View From The Bench: How Preamps Quietly Make Big Gains, Part 2

VIVA LA DIFFERENTIAL

One desired characteristic of microphone amp design is the rejection of external noise sources, so that when the audio signal receives amplification, the noise does not. In professional balanced systems we use what is called a differential input, and the easiest way to get one is to use an audio transformer.

A transformer consists of two coils of wire known as the Primary (input coil) and the Secondary (output coil). When a microphone sends its signal down a cable, it will do so in the form of a varying voltage between two wires. When that ‘difference voltage’ is connected across a transformer’s Primary, a varying current will flow through the coil. This causes a magnetic interaction with the Secondary, creating a facsimile of that same signal at the output.

The trick with a differential input is how it deals with noise. When radio signals and other delinquent atmospheric nasties attack our innocent microphone cable, they will inflict their voltage upon those two wires equally and in common, creating what is called ‘common mode noise’. At the transformer input the noise voltage on both wires will be equal, and therefore the difference voltage between them will be zero, producing no current flow through the primary, no magnetic interaction, and no reproduction of that noise at the output of the Secondary.

Think of it like a piano accordion: when one hand goes left and one hand goes right, the bellows expand or shrink, and music comes out. If both hands go left or right by the same amount, the bellows don’t change size, and the music stops. With the accordion, the differential movement produces audio, and the common mode movement prevents it.

The ability to suppress common mode noise is an important figure of merit. A typical common mode rejection ratio (CMRR) for a transformer might be 100dB or more. What that means is a microphone cable could have equal amounts of differential signal and common mode noise travelling down it, and therefore a disastrous signal to noise ratio of 0dB. However, when that mess exits our audio transformer’s Secondary, the differential signal will pass through whole, but the common mode noise will be attenuated by 100dB, and that would become the new signal to noise figure at the output! It must be said that transformers have a natural talent in this area.

I will just pause here for a moment to reiterate that what I have just described is a ‘balanced audio system’, and the whole reason it exists is for the cancellation of common mode noise — which is a hopeless pursuit in an unbalanced system. It is a widely mistaken belief that balanced audio means equal and opposite signals must appear on the two wires of an audio cable, which it does not. It refers to the balancing of output and input impedances seen by both wires, and in a properly matched system it makes absolutely no difference if you have a signal on one wire, and nothing on the other. The point is the audio signal will arrive at the balanced input in differential mode, and be amplified, and the unwanted noise will arrive in common mode, and be cancelled. That is balanced audio.

The downside with transformer balancing is that in the process of converting from one form of energy to another and back again, information gets lost, and errors accrue. A happier way of saying this is that there is colouration, which can be a good or bad thing depending on what you are after. Transformers traditionally sound a bit soft and furry in the bottom end, but also seem to have a signature of their own, much like microphones do. If a device has an audio transformer in the signal path, that sound can be hard to get away from.

Not all microphone amplifiers use an audio transformer, and it is possible to build a differential input exhibiting good common mode rejection purely from electronic componentry. The circuit I developed for the Prodigal channel strip is such an animal, using low-noise transistors at the front end, configured as a differential input with high common mode rejection and moderate gain.

ACCENTUATE THE NEGATIVE

In 1927, while going to work on the Hoboken ferry, Harold Black of Bell Laboratories scribbled a diagram and some equations onto his copy of the New York Times, showing how negative feedback — a system derived from nature — could be applied to electronic amplification. Negative feedback is a system where some of the output from a process is fed back to the input in such a way as to have a subtractive and controlling effect on proceedings. Think of what happens when you get hot — you sweat, and the heat exchange from evaporation lowers your temperature. If you didn’t have a sweat mechanism providing negative feedback, you might just get hotter and hotter till your head melts. Your body temperature would be at the mercy of external forces, and its ongoing state would be unpredictable. The negative feedback helps keep your body at an even 37°C, and so provides equilibrium, stability and predictability.

In an amplifier the transistors or valves will provide gain, but the vagaries of manufacture will ensure that no two components provide exactly the same amount except by accident. Furthermore, whatever gain they do exhibit will vary with temperature. In early amplifier designs these vagaries were a dominant influence on the overall gain — a cause of non-linearity, harmonic distortion and amplitude drift. It was just as well that stereo had not been invented, because a stable image would have been impossible with such equipment.

Black’s idea was to send some of an amplifier’s output back to its input, to be subtracted like a bead of sweat from the input signal, with the result that gain would finally become fixed, predictable and constant. If a bonfire under Black’s amplifier caused increase in gain within the circuit, it would be accompanied by an increase in the negative feedback signal, to be further subtracted from the input, which would in turn reduce the output, and equilibrium would be maintained. This system sacrifices some of an amplifier’s available gain for a more linear response and lower distortion, so it is a method for controlling gain, rather than creating it.

Black built the first negative feedback amplifiers for Western Electric in 1928, and published a paper in 1934. Today there is scarcely an amplifier in operation which fails to take advantage of Black’s world-changing idea. For two bucks anyone can buy an ‘Op Amp’ integrated circuit and, using two resistors, apply negative feedback to set that amplifier to almost any level of gain they might require.

One limitation of the negative feedback system is that there is always a miniscule delay between a signal arriving at the input, and any correction forced by the feedback. This can cause oscillations at radio frequencies, if the delay is longer than a signal’s rise or fall, and the negative feedback fails to constrain its amplitude. In audio devices, particle noise is for the same reason almost impossible to control with negative feedback, and that is why a good amplifier will try to attenuate very high frequencies and particle noise before amplification.

