Ask three people how to size the capacitor after a bridge rectifier and you will hear “1000 µF per amp” from two of them. That rule works well enough in a surprising number of low-power circuits, which is exactly why it survives — and why it is dangerous. A 1 A, 50 Hz full-wave rail with a 2 V ripple target needs roughly 5,000 µF, while the same rail at 60 Hz needs about 4,200 µF. The rule would have you fit 1,000 µF and leave the rail five times noisier than designed. This article gives the one formula that replaces the rule, walks a complete worked example, and then shows when the rule-of-thumb is fine and when it quietly breaks.
The One Formula: C = I / (n × f × Vr) and What Each Term Means
The smoothing capacitor after a rectifier is sized by a single relationship: the charge the rail needs between input pulses divided by how far the voltage is allowed to sag in that gap. In equation form:
C = I / (n × f × Vr)
- C is the capacitance in farads.
- I is the average load current in amperes the capacitor must carry between pulses.
- n is the number of pulses per AC cycle: 1 for half-wave, 2 for full-wave, 6 for a six-pulse three-phase bridge.
- f is the AC line frequency in hertz: 50, 60, or the relevant aircraft/industrial frequency.
- Vr is the peak-to-peak ripple voltage you are willing to tolerate across the DC rail.
The physics behind it is simple. A capacitor holds charge Q = C × V, so when the rectifier stops supplying current between pulses, the load draws down the stored charge and the voltage falls with it. The period the capacitor must bridge is roughly 1/(n × f) seconds, the fraction of a cycle between pulses. The larger the capacitance, the smaller the voltage drop for a given current drain. That is the entire derivation compressed into one line.
Two real-world details refine the formula. First, the discharge is closer to linear than exponential in most practical rails because the load is roughly constant current, so the simplified relationship holds well enough for design. Second, the formula sizes ripple ignoring the diode drops and the regulator downstream — it tells you capacitor size for a ripple budget, not the final rail voltage. The peak voltage available before the capacitor is a separate calculation, covered in the next section.
Worked Example: 1 A, 50 Hz, Full-Wave, 2 V Ripple
Work the formula with concrete numbers so the scaling is obvious. Suppose a bench power supply uses a full-wave bridge on 50 Hz mains, draws 1 A on average, and the design tolerates 2 V of peak-to-peak ripple.
C = I / (n × f × Vr) = 1 A / (2 × 50 Hz × 2 V) = 1 / 200 = 0.005 F = 5,000 µF.
Unless you want exactly 2 V of ripple, the nearest standard value is the answer: 4,700 µF gives slightly more ripple, 6,800 µF gives less. The formula’s linearity makes the trade-offs easy to see. Doubling the ripple budget to 4 V halves the capacitor to 2,500 µF. Halving the load to 0.5 A also halves it. Moving from 50 to 60 Hz shrinks the required value by the ratio 50/60, to about 4,200 µF for the original 1 A, 2 V case — enough of a difference that a design copied from a 60 Hz source onto a 50 Hz rail lands short.
The filtering decision sits behind the rectifier in a complete power stage, which is why the switching power supply guide is a useful companion — the rail after the bridge is the same rail the SMPS output stage has to keep clean. The same arithmetic scales to three-phase. A six-pulse bridge running 1 A at 50 Hz with a 2 V ripple budget needs 1 / (6 × 50 × 2) = 0.00167 F = 1,700 µF. This is why industrial and server rectifiers run noticeably less bulk capacitance: more pulses per cycle shorten the gap the capacitor must bridge. It is also why the “µF per amp” rule misleads — the rule ignores n entirely, and n changes the answer by a factor of six between half-wave and six-pulse.
Why “1000 µF per Amp” Works (and When It Fails)
The rule of thumb is not random; it corresponds to a specific design point. Solving the formula for Vr with C = 1,000 µF per amp: Vr = I / (n × f × C) = 1 / (2 × 50 × 0.001) ≈ 10 V peak-to-peak. So “1,000 µF per amp” quietly assumes full-wave rectification at 50 Hz and accepts about 10 V of ripple.
