Filters
Learning Objectives
- Classify filters by function (low-pass, high-pass, band-pass, band-stop) and by construction (passive vs. active).
- Calculate the cutoff frequency of an RC low-pass and RC high-pass filter.
- Explain the role of op-amps in active filters and why they're used over passive designs.
- Compare FIR and IIR digital filters at a conceptual level.
- Choose the appropriate filter type for a given real-world signal-processing problem.
- Connect filter behavior back to frequency response concepts (magnitude and phase).
Quick Answer
A filter is a circuit that selectively passes some frequencies while blocking or attenuating others. There are four basic types based on which frequencies they pass: low-pass (keeps low frequencies), high-pass (keeps high frequencies), band-pass (keeps a middle range), and band-stop (removes a middle range). Filters can be built from passive components (resistors, capacitors, inductors) or active components (op-amps plus RC networks), and increasingly are implemented digitally in software. Filters matter because virtually every real signal is contaminated with unwanted content — noise, interference, out-of-band frequencies — and filtering is the primary tool for cleaning that up, whether it's removing hum from an audio signal, isolating a radio station, or smoothing sensor data in a control loop.
Filter Types by Function
| Filter type | Passes | Blocks | Typical use |
|---|---|---|---|
| Low-pass (LPF) | Frequencies below f_c | Frequencies above f_c | Removing high-frequency noise, anti-aliasing |
| High-pass (HPF) | Frequencies above f_c | Frequencies below f_c | Blocking DC offset, removing low-frequency rumble |
| Band-pass (BPF) | A range between f_1 and f_2 | Everything outside the range | Radio tuning, isolating a specific signal |
| Band-stop (BSF) | Everything outside f_1–f_2 | A specific range | Removing mains hum (50/60Hz notch filter) |
Passive RC Low-Pass Filter
Circuit: resistor R in series, output taken across capacitor C.
Transfer function: H(s) = 1 / (1 + sRC), and the cutoff frequency is f_c = 1 / (2πRC).
Worked Example 1: Design an RC low-pass filter with a cutoff frequency of 1kHz using a 10kΩ resistor. Find the required capacitor value.
f_c = 1/(2πRC) → C = 1/(2π × f_c × R)
C = 1 / (2π × 1000 × 10,000) = 1 / 62,831,853 ≈ 1.59×10⁻⁸ F ≈ 15.9 nF
Passive RC High-Pass Filter
Circuit: capacitor C in series, output taken across resistor R (the components swap positions from the low-pass filter).
Transfer function: H(s) = sRC / (1 + sRC), same cutoff formula: f_c = 1 / (2πRC).
Worked Example 2: An audio preamp needs to block DC and hum below 20Hz using a 100nF capacitor. Find the required resistor value.
f_c = 1/(2πRC) → R = 1/(2π × f_c × C)
R = 1 / (2π × 20 × 100×10⁻⁹) = 1 / (1.257×10⁻⁵) ≈ 79,577 Ω ≈ 79.6 kΩ
Active Filters
Active filters add an op-amp to the RC network, which brings two big advantages over purely passive designs:
- Gain: the op-amp can amplify the passed signal, not just attenuate the blocked one.
- Buffering (isolation): the high input impedance and low output impedance of an op-amp stage prevent the filter's behavior from being affected by whatever load or source is connected next — passive filters can shift their cutoff frequency if loaded incorrectly, but a well-designed active filter stage does not.
An active low-pass filter's transfer function takes the general form H(s) = 1 / (1 + s/ωc), where ωc = 1/(RC) — mathematically similar to the passive case, but the op-amp supplies gain and isolation on top.
