Capacitor calculator
Almost every circuit board pairs a large electrolytic with a small ceramic across the same rail. In capacitance terms the ceramic adds almost nothing: 100 nF beside 100 µF is a rounding error. It is there for its low inductance at high frequency, which the big electrolytic cannot manage. The parallel sum is the right arithmetic and the wrong way to think about that pairing.
In a capacitor network the values add in parallel and reciprocate in series, the mirror image of resistors: two 10 µF in series give 5 µF. It also gives RC charge time, five time constants to 99%, so 1 kΩ with 470 µF is about 2.35 seconds, and inductive reactance with the LC resonance an inductor and capacitor set together.
How to use it
Capacitors behave backwards from resistors, and the reason is geometric, not arbitrary. Resistance opposes current, so an extra path reduces it. Capacitance stores charge on plates, so extra plate area increases it: parallel capacitors are effectively one capacitor with larger plates and the values add. Series stacks the separation between the plates instead, which lowers capacitance, so the reciprocal sum applies and the total falls below the smallest member.
Voltage rating cuts the other way
In parallel there is no trade at all: every capacitor sees the full supply voltage, so the lowest rating in the set is the limit for the whole bank. Series is how you get a rating higher than any single part offers, and it comes with a trap. The voltage only divides evenly if the parts are matched and their leakage currents are similar. In practice leakage differs, the string drifts, and one capacitor ends up carrying more than its share until it fails, which then puts the full voltage across the next one. Any series string handling real voltage needs balancing resistors across each capacitor to hold the division steady, which is why you see them on every high-voltage bank.
Paralleling buys two things beyond capacitance. Equivalent series resistance falls roughly as the reciprocal of the count, which is why a bulk bank is built from several smaller parts instead of one large one. And a small ceramic sitting beside a big electrolytic is not there for its capacitance at all; it has far lower inductance and handles the fast transients the electrolytic cannot follow.
Charging is exponential, so nothing is ever full
One time constant is R times C and reaches 63.2%. Five leaves under 1% remaining, which is smaller than component tolerance, and that is why the figure quoted here is five tau and never a time to "fully charged". The shape matters as much as the number: charging is fastest at the start and slows continuously, so the last few per cent take as long as the first sixty.
Nothing limits the initial current except the series resistance, because an empty capacitor behaves like a dead short at the instant of switch-on. On a large supply that is what welds relay contacts and blows fuses. An inrush limiter is a resistor placed in series to hold the first surge down, shorted out by a relay once the capacitor is mostly charged. The same arithmetic run backwards tells you how long a board takes to become safe to touch, and the honest advice there is to confirm discharge with a meter instead of trusting a calculation.
The inductor is the mirror again
A capacitor opposes changes in voltage and passes high frequencies. An inductor opposes changes in current and blocks them, so its reactance rises with frequency where capacitive reactance falls. That makes it a good series filter against switching noise at hundreds of kilohertz and a poor one against 50 or 60 Hz mains hum, where the reactance is simply too small to do anything. At the frequency where the two reactances cancel you have resonance, which is the whole basis of an LC filter or a tuned circuit.
The stored-energy figure is the one that bites in switching supplies. Interrupting an inductor carrying current forces that energy to go somewhere, and with no flyback path it appears as a voltage spike large enough to destroy whatever did the switching.
What people use it for
- Building up a bulk reservoir from several capacitors
- Raising the working voltage of a bank by putting capacitors in series
- Timing an RC power-on delay or sizing an inrush limiter
- Working out how long a board takes to become safe to touch
- Finding the resonant frequency of an LC tuned circuit
- Estimating the energy stored in a switching supply choke
- Combining the decoupling capacitors sitting on one rail
- Reaching a value you do not stock from two that you do
- Lowering effective series resistance by paralleling several parts
- Halving a value with two identical capacitors in series
- Working through a snubber or a voltage-multiplier stage
- Estimating how long a supply rail takes to come up at switch-on
- Designing an LC filter, and checking its reactance at the switching frequency
Questions
C_total = C₁ + C₂ + … Straight addition, unlike resistors.
1/C_total = 1/C₁ + 1/C₂ + … For two, C = C₁C₂ ÷ (C₁ + C₂).
Series capacitors act like one capacitor with a greater plate separation, which reduces capacitance.
No, they all see the full voltage. The lowest voltage rating in the set is the limit for the whole bank.
In principle, for two matched parts — but only with balancing resistors. Without them, leakage differences push one capacitor over its rating.
For high-frequency performance, not capacitance. The ceramic has far lower inductance and handles fast transients the electrolytic cannot. Paralleling helps in the same direction for a different reason: equivalent series resistance falls roughly as the reciprocal of the count, which is why bulk banks use several smaller parts rather than one large one.
Five time constants to 99% either way, one time constant being resistance times capacitance. Discharge follows the same curve through whatever resistance is across the capacitor rather than the one that charged it, which is often a very different number.
The curve is exponential and approaches the supply asymptotically. Five tau leaves under 1% remaining, which is below component tolerance.
A large one at high voltage certainly is. Confirm discharge with a meter rather than assuming, especially in mains equipment.
Only the series resistance. An empty capacitor behaves like a short circuit at the first instant, which is what inrush limiters exist to control.
XL = 2πfL, the AC opposition an inductor presents. It rises with frequency, unlike capacitive reactance which falls.
f = 1 ÷ (2π√(LC)). At that frequency the inductive and capacitive reactances cancel.
Half L times current squared. A 10 mH inductor at one amp holds 5 millijoules.
Inductance over resistance: how long the current takes to reach 63% of its final value. Add a series resistance and it appears alongside the reactance.
Because interrupting inductor current forces the stored energy out as a voltage spike. The diode gives it a path instead of your transistor.