How do you design a Butterworth low-pass filter?
Work out the order from n ≥ log10[(10^(As/10) − 1) ÷ (10^(Ap/10) − 1)] ÷ (2 log10(fs/fp)) and round up: 3 dB at 5 kHz and 40 dB at 20 kHz needs order 4. For an LC ladder, scale g_k = 2 sin((2k − 1)π ÷ 2n) with L = g Z0 ÷ (2π fc) and C = g ÷ (2π fc Z0). For an active filter, cascade second-order stages with pole frequency fc and Q = 1 ÷ g_k.
Butterworth filter formula and order equation
A Butterworth low-pass filter has the flattest possible passband for its order. The Butterworth filter equation for its power response has only one term that grows with frequency.
|H(f)| = 1 ÷ √(1 + (f / fc)^(2n)) A(f) = 10 log10(1 + (f / fc)^(2n))
- fc
- −3 dB cutoff frequency in Hz
- n
- filter order
- A(f)
- attenuation in dB
The order for no more than Ap dB of loss at the passband edge fp and at least As dB at the stopband edge fs:
n ≥ log10[(10^(As/10) − 1) ÷ (10^(Ap/10) − 1)] ÷ (2 log10(fs / fp))
Any cutoff between these two meets both limits:
fc = fp ÷ (10^(Ap/10) − 1)^(1/2n) fc = fs ÷ (10^(As/10) − 1)^(1/2n)
The doubly terminated ladder (source and load both Z0) comes from the normalized values:
g_k = 2 sin((2k − 1)π ÷ 2n) L = g Z0 ÷ (2π fc) C = g ÷ (2π fc Z0)
- g_k
- k-th normalized element, k = 1 to n
- Z0
- source and load resistance in Ω
An active filter factors the same response into second-order sections s² + a s + 1, each with pole frequency fc, plus one first-order section for odd orders. The a values match the first half of the g values:
a_k = 2 sin((2k − 1)π ÷ 2n) Q_k = 1 ÷ a_k k = 1 to ⌊n/2⌋
Butterworth filter design: worked examples
Example 1: what order for 40 dB two octaves above the passband?
With Ap = 3 dB at 5 kHz and As = 40 dB at 20 kHz, the exact order is 3.32 (the example in C.L. Wyatt’s ECE 3704 notes), so use n = 4. Setting the cutoff to meet the passband exactly gives fc = 5.003 kHz and 48.1 dB at 20 kHz, which is 8.1 dB of margin.
Example 2: third-order π ladder at 10 MHz in 50 Ω
g = 1, 2, 1. C1 = C3 = 1 ÷ (2π × 10 MHz × 50 Ω) = 318 pF and L2 = 2 × 50 Ω ÷ (2π × 10 MHz) = 1.59 µH. Rounding to E24 gives 330 pF, 1.6 µH and 330 pF, for 3.22 dB of loss at 10 MHz instead of 3.01 dB. At 20 MHz the ideal filter is 18.1 dB down.
Example 3: fourth-order active filter at 1 kHz
The stages have Q = 0.5412 and 1.3066, both with their poles at 1 kHz. With C1 = 10 nF, stage 1 takes C2 = 12 nF, R1 = 12.4 kΩ and R2 = 16.9 kΩ; stage 2 takes C2 = 82 nF, R1 = 3.57 kΩ and R2 = 8.66 kΩ. TI’s 1 % rule needs an op amp with about 119 kHz of unity-gain bandwidth, set by the second stage.
Butterworth filter table: g values, order 1 to 10
Normalized element values, g_k = 2 sin((2k − 1)π / 2n), for equal 1 Ω terminations and a 1 rad/s cutoff. Each row reads the same forward and backward.
| n | g1 | g2 | g3 | g4 | g5 | g6 | g7 | g8 | g9 | g10 |
|---|
| 1 | 2.0000 | | | | | | | | | |
| 2 | 1.4142 | 1.4142 | | | | | | | | |
| 3 | 1.0000 | 2.0000 | 1.0000 | | | | | | | |
| 4 | 0.7654 | 1.8478 | 1.8478 | 0.7654 | | | | | | |
| 5 | 0.6180 | 1.6180 | 2.0000 | 1.6180 | 0.6180 | | | | | |
| 6 | 0.5176 | 1.4142 | 1.9319 | 1.9319 | 1.4142 | 0.5176 | | | | |
| 7 | 0.4450 | 1.2470 | 1.8019 | 2.0000 | 1.8019 | 1.2470 | 0.4450 | | | |
| 8 | 0.3902 | 1.1111 | 1.6629 | 1.9616 | 1.9616 | 1.6629 | 1.1111 | 0.3902 | | |
| 9 | 0.3473 | 1.0000 | 1.5321 | 1.8794 | 2.0000 | 1.8794 | 1.5321 | 1.0000 | 0.3473 | |
| 10 | 0.3129 | 0.9080 | 1.4142 | 1.7820 | 1.9754 | 1.9754 | 1.7820 | 1.4142 | 0.9080 | 0.3129 |
Attenuation versus frequency for each Butterworth filter order
A(f) = 10 log10(1 + (f/fc)^(2n)), where fc is the −3 dB frequency. Read across for how far above the cutoff you need the attenuation.
