Circuits & Electrical Power

Georg Ohm published this relation in 1827 after painstaking experiments with wires of different lengths and thicknesses — and was initially ridiculed for reducing electricity to arithmetic. The idea is simple: voltage is the electrical push, resistance is the opposition, and current is what results. Double the push and you double the flow; double the opposition and you halve it. A 12 V car battery connected across a 6 Ω lamp drives 12/6 = 2 A through it; swap in a 3 Ω lamp and the current doubles to 4 A.

The law holds for ohmic conductors at constant temperature — metals, resistors, most wiring. Components like diodes, filament bulbs, and thermistors bend the rule because their resistance shifts as they heat up or as voltage changes. The classic V–I–R triangle mnemonic works because every rearrangement here is a single multiplication or division.

Current is not a thing that flows. It is a rate — the amount of charge passing a chosen cross-section of the conductor each second — and one ampere means one coulomb per second. Once that is clear, $Q = It$ needs no proof, because it is the definition read backwards: if charge crosses at a steady rate, the total that crossed is the rate multiplied by how long it kept up. The only condition the equation imposes is the word steady. A current that varies has to be integrated, $Q = \int I\,dt$, and this page is the special case where the integral collapses to a rectangle.

A 2 A charger running for one hour moves $2 \times 3600 = 7200\ \text{C}$. Battery ratings are the same arithmetic wearing different units: a phone cell marked 3000 mAh holds 3 Ah, and $3 \times 3600 = 10\,800\ \text{C}$ of deliverable charge. Divide by the elementary charge, $1.602 \times 10^{-19}$ C, and that is about $6.7 \times 10^{22}$ electrons — a number that only sounds absurd until you remember a gram of copper contains ten times as many free ones already sitting in the metal, drifting at well under a millimetre per second.

Since the 2019 redefinition of the SI, this relation is closer to the foundation than it used to be. The ampere is now fixed by declaring the elementary charge to be exactly $1.602176634 \times 10^{-19}$ C, which makes the coulomb a count of charges and the ampere a count per second. Michael Faraday got there experimentally in the 1830s: his laws of electrolysis measure the charge needed to plate out a mole of a substance, and that constant — 96 485 C per mole — is nothing but $Q = It$ run on a plating tank. Electroplating, anodising and battery capacity testing all still bill in ampere-hours for exactly this reason.

Two errors are worth naming, and one convention deserves an apology. The first error is treating milliamp-hours as energy. They are charge; a 3000 mAh cell at 3.7 V holds $3 \times 3.7 = 11.1\ \text{Wh}$, and the same 3000 mAh at 1.2 V holds a third of that, so comparing two batteries by mAh alone tells you very little. The second is applying the equation to a current that is not constant — a motor's inrush, a switching supply's chopped input, or anything on AC, where over a full cycle the net charge transferred is zero even though the current is real all along. As for the convention: current is drawn flowing from plus to minus, while in a metal the electrons actually travel the other way. Benjamin Franklin guessed the sign in the 1750s, a century before anyone knew a charge carrier existed, and he guessed wrong. Nothing in the physics breaks — a deficit of negatives moving left is indistinguishable from positives moving right — but it is a historical accident, not a discovery, and it is worth knowing that it is one.

This formula falls straight out of the definitions. A volt is a joule per coulomb — the energy each unit of charge carries — and an ampere is a coulomb per second — how many units of charge arrive each second. Multiply them and the coulombs cancel, leaving joules per second: watts. A 1500 W space heater on a 120 V household circuit draws 1500/120 = 12.5 A, which is why it crowds a standard 15 A breaker and shouldn't share the circuit with much else.

Unlike the resistor-specific forms P = I²R and P = V²/R, this version works for any component — ohmic or not — including motors, LEDs, and batteries, because it comes from the definitions of voltage and current rather than from Ohm's law. Electric utilities meter exactly this product, accumulated over time, when they bill you for energy.

This is the heat form of electrical power, and the square is the whole point of it. Start from $P = VI$, substitute Ohm's law for the voltage across the resistance, $V = IR$, and you get $P = I^2 R$. Because the current appears twice, the heat does not track the load — it tracks the square of the load. Double the current and a conductor dissipates four times the heat; triple it and nine times. Nothing else in ordinary wiring punishes a modest overload so hard, and it is the reason a circuit that runs warm at rated current runs dangerously hot at 150% of it.

Take a 30 m branch circuit run in 12 AWG copper. The conductor is about 3.31 mm², copper's resistivity is $1.68 \times 10^{-8}$ Ω·m, so each metre is roughly 5.1 mΩ — and the current has to go out and come back, so the loop is 60 m and about 0.30 Ω. At 15 A the copper dissipates $15^2 \times 0.30 = 68\ \text{W}$, spread along the run, and drops $15 \times 0.30 = 4.6\ \text{V}$ of the supply before it ever reaches the load. Drop the current to 5 A and the loss falls to 7.5 W, not to a third: that is the square doing its work.

James Joule established the law in 1841 by immersing coils in water and measuring the temperature rise, which was also his route to the mechanical equivalent of heat and thence to the first law of thermodynamics. The same relation is the reason the grid transmits at hundreds of kilovolts. Delivered power is $VI$, so raising the voltage a hundredfold cuts the current a hundredfold for the same power, and cuts the line loss by ten thousand. Every transformer between a generating station and a house exists to move a fixed quantity of watts into a lower current, purely so that this equation returns a smaller number.

Three traps, in rising order of consequence. First, $R$ is the resistance of the thing dissipating the heat, and $I$ is the current through that same thing — mixing the conductor's resistance with the load's current is fine only because they are in series, and on a branched circuit it is not fine at all. Second, on AC the current must be RMS. A peak reading gives twice the power for a sine wave, and using it is one of the most common ways to double an answer without noticing. Third, and the one that bites in the field: this equation uses resistance, and on an AC circuit the opposition to current is impedance. A long run of steel-armoured cable, a coil, or a motor feeder has reactance as well as resistance, and the voltage drop computed from DC resistance alone will understate the real drop. The heat, though, still comes only from the resistive part — reactance stores energy and hands it back, so it moves voltage around without ever warming the copper.

This is the same power as $P = VI$, written for the situation you actually meet at a wall outlet: the voltage is fixed and the resistance is what you choose. Substitute $I = V/R$ into $P = VI$ and the current disappears, leaving $P = V^2/R$. Read it carefully, because the two variables behave in opposite directions. Power rises with the square of the voltage but falls inversely with resistance — so halving the resistance doubles the power, while doubling the voltage quadruples it.

A 1500 W kettle on a 120 V supply must therefore be built with $R = V^2/P = 14\,400/1500 = 9.6\ \Omega$ of element. The same element plugged into 240 V would try to deliver $57\,600/9.6 = 6000\ \text{W}$ — four times its rating — which is why travel appliances fail spectacularly rather than gradually. Run the arithmetic the other way for a 1000 W element on 230 V mains and you need about 53 Ω. Notice that the element's resistance is a design choice made to hit a wattage at one particular voltage; the wattage on the label is not a property of the element, it is a property of the element and the supply it was designed for.

The three power forms, $P = VI$, $P = I^2R$ and $P = V^2/R$ — are one equation seen from three sides, and picking the right one is mostly about which quantity is being held constant. Devices in parallel across a fixed supply are the $V^2/R$ case: every extra appliance you plug in adds power, because it adds a path, and lowering the effective resistance raises the total draw. Devices in series carrying a common current are the $I^2R$ case. Only $P = VI$ makes no assumption at all, which is why it is the one that still works for motors, LEDs and batteries where Ohm's law does not.

