If you need to strip high-frequency noise from a sensor signal or smooth a PWM waveform into a usable DC voltage, the first-order RC low pass filter is your baseline tool. The core low pass formula for calculating the -3dB cutoff frequency is fc = 1 / (2πRC). While the algebra is simple, misapplying it on the bench leads to distorted audio, sluggish control loops, and unexpected signal attenuation. This guide breaks down the derivation, tracks units through solved problems, and explores a real-world failure mode you will not find in textbook summaries.
The Core Low Pass Formula and Symbol Definitions
The cutoff frequency (also called the corner frequency or break frequency) is the point where the output signal power drops to half its passband value, corresponding to a voltage drop of -3.01 dB (roughly 70.7% of the input voltage). For a standard passive first-order resistor-capacitor network, the formula is:
Below is the spec-sheet definition for every symbol in the equation. Always convert these to base SI units before calculating.
| Symbol | Parameter | Base SI Unit | Typical Bench Range |
|---|---|---|---|
| fc | Cutoff Frequency (-3dB point) | Hertz (Hz) | 0.1 Hz (sensors) to 100 MHz (RF) |
| R | Resistance | Ohms (Ω) | 100 Ω to 1 MΩ |
| C | Capacitance | Farads (F) | 10 pF to 1,000 μF |
| π | Pi (mathematical constant) | Dimensionless | ≈ 3.14159 |
Rearranged Forms: Solving for R and C
On the bench, you rarely know all three variables and need to find the fourth. Usually, you have a target cutoff frequency and a limited inventory of resistors or capacitors. Here are the algebraically rearranged forms of the low pass formula, solved for each component:
- To find Resistance (R):
R = 1 / (2 × π × fc × C) - To find Capacitance (C):
C = 1 / (2 × π × fc × R)
Assumptions, Limits, and Unit Traps
The low pass formula is elegant, but it relies on idealized assumptions that break down in physical circuits. Understanding these limits is the difference between a simulation that works and a board that fails.
When the Formula Applies (and When It Does Not)
This formula strictly applies to a first-order, passive RC filter driven by an ideal voltage source (zero output impedance) and feeding an infinite load impedance. If your signal source has a high output impedance (like a piezo sensor or a high-impedance voltage divider), that source resistance (Rs) adds directly to your filter resistor. If your load has a low input impedance (like an unbuffered ADC pin or a 50Ω oscilloscope input), it forms a parallel equivalent resistance with your filter capacitor, shifting the cutoff frequency higher and attenuating the passband gain. For complex impedance interactions, refer to the filter design guidelines in the Texas Instruments Application Report SLOA093.
The Unit Mistakes That Break Calculations
The most common reason a calculated filter fails on the bench is a unit conversion error. The formula demands base SI units: Ohms and Farads.
If you plug 10 into the C variable for a 10 μF capacitor, your calculated fc will be off by a factor of one million. Always use scientific notation during the calculation step: 10 μF must be entered as 10 × 10-6 or 0.00001. Similarly, 4.7 kΩ must be entered as 4700, not 4.7.
Realistic Answer Magnitudes
If your calculator spits out an answer, sanity-check it against these typical domains:
- DC Sensor Filtering (Thermocouples, Load Cells): 0.1 Hz to 10 Hz
- Audio Crossovers and DAC Smoothing: 20 Hz to 20,000 Hz (20 kHz)
- EMI/RF Snubber and PWM Carrier Filtering: 100 kHz to 50 MHz
If you are designing an audio filter and your math yields 4.5 MHz, you forgot to convert microfarads to farads.
Worked Problem 1: Calculating Cutoff Frequency
Scenario: You are building an anti-aliasing filter for an audio line-in. You have a 4.7 kΩ resistor and a 100 nF ceramic capacitor in your parts bin. What is the cutoff frequency?
