Low-Pass vs High-Pass Filter: Differences and Examples

Quick answer

A low-pass filter passes DC and slow signal changes while attenuating higher-frequency content. A high-pass filter attenuates DC and slow signal changes while passing higher-frequency content.

Choose low-pass for analog filtering, such as baseband ADC conditioning, PWM-to-analog smoothing, and supply-noise reduction. Choose high-pass for audio AC coupling, sensor baseline removal, and a speaker crossover's treble branch.

Low-pass vs high-pass filter: comparison

Picture frequency increasing to the right. A low-pass trace starts flat on the left and bends downward toward the right, while a high-pass trace rises from the left and becomes flat on the right. Both filters attenuate frequencies gradually around cutoff.

Table 1. Low-pass and high-pass filter comparison

Choose low-pass filter if…

• A baseband analog-to-digital converter (ADC) measures a slowly changing quantity. Specify required rejection of interference and acquisition settling together. Anti-aliasing reduces frequencies that would fold into the sampled band.
• A microcontroller (MCU) produces a pulse-width modulation (PWM) output. Preserve the duty-cycle average voltage while attenuating PWM-frequency harmonics, then balance ripple against response time under load.
• An analog power rail carries switching noise. Check current, voltage drop, damping, and regulator stability before applying a signal-filter circuit.
• A speaker crossover needs a low-frequency electrical branch. Speaker impedance and acoustic response still determine the final result.
• A button input needs bounce suppression before threshold detection. Include hysteresis or firmware qualification to avoid repeated transitions.

Choose high-pass filter if…

• An audio input must reject upstream DC. Provide a receiver bias resistor and assess connection and startup transients.
• A vibration sensor measures vibration rather than absolute displacement. Decide which low-frequency motion can be discarded.
• An amplifier needs a different DC operating point from its source. A coupling capacitor blocks the source offset, while a resistor sets the new bias.
• A speaker crossover needs a high-frequency electrical branch. The speaker's acoustic response determines final crossover behavior.
• When baseline drift dominates a measurement, verify that useful waveform content does not occupy the same low-frequency region.

Cutoff formulas, slope, phase, and step response

Assume a low-impedance voltage source, negligible output loading, linear components, and a common AC reference. Analog Devices' Analog Filter Design Demystified, dated February 13, 2003, gives the first-order RC transfer functions:

$$\tau =RC,f_c=\frac{1}{2\pi RC}$$

$$H_{LP}(s)=\frac{1}{1+sRC},H_{HP}(s)=\frac{sRC}{1+sRC}$$

Here s is the Laplace variable. At cutoff, output amplitude is $1/\sqrt{2}$ of its passband value, giving:

$$20{\log }_{10}(1/\sqrt{2})=-3.0103\mathrm{dB}$$

The −3 dB point means output voltage is 70.7% of passband amplitude and power is halved at equal impedance. A connected load changes the circuit impedance and may move the cutoff.

For x equal to frequency divided by cutoff, the calculated phase responses are:

$${\varphi }_{LP}=-{\tan }^{-1}(x),{\varphi }_{HP}=90^{\circ }-{\tan }^{-1}(x)$$

At the cutoff frequency, the phase equations evaluate to −45° and +45°. The same equations give 20 dB of additional attenuation per tenfold change far beyond cutoff. Low-pass attenuation increases above cutoff, and high-pass attenuation increases below cutoff. Near cutoff the slope is gradual.

Analog Devices' A Filter Primer, dated September 18, 2002, relates the RC pole to exponential settling. Inverting the transfer functions for a normalized unit step, with zero initial energy, gives:

$$v_{LP}(t)=1-e^{-t/\tau },v_{HP}(t)=e^{-t/\tau },t\ge 0$$

After an input step, low-pass output rises gradually, while high-pass output briefly rises and decays. Check receiver thresholds, pulse duration, hysteresis, and input slew because filtering can change logic timing.

Main circuit examples: RC low-pass and high-pass

Changing where R and C connect produces different responses from the same parts.

Use the same exact library parts in both examples:

• Vishay CRCW060310K0FKEA chip resistor: nominal 10 kΩ, ±1%.
• Murata GRM2195C1H103FA01D ceramic capacitor: nominal 10,000 pF, equivalent to 10 nF. Its dielectric is C0G and tolerance is ±1%.

Calculation using those nominal values gives:

$$\tau =100\mu \mathrm{s},f_c=1591.55\mathrm{Hz}\approx 1.59\mathrm{kHz}$$

The result assumes negligible source and load impedances. Use small signals and a high-impedance receiver or buffer. Include actual source and load impedance in the calculation if loading is significant.