| Order | 1.2 × fc | 1.5 × fc | 2 × fc | 3 × fc | 5 × fc | 10 × fc |
|---|
| 1 | 3.9 dB | 5.1 dB | 7.0 dB | 10.0 dB | 14.1 dB | 20.0 dB |
| 2 | 4.9 dB | 7.8 dB | 12.3 dB | 19.1 dB | 28.0 dB | 40.0 dB |
| 3 | 6.0 dB | 10.9 dB | 18.1 dB | 28.6 dB | 41.9 dB | 60.0 dB |
| 4 | 7.2 dB | 14.3 dB | 24.1 dB | 38.2 dB | 55.9 dB | 80.0 dB |
| 5 | 8.6 dB | 17.7 dB | 30.1 dB | 47.7 dB | 69.9 dB | 100.0 dB |
| 6 | 10.0 dB | 21.2 dB | 36.1 dB | 57.3 dB | 83.9 dB | 120.0 dB |
| 7 | 11.4 dB | 24.7 dB | 42.1 dB | 66.8 dB | 97.9 dB | 140.0 dB |
| 8 | 12.9 dB | 28.2 dB | 48.2 dB | 76.3 dB | 111.8 dB | 160.0 dB |
Butterworth polynomial table: second-order stages for active filters
All sections have pole frequency fc. Values are listed in rising Q, the order the stages should be built in. They agree with TI SLOA049 Table 9-1 and Op Amps for Everyone Table 16-5.
| Order | First-order RC section | a (s² + a s + 1) | Q |
|---|
| 2 | no | 1.4142 | 0.7071 |
| 3 | yes | 1.0000 | 1.0000 |
| 4 | no | 1.8478, 0.7654 | 0.5412, 1.3066 |
| 5 | yes | 1.6180, 0.6180 | 0.6180, 1.6180 |
| 6 | no | 1.9319, 1.4142, 0.5176 | 0.5176, 0.7071, 1.9319 |
| 7 | yes | 1.8019, 1.2470, 0.4450 | 0.5550, 0.8019, 2.2470 |
| 8 | no | 1.9616, 1.6629, 1.1111, 0.3902 | 0.5098, 0.6013, 0.9000, 2.5629 |
| 9 | yes | 1.8794, 1.5321, 1.0000, 0.3473 | 0.5321, 0.6527, 1.0000, 2.8794 |
| 10 | no | 1.9754, 1.7820, 1.4142, 0.9080, 0.3129 | 0.5062, 0.5612, 0.7071, 1.1013, 3.1962 |
Where the ideal filter and the real one part ways
The ladder model in this Butterworth calculator uses lossless L and C and exact resistive terminations. A real inductor adds series resistance, which turns the sharp −3 dB corner soft and adds passband loss, and a self-resonance above which it behaves like a capacitor, which limits the stopband. Capacitors have ESL for the same reason, so stopband rejection at high frequency is usually set by parts and layout, not by the order. Mismatched terminations also distort the response: an active stage or an attenuator pad is the fix if the source or load is not really Z0. In an active cascade, the highest-Q stage (3.20 at order 10) is the most sensitive to part tolerance and needs the most op amp bandwidth, and C0G/NP0 capacitors with 1 % resistors keep each stage on its target. At high orders the tolerance stack and the op amp count usually favor a lower-order Chebyshev design if some passband ripple is acceptable.
Glossary
- Butterworth response
- The maximally flat low-pass response: no ripple, −3 dB at fc, and a monotonic fall.
- Order (n)
- The number of poles, equal to the number of L and C elements in a ladder.
- g values
- Normalized ladder element values for 1 Ω terminations and a 1 rad/s cutoff.
- π and T ladders
- The two ways to start a ladder: a shunt capacitor first (π) or a series inductor first (T).