- Identify and convert to base SI units:
R = 4.7 kΩ = 4,700 Ω
C = 100 nF = 100 × 10-9 F = 0.0000001 F - Substitute into the formula:
fc = 1 / (2 × 3.14159 × 4700 × 0.0000001) - Calculate the denominator:
2 × 3.14159 × 4700 × 0.0000001 = 0.002953 - Divide 1 by the denominator:
fc = 1 / 0.002953 = 338.6 Hz
Result: The filter will begin attenuating frequencies above 338.6 Hz at a rate of -20 dB/decade. This is suitable for a subwoofer crossover, but will kill midrange audio frequencies.
Worked Problem 2: Sizing a Capacitor for a Target Cutoff
Scenario: You need to filter out 60 Hz mains hum from a slow-moving temperature sensor. You want a cutoff frequency of 15 Hz to preserve the sensor's step response. You decide to use a 10 kΩ resistor to minimize current draw. What capacitor do you need?
- Rearrange the formula to solve for C:
C = 1 / (2 × π × fc × R) - Identify and convert to base SI units:
fc = 15 Hz
R = 10 kΩ = 10,000 Ω - Substitute and calculate:
C = 1 / (2 × 3.14159 × 15 × 10000)
C = 1 / 942477.7
C = 1.061 × 10-6 F - Convert back to practical units:
1.061 × 10-6 F = 1.061 μF
Result & Component Selection: You need a 1.061 μF capacitor. Since this is not a standard value, you would select a 1 μF X7R ceramic capacitor (yielding an actual fc of 15.9 Hz) or a 1.2 μF film capacitor (yielding 13.2 Hz). For precision DC sensor filtering, avoid high-K dielectrics like Y5V; use C0G/NP0 ceramics or polypropylene film to prevent dielectric absorption and microphonic noise.
Real-World Bench Scenario: ESP32 PWM Smoothing Filter
Textbook formulas assume infinite load impedance. Here is what happens when that assumption collides with a microcontroller bench project.
The Setup
We needed to generate a clean 0-3.3V analog control voltage from an ESP32 (DevKit v1) to drive a third-party laser module. The ESP32 lacks a true DAC on most pins, so we configured GPIO 18 to output a 5 kHz PWM signal. To convert this digital square wave into a smooth DC voltage, we designed a passive RC low pass filter. We wanted the cutoff frequency to be at least two decades below the 5 kHz carrier to ensure minimal ripple.
The Numbers
We targeted an fc of 50 Hz. Using a standard 10 kΩ resistor, the rearranged low pass formula dictated a capacitor value:
C = 1 / (2 × π × 50 × 10000) = 318 nF
However, we wanted an even smoother output and decided to oversize the capacitor, swapping in a 10 μF electrolytic capacitor. This pushed our theoretical fc down to 1.59 Hz. The time constant (τ = R × C) was 100 milliseconds.
The Outcome
On the oscilloscope, the 5 kHz PWM ripple was completely gone; the output was a flat, clean DC line. But when we sent a command to the ESP32 to step the voltage from 1.0V to 3.0V, the laser module reacted erratically. The scope revealed that the voltage took nearly 500 milliseconds (5τ) to settle at the new target. Furthermore, the steady-state DC voltage was sagging to 2.8V instead of the expected 3.0V.
What Went Wrong
Two distinct failures occurred, both rooted in ignoring the physical realities behind the low pass formula:
- Time-Domain vs. Frequency-Domain Tradeoff: By pushing the cutoff frequency down to 1.59 Hz to kill the ripple, we destroyed the step response. A 100ms time constant is far too slow for a control loop that expects rapid voltage updates. The formula only tells you about frequency attenuation; it does not warn you about the resulting phase shift and settling time.
- Load Impedance and Capacitor Leakage: The laser module's control input was not infinite impedance; it had an internal 50 kΩ pull-down resistor. This 50 kΩ load formed a voltage divider with our 10 kΩ filter resistor, causing the steady-state voltage sag (3.3V × [50k / (10k + 50k)] = 2.75V). Additionally, the cheap 10 μF electrolytic capacitor had high leakage current, exacerbating the voltage drop.