Low-pass: preserve the baseline

For low-pass, connect source → R → OUT with C from OUT to return. Measure OUT relative to return. At DC the capacitor is effectively open. At higher frequencies its lower impedance diverts more current into return.

The RC example illustrates signal conditioning before a buffer or ADC. ADC sampling draws charging current, so verify acquisition settling and any required capacitor at the input. For detailed sizing, see Flux's RC low-pass filter guide.

High-pass: remove the baseline

For high-pass, connect source → C → OUT with R from OUT to the receiver's reference. Measure OUT relative to that reference. On a single-supply board, the reference may be a quiet DC bias voltage instead of ground.

The same 1.59 kHz cutoff attenuates frequencies below that point. High-pass filtering suits changing sensor signals or AC coupling when DC level is unnecessary. Check capacitor voltage stress from source/receiver bias differences, input bias current, startup charging, and protection. Keep the receiver's input impedance high relative to R to preserve the nominal calculation.

Beyond RC: RL, LC, and Sallen–Key

RL and LC networks use inductors. Sallen–Key circuits add an op-amp for second-order active filtering.

RL: choose the output deliberately

For a series resistor-inductor (RL) chain, measure voltage across R for low-pass behavior or across L for high-pass behavior.

A voltage-divider derivation using inductive impedance (sL) gives:

$$H_{LP}=\frac{R}{R+sL},H_{HP}=\frac{sL}{R+sL},f_c=\frac{R}{2\pi L}$$

For a small-signal numerical example, use Vishay's 100 Ω resistor CRCW0603100RFKEA with the Bourns SDR0604-100ML 10 µH inductor. Vishay's ordering guide identifies R as the resistance decimal marker and F as ±1%. Decoding 100R therefore gives nominal 100 Ω. With nominal L = 10 µH, calculation gives fc = 1.59155 MHz. Use a low-impedance source and high-impedance measurement. Measure the voltage across R for low-pass and L for high-pass. Bourns specifies minimum self-resonant frequency of 23 MHz in revision 04/26, so the calculated cutoff falls below self-resonance. Winding resistance, parasitics, and termination still require a measured sweep.

LC: natural frequency is not automatically −3 dB cutoff

To damp LC resonance, reuse the 100 Ω resistor: low-pass connections are source → R → L → OUT with C from OUT to return. The high-pass dual uses source → R → C → OUT with L from OUT to return. Assume a low-impedance source and high-impedance receiver. Include actual terminal impedances when loading is significant.

TI's LC Filter Design, SLAA701B, revised June 2026, gives the LC natural-frequency relationship. Voltage-divider analysis of these series-damped, high-impedance reference circuits gives:

$$f_0=\frac{1}{2\pi \sqrt{LC}},Q=\frac{\sqrt{L/C}}{R}$$

Here R is the series resistor that controls damping. With Bourns SDR0604-100ML, nominal 10 µH, ±20%, from the manufacturer's SDR0604 datasheet revision 04/26, the Murata capacitor, and R = 100 Ω, calculation gives nominal f0 = 503.29 kHz and Q = 0.316. The nominal values estimate resonance and damping. The loaded circuit determines the actual −3 dB cutoff.

Loading, losses, and Q determine resonance peaking and cutoff. TI's RLC Filter Design for ADC Interface Applications explicitly distinguishes characteristic frequency from −3 dB bandwidth. For power rails, see Flux's pi-filter guide for damping and regulator stability.

Active Sallen–Key: isolate loading and control response

An op-amp buffers the two-stage RC network.

TI's Analysis of the Sallen-Key Architecture, SLOA024B, revised September 2002, gives equal-component relationships:

$$f_0=\frac{1}{2\pi RC},Q=\frac{1}{3-K}$$

With a unity-gain follower, K = 1 and the calculated Q is 0.5, below a Butterworth filter's 0.707. Reuse the RC example's 10 kΩ resistors and 10 nF capacitors as matched nominal pairs. For low-pass, connect two series resistors between IN and noninverting input B, with midpoint A. One capacitor connects B to return, and another connects A to amplifier OUT. Connect the inverting input to OUT. The high-pass dual uses two series capacitors, R from B to return, and R from A to OUT.

For the active stage, use TI OPA197IDBVR within its datasheet supply and signal ranges, with recommended bypassing and a light load. For ground-referenced bipolar signals, dual supplies simplify biasing. Single-supply operation needs a DC bias reference.

Both Sallen–Key sections have nominal f0 = 1.59 kHz. Solving TI's second-order response at Q = 0.5 gives calculated −3 dB frequencies of 1.02431 kHz for low-pass and 2.47291 kHz for high-pass. At Q = 0.5, the −3 dB cutoff differs from $f_0$. Component mismatch and finite amplifier gain alter the response.