Where this goes wrong. The resistance to use is the resistance at operating temperature, and for anything that glows that is not what an ohmmeter reads on the bench. A tungsten filament's resistance climbs by roughly a factor of fifteen between room temperature and incandescence, so a 60 W lamp measuring 20 Ω cold does not draw 720 W — it draws a large inrush for a few milliseconds and then settles near 240 Ω. The second error is the mains one: use RMS voltage, always. Plugging the 170 V peak of a 120 V circuit into this formula doubles the answer. The third is applying it to a load that is not resistive. A motor at 120 V drawing 5 A is not a 24 Ω resistor; most of that opposition is reactance, which stores and returns energy rather than turning it into heat, and $V^2/R$ with an impedance in the denominator will overstate the watts by exactly the power factor.

Two resistors wired end to end sit on one path, and that single fact settles everything else. Charge has nowhere to go but forward, so the same current passes through both — it is not divided between them and it is not shared out according to size. Each resistor then takes its own voltage drop, $V_1 = I R_1$ and $V_2 = I R_2$, and the two drops must add up to whatever the supply provides. Divide that sum by the current they have in common and the resistances add: $R_t = R_1 + R_2$. This is not a rule to memorise. It is Ohm's law applied twice in a circuit that has only one loop.

Put a 47 Ω resistor in series with a 220 Ω resistor across a 12 V supply. The total is 267 Ω, so the current is $12/267 = 45\ \text{mA}$, and it is 45 mA at every point in the loop — before the first resistor, between them, and after the second. The 47 Ω part drops $0.045 \times 47 = 2.1\ \text{V}$; the 220 Ω part drops $0.045 \times 220 = 9.9\ \text{V}$; and 2.1 + 9.9 gives back the 12 V we started with. Adding the drops as a check costs nothing and catches most arithmetic errors on the spot.

The bookkeeping behind that check is Kirchhoff's voltage law, published by Gustav Kirchhoff in 1845 while he was still a student: go once around any closed loop and the voltage rises equal the voltage falls, because the loop returns you to the potential you started at. Every series result descends from it. The voltage divider is the same equation rearranged — each resistor claims the fraction $R_1/(R_t)$ of the supply — and that is how a potentiometer, a thermistor bridge and a sensor's biasing network all work. Solving this page backwards for one resistor is subtraction, $R_1 = R_t - R_2$, which is why the total must exceed the branch you already know.

The mistakes cluster in three places. The first is reaching for the wrong combination rule: series resistors add, parallel resistors combine as product over sum, and capacitors do exactly the reverse — series capacitors are the reciprocal case. Whenever you find yourself using product-over-sum on a series string, stop and ask which component you are holding. The second is assuming the larger resistor gets the larger current; it gets the larger voltage drop at the same current, and confusing those two makes a mess of any divider. The third only appears on AC: a coil or a capacitor in series with a resistor cannot be added arithmetically to it. Reactance is 90° out of phase with resistance, so a 30 Ω resistor in series with 40 Ω of reactance presents $\sqrt{30^2 + 40^2} = 50\ \Omega$, not 70. This page adds resistances, and resistances are what it will add — the quadrature sum belongs on the impedance pages.

Wired side by side, two resistors give the current two paths at once, so more current flows for the same voltage and the combination resists less than either branch alone. The tidy product-over-sum form is just 1/R_t = 1/R₁ + 1/R₂ rearranged — conductances, not resistances, are what add in parallel. Worked example: 100 Ω in parallel with 25 Ω gives (100 × 25)/(100 + 25) = 2500/125 = 20 Ω, comfortably below the smaller branch.

Two handy special cases: equal resistors in parallel halve (two 100 Ω resistors make 50 Ω), and a much smaller resistor dominates — 10 Ω in parallel with 10 kΩ is essentially 10 Ω. Household outlets are wired in parallel so every appliance sees full mains voltage. When solving for a branch, the other resistance must exceed the total, since the total is always the smallest value in the circuit.

Series resistors carry the same current, so each takes a share of the voltage in proportion to its resistance — the divider is Ohm's law applied twice. Twelve volts across 10 kΩ over 5 kΩ puts 12 × 5/15 = 4 V at the junction. It is the most-used circuit in electronics: bias networks, level shifters, the feedback string that sets a regulator's output, and every potentiometer, which is just a divider with a movable tap.

The trap is loading. This formula assumes nothing draws current from the tap; connect a load comparable to R₂ and the output sags, because the load parallels R₂. The rule of thumb is to make the divider current at least ten times the load current — but not so low-resistance that it wastes power. That trade-off is why high-impedance dividers pair with op-amp buffers, and why measuring a 1 MΩ divider with a cheap 1 MΩ-input meter reads badly wrong.

Parallel branches share a voltage, so current splits inversely with resistance — and that inversion is why the opposite resistor appears on top. Six amps entering a 4 Ω branch paralleled with a 2 Ω branch sends 6 × 2/6 = 2 A through the 4 Ω path and 4 A through the 2 Ω path: the easier road takes the most traffic.

Electricians meet this as the reason parallel feeders must be identical in size, length and material — a run even slightly shorter hogs current and overheats while its twin loafs, which is exactly what the NEC's paralleling rules exist to prevent. Benchtop electronics meets it as the ammeter shunt: send 99% of the current through a milliohm resistor and measure the small remainder. Note that for more than two branches the tidy two-resistor form fails; use conductance ratios instead.

Resistance behaves like a road: make it longer and the journey costs more; make it wider and traffic moves more easily. Doubling the length of a conductor puts twice as many collisions between the electron and the far end, so resistance doubles. Doubling the cross-sectional area gives the current twice as many parallel paths, so resistance halves. What is left over — the part that depends on the material rather than its shape — is the resistivity $\rho$, and $R = \rho L/A$ is nothing more than those three statements written together. Copper's $1.68 \times 10^{-8}$ Ω·m is the reason it wires the world; aluminium is about 1.6 times higher, silver marginally lower, and a good insulator is some twenty orders of magnitude higher than any of them.

A hundred metres of 2.5 mm² copper comes to $(1.68 \times 10^{-8} \times 100)/(2.5 \times 10^{-6}) = 0.67\ \Omega$. Carrying 16 A, that single conductor drops $16 \times 0.67 = 10.7\ \text{V}$ and dissipates 172 W along its length. Both figures matter, and both are why a long run is specified thicker than the load current alone would suggest — the ampacity keeps the cable from overheating, but it is the voltage drop that usually decides the size.

Resistivity is a bulk property of the material, and the reciprocal quantity, conductivity, is what materials people usually quote. The relation also underpins the strain gauge, whose resistance changes as stretching lengthens it and narrows it; the shunt, a precisely dimensioned low-resistance bar used to measure large currents; and the whole logic of high-voltage transmission, where a fixed $\rho L/A$ in the line means the only way to cut the $I^2R$ loss is to cut $I$.

Four errors, and the first one is nearly universal. When you are computing voltage drop on a circuit, $L$ is the total length of conductor the current traverses, which for an ordinary two-wire supply is twice the run distance — out along one conductor and back along the other. A 30 m run is 60 m of copper, and using 30 halves the answer. Second, resistivity is strongly temperature-dependent: copper rises about 0.39% per degree, so a conductor at 75 °C has roughly 22% more resistance than the 20 °C handbook figure, and electrical codes tabulate drop at operating temperature for exactly this reason. Third, watch the area units. Wire gauges are not areas — 12 AWG is 3.31 mm², 10 AWG is 5.26 mm² — and a diameter must be squared and taken through $\pi d^2/4$ before it can go in the denominator. Fourth, this is a DC resistance. On AC the current crowds toward the surface of the conductor, and above a few hundred hertz, or in large conductors at mains frequency, that skin effect raises the effective resistance measurably; and a long AC run has reactance as well, so the drop you actually measure will exceed what $\rho L/A$ alone predicts.

