Scientia Potentia Est
- “Another damned thick book! Always scribble, scribble, scribble! Eh, Mr. Gibbon?”
- Attributed to Prince William Henry, Duke of Gloucester, in 1781 upon receiving the second volume of The History of the Decline and Fall of the Roman Empire from its author.
This book deals with small-signal audio design; the amplification and control of audio in the analogue domain, where the processing is done with opamps or discrete transistors, usually working at a nominal level of a volt or less. It constitutes a major update of the second edition. “Small-signal design” is the opposite term to “large-signal design”, which in audio represents power amplifiers driving loudspeakers rather than the electricity distribution grid or lightning.
As stated on the back cover of this book, the publication of Electronics For Vinyl  allowed the vinyl-oriented material in this book to be much reduced, so the space freed can be used for new material. All the phono material that was in the second edition of Small Signal Audio Design is in Electronics For Vinyl, plus a great deal more. Therefore the chapters on moving-magnet inputs have been reduced to one (Chapter 9). This cannot give comprehensive coverage of a very big subject, but it does give the most important information with many pointers to where very much more can be found in Electronics For Vinyl. Chapter 9 also contains new information that has been acquired since EFV was written.
So much has been added to this edition that it is difficult to summarise, but the new material includes:
- A new chapter on tape machines. There is a revival of interest in these devices.
- A new chapter on electronics for electric guitars.
- More on noise gain and the summation of noise sources
- How the accumulation of roll-offs defines frequency response.
- Phase perception and why there isn’t any.
- Combining filters and gain controls in one stage
- Elliptical subsonic filters for vinyl reproduction
- New developments in the Baxandall active volume control
- More on loudness controls. Loudness controls are currently unfashionable, but the thinking behind them is intriguing. I include a possible solution to the mystery of why almost everyone disliked them, when consensus of any sort is rare in the hi-fi business.
- Combining tone and balance controls in one stage
- Adjusting the Q of mid tone controls
- Electronics for ribbon microphones
- Bootstrapped balanced line inputs
- More on ground-cancelling outputs
- Improving fader offness
- Crosstalk cancellation in mixers
There is unquestionably a need for high-quality analogue circuitry. For example, a good microphone preamplifier needs a gain range from 0 to +80 dB if it is to get any signal it is likely to encounter up to a workable nominal value. There is clearly little prospect of ever being able to connect an A-to-D converter directly to a microphone. The same applies to other low-output transducers such as moving-coil and moving-magnet phono cartridges. If you are starting at line level and all you need is a simple but high-quality tone control, there is little incentive to convert to digital via a relatively expensive ADC, perform the very straightforward arithmetic manipulations in the digital domain, then go back to analogue via a DAC; there is also the need to implement the actual controls as rotary encoders and have those overseen by a microcontroller. All digital processing involves some delay, because it takes time to do the calculations; this is called the latency and can cause serious problems if more than one signal path is involved.
The total flexibility of digital signal processing certainly allows greater scope –you might contemplate how to go about implementing a 1-second delay in the analogue domain, for example –but there are many times when greater quality or greater economy can be obtained by keeping the signal analogue. Sometimes analogue circuitry connects to the digital world, and so a complete chapter of this book deals with the subtleties of analogue/digital interfacing.
Therefore analogue circuitry is often the way to go. This book describes how to achieve high performance without spending a lot of money. As was remarked in a review of my recent book Active Crossover Design, duplicating this performance in the digital domain is not at all a trivial business. You can of course start off in analogue, and when you have identified the filter slopes, equalisation curves, and whatnot that you want, it is relatively easy to move it over to the DSP world.
I have devoted the first few chapters to the principles of high-quality small-signal design, moving on to look closely at first hi-fi preamplifiers and then mixing consoles. These two genres were chosen partly because they are of wide interest in themselves but mainly because they use a large number of different functional blocks, with very little overlap between them. They cover a wide range of circuit functions that will be useful for all kinds of audio systems. You will find out how to adapt or design these building blocks for audio and how to put them together to form a system without bad things happening due to loading or interaction. You should then be able to design pretty much anything in this field.
In the pursuit of high quality at low cost, certain principles pervade this book. Low-impedance design reduces the effects of Johnson noise and current noise without making voltage noise worse; the only downside is that a low impedance requires an opamp capable of driving it effectively, and sometimes more than one. The most ambitious application of this approach so far has been in the ultra-low noise Elektor 2012 Preamplifier.
Another principle is that of using multiple components to reduce the effects of random noise. This may be electrical noise, in which case the outputs of several amplifiers are averaged (very simply with a few resistors) and the noise from them is partially cancelled. Multiple amplifiers are also very useful for driving the low impedances just mentioned. Alternatively, it may be numerical noise, such as tolerances in a component value; making up the required value with multiple parts in series or parallel also makes errors partially cancel. This technique has its limits because of the square-root way it works; four amplifiers or components are required to halve the noise, 16 to reduce it to a quarter and so on. Multiple parts also allow very precise non-standard values to be achieved.
There is also the principle of optimisation, in which each circuit block is closely scrutinised to see if it is possible to improve it by a bit more thinking. One example is the optimisation of Recording Industry Association of America (RIAA) equalisation networks. There are four ways to connect resistors and capacitors to make an RIAA network, and I have shown that one of them requires smaller values of expensive precision capacitors than the others. This finding is presented in detail in Chapter 9, along with related techniques of optimising resistor values to get convenient capacitor values.
In many places hybrid amplifiers combining the virtues of discrete active devices and opamps are used. If you put a bipolar transistor before an opamp, you get lower noise, but the loop gain of the opamp means the distortion is as good as the opamp alone. This is extremely useful for making microphone amplifiers, tape replay amplifiers, and virtual-earth summing amplifiers. If you reverse the order, with an opamp followed by bipolar transistors, you can drive much heavier loads, with the opamp gain once again providing excellent linearity. This latter technology, among others, is explained in a chapter on headphone amplifiers.
However, what you most emphatically will not find here is any truck with the religious dogma of audio subjectivism; the directional cables, the oxygen-free copper, the World War I vintage triodes still spattered with the mud of the Somme, and all the other depressing paraphernalia of pseudo- and anti-science. I have spent more time than I care to contemplate in double-blind listening tests –properly conducted ones, with rigorous statistical analysis –and every time the answer was that if you couldn’t measure it, you couldn’t hear it. Very often if you could measure it you still couldn’t hear it. However, faith-based audio is not going away any time soon, because few people (apart of course from the unfortunate customers) have any interest in it so doing; you can bet your bottom diode on that. If you want to know more about my experiences and reasoning in this area, there is a full discussion in my book Audio Power Amplifier Design (sixth edition).
A good deal of thought and experiment has gone into this book, and I dare to hope that I have moved analogue audio design a bit further forward. I hope you find it useful. I hope you enjoy it too.
I have a website at www.dself.dsl.pipex.com, where I will be adding supplementary material to this book.
Further information and PCBs, kits and built circuit boards of some of the designs described here, such as phono input stages and complete preamplifiers, can be found at: www.signaltransfer.freeuk.com.
London, October 2019