- Passband edge and stopband edge
- The frequency up to which the loss must stay under a limit, and the frequency from which it must exceed one.
- Insertion loss
- How much less power reaches the load with the filter in place than with a direct connection.
Frequently asked questions
How do I calculate the Butterworth filter order?
n ≥ log10[(10^(As/10) − 1) / (10^(Ap/10) − 1)] / (2 log10(fs/fp)), rounded up. For 3 dB at 5 kHz and 40 dB at 20 kHz the exact value is 3.32, so the filter needs order 4. The Butterworth filter order calculator in the first tab does this from your passband and stopband limits.
What are Butterworth filter coefficients and g values?
The g values are the normalized ladder elements for 1 Ω terminations and a 1 rad/s cutoff, g_k = 2 sin((2k − 1)π / 2n). For order 3 they are 1, 2, 1, and for order 4 they are 0.7654, 1.8478, 1.8478, 0.7654. An active filter uses the a coefficient of each s² + a s + 1 section, and those match the first half of the g values.
How do I turn g values into L and C?
L = g Z0 / (2π fc) for each series element and C = g / (2π fc Z0) for each shunt element, with equal source and load resistance Z0. Alternate them: a π ladder starts with a shunt capacitor and a T ladder with a series inductor.
What is the attenuation of a Butterworth filter at the cutoff frequency?
3.01 dB for every order, because the response is 1 / √(1 + (f/fc)^(2n)) and f = fc makes that 1 / √2. Order only changes how fast it falls after that.
How fast does a Butterworth filter roll off?
20 dB per decade per order well above the cutoff. A 5th-order filter is 100 dB down at 10 × fc. Near the cutoff it is less: the same filter is 30.1 dB down at 2 × fc.
What Q does each stage of an active Butterworth filter need?
A 2nd-order filter uses one stage with Q = 0.7071. A 4th-order uses 0.5412 and 1.3066. A 6th-order uses 0.5176, 0.7071, 1.9319. Every stage has its pole frequency at fc, and odd orders add one first-order RC section.
Should I use a π or a T ladder?
Both give the same response with equal terminations. For odd orders the π form has one inductor fewer, which usually means lower cost and loss, so it is the common choice. Even orders use the same number of each, so pick the form that suits what is at each end.
Butterworth, Chebyshev or Bessel: which should I use?
TI SLOA049 summarizes it: Butterworth has the flattest passband and a moderate roll-off, Chebyshev rolls off faster but ripples and rings, and Bessel keeps a constant group delay with no overshoot but rolls off slowest. Butterworth is the usual default when a flat passband matters and the transient response is not critical.
More free PCB tools
See all free electronics tools →
- Johnson noise calculator (resistor thermal noise)Johnson-Nyquist noise voltage and current of a resistor for a bandwidth and temperature, in nV/√Hz, µVrms and dBm.
- Notch filter calculator (twin-T)Notch frequency and component values for a twin-T notch filter, plus the Q from an active feedback version.
- Transistor base resistor calculatorBase resistor to switch a load fully on with an NPN or PNP transistor, using a forced beta, with base current and power.
- Multiple feedback band-pass filter calculatorResistors for a multiple-feedback (MFB) active band-pass filter from center frequency, Q, gain and capacitor value.
- Common emitter calculator (BJT biasing)Q-point of a voltage-divider biased common-emitter stage: base, emitter and collector voltages, currents and gain.
- Sallen-Key filter calculatorComponent values for a unity-gain Sallen-Key low-pass or high-pass filter with Butterworth, Bessel or Chebyshev response.
Sources
- M. Steer, Microwave and RF Design: Modules, section 2.7, Butterworth and Chebyshev Filters (g values and table)
- C.L. Wyatt, ECE 3704 Lecture 16: Filter Design in CT (Butterworth magnitude, order formula and example)
- National Semiconductor AN-779 (TI SNOA224), A Basic Introduction to Filters (response equation, Tables 1 and 2)
- Texas Instruments, Op Amps for Everyone, Chapter 16 (Table 16-5, Butterworth coefficients, op amp bandwidth rules)
- Texas Instruments SLOA049, Active Low-Pass Filter Design (Table 9-1, stage Q; filter type trade-offs)
Design the whole board, not one number
ProtoFlow is free AI PCB design software: describe a circuit and get a schematic with real parts, PCB layout with an autorouter, SPICE simulation and Gerber files. Native KiCad, Altium and EAGLE import and export. Free on macOS, Windows and Linux.