Metals conduct worse when hot: heat jostles the lattice and scatters the electrons. Copper's coefficient is about 0.00393 per kelvin referenced to 20 °C, aluminium's about 0.00403, so a 10 Ω copper winding at 20 °C measures 10 × (1 + 0.00393 × 55) = 12.16 Ω at 75 °C — a 22% rise. Because a kelvin and a Celsius degree are the same size, only the temperature difference matters; if your α is quoted per °F, choose that unit and the solver handles it.

This relation is quietly one of the most useful in the trade. It turns a winding's cold and hot resistance into a motor's average winding temperature, the basis of the standard heat-run test — no thermocouple can reach the middle of a coil, but its resistance always can. It is also why cable ampacity tables assume a conductor temperature, why voltage-drop constants differ between 20 °C and 75 °C, and why platinum's clean, repeatable version of this curve makes the RTD the workhorse of industrial temperature measurement.

Gustav Kirchhoff published this in 1845 while still a student, and it is nothing more than energy conservation dressed for a circuit: carry a charge all the way around a loop and back to where it started, and it must return to the same potential. Every rise has to be paid for by drops. So a 24 V supply feeding three series elements that drop 5.5 V and 12.1 V must be dropping the remaining 6.4 V across the third, whatever that third element happens to be.

The trap is sign, not arithmetic. Two sources fighting each other in a loop — a battery charging another battery, a generator paralleled slightly out of phase — subtract rather than add, and a loop containing a reversed source needs $V_s = V_1 + V_2 - V_3$. This solver assumes all three elements drop in the same direction, so enter an opposing source as a negative number and the algebra still comes out right.

What people rarely notice is that KVL is an approximation, and a good one only because circuits are small. The law follows from the electric field being conservative, which stops being true the moment a changing magnetic field threads your loop. Move a scope probe's ground lead near a transformer and the "same" two points read different voltages — not a bad measurement, but a real induced EMF in the loop the probe lead just formed.

The node rule is the charge-conservation twin of the loop rule: charge cannot pile up at a junction, so whatever flows in must flow out. Ten amps arriving at a splice feeding three branches, two of which are carrying 2.5 A and 3.5 A, leaves exactly 4 A for the third. No knowledge of the loads is needed, which is what makes it so useful — it is a constraint you get for free before any analysis begins.

The classic mistake is treating current as something a source "sends" to a load rather than something that circulates. A branch drawing less current does not leave the surplus somewhere; the source simply produces less. That is why removing a lamp from a parallel lighting circuit dims nothing else, and why removing one from a series string kills everything: the series case has one node current, so it is KCL that says they all must be equal.

In the field this is the arithmetic behind every panel schedule and every neutral calculation. It is also the basis of nodal analysis, the method every circuit simulator actually runs — SPICE writes one KCL equation per node, solves the matrix, and never touches KVL at all. And the counter-intuitive part for electricians: on a balanced three-phase four-wire system, three currents of 100 A each sum to zero in the neutral, because KCL sums phasors, not magnitudes.

Léon Thévenin's 1883 theorem says something extravagant: any tangle of sources and resistors, seen from two terminals, behaves exactly like one voltage source behind one resistor. You never have to know what is inside. And you can measure both numbers with a voltmeter and one known resistor — read the open-circuit voltage, hang the load, read the sagged voltage, and the internal resistance falls out of the droop.

Work an example the long way and the formula becomes obvious. A 9 V cell reads 9.00 V open and 8.10 V across 100 Ω. That load is drawing $8.10/100 = 0.081$ A, and the 0.90 V that went missing was dropped inside the cell, so $R_{Th} = 0.90/0.081 = 11.1$ Ω. A fresh alkaline should be well under 2 Ω, so this one is tired — which is exactly how a battery tester works.

The mistake to avoid is shorting the terminals to find the internal resistance directly. On a cell that is merely rude; on a car battery or a lithium pack it is a welding accident. Use a load that draws a sensible current and let the droop tell you. One caution on the theorem itself: it holds for linear circuits only. Put a diode, a lamp filament or a switching supply behind the terminals and R_Th becomes a number that changes with the load you used to measure it.

Edward Norton at Bell Labs described the dual of Thévenin's result in 1926: the same two terminals can equally be modelled as a current source with a resistor in parallel. The two models are interchangeable, and the bridge between them is Ohm's law — a 12 V source behind 4 Ω is a 3 A source across 4 Ω, and no measurement made at the terminals can tell them apart.

Which one you pick is pure convenience. Norton form makes parallel branches trivial, because current sources in parallel simply add, so it is the natural language of nodal analysis and of transistor models. Thevenin form makes series chains trivial. Experienced analysts flip back and forth mid-problem — that flip is called a source transformation, and it is often the single step that turns an ugly network into an obvious one.

The number I_N is the current the source would deliver into a dead short, which is worth respecting. A 12 V car battery with 5 mΩ internal resistance has a Norton current of 2400 A, and that is why a dropped spanner across the terminals vaporises. Note also that both models are fictions from the terminals only: a Norton equivalent burns power in its resistor even at no load, so it tells you nothing true about what the real box is dissipating inside.

Sweep a load resistance from zero to infinity across a source and the power it receives rises, peaks, and falls again. Too small and the voltage collapses; too large and the current does. The peak sits exactly where $R_L = R_{Th}$, and there the load takes $V_{Th}^2/4R_{Th}$. Twenty volts behind 5 Ω, matched by a 5 Ω load: the loop carries 2 A, and the load absorbs $2^2 \times 5 = 20$ W.

Notice what else happened in that example — the source dissipated 20 W too. Matched transfer is 50% efficient, always, by construction. That is the classic misreading of this theorem: it maximises power delivered, not power saved. No utility matches its generators to the grid; a power system deliberately runs with source impedance far below load impedance so efficiency approaches 100% and voltage stays stiff. Matching belongs where the signal matters and the watts do not.

So it lives in RF and audio and instrumentation: antennas into 50 Ω feedlines, transmission lines terminated to stop reflections, a solar panel held at its maximum-power point by a tracking converter. And the AC version wants the load impedance to be the complex conjugate of the source, so a source that looks inductive must be met with capacitance — reactance cancelled first, resistance matched second.

Some networks refuse to be reduced. A Wheatstone bridge has no two resistors in plain series and no two in plain parallel, so the usual tools stall. Arthur Kennelly's 1899 transformation is the way out: swap any triangle of three resistors for an electrically identical star, and the series-parallel structure reappears. The arm of the star at node A is the product of the two triangle legs that touch A, divided by the sum of all three.

The balanced case is worth memorising because it is so clean: three equal 30 Ω legs become three equal 10 Ω arms, $R_Y = R_\Delta/3$. For the unbalanced case take a 10/20/30 Ω delta — the sum is 60, so the arm at A, touching the 10 and the 30, is $10 \times 30/60 = 5$ Ω. Every star arm always comes out smaller than either delta leg it touches, which is a useful sanity check on your own arithmetic.

The trap is bookkeeping: it is dangerously easy to pair an arm with the wrong two legs. Label the nodes on the drawing before you start, and remember the rule in words — the arm at a node uses the two legs meeting at that node, and the leg opposite a node never appears in the numerator. Note also that this is a resistance identity, but it works unchanged for complex impedances, which is how three-phase engineers convert a delta-connected load to its wye equivalent before applying per-phase analysis.

