How to Design an Op Amp as Filter: Active Low-Pass Guide

Why Use an Op Amp as Filter Instead of Passive RC?

If you just need to roll off high-frequency noise, a passive resistor-capacitor (RC) network seems like the obvious choice. But when you use an op amp as filter, you solve the two biggest headaches of passive designs: loading effects and signal attenuation.

A passive RC filter suffers from a gradual -20dB/decade roll-off and drops your signal voltage. If you connect a low-impedance load (like an ADC input or a speaker amplifier) to the output of a passive RC filter, the load impedance parallels your capacitor, shifting your cutoff frequency and ruining the response curve. An active filter uses an operational amplifier to buffer the output. The op amp's high input impedance prevents the filter network from being loaded down, while its low output impedance drives the next stage effortlessly. Furthermore, active topologies like the Sallen-Key allow you to cascade stages without them interacting, giving you sharp -40dB/decade or -60dB/decade roll-offs while maintaining unity gain or even amplifying the signal.

Anatomy of the IC: Pinout and Operating Regions

Before wiring up feedback networks, you need to understand the physical package and how the silicon behaves under different conditions. Most hobbyist and prototyping op amps come in an 8-pin DIP (Dual In-line Package) or SOIC format containing two independent op amp channels.

Standard 8-Pin Dual Op Amp Pinout

• Pin 1: Output A
• Pin 2: Inverting Input A (-)
• Pin 3: Non-Inverting Input A (+)
• Pin 4: V- (Negative Supply or GND)
• Pin 5: Non-Inverting Input B (+)
• Pin 6: Inverting Input B (-)
• Pin 7: Output B
• Pin 8: V+ (Positive Supply)

When configuring an op amp as filter, you are forcing the device to operate in specific regions depending on the input signal and frequency. Here is what happens inside the silicon:

The Go-To Parts: Safe Defaults and Selection

Not all op amps are created equal. Selecting the wrong part for a filter circuit will result in phase shift errors, oscillation, or excessive noise. Here are the safe defaults for 90% of workbench projects, complete with their critical ratings.

1. LM358 (The Single-Supply Workhorse)

• Cost: ~$0.30 per IC
• Supply: 3V to 32V (Single or Dual)
• GBW (Gain-Bandwidth Product): 1 MHz
• Slew Rate: 0.3 V/µs
• Best For: DC to 10kHz low-pass filters, sensor conditioning, battery-powered single-supply circuits.

2. TL072 (The Audio Standard)

• Cost: ~$1.10 per IC
• Supply: ±5V to ±18V (Dual Supply Required)
• GBW: 3 MHz
• Slew Rate: 13 V/µs
• Best For: Audio-frequency active filters (20Hz - 20kHz), low-noise applications, dual-rail bench setups.

3. MCP6002 (The Modern Logic-Level Choice)

• Cost: ~$0.80 per IC
• Supply: 1.8V to 6V (Single Supply)
• GBW: 1 MHz
• I/O: Rail-to-Rail Input and Output (RRIO)
• Best For: Interfacing directly with 3.3V ESP32 or 5V Arduino ADCs where you need the output to swing all the way to the supply rails.

Building a 1kHz Active Low-Pass Filter

Let's build a unity-gain Sallen-Key low-pass filter with a Butterworth response (maximally flat passband). A Butterworth response requires a Quality Factor (Q) of 0.707. In a unity-gain Sallen-Key topology with equal resistors (R1 = R2), achieving Q = 0.707 requires the feedback capacitor (C2) to be exactly twice the value of the input capacitor (C1).

Component Selection and Math

We want a cutoff frequency ($f_c$) of roughly 1kHz. Using the formula $f_c = \frac{1}{2\pi R \sqrt{C_1 C_2}}$:

• Choose C1 = 10nF and C2 = 20nF (satisfying the C2 = 2*C1 rule).
• Solving for R yields 11,254Ω. The nearest standard E24 value is 11kΩ.
• Recalculating with 11kΩ gives an actual $f_c$ of 1,023 Hz.

Critical Capacitor Note: You must use C0G/NP0 ceramic or film capacitors for C1 and C2. Do not use X7R or Y5V dielectrics. X7R capacitors exhibit severe voltage coefficients and microphonic effects that will distort your filter's cutoff frequency as the signal amplitude changes.