Worked Example 3: An active low-pass filter is built with an inverting op-amp stage where the feedback resistor is 10kΩ, feedback capacitor is 15.9nF, and input resistor is 5kΩ. Find the cutoff frequency and the passband gain.
f_c = 1/(2π × R_feedback × C_feedback) = 1/(2π × 10,000 × 15.9×10⁻⁹) ≈ 1000 Hz
Passband gain = −R_feedback/R_input = −10,000/5,000 = −2 (gain magnitude of 2, inverted)
Passive vs. Active Filters
| Property | Passive filters | Active filters |
|---|---|---|
| Components | R, L, C only | R, C plus op-amps |
| Gain | Never above 1 (can only attenuate) | Can amplify the passband |
| Loading sensitivity | High — cutoff shifts if loaded | Low — op-amp buffers isolate stages |
| Power requirement | None (passive) | Needs a power supply for the op-amp |
| Cost/complexity | Simple, cheap | More complex, but more capable |
Digital Filters (Conceptual Overview)
Digital filters process sampled signals numerically rather than with physical components, and dominate modern signal processing because they're programmable and highly precise.
- FIR (Finite Impulse Response): output depends only on a finite window of past inputs; always stable; can achieve linear phase (no phase distortion), at the cost of needing more coefficients (more computation) for a sharp cutoff.
- IIR (Infinite Impulse Response): output depends on past inputs and past outputs (feedback); can achieve a sharp cutoff with far fewer coefficients than FIR, but may introduce phase distortion and requires care to ensure stability.
Key Terms
| Term | Definition |
|---|---|
| Passband | The range of frequencies a filter allows through with little attenuation |
| Stopband | The range of frequencies a filter blocks or heavily attenuates |
| Cutoff frequency (f_c) | The frequency marking the boundary between passband and stopband (typically the -3dB point) |
| Passive filter | A filter built only from resistors, capacitors, and/or inductors |
| Active filter | A filter that includes an amplifying element (typically an op-amp) alongside R and C |
| FIR filter | A digital filter whose output depends only on a finite set of past input samples |
| IIR filter | A digital filter whose output depends on past inputs and past outputs (feedback) |
Common Mistakes
-
Misconception: "A passive filter can also amplify the passband signal if you choose good component values." Why it's wrong: Passive components (R, L, C) only store or dissipate energy — they cannot add energy to a signal, so a passive filter's gain can never exceed 1 (0dB) anywhere in its passband. Correct: To get gain along with filtering, you need an active filter with an op-amp, which supplies the extra energy from its power supply.
-
Misconception: "Swapping the resistor and capacitor positions in an RC filter doesn't change anything important." Why it's wrong: Swapping positions changes a low-pass filter into a high-pass filter (or vice versa) — the cutoff frequency formula stays the same, but which frequencies pass and which are blocked reverses completely. Correct: In an RC low-pass, output is taken across C; in an RC high-pass, output is taken across R (with C in series). The physical arrangement, not just the components used, determines the filter type.
-
Misconception: "A band-stop filter is the same as putting a low-pass and high-pass filter in series." Why it's wrong: A low-pass and high-pass filter in series (with overlapping ranges) actually forms a band-pass filter (passing only the overlap), not a band-stop filter. Correct: A band-stop (notch) filter is typically built as a parallel combination or a dedicated resonant circuit designed to reject one specific frequency band while passing everything else.
Comparison and Connections
| Concept | Low-pass | High-pass | Band-pass | Band-stop |
|---|---|---|---|---|
| Passes | Below f_c | Above f_c | Between f_1, f_2 | Outside f_1, f_2 |
| Simple RC form | R then C (output across C) | C then R (output across R) | LP + HP in series | Parallel/notch design |
| Typical application | Anti-aliasing, audio bass | DC blocking, rumble removal | Radio tuning | Mains hum rejection |
Practice Questions
Recall 1: What is the cutoff frequency formula for a simple RC filter (low-pass or high-pass)? Answer guidance: f_c = 1/(2πRC).
Recall 2: Name the four basic filter types by function. Answer guidance: Low-pass, high-pass, band-pass, band-stop.
Understanding 1: Why can't a purely passive RC filter provide gain? Answer guidance: Passive components (R, C) can only dissipate or store energy, never add it, so the output can never exceed the input in amplitude anywhere in the passband.
Understanding 2: Explain the practical benefit of the buffering effect in active filters. Answer guidance: The op-amp's high input impedance and low output impedance prevent whatever is connected before or after the filter from loading it and shifting its cutoff frequency, unlike a passive filter whose behavior can change depending on the load.