Going back the other way, the delta leg between two nodes is the sum of the three pairwise products of the star arms, divided by the arm at the node the leg does not touch. A balanced 10 Ω star becomes a 30 Ω delta, the mirror of $R_\Delta = 3R_Y$. Unbalanced: arms of 5, 10 and 20 Ω give pairwise products 50, 200 and 100, summing to 350, so the leg opposite the 20 Ω arm is $350/20 = 17.5$ Ω.

That opposite-arm divisor is where people go wrong. The numerator is the same for all three legs — compute it once — and only the denominator changes, so the leg opposite the smallest arm comes out largest. Another built-in check: every delta leg must exceed the sum of the two star arms it spans, because the direct path has to imitate both the series path and the roundabout one. Here 17.5 Ω comfortably beats $5 + 10 = 15$ Ω.

Practically this direction shows up when a network has a star that blocks reduction, and in three-phase work when a wye-connected load has to be restated as delta for a source that is delta-connected. It is also the reason a delta-connected motor winding draws three times the current of the same windings in wye at the same line voltage, which is the whole basis of wye-delta starting.

An LED is not a resistor — its current climbs almost vertically once forward voltage is reached, so a tiny voltage excess becomes a huge, fatal current. The resistor is what makes the circuit predictable: it eats the difference between supply and forward voltage, and Ohm's law sets the current. Running a red LED (V_f ≈ 2.1 V) at 20 mA from 5 V needs (5 − 2.1)/0.02 = 145 Ω, so you fit a 150 Ω standard value.

Two things beginners miss. First, check the resistor's power: here it dissipates (5 − 2.1) × 0.02 ≈ 58 mW, comfortable for a quarter-watt part, but a 12 V supply on the same LED wastes 198 mW and argues for a proper constant-current driver. Second, forward voltage depends on colour and on temperature — roughly 1.8–2.2 V for red, 3.0–3.4 V for blue and white — and it drops about 2 mV per °C as the die heats, which is precisely why high-power LEDs must never be run from a resistor alone.

Capacitance is charge-storing capacity: how many coulombs a capacitor soaks up for every volt you apply across it. The unit, the farad, honors Michael Faraday — and it is a giant. A full farad would have been a bench-filling curiosity for most of electronics history, so practical parts are marked in microfarads, nanofarads, and picofarads; only modern supercapacitors reach whole farads. Doubling the applied voltage doubles the stored charge, but C itself stays fixed — it depends only on geometry and the insulating material between the plates.

Worked example: a 100 µF capacitor charged to 12 V holds Q = CV = 100×10⁻⁶ × 12 = 1.2 mC of charge — about 7.5×10¹⁵ electrons parked on one plate and missing from the other. This defining relation applies to every capacitor, from the tuning capacitor in a radio to the DRAM cell storing one bit in your computer.

Why the half? Charging a capacitor is like stretching a spring: the first coulomb slides on easily, but every later coulomb must be pushed against the voltage the earlier ones built up. The voltage ramps linearly from 0 to V as charge accumulates, so the average push is V/2, and the total work is Q×V/2 = ½CV² — the triangular area under the Q–V line. Equivalent forms E = Q²/(2C) and E = QV/2 follow directly from C = Q/V.

Worked example: a camera-flash capacitor of 1000 µF charged to 300 V stores E = ½ × 0.001 × 300² = 45 J, dumped through the xenon tube in about a millisecond — a burst of tens of kilowatts from a pocket battery. The same math sizes defibrillators and grid-scale supercapacitor banks. Solving for V takes the positive square root, since the formula gives the voltage magnitude.

Capacitors in series combine the way resistors in parallel do, and the reason is worth following rather than memorising. Wire two capacitors end to end and the middle section — the bottom plate of the first joined to the top plate of the second — is isolated from everything else. Whatever charge leaves that plate must arrive at this one, so every capacitor in a series string carries the identical charge $Q$. Each then holds its own voltage, $V_1 = Q/C_1$ and $V_2 = Q/C_2$, and those add to the applied voltage. Divide through by the common $Q$ and you get $1/C_t = 1/C_1 + 1/C_2$, which for two parts is the product-over-sum form on this page.

Two equal 10 µF capacitors in series make 5 µF, not 20. A 10 µF with a 1 µF gives $(10 \times 1)/(10 + 1) = 0.91\ \mu\text{F}$ — barely less than the small one, because the small capacitor needs the most volts to accept the shared charge and therefore dominates the total. That is the general behaviour: the result is always smaller than the smallest member, and a large capacitor in series with a small one is very nearly the small one alone. Voltage divides in inverse proportion to capacitance, so in that pair, 100 V applied puts about 91 V across the 1 µF part and only 9 V across the 10 µF.

The geometric picture explains why it must be so. Capacitance for parallel plates is $C = \varepsilon A/d$, and stacking capacitors in series is effectively increasing $d$ — putting more insulating distance between the outermost plates — which lowers capacitance. Stacking them in parallel increases $A$, which raises it. That is also the practical use of a series string: the working voltage adds even as the capacitance falls, so two 400 V parts in series will stand 800 V, and high-voltage DC links are routinely built this way.

The famous error is the one this page exists to prevent: series capacitors are not added. The mirror-image mistake — treating parallel capacitors with product-over-sum — is just as common. If you can only remember one anchor, remember that a series connection always makes capacitance worse and parallel always makes it better, then check your answer against that. The second error is far more hazardous and almost never taught. The voltage division above holds only at DC in an ideal world; real capacitors, and electrolytics in particular, have leakage currents that differ from part to part, and over minutes the string will redistribute itself until one capacitor carries most of the applied voltage and fails. Series electrolytic banks therefore need balancing resistors across each part, sized so their current is many times the worst-case leakage. Finally, do not assume two parts marked 400 V in series give a comfortable 800 V rating — they give it only while they share, and sharing is what the balancing resistors are for.

Capacitors side by side across the same two nodes see the same voltage, and each stores the charge that voltage buys it: $Q_1 = C_1 V$ and $Q_2 = C_2 V$. The total charge held is the sum, so the total capacitance — charge per volt — is also the sum, $C_t = C_1 + C_2$. Physically the two components have simply become one larger capacitor: parallel connection joins their plates, and joining plates adds plate area. Since capacitance for a parallel-plate part is $C = \varepsilon A / d$, more area at the same separation means proportionally more capacitance, and the arithmetic could hardly come out any other way.

A 4.7 µF and a 2.2 µF part in parallel behave as 6.9 µF. Three 1000 µF electrolytics across a power-supply rail behave as 3000 µF. The additions are exact and there is no upper limit beyond what will fit on the board, which is why designers reach for parallel banks to hit values no manufacturer stocks — 4.7 with 2.2 to make 6.9 — and why a supply that needs 10 000 µF of bulk storage is built from several parts rather than one.

There is a second reason for the bank that has nothing to do with capacitance. Every real capacitor carries series resistance and series inductance in its own leads and foil, and putting parts in parallel divides both. Ten capacitors in parallel have a tenth of the equivalent series resistance, so they run cooler under ripple current and hold the rail steadier under a sudden load step. This is also why you find a 100 nF ceramic beside a 470 µF electrolytic beside a chip: the electrolytic supplies the bulk energy, the ceramic — low inductance, fast — supplies the first microsecond, and in parallel they cover a band neither covers alone.

The one big mistake is the combination rule itself, in mirror image: parallel capacitors add, series capacitors do not. Reaching for product-over-sum on a parallel pair gives an answer smaller than either part, which is your signal that the wrong rule has been applied — parallel connection can only ever increase capacitance. Two subtler points matter in practice. Ripple current in a parallel bank does not divide by capacitance; it divides by impedance, so a bank of mismatched parts overloads whichever has the lowest ESR while the others idle, and a bank should be built from identical parts. And the big-plus-small pairing described above can, at some megahertz frequency, form a parallel resonance between the small capacitor's capacitance and the large one's lead inductance, producing an impedance peak exactly where you wanted a low one. At mains and audio frequencies none of this matters and the addition is all you need.