That ripple is invisible in some circuits and fatal in others. If the load is a resistive heater or a battery charger with no regulator, 10 V ripple may be irrelevant. If the load is a 5 V logic rail or a precision analog circuit, 10 V ripple is a whole rail of noise and the rule fails completely. The failure is not gradual — it is a mismatch of intent. The rule sizes a capacitor for “someone’s 50 Hz full-wave supply” and stays silent about the rest of the duty, which is exactly the information the formula makes explicit.
Two additional constraints bracket the rule from below. The formula produces a lower bound for ripple; real capacitors add ESR, series inductance, and aging, so designs that depend on tight ripple add margin. And the smoothing capacitor carries the ripple current, not just the DC load — a point developed in the final section. For quick bench work where 10 V ripple on a tolerant load is acceptable, “1,000 µF per amp” is a legitimate shortcut, but it should be recognized as the specific design point it is, not as a law of electronics.
Two Drops After the Bridge: Peak Voltage and Regulator Headroom
Sizing the capacitor for ripple is only half the design; the rail has to reach the right voltage first. The peak DC voltage out of a full-wave bridge, before the capacitor and ignoring losses, is:
Vpeak = √2 × Vrms − 2 × VF
where VF is the forward voltage drop of each diode in the conducting path. Two diodes conduct simultaneously in a full-wave bridge, hence the factor of 2. For a 12 Vrms transformer feeding a rectifier whose diodes drop 0.7 V each, the peak is 1.414 × 12 − 1.4 ≈ 15.6 V. That is the ceiling the capacitor can ever reach; the regulator or load sees this peak minus the ripple the capacitor allows.
This is where the diode choice shows up in a calculation that is otherwise all about the capacitor. A Schottky diode at 0.3 V forward drop raises the same rail to 1.414 × 12 − 0.6 ≈ 16.3 V — a real, if modest, headroom gain. On a 5 V rail where the transformer ratio is tight and every tenth of a volt matters, the low-drop diode comparison starts to pay for itself. The regulator headroom rule completes the picture: the average DC voltage after the capacitor must sit far enough above the regulator’s set point that the deepest ripple trough never pulls it below the regulation range. That is the coupling between the two calculations — ripple sizing sets the trough depth, peak voltage sets the ceiling.
The standard formulation of this relationship is the ripple factor in traditional rectifier textbooks, but the working version for a bench rail is the one above. Redesigning for a lower VF diode always trades against leakage and cost, which is why the comparison is between complete diode families rather than between single numbers.
Capacitor Realities: Voltage Rating, Ripple Current, and ESR
The formula assumes a perfect capacitor; the shelves do not stock perfect capacitors. Three parameters turn a computed value into a real part, and each one has a distinct failure mode when ignored.
Voltage rating is the first. The capacitor must be rated for the peak rail voltage plus margin — never the RMS transformer voltage, which is usually about 30% lower than the peak the capacitor actually sees. A capacitor rated for a 12 Vrms line at only 16 V floats barely above the 15.6 V peak calculated above, leaving no aging margin. The practical habit is to rate for the peak and add 20–50% depending on how much ripple and temperature the part will see.
Ripple current is the second and most-often-missed parameter. The capacitor charges and discharges every half cycle, so it conducts an AC current equal in scale to the load current even though the DC load current flows to the load. That ripple current heats the capacitor and ages its dielectric; a part selected only on capacitance and voltage can still run hot. For heavily loaded rails, choosing a capacitor rated for the ripple current, or splitting the capacitance between parallel parts, is a real reliability decision.
ESR and equivalent-series inductance cap the performance in the other direction. A capacitor with high ESR converts ripple current into heat and limits how low the ripple can actually go regardless of its farad rating; electrolytics also age, losing capacitance and raising ESR over years. On rails where ripple is a spec, the ripple and noise troubleshooting guide shows how these capacitor realities show up as field symptoms, from hot cans to regulator hunting. Together, the four numbers — capacitance, voltage, ripple current, and ESR — are the complete capacitor selection, and the fifth thing to check is the package size in the bridge rectifier category whose output stages you are smoothing.