RC filter — cutoff frequency

Low-pass filter — the frequency at which the output falls to −3 dB (0.707×).

Diagram: RC filter — cutoff frequency
Ω
F

SI prefixes accepted (4k7 / 1M / 10m / 220). Upper-case M = mega, lower-case m = milli

Cutoff frequency
—Hz
Time constant
—µs
History
Press Enter in any field to keep a snapshot here.

Formula

Cutoff frequency: fc = 1 / (2π × R × C) [Hz]
Time constant: τ = R × C [s]
At the cutoff frequency the output is −3 dB (≈0.707×)

Design notes

An RC filter is the simplest filter there is — one resistor and one capacitor. It turns up everywhere: noise removal, anti-alias filtering ahead of an ADC, supply decoupling and more.

Design notes:
• To raise the cutoff tenfold, divide either R or C by ten
• The fields accept SI prefixes (10k, 100n and so on)
• A low-pass filter passes everything below the cutoff and attenuates above it

The same R and C make a high-pass filter — just swap their positions.

When you need this

For single-pole filters made from one resistor and one capacitor — ADC anti-aliasing, sensor noise removal, reset timing, PWM smoothing. It reports cutoff frequency and time constant together, so you can reason in the frequency domain or the step response as suits the problem.

What cutoff frequency means

The cutoff fc = 1/(2πRC) is where the amplitude falls to −3dB (about 70.7%). It is not a wall; it is where the roll-off begins.

A single pole rolls off at 6dB per octave, or 20dB per decade. Ten times the cutoff frequency only gets you down to a tenth. Strong rejection needs more poles or a much lower cutoff.

Time constant and step response

Time constant τ = RC; settling towards a step input.
ElapsedSettled toRemaining error
1 τ63.2 %36.8 %
2 τ86.5 %13.5 %
3 τ95.0 %5.0 %
5 τ99.3 %0.7 %
7 τ99.9 %0.1 %
9 τ99.99 %0.01 %

With a filter in front of an ADC, this table sets the minimum settling time between channels. For 12-bit accuracy (1/4096 ≈ 0.024%) you need at least 8τ after switching. On multiplexed designs this often becomes the limiting factor.

Worked example: ADC anti-aliasing

  1. Sampling a sensor at 1kHz, so Nyquist is 500Hz.
  2. To limit aliasing, target a cutoff around 150Hz.
  3. R = 10kΩ and C = 100nF give fc = 159.15Hz and a time constant of 1000µs.
  4. At 500Hz the attenuation is only about −10dB (a third) — that is all a single pole achieves.
  5. For better rejection, go to a higher-order filter, or oversample and decimate digitally, which is often more practical.

Choosing values

  • Fix the resistor first. It follows from what the previous stage can drive and the input impedance of the next. In front of an ADC, 1–10kΩ is a comfortable range.
  • Watch the capacitor dielectric. High-permittivity ceramics (X5R / X7R) lose capacitance under DC bias — some parts halve at half their rated voltage. That detunes the filter. Where accuracy matters, use C0G / NP0 or a part with substantial voltage headroom.
  • Reconcile the filter resistor with the ADC. A larger resistor makes a low cutoff easier but starves the ADC sampling capacitor. If you raise the resistance, raise the capacitance enough that it acts as the charge source the ADC sees.
  • Low-pass or high-pass is set by topology. fc is the same formula either way: capacitor to ground gives low-pass, capacitor in series gives high-pass.

Frequently asked questions

What does −3dB mean?
The amplitude has fallen to 1/√2 (about 0.707). In power terms that is half, which is why it is also called the half-power point. It is a conventional reference, not a point where the response suddenly stops.
How much attenuation at ten times the cutoff?
About −20dB, a factor of ten, for a single pole. A hundred times gives −40dB. For a steeper roll-off you need more stages, but cascading naively lets the stages interact — buffer between them or design a proper active filter.
Does cascading two RC stages give a sharper cutoff?
The slope steepens, but the first stage's output impedance loads the second, so it does not follow the ideal calculation. Making the second resistor at least ten times the first limits the interaction. For a controlled response, use an op-amp active filter.
How bad is ceramic DC bias derating?
It depends on the part, but a 0603 X5R 10µF 6.3V device at 5V can retain only 30–40% of its nominal capacitance. For filters and timing that is fatal. Check the DC bias curve in the datasheet and use C0G / NP0 where accuracy matters.

Standards and references

  • IEC 60384-1 — Generic specification for fixed capacitors.
  • EIA-198 — Temperature characteristic codes (C0G / X5R / X7R).

Last updated: 2026-08-29