Ohms multiplied by farads give seconds, and that is not a coincidence. A farad is a coulomb per volt — how much charge a volt buys — and an ohm is a volt per ampere, an ampere being a coulomb per second. Multiply and the coulombs and volts cancel, leaving time. The physical reading is just as direct: $C$ says how much charge the capacitor must accumulate to reach a given voltage, $R$ says how slowly that charge is allowed to arrive, and $\tau = RC$ is the ratio of the job to the rate at which it can be done.

A 10 kΩ resistor with a 100 µF capacitor gives $\tau = 10\,000 \times 100 \times 10^{-6} = 1\ \text{s}$. Charging from a supply, the capacitor covers 63.2% of the remaining gap in the first second, 86.5% by two seconds, 95% by three, and 99.3% by five — which is why 5τ is the working rule for "finished". The step sizes shrink because the driving voltage is the difference between the supply and what the capacitor has already reached, so as the gap closes the current that closes it falls in proportion. That self-limiting behaviour is what makes the curve exponential rather than a straight ramp.

The exponential is not asserted, it falls out of one line of calculus: the current through the resistor is $(V_s - V_C)/R$ and it must equal $C\,dV_C/dt$, giving $RC\,dV_C/dt = V_s - V_C$, whose solution is $V_C = V_s(1 - e^{-t/RC})$. Set $t = \tau$ and the bracket is $1 - 1/e = 0.632$. The inductive version on this site, $\tau = L/R$, is the same equation with the roles of the storage element and the resistor exchanged, and the RC discharge page is this one running downhill. Timing circuits, debounce networks, RC snubbers and the anti-aliasing filter in front of an ADC are all this constant chosen deliberately.

The unit slip is the most common failure by a wide margin. Capacitance is almost never quoted in farads — microfarads, nanofarads and picofarads are what appear on parts — and 10 kΩ with "100" entered as farads rather than microfarads produces a time constant of $10^6$ seconds, or eleven days. If an answer looks absurd, check the prefix first. The second mistake is misreading what τ measures: it is not the time to charge, and the capacitor never mathematically arrives at all. The third is using the wrong resistance. What matters is the total resistance in the charging path, including the source's internal resistance and anything you have connected to watch it — a 10 MΩ oscilloscope probe across a 1 µF capacitor imposes its own 10-second discharge whether you wanted one or not. And a circuit's charge and discharge paths are often different, through a diode or a separate bleeder, in which case it has two time constants and only one of them is $RC$ as drawn.

A capacitor discharging through a resistor loses the same fraction of what remains in every equal interval, not the same number of volts. The reason is a short feedback loop: the voltage on the capacitor is what drives current through the resistor, that current is what removes charge, and removing charge is what lowers the voltage. As the voltage falls the current falls with it, so the discharge slows exactly in step with its own progress. Any quantity whose rate of decrease is proportional to itself decays as $e^{-t/\tau}$, and here $\tau = RC$. After one time constant 36.8% of the original voltage is left, after two 13.5%, after three 5%, after five 0.7%.

Real numbers make the point better than percentages. A camera flash or a switch-mode supply may hold 470 µF at 400 V, bled off through a 1 MΩ resistor. That gives $\tau = 470\ \text{s}$, close to eight minutes. Five minutes after the unit is unplugged the capacitor is still at $400\,e^{-300/470} = 211\ \text{V}$; after a full ten minutes it is at 111 V, which will still hurt you. Inverting the relation gives the useful form: the time to fall to a chosen voltage is $t = \tau \ln(V_0/V)$, so reaching a nominally safe 50 V from 400 V takes $470 \times \ln 8 = 977\ \text{s}$, sixteen minutes.

The mathematics is identical to radioactive decay, and the half-life language transfers directly: the voltage halves every $\tau \ln 2 = 0.693\tau$, regardless of where you start counting. That is often the easier mental model — six and a bit half-lives to reach 1%. The same exponential governs the RC charging curve, thermal cooling under Newton's law, and the settling of a pressure transient in a pipe; whenever a store discharges through a restriction, this is the shape you get.

The dangerous misunderstandings here are all about "empty". Mathematically the capacitor never reaches zero, so any statement that it is discharged is a statement about a threshold someone chose. Worse, a large capacitor that has been shorted out and released will climb back up on its own — often to tens of volts — as charge trapped in the dielectric relaxes out of it. That effect is called dielectric absorption, or soakage, and it is why service procedures call for a bleeder resistor left in place rather than a screwdriver across the terminals, and why you measure before you touch rather than assuming. Two smaller traps: the exponential assumes the only discharge path is $R$, so a real capacitor's own leakage and any parallel load shorten τ below the value you calculated; and when solving this page for $t$ or $\tau$ the logarithm demands $V$ below $V_0$, because a discharging capacitor only ever loses voltage.

Close a switch on a coil and the current does not jump to its final value — it climbs. The inductor generates a back-EMF proportional to how fast the current is changing, and that back-EMF eats into the voltage available to drive the current, so the rise is self-limiting in exactly the way an RC charge is. The current covers 63.2% of its remaining gap in each interval $\tau = L/R$, and is within 1% of its final $V/R$ after 5τ. The units check out because a henry is a volt-second per ampere and an ohm is a volt per ampere: divide and the volts and amperes cancel, leaving seconds.

A 100 mH relay coil wound to 50 Ω has $\tau = 0.1/50 = 2\ \text{ms}$, so the current is essentially established after about 10 ms and the relay clicks in on that timescale — which is why mechanical relays are measured in milliseconds while transistors are measured in nanoseconds. Reduce the resistance and the coil gets slower, which is the counterintuitive part: τ is $L/R$, so halving $R$ doubles the time constant even though it doubles the final current. A superconducting loop has $R = 0$ and a time constant that is formally infinite, which is precisely why persistent-mode magnets hold their current for years.

The same constant governs the decay when the supply is removed, and that case is where the practical importance lies. The current will not stop, so it will find a path; if the only path is the opening switch contact, the inductor drives the voltage up until the air breaks down and the current continues as an arc. A flyback diode across the coil gives it a legal route instead. Note what the diode does to the timing: the decay's τ is set by the resistance in whatever loop the current ends up in, so a plain diode across the coil leaves only the winding resistance and the current takes a long time to die — which delays relay dropout — while adding a resistor or a Zener in that path shortens τ at the cost of a higher clamp voltage. Designers trade those two off deliberately.

Get the resistance right and the order of the division right. The $R$ that matters is every ohm in the current's loop — the winding's own DC resistance, the source's internal resistance, the sense resistor, the switch — and using only the external resistor when the coil itself has more resistance than it will give a τ that is badly wrong. Second, the arrangement is $L/R$, not $RL$; the RC constant is a product and the RL constant is a quotient, and swapping them is the most common slip on this page. A quick sanity check: more inductance should mean slower, more resistance should mean faster. Third, inductance is often given in millihenries or microhenries and must be converted before dividing. And as with any inductor equation, an iron or ferrite core makes $L$ a function of current, so a coil approaching saturation has a shrinking time constant and its current rises faster than the calculation predicts.

