RC Circuit Calculator (Time Constant & Cutoff Frequency)
An RC circuit pairs a resistor with a capacitor in series: connected to a source, the capacitor charges gradually, never instantly. How fast it charges — and so the frequency at which the circuit filters a signal — comes down to a single number: the time constant τ = R·C.
Enter R and C: the calculator gives τ, the cutoff frequency of the matching RC filter, and the time needed to reach 50%, 63%, 95% and 99% of the final charge.
For sizing: pick a common capacitor value first (100 nF, 1 µF, 100 µF…), then adjust R to hit the τ or fc you're after — resistors from a few kΩ to a few hundred kΩ stay the easiest to source and the least sensitive to noise.
The circuit
What the time constant τ actually means: the physical intuition
τ is how sluggish the pair is. Charge through R into C (or the other way) and after one τ the gap to the final voltage has shrunk to 1/e ≈ 36.8 % — so you are at 63.2 % of the way on a charge, or down to 36.8 % on a discharge.
τ = R · C (R in Ω, C in F, τ in s)
Big τ: slow. Small τ: snappy. 10 kΩ and 100 nF is 1 ms. That one number is the circuit.
Charging and discharging a capacitor: the two exponential curves
A step E on a series RC, cap initially empty:
VC(t) = E · (1 − e−t/τ)
Discharge from E through R:
VC(t) = E · e−t/τ
At 5τ you are past 99 %. Engineers treat 5τ as “done”. The 50 % point is ln(2)·τ ≈ 0.69τ, which is also why a 555 timer is full of 0.693 factors.
RC low-pass and high-pass filters: same formula, two circuits
Same two parts. Output across C: low-pass (C follows slow things, shorts the fast ones). Output across R: high-pass (C blocks DC, passes edges). Cut-off is the same number:
fc = 1 / (2π · R · C)
First order: −20 dB/decade past the knee, −3 dB on it, phase heading toward ±90°.
Pick C first, then R: the method that avoids impractical component values
Capacitors come in clumsy E6 / E12 steps and sloppy tolerances. Resistors come in E24 or E96. Fix a convenient C (1 nF, 10 nF, 100 nF, 1 µF), then compute R. Keep R in the 1 kΩ–1 MΩ band if you can: below that you waste current, above that leakage and bias currents start to argue.
- Choose a standard C.
- R = τ / C, or R = 1 / (2π fc C).
- Round R, recompute fc or τ, live with it or pick the next C.
Time constant and bandwidth: the link to −3 dB
fc is the frequency where |Vout/Vin| = 1/√2 ≈ 0.707, i.e. −3 dB. For a first-order low-pass that is also the bandwidth from DC.
τ = 1 / (2π fc) · fc = 1 / (2π τ)
A 1 ms τ is about 159 Hz. A 20 kHz audio knee is about 8 µs. Same coin, two faces.
Series RC versus parallel RC: what changes
Series RC, same current: filters, timing, integrators. Parallel RC, same voltage: decoupling, snubbers, the load on a current source. A cap across a supply is not “τ = R·C” unless you have named the R (the load, the ESR, the Thevenin of the regulator). Do not paste the series-filter formula onto a bypass cap and expect a cut-off.
Common uses: debouncing, delay circuits, power-supply filtering, integrators
- Debounce — 10–50 ms into a Schmitt trigger. Mechanical bounce is milliseconds, not microseconds.
- Delay — charge to a comparator threshold. t = −τ ln(1 − Vth/E) if you need the exact crossing, not 0.69τ by habit.
- Supply filter — R (or a ferrite) plus C after a regulator, or the reservoir on a rectifier.
- Integrator — series R, C to a virtual ground, τ long compared with the signal period, or you have a leaky average, not an integral.
Component tolerances and their real effect on τ and fc
±5 % resistor and ±10 % capacitor: τ and fc wander ~15 %. Audio crossovers and timing want C0G / NP0 ceramics or polypropylene, and a trimmer if the number has to be right. An X7R 100 nF can be 80 nF at 5 V DC bias. The formula does not know that; you do.
RC versus RL circuits: the swapped role of capacitor and inductor
RL stores current in a magnetic field, τ = L/R. High-pass across L, low-pass across R — swapped versus RC. In practice we use RC: inductors are large, saturable, and pick up hum. RL shows up in power filters and RF, not in a 1 kHz lab exercise, unless someone is making a point.
Worked examples: an audio filter, a button debouncer, a simple delay
- 20 kHz low-pass. C = 1 nF → R = 1/(2π·20e3·1e−9) ≈ 7.96 kΩ → 8.2 kΩ, fc ≈ 19.4 kHz.
- 20 ms debounce. C = 1 µF → R = 0.02/1e−6 = 20 kΩ. Follow with a Schmitt gate so the threshold is clean.
- ~2 s LED soft-start. R = 100 kΩ → C = 2/1e5 = 20 µF → 22 µF, τ ≈ 2.2 s. An electrolytic is acceptable here; it would not be in example 1.
The 5τ rule, and when 3τ is enough
5τ ≈ 99.3 %. 3τ ≈ 95 %. Digital timing that only needs “definitely settled” can use 3τ and save parts. Analog hold that needs a millivolt wants 5τ or more, plus a look at leakage. Sample-and- hold designers count both.
Source resistance: the Thevenin you forgot
The R in τ = RC is every ohm in the charge path: the resistor you placed, the output impedance of the previous stage, the switch. A 50 Ω generator plus 50 Ω of intended R is 100 Ω, and fc is half what you typed. Include the Thevenin, or the scope will.
Why electrolytics are the wrong C for an audio filter
Polar, leaky, wide tolerance, voltage-dependent, inductive at high frequency. Fine as a coupling cap (large, not on the cut-off) or a reservoir. Wrong as the C that sets 1 kHz. Use film or C0G, and keep electrolytics where τ is seconds and nobody is matching a note.
τ = RC needs ohms and farads. 100 nF is 1×10−7 F, not 100. A power-of-ten slip is the usual way to get a 1 Hz filter when you wanted 1 kHz.