PHANTOM POWER

In the olden days, condenser mics would have their own power supplies, and output their signal down the wire to a mic amp. At some point, putting the mic power supply inside the mic amp was deemed more convenient. Power was then sent up the wire to the microphone, the same way as the terrestrial telephone system worked. The practice caught on, and now every microphone amplifier in the universe sports a 48V DC phantom supply.

About 10 years ago I downloaded a copy of the IEC standard and got a shock. Yes, a voltage shock. It stated that although 48 volt systems are in use, new systems should be confined to 24 volts! Was I reading correctly? Yes, I was. Well it actually made great sense, especially if you are building a preamp that runs on batteries. In fact, most condenser mics will work fine on as little as 12 volts, but I think the standards committee just made that decision too late in history and no manufacturers wanted to change. I for one have never laid eyes on a 24 volt system, and while I momentarily considered making the first one myself was basically too spineless to risk a move of such flamboyant extravagance. So why was 48 volts chosen in the first place? Simple — it was what AT&T used to power their telephones.

ADJUSTABLE GAIN

Some microphones are louder than others, so it helps to have a Gain knob, mainly to wind backwards so that the loud ones aren’t driven into distortion. Gain is different from volume. A gain control might form part of the negative feedback loop, and so changes the amount of amplification being applied to a signal. A volume knob usually appears before or after amplification and merely attenuates the signal passing through. For best noise performance and headroom management, most channel strips employ a gain control at the input, and volume control (the fader) at the output.

INPUT PAD

When the mic signal is really loud — say a condenser with a kick drum pounding through it — a microphone amplifier’s lowest gain setting might not be low enough to allow the signal to reach the output without having its peaks clipped. In those instances it is common to attenuate the signal before it enters the mic amp. A switchable pad of 15 to 20dB can be a very useful inclusion to a mic amp design, and the process can be done quite simply and cleanly with a voltage divider made from a few resistors. Even these days not all mic amps have input pads, and I still carry a couple of home made XLR pads in my kit.

PHASE REVERSAL SWITCH

This is another useful widget that not all mic amps feature. It can be used to correct a bottom snare drum mic that is physically 180° out of phase with the top snare mic, or to check the balance of a stereo pair — maximum cancellation with one mic flipped equals perfect balance when you un-flip it.

TINKERING WITH THE GHOST OF FRANKENSTEIN

If you are designing a mic amp, the entire laundry list of features above needs to be considered, and it all looms like a crazy jigsaw puzzle before you even begin. By the time you finish, pieces will have been thrown in the air or rearranged, and there will be dead components with black holes in them where the smoke has come out. A bearded sage once spake unto me that all electronic devices run on smoke, because when they break the smoke comes out. He said that when mankind works out how to put the smoke back in, electronic devices will last forever. He had a point. Design development can be quite a circus, and not every interesting idea you try will survive. But that’s the fun of it.

And here’s the thing — if you are really looking, then you are also listening. Not just staring at a computer screen, but fiddling with an actual circuit and seeing how you can manipulate the sound of it. With the Prodigal, I set out to make a transformerless differential input, and I wanted the microphone amplifier to have its maximum possible gain only slightly higher than when the negative feedback is connected, to help keep the noise down. After satisfying such matters of housekeeping I was ready to open my ears.

Changing around electronic components can be a bit like changing a microphone. The exact same circuit can be made to exhibit variations in colour simply by swapping say, a transistor, a capacitor or an op amp to a different kind with the same value. If you own an electric guitar, and the tone control uses a ceramic disc capacitor, changing it to a polyester greencap of the same value will probably yield a difference you can hear. Also, if you haven’t tried it, interchanging a TL072 op amp with an NE5532 will aurally change the flavour of a circuit, but I’ll be jiggered if I can see why with scopes and probes. Altering the architecture within which these components interact will of course have even greater potential to affect the sound. I have to conclude that the human ear is a very sensitive instrument and, for audio, the most important tool of reference.

Throughout the Prodigal Frankensteining process I did all the required stuff with lab equipment, but also performed critical listening tests with actual music. One of the albums I was fond of using was Aaron Neville’s Warm Your Heart which was also my favourite PA tuning reference disc for many years. Besides sporting a full rich bandwidth with plenty of percussive edges, it also has the added wrinkle of Aaron’s beautiful voice which contains a karate chop at 2.5kHz that can sound really nasty if your mids are crap.

So I was inching my circuit forward, getting it to run on lower and lower currents, messing with the input impedance, swapping out transistors, etc. At some point Aaron’s voice got all three dimensional, and while I was thinking ‘that sounds good’, Mrs Tech Bench stuck her head around the corner and said ‘that sounds good’. Wow. So what makes something sound good? I can’t really quantify an answer, except to say if you have sound A and sound B, and you would rather hear sound B, then sound B is the good one. After some more tinkering, it was a variation of that version of the mic preamp that finally made it into production.

GOODNESS GRACIOUS ME

So in the end I wound up with something I think sounds great. I also did a bunch of tests and wrote down the results, which I hoped would be impressive, so that people with no ears could enjoy my work as well. Olive Gherkin appeared in my dreams dressed as a bearded sage, and banged on about metaphysics while waving her golden staff at the heavens. She told me I was a phoney, and that Edison never tinkered like me, but received divine design instructions from a winged cherub with a mohawk. I woke up in the night all full of self-doubt, wondering about the blunt instrument and the brute force, and the complex interactions between all the constituent parts. Does my gadget fly? Does it have a sound, and is that sound the important bit? I had been on a journey around the big design table, exploring the subtle and nuanced mystery of what is good, and discovered that while brute force may be quantified, goodness really is a mystery and probably always will be.