An inductor stores energy in the magnetic field it creates, and the half in $E = \tfrac{1}{2}LI^2$ has the same origin as the half in a capacitor's $\tfrac{1}{2}CV^2$ or a spring's $\tfrac{1}{2}kx^2$. Building the current is not free: the coil opposes any change in current with a back-EMF of $L\,di/dt$, so the source must push against it the whole way up. That opposition grows in step with the current, from nothing at the start to its full value at the end, so the average is half the maximum and the work done is half of what a naive product would give. Integrate $P = vi = Li\,(di/dt)$ from zero to $I$ and the half appears formally.

A 100 mH choke carrying 3 A holds $0.5 \times 0.1 \times 9 = 0.45\ \text{J}$. That sounds negligible until you ask how fast it can be released: interrupt that current in 10 µs and you are dissipating 45 kW while it lasts. Because the energy goes as the square of the current, a coil at twice the current holds four times as much — an important sizing fact, since it means an inductor's usable rating is set by the current it can carry without saturating far more than by its inductance.

The release is what makes inductors interesting. Open a switch on an energised coil and the current cannot stop instantly, because stopping it instantly would require infinite $di/dt$; instead the collapsing field drives the terminal voltage to whatever value is needed to keep charge moving — through the air across the opening contacts, if that is the only path left. A car's ignition coil is this effect engineered on purpose, turning 12 V into a 30 kV spark by breaking a primary current. A switch-mode power supply does the same thing gently and tens of thousands of times a second, filling an inductor from the input and emptying it into the output, which is how a buck converter changes DC voltage with efficiency a resistor could never approach.

The mistakes here are mostly about symmetry that is not really there. A capacitor holds its energy at rest and will still bite you next week; an inductor's store exists only while the current flows, and cutting the current does not park the energy, it forces it out somewhere in the next few microseconds. Design for where it goes — a flyback diode, a snubber, a clamp — or the switch contacts and the transistor will volunteer. Second, $L$ in an iron-cored or ferrite-cored coil is not constant with current: as the core approaches saturation the inductance falls, sometimes by half, so $\tfrac{1}{2}LI^2$ computed with the datasheet's small-signal inductance overstates the energy a saturating part actually holds, and the current rises much faster than expected. Third, on AC the current to use is the peak, not the RMS, because it is the instantaneous current that sets the instantaneous stored energy. And do not read the stored energy as a loss — it is returned to the circuit, unlike $I^2R$ heat, which is gone.

An alternating voltage spends most of its time somewhere below its peak and averages zero over a cycle, so "the" voltage of an AC supply needs defining before it means anything. The definition that earns its keep is the heating-equivalent one: the DC voltage that would deliver the same power to a resistor. Since power goes as $V^2$, you square the waveform, take the mean of the square over a cycle, and take the root of that — root-mean-square, done in that order. For a sine wave the mean of $\sin^2$ over a cycle is exactly $1/2$, so the RMS value is $V_{peak}/\sqrt{2}$, about 0.707 of the peak. The $\sqrt{2}$ is not a convention or a fudge; it is the square root of that one-half, and it belongs to the sine and to nothing else.

North American mains at 120 V RMS actually swings to $120 \times 1.414 = 170\ \text{V}$ either side of neutral, 340 V peak to peak. A 230 V European supply reaches 325 V peak, 650 V peak to peak. A 24 V control transformer delivers 34 V peaks. Those peak numbers are the ones that matter when you choose an insulation rating, a rectifier's reverse voltage, or the working voltage of a filter capacitor — a 200 V capacitor across a 120 V circuit is not a comfortable margin, it is already an underrating.

The whole reason for the convention is that it makes the DC power formulas keep working. Feed 120 V RMS into $P = V^2/R$ and the answer is the true average power in watts, no correction needed, which is precisely the property RMS was constructed to have. That is why meters, nameplates and every distribution standard quote RMS, and it is why the RMS value is what a thermal instrument naturally measures — an old thermocouple-type meter reads the heating effect directly and is right by construction.

Now the traps, and there are several worth knowing. The $\sqrt{2}$ applies to a sine wave only. A square wave's RMS equals its peak; a triangle wave's is peak over $\sqrt{3}$; a rectified or chopped waveform is something else again. Second, and this catches working electricians: an ordinary averaging multimeter does not measure RMS at all. It rectifies, takes the average, and multiplies by 1.11 — a factor that assumes a sine. On the distorted current drawn by LED drivers, variable-frequency drives and switch-mode supplies, that meter can read 20 to 40% low, and only an instrument marked true RMS will tell you the truth. Third, keep peak and peak-to-peak straight: an oscilloscope shows peak-to-peak, which is twice the peak and $2\sqrt{2}$ times the RMS. Fourth, RMS current squared times resistance gives real heating, but RMS volts times RMS amps gives volt-amperes, not watts — the power factor still has to be applied. And never mix them within one calculation: peak volts with RMS amps produces an answer that is wrong by 41% and looks entirely reasonable.

An inductor opposes change in current, not current itself, so how hard it opposes depends on how fast you are asking the current to change. On a sine wave of frequency $f$ the current has to reverse $2f$ times a second, and the steepness of that reversal rises in proportion to the frequency — hence $X_L = 2\pi f L$, climbing linearly. The $2\pi$ is there because a full cycle is $2\pi$ radians, and the group $2\pi f$ is angular frequency $\omega$, which is why the relation is more compactly written $X_L = \omega L$. At DC, where nothing changes, the reactance is zero and the coil is just its own winding resistance.

A 10 mH coil offers $2\pi \times 60 \times 0.01 = 3.8\ \Omega$ at 60 Hz mains, 63 Ω at 1 kHz, and 628 Ω at 10 kHz. That strong frequency preference makes an inductor a natural low-pass element: put it in series with a load and the low frequencies get through while the high ones are held back. A loudspeaker crossover does exactly this, an inductor in series with the woofer passing bass and blocking treble, while a capacitor does the complementary job for the tweeter. The same principle is at work in a mains filter choke and in the smoothing inductor of a power supply.

Reactance is what an inductor contributes to the circuit's impedance, $Z = R + jX_L$, where the $j$ records that the voltage across an inductor leads its current by 90°. This quarter-cycle lag is the physical heart of the matter: current lags voltage in a coil because the coil resists getting started, and it is the reason an inductive motor load drags a plant's power factor down and needs correcting with capacitors. On the site's LC resonance page the inductive and capacitive reactances become equal and, because they are opposite in sign, cancel — the whole subject of tuning is contained in that cancellation.

Reactance is measured in ohms, and that is the single most misleading fact about it. It is not resistance and it does not behave like resistance in three important ways. First, it dissipates no power: the energy goes into the magnetic field on one quarter-cycle and comes back out on the next, so a pure reactance heats nothing, and a wattmeter across it reads zero. Second, it does not add arithmetically to resistance. A coil with 30 Ω of winding resistance and 40 Ω of reactance presents $\sqrt{30^2 + 40^2} = 50\ \Omega$, never 70, because the two are 90° apart and combine in quadrature. Third, it is frequency-specific, so a single number is meaningless without the frequency it was computed at — and on a supply carrying harmonics, the fifth harmonic sees five times the reactance the fundamental does, which is why harmonic currents can produce voltage distortion far out of proportion to their size. Finally, this page gives the reactance alone; to find the current from a supply voltage you need the full impedance, resistance included.

No charge crosses the gap between a capacitor's plates, yet a capacitor plainly passes alternating current. What actually happens is that charge piles onto one plate and off the other, and if the supply reverses before much has accumulated, the opposing voltage never gets large and current keeps flowing freely. The faster the alternation, the less charge accumulates per half-cycle and the less the capacitor pushes back — so its opposition falls as frequency rises, $X_C = 1/(2\pi f C)$. This is the exact mirror of the inductor, which opposes more as frequency rises, and the two mirror each other in sign as well as in shape.

