showstack

The pair everybody meets first

Dynamic and condenser, as signal problems

Both start identically: sound moves a diaphragm. What differs is how that movement is turned into electricity, and every practical consequence follows from that one choice.

A dynamic microphone attaches a coil of wire to the diaphragm and suspends it in a magnetic field. Move a conductor through a field and a voltage appears — this is a generator, and it generates its own signal from the energy in the sound itself. Nothing has to be supplied. The moving mass is comparatively large, so it is slower to respond and less sensitive, and it will take an enormous amount of level before anything gives up.

A condenser makes the diaphragm one plate of a capacitor with a fixed backplate behind it. Sound changes the spacing, which changes the capacitance, which — with a fixed charge held across it — changes the voltage. The moving mass is tiny, so it responds far more precisely to detail and transient.

But that capacitance is minute, and its output impedance is enormous. Connect it directly to a cable and the signal is gone. So every condenser has an impedance converter built in, a few millimetres behind the capsule — and that circuit needs a supply, which is what phantom power is for. Phantom is not powering the diaphragm. It is powering the amplifier that makes the capsule's output usable at all.

The journey after that

Why a balanced line works, in one sentence

A microphone produces something in the order of a thousandth of a volt. It then has to travel fifty metres through a building full of dimmers, motors and mains cable. That should be hopeless, and it is not, because of one idea.

Send the signal down two conductors, one of them inverted. Interference picked up along the way lands on both conductors almost identically — the two wires are twisted together and are in the same field. At the far end the receiver takes the difference between them. The wanted signal, being opposite on the two legs, doubles. The interference, being identical on both, subtracts to nothing.

That is common-mode rejection, and it is the whole trick. It is also why a balanced line stops working the instant the two legs stop being treated identically: an unbalanced adapter, a broken cold leg, a badly wired patch. The rejection is not a property of the cable. It is a property of the symmetry.

The hum

Ground loops, and why the fix is isolation

Earth is supposed to be one thing at one potential. In a building it is not: two sockets on different circuits sit at slightly different potentials, and the difference is small but real.

Now connect two devices to those two sockets, and also connect them to each other with a cable whose shield joins their chassis. You have made a loop — a closed conducting ring, often several metres across, threading through a building full of mains fields. A loop in a changing magnetic field has a current induced in it, and that current is at mains frequency and its harmonics. It flows through your signal grounds, and you hear it.

The cure is to break the loop, and the honest options are limited. An isolation transformer — the thing inside a DI box's ground-lift position — passes the signal magnetically while leaving no conductive path, which removes the loop entirely and is the only fully general answer. Lifting a shield at one end removes the ring while keeping the screen doing its job for the rest of the run. Feeding everything from one distribution point stops the two earths differing in the first place.

What is not an option is lifting a mains earth. That earth is a safety conductor, it is there so that a fault current trips a breaker instead of finding a person, and removing it turns a hum into a hazard. This is the one place in this whole site where the shortcut is genuinely dangerous.

Where the noise floor puts the ceiling

A transducer's output level and its self-noise together decide how much of the range downstream is actually usable. Change the source level and watch what is left.

Sensing position

Two kinds of encoder, and why the difference is a safety question

An encoder turns rotation into numbers, and there are two families that answer completely different questions.

An incremental encoder produces pulses as it turns. Two channels, A and B, are offset a quarter cycle from each other, so which one leads tells you the direction and counting the edges tells you how far. It is cheap, precise and — crucially — it only knows change. Power it down and it has no idea where anything is. Power it back up and it starts counting from wherever it happens to be, believing that to be zero.

An absolute encoder puts a unique code at every position, usually as concentric tracks read all at once. Ask it at any moment and it tells you where it actually is. Power-cycle it and it still knows, because the answer is written on the disc rather than accumulated in a counter.

On a show that is not an efficiency question. An incremental system has to be homed — driven to a known reference — before it can be trusted, every time it loses power, and a machine that skips homing will happily drive a piece of scenery to a position it has calculated from a false zero. That is why automation uses absolute feedback for anything whose position matters, and why "we just power-cycled the rack" is a sentence a head of automation reacts to.

The family

Everything else is the same idea again

Once the pattern is visible, the rest of the sensor world stops needing memorising.

A load cell is a piece of metal with strain gauges bonded to it: it deforms minutely under load, the gauges change resistance, and a bridge circuit turns that into a voltage. That is what tells a rigging monitor what a point is really carrying.

A Hall sensor gives a voltage in the presence of a magnetic field — used for limit switches, speed sensing and anything that must detect a passing magnet without touching it. A photodiode turns light into current, which is what is behind an opto-isolator, a smoke detector and a laser safety interlock. A thermistor or thermocouple turns temperature into resistance or voltage, and is inside every amplifier and dimmer that protects itself.

And a loudspeaker is a microphone driven backwards, which is not a party trick — it is the same electromagnetic principle with the energy flowing the other way.

Four questions answer all of them. Does it need power? What is its full range, and what happens past it? What is its noise floor? Does it report an absolute value or a change? Those four determine everything downstream, including how the system behaves when the sensor fails.