waste your time and money.

Single point impedance numbers are useless. Input and output impedance are important, but need to be described over the entire audio range and beyond. It's necessary to know the equivalent circuits of all inputs and outputs. Once this is known, understanding interconnect and compatibility issues is trivial. (Actually, it isn't if you have to do it by hand, but any cheap or free Spice program reveals all very quickly.)

First rate speakers use first rate drivers. No amount of crossover tweaking, zobel networks, or cabinet design will compensate for peaky response or non-linear operation. There are a tremendous number of drivers out there that seem to have no known application, at least for quality audio. Perhaps the market for sirens and buzzers is larger than I thought.

Great speaker design cannot be done solely by ear. Merely OK speaker design maybe, but to do better than that you have to be able to characterize drivers and measure the finished system in time, frequency and spatial domains. There are just too many variables to sort out without measurement. Even the best speaker systems have significant flaws, so you're left with the question of what combination of flaws sounds best. In the end, sound is what matters, so measurements have to be combined with countless listening sessions over a significant period of time to achieve success.

It is impossible to solve most room problems with an equalizer. Sorry, but response is rarely the reason that a room "sounds bad". Standing waves, reflections, and reverberation time are usually the culprits and fixing them requires that physical changes be made to the room and speaker position. An equalizer can solve other problems and be beneficial, but fix the room first.

Well, that's a quick summary of a few things I've found to be true. Perhaps I've given you enough of an outline to get you thinking in a critical manner. Or perhaps subjectivity has taken such a deep root that you reject all measurement. Send me an email and tell me what you believe and why.

edit- I wrote this several years ago and haven't seen, heard or measured anything since that would change my opinions in the slightest. I want to reemphasize the difficulty of making comparisons. The brain comes with a set of biases that seem specifically designed to muddy the waters of audio comparisons. Almost any knowledge about the signal chain contaminates the comparison and biases the results. Many (most?) audio businesses are based on this fact whether they can acknowledge it, even internally, or not. Everybody has a horse in the race, so take most claims of audio superiority with a grain of salt. CRH 4/29/13