A 1 µF capacitor presents $1/(2\pi \times 60 \times 10^{-6}) = 2.65\ \text{k}\Omega$ at 60 Hz, 159 Ω at 1 kHz, and 0.16 Ω at 1 MHz — from a substantial obstacle to a near short circuit across that range. That is the whole basis of bypassing: a 100 nF capacitor from a supply rail to ground is 26 kΩ at mains frequency, so it draws nothing, but a fraction of an ohm at the megahertz where a digital chip's switching noise lives, where it acts as a local reservoir. In a loudspeaker crossover the same behaviour puts a capacitor in series with the tweeter, blocking bass and passing treble.

At DC the relation gives infinity, and that is the honest answer: once charged, a capacitor is an open circuit, which is why this page refuses $f = 0$ rather than returning a number. That property is as useful as the frequency dependence — a coupling capacitor passes an audio signal while blocking the DC bias on either side of it, and every AC-coupled amplifier stage depends on it. The complementary behaviours of $X_L$ and $X_C$ meet on the LC resonance page, where at one frequency they are equal and, being opposite in sign, cancel entirely.

The first trap is the one shared with inductive reactance: ohms are not resistance. A capacitor dissipates no power — the current leads the voltage by 90°, energy flows in and back out each quarter-cycle, and a wattmeter reads zero. It follows that reactance never adds arithmetically to resistance: 30 Ω of resistance in series with 40 Ω of capacitive reactance is 50 Ω of impedance. The second trap is specific to this page. Inductive and capacitive reactances are opposite in sign, $+jX_L$ against $-jX_C$ — so in a series circuit they subtract, and the net is $X = X_L - X_C$ before you take the quadrature sum with $R$. Add them and you will get an answer that is not merely wrong but wrong in the direction that hides resonance completely. Third, a real capacitor is not pure: its equivalent series resistance does dissipate, which is what heats an electrolytic under ripple current and eventually kills it, and above its self-resonant frequency a capacitor's own lead inductance takes over and it starts behaving inductively — a bypass capacitor used above that point is doing nothing you intended.

A resistor's voltage is in phase with its current, a reactance's is a quarter cycle away, so their oppositions add as the legs of a right triangle: 40 Ω of resistance with 30 Ω of reactance gives 50 Ω, not 70 Ω. This is the everyday impedance calculation — the winding resistance and inductance of a relay coil, a contactor, a transformer primary, a loudspeaker crossover leg.

The same triangle hands you the phase angle, φ = arctan(X/R), and hence the power factor cos φ = R/Z. That is worth remembering when you meet an inductive load with no nameplate: measure the DC resistance, measure the AC current at a known voltage to get Z, and the power factor falls out of the ratio. Rough numbers are enough to tell a mostly resistive load from a mostly reactive one.

Reactances are 90° out of phase with resistance and 180° out of phase with each other, so they subtract first and then join R by Pythagoras. With R = 30 Ω, X_L = 80 Ω and X_C = 40 Ω, the net reactance is 40 Ω and $Z = \sqrt{30^2 + 40^2} = 50$ Ω. Note that reactances can each be far larger than the total impedance — at resonance they cancel exactly and Z collapses to R alone, which is how a tuned circuit picks one station out of the air.

Because the square root discards sign, solving for a reactance has two answers; this brain returns the inductive-dominant branch (X_L above X_C), so subtract the root instead if you know the circuit is capacitive. The practical warning: never add reactances arithmetically to resistance. A 300 Ω coil in series with 400 Ω of resistance is 500 Ω, not 700 Ω, and a meter that reads otherwise is measuring DC.

Power factor and phase angle are the same fact in two languages. A PF of 0.8 means the current waveform is 36.87° out of step with the voltage — the 3-4-5 triangle again, since cos 36.87° = 0.8 and sin 36.87° = 0.6. Once you have φ you can get the reactive burden directly as Q = P tan φ, which is how correction tables are built.

The number alone does not say which way: motors and transformers pull lagging current (current behind voltage), while capacitor banks and lightly loaded long cables push leading current. Both give the same cos φ, so meters mark them separately, and it matters — leading power factor on a weak system raises voltage rather than lowering it. Note this calculator works in radians internally and displays degrees.

Connect a charged capacitor across an inductor and the energy does not simply drain away — it sloshes. The capacitor discharges into the coil, building a magnetic field; when the capacitor is empty the field is at its peak, and a collapsing field keeps the current going, so it charges the capacitor back up the other way round. The cycle repeats at one natural frequency. It is the electrical twin of a mass bouncing on a spring, with inductance playing the part of mass — the reluctance to change velocity, or current — and $1/C$ playing stiffness. The frequency comes from setting the two reactances equal: $2\pi f L = 1/(2\pi f C)$, solve for $f$, and you have $f = 1/(2\pi\sqrt{LC})$.

A 100 µH coil with a 250 pF capacitor rings at $1/(2\pi\sqrt{10^{-4} \times 2.5 \times 10^{-10}}) = 1.01\ \text{MHz}$, the middle of the AM broadcast band. The square root makes the circuit reassuringly insensitive: to double the frequency you must quarter the product $LC$, so a 10% error in a capacitor moves the resonance by only 5%. Turning the tuning knob on an old radio physically rotated the vanes of a variable capacitor, sliding this frequency across the dial, and the whole receiver amounted to that one circuit picking one station out of the air.

Every radio transmitter, every RFID tag, every induction hob and every crystal oscillator descends from this arrangement. Heinrich Hertz used a spark-excited LC circuit in 1887 to generate and detect the first deliberate radio waves, confirming Maxwell's prediction; Marconi turned the same circuit into a business within a decade. The relation also explains why antennas have a length: a resonant antenna is an LC circuit whose inductance and capacitance are distributed along the conductor rather than lumped into parts.

Four things go wrong. The commonest by far is the factor of $2\pi$: $f$ is in hertz, $\omega = 2\pi f = 1/\sqrt{LC}$ is in radians per second, and the two differ by 6.28. Half the confusion in filter design traces to a formula quoted in one and used as the other. Second, prefixes — microhenries and picofarads must both be converted before multiplying, and an error here moves the answer by orders of magnitude, not percent. Third, this is the undamped natural frequency. Real circuits have resistance, which shifts the actual resonance slightly lower and, more importantly, sets how sharp it is: the quality factor $Q = (1/R)\sqrt{L/C}$ governs the bandwidth, and a lossy coil gives a broad, useless peak at the right frequency. Fourth, series and parallel LC circuits share this formula and behave in opposite ways at it. A series LC becomes a near short circuit at resonance, with the voltage across each component rising to $Q$ times the supply — a genuine hazard, since a modest input can put hundreds of volts across a capacitor. A parallel LC becomes a near open circuit. Choosing the wrong one gives a circuit that does the exact opposite of what was intended at precisely the frequency you designed for.

Q is the ratio of energy stored to energy lost per radian of oscillation, and for a series circuit that works out to $\sqrt{L/C}/R$. It sets the bandwidth — BW = f₀/Q — so a 100 µH coil with 250 pF and 5 Ω of loss gives Q ≈ 126, meaning a 1 MHz resonance only 8 kHz wide. That selectivity is what lets a receiver hear one station while ignoring its neighbour.

The startling part is voltage magnification: at resonance the voltage across the inductor (and across the capacitor) is Q times the applied voltage, even though they cancel each other in the sum. Feed 10 V into that circuit and the coil sits at 1.26 kV. Bench technicians have destroyed capacitors this way, and power engineers watch for the same effect when a capacitor bank resonates with system inductance at a harmonic frequency.