Band-pass vs low-pass and band-stop filters

Low-pass passes DC, whereas band-pass selects a bounded frequency range. Cascading the earlier high-pass and low-pass stages with a buffer between them yields:

$$H_{BP}(s)=\frac{s\tau }{(1+s\tau {)}^2}$$

The calculated nominal peak is at 1.59155 kHz with amplitude 0.5 relative to input, equivalent to −6 dB at the peak. Without the buffer, stage loading changes the response.

For a numerical band-stop example, connect a 100 Ω resistor from a low-impedance source to OUT, with the Bourns inductor and Murata capacitor in series from OUT to return. For a high-impedance receiver, ideal series resonance is 503.29 kHz and calculated Q = sqrt(L/C)/R = 0.316: Q = 0.316 predicts broad rejection. Actual winding loss limits depth, tolerances move the rejection frequency, and parasitics and loading affect bandwidth. Measure the response. The inductor's self-resonance constrains the upper passband. Measure rejection depth and bandwidth.

Design considerations: schematic, BOM, and PCB

Real impedances and component behavior affect the final filter response.

Impedances first. Model source resistance and receiving impedance. Source resistance adds to the passive low-pass series R, and the receiver loads C. For AC coupling, the receiver's bias resistor affects cutoff.

Dielectric and tolerance. Murata's C02E catalog distinguishes temperature-compensating capacitors from high-dielectric-constant types and describes DC-bias capacitance change. Capacitance changes move cutoff. Prefer C0G for stable signal-filter timing. For X7R capacitors, obtain the exact part's bias curve instead of equating nominal capacitance or temperature classification with capacitance under operating bias. Recalculate frequency and Q across component tolerances, voltage, and temperature. Nominal tolerance does not account for dielectric effects.

Preferred values. Vishay document 28372, revision 22-Aug-12, lists E24 and E96 according to IEC 60063. Choose available preferred values, then recompute response. Check each part's tolerance and availability.

Op-amp bandwidth. OPA197 has typical 10 MHz unity-gain bandwidth under its stated supply test conditions. Gain-bandwidth product (GBW) affects the active filter's response. Evaluate loop gain, Q sensitivity, noise, slew rate, headroom, and capacitive loading rather than using a universal bandwidth multiplier.

Placement and returns. Put input interference filtering near the connector, but place ADC filters where acquisition settling requires. Keep shunt paths short and referenced to the receiver's low-impedance return. Avoid coupling switching traces into the filtered node, shared noisy return paths, and inductor fields near sensitive analog nodes. Preserve short op-amp feedback geometry. See Flux's signal integrity guide for routing guidance.

FAQ

What is the difference between a high pass vs low pass filter?

Is a low pass filter vs high pass filter better for an ADC?

Does a high-pass filter remove DC from a sensor?

Can low-pass filtering convert PWM to analog?

What is band pass vs low pass filter selection based on?

Will an RC filter debounce a button?

Sources

• Analog Devices, Analog Filter Design Demystified, February 13, 2003: https://www.analog.com/en/resources/technical-articles/analog-filter-design-demystified.html
• Analog Devices, A Filter Primer, September 18, 2002: https://www.analog.com/en/resources/technical-articles/a-filter-primer.html
• TI, Analysis of the Sallen-Key Architecture, SLOA024B, July 1999, revised September 2002: https://www.ti.com/lit/an/sloa024b/sloa024b.pdf
• TI, RLC Filter Design for ADC Interface Applications, SBAA108A, December 2003, revised January 2005: https://www.ti.com/lit/an/sbaa108a/sbaa108a.pdf
• TI, LC Filter Design, SLAA701B, October 2016, revised June 2026: https://www.ti.com/lit/pdf/slaa701
• Vishay, D/CRCW e3, document 20035, revision 14-Apr-2026: https://www.vishay.com/docs/20035/dcrcwe3.pdf
• Vishay, D/CRCW e3, document 20035, revision 14-Apr-2026, for CRCW0603100RFKEA: https://www.vishay.com/docs/20035/dcrcwe3.pdf
• Murata, Chip Multilayer Ceramic Capacitors for General, C02E.pdf, Nov.27,2017:
• Bourns, SDR0604 Series, revision 04/26: https://www.bourns.com/docs/product-datasheets/sdr0604.pdf
• TI, OPA197 / OPA2197 / OPA4197, SBOS737C, January 2016, revised March 2018: https://www.ti.com/lit/gpn/opa197
• Vishay, Standard Series Values in a Decade for Resistances and Capacitances, document 28372, revision 22-Aug-12: https://www.vishay.com/docs/28372/e-series.pdf