An RC filter is one resistor and one capacitor. Put the capacitor to ground after the resistor and you get a low-pass filter, which lets slow signals through and shunts fast ones away. Swap the two parts and it becomes a high-pass filter. Either way, one number describes it: the cutoff frequency.
fc = 1 / (2π · R · C)
with R in ohms, C in farads and fc in hertz.
A worked example
R = 1 kΩ, C = 0.1 µF (the "104" ceramic capacitor):
fc = 1 / (2π × 1000 × 0.0000001) = 1 / 0.000628 = 1,591.55 Hz
Common pairs and what they give:
| R | C | fc | Time constant τ = RC |
|---|---|---|---|
| 1 kΩ | 0.1 µF | 1,591.55 Hz | 0.1 ms |
| 10 kΩ | 10 nF | 1,591.55 Hz | 0.1 ms |
| 100 Ω | 1 µF | 1,591.55 Hz | 0.1 ms |
| 1.6 kΩ | 0.1 µF | 994.72 Hz | 0.16 ms |
| 4.7 kΩ | 1 µF | 33.86 Hz | 4.7 ms |
| 10 kΩ | 1 µF | 15.92 Hz | 10 ms |
Only the product RC matters for the frequency, so the first three rows are the same filter as far as fc goes. Which one you pick depends on what drives and loads it (see below).
What "cutoff" actually means
The cutoff is not a wall. At fc the output is −3 dB, about 70.7% of the input amplitude (half the power), and the output lags the input by 45°. For the 1,591.55 Hz filter:
| Input frequency | Gain | Output amplitude | Phase |
|---|---|---|---|
| 100 Hz | −0.02 dB | 99.8% | −3.6° |
| 1,591.55 Hz (fc) | −3.01 dB | 70.7% | −45° |
| 15,915.5 Hz (10 × fc) | −20.04 dB | 9.95% | −84.3° |
| 159,155 Hz (100 × fc) | −40.00 dB | 1.0% | −89.4° |
Well above fc, the output falls by 20 dB per decade (6 dB per octave): every tenfold rise in frequency divides the output by ten. That is gentle. If you need a 1 kHz signal kept and a 2 kHz one removed, a single RC stage won't do it; you need a higher-order filter.
A Bode plot draws exactly this table as two curves, flat then sloping at −20 dB/decade, with the corner at fc.
Low-pass vs high-pass
| Low-pass | High-pass | |
|---|---|---|
| Layout | R in series, C to ground | C in series, R to ground |
| Passes | below fc | above fc |
| At fc | −3 dB, −45° | −3 dB, +45° |
| Typical use | smoothing PWM, anti-aliasing before an ADC, debouncing | blocking DC, AC coupling an audio input |
The formula for fc is the same for both.
Picking R and C for a target frequency
- Choose C first. Capacitors come in fewer values than resistors, and small ceramics (1 nF–1 µF) are cheap and stable.
- Solve for R: R = 1 / (2π · fc · C). For fc = 1 kHz and C = 0.1 µF, R = 1,591.5 Ω.
- Round to a real resistor. The nearest E24 value is 1.6 kΩ, which gives fc = 994.72 Hz. A 0.5% error is well inside the 5–20% tolerance of most capacitors anyway.
Keep R roughly between 1 kΩ and 100 kΩ:
- Too small and the source has to supply a lot of current into the capacitor at high frequencies.
- Too large and whatever reads the output (an ADC input, the next stage) loads it and shifts the cutoff. The load should be at least ten times R.
Time constant and step response
τ = RC is the same filter seen in the time domain. Feed a step into a low-pass RC and the output reaches 63% of the final value after one τ and about 99% after five. For 10 kΩ and 1 µF, τ = 10 ms, so a button debounced this way settles in about 50 ms.
Common mistakes
- Forgetting to convert units. 0.1 µF is 1 × 10⁻⁷ F, and 10 nF is 1 × 10⁻⁸ F. An error of 1,000× here is the most common wrong answer.
- Using 1/(RC) instead of 1/(2πRC). 1/(RC) is the angular frequency ω in rad/s, not hertz.
- Expecting a sharp cutoff. At 2 × fc the signal is only down about 7 dB.
- Ignoring the load. A 10 kΩ filter feeding a 10 kΩ input is a different circuit with a different fc.
Try it
The RC filter calculator gives fc from any R and C. To see the whole response, plot it with the Bode plot generator, and convert between gain ratios and decibels with the decibel converter. If the capacitor only says "104" or "223", the capacitor code decoder turns that into a value.
For a related walkthrough, read Bode Plot: Gain and Phase Margin Example.