DC vs AC

Intuition

Both strokes of the saw cut

Cutting a plank with a hand saw, you do not push the blade in one direction and then carry it back for another go. You push and pull, push and pull, and sawdust comes out on every stroke. Motion that reverses can still do work, as long as something useful happens each way.

Current comes in two kinds, and only one of them behaves like a conveyor belt. Direct current — DC — is charge moving steadily one way round the circuit, out of one battery terminal and back into the other, the direction unchanged for as long as the battery lasts. Alternating current — AC — reverses over and over, pushing then pulling, dozens of times every second. The lamp filament does not care which way the charge went through it. It heats up either way, and the light looks perfectly steady.

Almost everything with a battery in it runs on DC: torches, phones, cars, laptops, every chip ever made. What plugs into a wall arrives as AC instead, and is then quietly turned into DC inside the case, which is what the heavy brick on a laptop lead is for.

Two systems survive because AC has one property DC lacks: its voltage can be raised or lowered with a transformer, cheaply and with very little loss. Electricity is generated far from where it is used, and it travels efficiently only at very high voltage. The voltage can be stepped up for the journey and back down again at the socket, and that is why the world's grids distribute AC.

So the two share the work rather than compete for it. AC brings the energy to the building, and DC takes over the moment anything precise has to be done with it.

Practitioner

Describing a wave with one number

A DC voltage is easy to state: one number that does not move. An AC voltage changes continuously, so quoting it takes a convention, and several are in use. The peak is the largest value the waveform reaches, and the peak-to-peak value spans the extremes in both directions. The average over a whole cycle of a symmetrical wave comes out at zero, which is why nobody quotes it. The figure that does get quoted is the RMS value, because it is the DC voltage that would deliver the same power to a resistor.

For a sine wave — the shape the grid actually produces — RMS and peak are locked together by a fixed ratio:

A great many mistakes come from comparing a peak figure with an RMS one, so the convention has to travel with the number. Oscilloscopes naturally show peak and peak-to-peak, multimeters naturally report RMS, and datasheets specify whichever suits the parameter and say so in the small print.

Not every meter measures the same thing on AC. An inexpensive multimeter rectifies the waveform, averages it, and multiplies by a fixed constant that is only correct for a clean sine. Feed it the chopped waveform from a dimmer, a switch-mode supply or a variable-speed drive and the reading is wrong, sometimes badly so. A true RMS multimeter computes the real value, and that is most of what the extra money buys.

Frequency is the second number an AC quantity needs, counting complete cycles per second in hertz. India, Europe and most of Asia and Africa run at 50 Hz; North America and much of South America run at 60 Hz. Given amplitude and frequency, everything else about a sine wave follows — its period, its rate of change, how a capacitor or inductor responds to it. The sine wave lesson develops it properly.

Real waveforms are rarely purely one or the other. A supply rail is nominally DC with a small AC ripple riding on it, and an audio signal is AC that a circuit has deliberately shifted onto a DC bias. Splitting a measured waveform into a DC component plus an AC component is the standard way to analyse it.

Engineer

What RMS actually is, and what the grid does with it

RMS stands for root mean square, and the name is the recipe read backwards. Square the waveform at every instant, take the mean of those squares over a whole cycle, then take the square root. Squaring is what makes it meaningful: power in a resistance goes as the square of voltage, so the mean of the square is proportional to the mean power, and its square root is the equivalent steady voltage. RMS is therefore a statement about heating rather than about the shape of the curve, and that is what makes it a fair basis for comparing an alternating quantity with a steady one.

The particular ratio in Layer 2 belongs to the sine wave alone. A square wave's RMS equals its peak, a triangle's is its peak divided by the square root of three, and a narrow pulse train's sits far below its peak. Quoting "divide by 1.414" for an arbitrary waveform is one of the most common errors in AC work, and RMS value works through the general case.

A capacitor rated 250 V DC placed across 230 V mains is over its rating at the top of every cycle. It will not fail on the spot, which is what makes the mistake easy to keep: the dielectric degrades a little each time it is pushed past what it was designed for, and the part fails somewhere between immediately and months later, with nothing in its behaviour beforehand to say which. Parts intended to sit across the mains are specified against the peak and against the surges the supply carries on top of it, which is why they are a separate product category rather than an ordinary capacitor with a big number on it.

At 50 Hz, one complete cycle takes 20 ms, since period is the reciprocal of frequency. The voltage therefore passes through zero a hundred times every second. A filament lamp's thermal mass smooths that out, but an LED driven directly from a rectified supply flickers at exactly that rate, and a switch interrupting an AC circuit is handed a free chance to extinguish its arc every time the current crosses zero.

Reversal also changes what the passive components do. With DC, capacitors and inductors settle and then behave as an open or a short. With AC they never settle: a capacitor passes more current the faster the voltage changes, an inductor opposes any change in current, and both introduce a phase shift between voltage and current. Resistance generalises to impedance, and the algebra becomes complex-valued.

That phase shift breaks the simple power calculation as well. When voltage and current are out of phase, part of the current does no net work at all, flowing out and back again each cycle. The ratio of real power to apparent power is the power factor, and it is why an industrial electricity bill carries more numbers than a domestic one.

Even a plain wire stops behaving the way its DC resistance says it should. Alternating current crowds toward the surface of a conductor and away from its centre, more strongly the higher the frequency, so the effective resistance of a wire rises with frequency while its DC figure sits unchanged.

Professional

Where the choice is actually made

The transformer settled the grid argument in the nineteenth century, and its verdict still holds for distribution and for every domestic supply. DC has been taking parts of it back, though. High-voltage DC links now carry bulk power over very long distances and between grids that are not synchronised, because at that scale the dominant losses are not the ones a transformer solves. Power distribution covers the system view.

Inside equipment there is no argument left, because every semiconductor needs a stable DC rail. The incoming mains is rectified — see half-wave and full-wave rectifiers — then smoothed by a bulk capacitor and regulated, so everything downstream sees DC with some ripple on it. What the designer chooses is how much of that ripple the load tolerates, and what removing the rest is worth paying for.

An averaging meter has a sine wave baked into its calibration. Its scaling constant is the form factor, the ratio of a sine's RMS value to its rectified mean, which works out to 1.11. Present it with a dimmer's chopped output or a switch-mode supply's spiky input current and the number on the display has no defined relationship to anything. On distorted waveforms a true-RMS instrument is the only kind worth reading.

Switching is where DC is genuinely harder. An AC arc gets a natural extinction point twice per cycle, when the current passes through zero. A DC arc gets none, and simply continues until the contacts have separated far enough to break it. A switch or relay rated at a comfortable AC voltage may therefore carry a far lower DC rating, and DC circuit breakers are physically larger for the same duty. Battery and photovoltaic systems need components specified for DC rather than borrowed from mains practice.

Corrosion is a DC problem far more than an AC one. A steady current through a damp joint drives electrolysis, moving metal from one electrode to the other, so buried DC systems need cathodic-protection thinking that an AC system at the same voltage does not.

Frequency also has a safety dimension. Body impedance falls as frequency rises, and the heart is most easily disrupted in exactly the range mains occupies, which puts mains frequency close to the worst case for electric shock. DC of the same magnitude produces a different effect, generally less arrhythmogenic, though the let-go behaviour differs too. Both are dangerous, and electrical safety fundamentals covers them.

Those zero crossings are also the reason for a whole component family. Triacs and diacs exist because a crossing gives a cheap moment to switch on partway through each half-cycle, leaving the load a controlled fraction of the available energy. The chopped waveform that comes out is the one that defeats an averaging meter.