Breadboard Wiring Steps

1. Insert the LM358 across the breadboard center trench. Connect Pin 8 to your +9V rail and Pin 4 to GND.
2. Place a 100nF bypass capacitor directly across Pin 8 and Pin 4 to prevent high-frequency rail noise.
3. Connect R1 (11kΩ) from your input signal source to Pin 3 (Non-Inverting Input).
4. Connect R2 (11kΩ) from the junction of R1/Pin 3 to one leg of C1 (10nF).
5. Connect the other leg of C1 to Pin 1 (Output).
6. Connect C2 (20nF) from the junction of R2/C1 to GND.
7. Wire Pin 1 (Output) directly to Pin 2 (Inverting Input) to configure it for unity gain.

For a comprehensive breakdown of the transfer functions and pole-zero placement for this topology, refer to the classic Texas Instruments SLOA024B application note on active filter design.

Bench War Story: When the Filter Oscillates

Theory is clean; the workbench is not. A few months ago, I needed to filter a 25kHz PWM signal from a microcontroller down to a clean DC voltage. I designed a 2kHz active low-pass filter using the Sallen-Key topology and grabbed an LM358 from my parts bin because it was handy.

The Setup: R = 15kΩ, C1 = 5.6nF, C2 = 11nF. Target $f_c$ = 1.9kHz. I fed it a 3.3V PWM signal and hooked my oscilloscope to the output.

The Outcome: Instead of a smooth 1.65V DC line, the scope showed a violent 350kHz square wave oscillating between 0V and 3V. The op amp was acting as a high-frequency oscillator, not a filter.

What Went Wrong: I ignored the Gain-Bandwidth Product (GBW) and phase margin. The LM358 has a GBW of 1MHz. At my PWM frequency of 25kHz, the open-loop gain of the op amp was already dropping. More importantly, the feedback capacitor (C1) introduced a high-frequency pole in the feedback loop. Because the LM358's internal phase margin was marginal at those frequencies, the capacitive load of the feedback network pushed the phase shift past 180 degrees, turning my negative feedback into positive feedback.

The Fix: I swapped the LM358 for a TL072 (which has better high-frequency phase characteristics and a 3MHz GBW) and added a 100Ω series resistor between Pin 1 and the feedback node to isolate the internal output stage from the capacitive load. The oscillation vanished, leaving a clean DC output. Always check your op amp's GBW and ensure it is at least 50 to 100 times higher than your filter's cutoff frequency.

Troubleshooting: How Op Amps Fail and How to Test Them

Op amps rarely fail gracefully. If your filter circuit is passing DC but killing the AC signal, or outputting a stuck rail voltage, the IC might be dead. Here is how they fail and how to verify with a multimeter.

Common Failure Modes

• Input Stage Blowout: Caused by applying a voltage to the input pins that exceeds the supply rails (violating the absolute maximum ratings). This destroys the input differential pair. The op amp will output a stuck voltage near one of the rails.
• Output Short-Circuit Thermal Shutdown: If you accidentally short the output pin to ground or Vcc while sourcing current, the internal thermal protection kicks in. The IC will get hot to the touch and output zero. It usually recovers once cooled, but repeated abuse degrades the silicon.
• Latch-Up: Common in CMOS op amps (like the MCP6002) when input signals spike above V+ due to inductive kickback or hot-plugging. The IC draws massive current (often >50mA) and acts as a dead short across the power rails until power is completely cycled.

The Multimeter Dead-Test

Before applying power, pull the IC out of the circuit (or ensure power is completely disconnected) and set your DMM to Diode Test mode.

1. Place the red probe on Pin 4 (V-) and the black probe on the input pins (2, 3, 5, 6). You should read a standard silicon diode drop (0.6V to 0.8V). If it reads 0.00V (short) or OL (open), the input protection diodes or junctions are blown.
2. Check for continuity between V+ (Pin 8) and V- (Pin 4). It should read OL. A dead short here means the internal bias network is fried.

The Live Bias Verification

If the dead-test passes, power the circuit and use your DMM in DC Voltage mode. With no AC input signal applied, measure the voltage at the output pin. In a properly biased single-supply filter, the output should sit exactly at your virtual ground voltage (e.g., 4.5V on a 9V supply). If the output is pegged to V+ or V-, and your feedback wiring is correct, the internal output transistor is shorted. For deeper troubleshooting on active filter response anomalies, Analog Devices provides excellent primers on component parasitics that ruin high-frequency filter performance.

Designing an active filter is one of the most rewarding analog exercises on the bench. By respecting the GBW limits, choosing the right capacitor dielectrics, and understanding the biasing requirements, you can build precision signal-conditioning circuits that outperform expensive digital DSP implementations in both latency and simplicity.