Bandwidth and quality factor are two ways of saying how sharp a resonance is: $BW = f_0/Q$, measured between the half-power points either side of centre. A 1 MHz tuned circuit with Q of 125 passes an 8 kHz-wide slice, which is roughly what an AM broadcast channel needs. Raise Q and the passband narrows in exact proportion.

The reason a broadcast receiver's intermediate-frequency strip exists is buried in that ratio. Getting a 10 kHz window at 1 MHz needs Q = 100; getting the same window at 100 MHz would need Q = 10,000, which no ordinary LC circuit reaches. So a superheterodyne mixes everything down to a fixed 455 kHz, where Q of only 45.5 does the job, and the selectivity stops depending on where you tuned. That is Edwin Armstrong's 1918 insight and it is in every radio since.

Two cautions. Q is not a property of the coil alone — connect a load across a tank circuit and you have added loss, so the loaded Q is lower and the bandwidth wider than the components suggest. And selectivity is not free: a filter narrower than the signal chews the sidebands off, so an over-sharp IF makes speech muffled and, in a data link, smears symbols into each other. Bandwidth and rise time are the same constraint viewed from two ends.

The corner frequency of an RC filter is where the capacitor's reactance $1/2\pi f C$ has fallen to equal the resistance. There the output is 0.707 of the input — half the power, hence "−3 dB" — and the phase has shifted 45°. A 1 kΩ resistor with 0.1 µF gives $1/(2\pi \times 10^{-4}) \approx 1592$ Hz. Above that corner a low-pass output rolls off at 6 dB per octave and keeps rolling forever.

The name "cutoff" is the misleading part: nothing is cut off. At the corner you still have 71% of the signal, and one octave above you still have about half. A single RC section is a gentle slope, not a wall, which is why anti-aliasing filters ahead of an ADC use multiple poles and why putting a filter corner right at the edge of the wanted band always disappoints. Set the corner well clear of what you want to keep.

The other trap is loading. This formula assumes nothing draws current from the output; a following stage with an impedance comparable to R shifts the corner and shrinks the passband gain. And note the same components in the time domain give the time constant $\tau = RC$, so $f_c = 1/2\pi\tau$ — the corner frequency and the 63% charging time are the same fact told twice. Swap which component the output is taken across and the identical corner becomes a high-pass instead.

An RL filter corners where the inductive reactance $2\pi f L$ has grown to equal the resistance, giving $f_c = R/2\pi L$. Fifty ohms with a 10 mH choke corners at about 796 Hz. Above that the inductor dominates and current falls; below it the resistor dominates and the inductor is barely there. The behaviour mirrors the RC case exactly, with the roles of "rising" and "falling" reactance swapped.

Notice that raising the resistance raises the corner, which feels backwards until you see it as a race: the corner is where reactance catches up with resistance, so a bigger target takes a higher frequency to reach. The same reasoning gives the time-domain view, since $\tau = L/R$ and $f_c = 1/2\pi\tau$ again — a larger R makes the inductor's current settle faster, not slower.

In practice RC is preferred wherever it will do the job, because inductors are big, expensive, magnetically noisy and never ideal: a real coil has winding resistance, interwinding capacitance and a self-resonance above which it stops being an inductor at all. RL earns its place where high current makes capacitors impractical — mains filters, switching supply output chokes, motor drive dV/dt filters — and in the humbler case of a relay coil, where the same L/R that defines this corner is what decides how long the contacts take to drop out.

Power goes as voltage squared, and the log of a square is twice the log — so a voltage ratio wears a 20 where a power ratio wears a 10. Ten volts out for one volt in is 20 dB; a 40 dB preamp multiplies 50 mV up to 5 V. Handy landmarks: 6 dB is a doubling of voltage, 20 dB is ten times, −3 dB is the 0.707 point that defines a filter's corner frequency.

Strictly, equating a voltage ratio to a power ratio in dB assumes the input and output see the same impedance — an assumption from the days when everything was 600 Ω. Modern audio and instrumentation ignore it and use 20 log of voltage as a convention, which is fine as long as everyone in the conversation agrees. Where absolute level is meant, look for the suffix: dBV references 1 V, dBu references 0.7746 V (1 mW into 600 Ω).

Bell Telephone engineers needed a way to add up losses along a line instead of multiplying them, so they took logarithms; the bel proved too coarse and the decibel — a tenth of one — stuck. Ten times log of the power ratio means 3 dB is a doubling, 10 dB is ten times, 20 dB is a hundred times, and 100 W out of 1 W in is a 20 dB gain. Cascade stages and the decibels simply add.

The trap is the reference. A plain dB is a ratio and says nothing about absolute level; dBm fixes P₁ at 1 mW, so 30 dBm is exactly 1 W and 0 dBm is 1 mW. Mixing the two — treating a gain in dB as a level in dBm — is a classic RF blunder. Remember also that this 10-log form is only for power; voltage and current ratios use 20 log, since power goes as the square.

On DC, watts are simply volts times amps. On AC the current can lag or lead the voltage, and only the in-phase component delivers energy; the power factor cos φ is the bookkeeping for that. A 240 V single-phase welder pulling 20 A at 0.95 PF consumes 240 × 20 × 0.95 = 4560 W, not the 4800 VA its supply cable and breaker must actually carry.

Resistive loads — heaters, incandescent lamps, kettles — sit at PF = 1, which is why the clamp-meter product matches the wattmeter on those and not on a motor. Watch out for modern electronics: a switching power supply may draw badly distorted current, giving a total power factor well below the displacement cos φ this formula assumes, so the meter reading and the calculation drift apart.

Power factor is an efficiency of delivery, not of conversion: it says how much of the capacity you rented from the utility is producing work. A load reading 8 kW on the wattmeter but 10 kVA on the volt-amp meter runs at PF = 0.8, meaning a fifth of the supplied capacity is sloshing back and forth building magnetic fields instead of turning shafts. Both fields here are watt fields; enter kVA in the kW box, since VA and W are the same dimension.

Utilities care because that reactive current still heats their conductors and transformers, so large customers face a power-factor penalty — often triggered below 0.9 or 0.95. Lightly loaded induction motors are the usual culprit, drifting to 0.3–0.5 PF at no load; the standard cures are capacitor banks and, better, not oversizing motors in the first place.

Real power P and reactive power Q are 90° apart, so they add like the legs of a right triangle with apparent power S as the hypotenuse: S² = P² + Q². A load pulling 10 kVA while doing 8 kW of work is carrying $\sqrt{10^2 - 8^2} = 6$ kvar of reactive burden — the classic 3-4-5 triangle in electrical dress. Vars are watts dimensionally, so enter kvar and kVA in the kW fields here.

Reactive power does no net work over a cycle; it is energy borrowed to build a magnetic field and returned half a cycle later. The borrowing is not free — the round trip flows as real current through your conductors and the utility's transformers. That is why the fix is local: put capacitors near the motor and the vars shuttle between motor and capacitor instead of travelling back to the substation.

The √3 is the whole story of three-phase power. Each of the three phases carries V_phase × I_phase, but the meter and the nameplate quote line quantities, and the 120° spacing between phases turns the sum of three phase powers into √3 × V_L × I_L. Multiply by the power factor and you have the real, billable, heat-and-torque-producing watts. A 480 V feeder carrying 100 A at 0.85 PF delivers 1.732 × 480 × 100 × 0.85 ≈ 70.7 kW.

The trap is mixing line and phase values: the √3 belongs in the formula only when V_L and I_L are both line quantities. Drop the power factor and you get apparent power in VA, not watts — the two differ by 15% on that feeder, which is exactly why a 100 kVA transformer will not carry 100 kW of motor load. Nikola Tesla's polyphase patents of 1888 made this arithmetic the foundation of every industrial plant since.