Application 1: Design an RC low-pass filter with cutoff frequency 500Hz using a 4.7kΩ resistor. Find the capacitor value. Answer guidance: C = 1/(2π×500×4700) = 1/(14,765,485) ≈ 6.77×10⁻⁸ F ≈ 67.7 nF.
Application 2: An RC high-pass filter uses R=1kΩ and C=100nF. Find the cutoff frequency. Answer guidance: f_c = 1/(2π×1000×100×10⁻⁹) = 1/(6.283×10⁻⁴) ≈ 1592 Hz.
Analysis 1: A student builds a passive RC low-pass filter, then connects a low-impedance load directly at the output. Explain what happens to the actual cutoff frequency compared to the design value, and why an active filter would avoid this problem. Answer guidance: The load resistance effectively appears in parallel with the capacitor (or interacts with the output impedance of the RC network), changing the effective RC time constant and shifting the cutoff frequency away from the designed value. An active filter's op-amp output stage has low output impedance, so it can drive a load without altering the filter's internal RC-determined cutoff.
Analysis 2: Compare FIR and IIR digital filters for an application requiring a very sharp cutoff with minimal computational resources, versus an application requiring zero phase distortion. Which filter type suits each requirement, and why? Answer guidance: For a sharp cutoff with minimal computation, IIR is preferred because it achieves steep roll-off with fewer coefficients (using feedback). For zero phase distortion, FIR is preferred because it can be designed with exactly linear phase, which IIR generally cannot guarantee due to its feedback structure.
FAQ
Q1: What does "order" mean for a filter? The order refers to the number of reactive elements (or poles) in the filter, which determines how steeply it attenuates frequencies beyond the cutoff. A first-order RC filter rolls off at 20dB/decade; a second-order filter rolls off at 40dB/decade, and so on.
Q2: Why do audio systems use low-pass filters for subwoofers and high-pass filters for tweeters? Subwoofers reproduce low frequencies well but can be damaged or distort trying to reproduce high frequencies, so a low-pass filter sends only bass to them. Tweeters are the opposite — small and fast, ideal for highs but easily damaged by large low-frequency excursions — so a high-pass filter protects them by blocking bass.
Q3: Is a band-stop filter the same as a "notch" filter? Essentially yes — "notch filter" usually refers to a band-stop filter with a very narrow rejection band, commonly used to eliminate a single troublesome frequency like 50Hz or 60Hz mains hum.
Q4: Can filters change the shape of a signal, not just its amplitude? Yes — because filters have frequency-dependent phase shifts as well as magnitude changes, they can distort the shape of complex waveforms (like square waves) even while only intending to remove certain frequency components, since a square wave is built from many harmonics that each get shifted differently.
Q5: Why are digital filters replacing analog filters in many modern designs? Digital filters are programmable (the same hardware can implement different filter designs just by changing coefficients), don't drift with component aging or temperature the way analog R/L/C values can, and can achieve very precise, complex responses that would be impractical to build with discrete components.
Quick Revision
- Four basic filter types: low-pass, high-pass, band-pass, band-stop.
- RC filter cutoff frequency: f_c = 1/(2πRC), same formula for both low-pass and high-pass.
- Low-pass: output taken across C (R in series first). High-pass: output taken across R (C in series first).
- Passive filters (R, L, C only) can never amplify — gain is always ≤ 1.
- Active filters add an op-amp for gain and buffering (isolation from load effects).
- Passive filters are sensitive to loading; active filters are largely immune due to op-amp buffering.
- Band-pass = overlapping low-pass + high-pass ranges; band-stop (notch) rejects one specific range.
- FIR digital filters: always stable, can have linear phase, need more coefficients for sharp cutoffs.
- IIR digital filters: sharper cutoff with fewer coefficients, but may distort phase and need stability checking.
- Filter "order" determines roll-off steepness: first-order = 20dB/decade, second-order = 40dB/decade, etc.
Related Topics
Prerequisites: Frequency Response, AC and DC Circuits, basic op-amp concepts.
Related Topics: Frequency Response, Signal and Noise Analysis, Network Theorems.
Next Topics: Signal and Noise Analysis, Power Calculations.