What Is Input Impedance in Signal Integrity?
Input impedance ($Z_{in}$) is the complex opposition—comprising both resistance and reactance—that a circuit presents to a signal source at its input terminals. In direct current (DC) and low-frequency analog design, we often simplify this to input resistance. However, in signal integrity and noise control, the reactive component (capacitance and inductance) is usually the culprit behind unexplained noise floors and oscillations.
When diagnosing what is input impedance doing to your signal chain, you must look beyond power transfer. In audio, sensor, and DC measurement circuits, we use voltage bridging: the source has a low output impedance ($Z_{out}$), and the receiver has a high input impedance ($Z_{in}$). This maximizes voltage transfer but turns high-impedance nodes into highly efficient antennas for electric fields. Conversely, in RF and high-speed digital design, we use impedance matching ($Z_{in} = Z_{out}$, typically 50 Ω or 75 Ω) to prevent signal reflections, which makes the circuit immune to capacitive noise but vulnerable to magnetic fields and ground loops.
Identifying the Dominant Noise Coupling Path
To fix a noisy input, you must first answer: which coupling path is dominant here? The input impedance of your receiver dictates which physical noise mechanism will successfully corrupt your signal.
1. Capacitive Coupling (Electric Fields)
Dominant in: High-impedance nodes (>10 kΩ).
Source: 50/60 Hz mains wiring, switching power supplies, digital clock lines.
Mechanism: Parasitic capacitance ($C_{parasitic}$) between the noise source and your high-Z trace forms a voltage divider. The coupled noise current ($I_{coupled}$) flows through your input impedance, generating a noise voltage: $V_{noise} = I_{coupled} \times Z_{in}$. If $Z_{in}$ is 1 MΩ, even picoamps of coupled current result in millivolts of noise.
2. Radiated / Inductive Coupling (Magnetic Fields)
Dominant in: Low-impedance loops (<1 kΩ) and large physical loop areas.
Source: Transformers, motor variable frequency drives (VFDs), relay coils.
Mechanism: Changing magnetic fields induce a voltage in the loop formed by your signal and ground return. Lowering $Z_{in}$ does not protect against this; in fact, low-impedance circuits draw more induced current, exacerbating the voltage drop across shared ground paths.
3. Conductive Coupling (Common Impedance / Ground Loops)
Dominant in: Multi-device systems sharing a ground return.
Source: High-current loads (motors, heaters) sharing a ground plane or wire with sensitive analog inputs.
Mechanism: The return current from a noisy load flows through the finite resistance of the ground path, creating a voltage offset ($V = I_{load} \times R_{ground}$) that appears in series with your low-impedance input signal.
The Fix List: Ranked by Cost and Effectiveness
There is no universal cure for noise. Ferrite beads, for instance, are largely ineffective against low-frequency 60 Hz capacitive hum because their impedance at 60 Hz is negligible. Here are the actual fixes, ranked from the cheapest immediate band-aid to the most robust architectural solution.
- Pull-Down / Pull-Up Resistor (Cost: $0.01 | Effectiveness: Moderate)
The cheapest fix that actually works for floating high-Z inputs is bleeding off accumulated capacitive charge. Adding a 10 kΩ to 100 kΩ resistor to ground lowers the AC/DC impedance of the node, shunting high-frequency displacement currents away from the op-amp gate. Trade-off: Increases source loading and thermal noise. - RC Low-Pass Filter (Cost: $0.05 | Effectiveness: High)
Place a 1 kΩ series resistor followed by a 100 nF ceramic capacitor (C0G/NP0 dielectric for low distortion) to ground. This maintains high DC input impedance for bridging but drops the AC impedance to near-zero above 1.5 kHz, killing RF and switching noise. Use Yageo or Murata C0G caps to avoid piezoelectric microphonics. - Common-Mode Choke + Pi Filter (Cost: $0.30 | Effectiveness: High for EMI)
For differential inputs facing radiated EMI, a common-mode choke (e.g., Wurth Elektronik 744232090) followed by a Pi-filter rejects high-frequency common-mode noise without attenuating the differential signal. - Differential Instrumentation Amplifier (Cost: $3.50 - $8.00 | Effectiveness: Maximum)
If conductive ground loops or massive common-mode noise are the issue, abandon single-ended high-Z inputs entirely. Use an instrumentation amp like the Texas Instruments INA828. It provides high input impedance to both lines while rejecting common-mode noise with a CMRR of >100 dB.
Decision Tree: Picking Your Input Impedance Strategy
Use this decision matrix to terminate your signal chain correctly. Do not guess; match the termination to the physical reality of the cable and the noise environment.
| Signal Type & Environment | Dominant Coupling Path | Required $Z_{in}$ Strategy | Concrete Pick / Part Number |
|---|---|---|---|
| DC / Low-Freq Analog (<10 kHz) over long unshielded cables | Capacitive (Mains Hum) | High DC Z, Low AC Z (RC Filter) | 1 kΩ series + 100 nF C0G Cap (e.g., Kemet C0805C104J5GACTU) |
| RF / High-Speed Digital (>1 MHz) over coax or controlled impedance trace | Reflections & Radiated | Matched Impedance (50 Ω or 75 Ω) | 50 Ω 0402 Resistor (e.g., Yageo RC0402FR-0750RL) |
| Precision Sensor (Thermocouple/Strain Gauge) in industrial VFD environment | Conductive (Ground Loops) & Magnetic | Matched High Z, Differential | INA828 Instrumentation Amp + 100 Ω input protection resistors |
| Electrometer / pH Sensor (Ultra-low current) | Capacitive & Dielectric Absorption | Ultra-High Z with Active Guarding | LMP7721 Op-Amp with PCB guard ring driven by buffer |
Default Recommendation: If you are designing a general-purpose microcontroller ADC input (like on an ESP32 or STM32) and the source is a slow-moving analog sensor, always default to a 10 kΩ series resistor and a 100 nF capacitor to ground. This creates a 159 Hz cutoff, provides a charge reservoir for the ADC's internal sample-and-hold capacitor, and shunts RF noise.
How to Prove the Fix with a Scope or Meter
You cannot manage what you do not measure. Here is the exact before-and-after method to prove your impedance fix is working, avoiding the most common measurement trap: probe loading.
Step-by-Step Verification:
- Establish Baseline (The Floating Test): Disconnect the signal source. Leave the receiver input floating. Connect your 10x scope probe. Set the scope to AC Coupling and measure the RMS voltage. If you read >10 mV RMS of 50/60 Hz hum, your node is highly susceptible to capacitive coupling.
- Apply the Fix: Solder the chosen termination (e.g., the 10 kΩ pull-down or the RC filter) directly at the input pin. Keep lead lengths under 3 mm to prevent parasitic inductance.
- Re-Measure: With the source still disconnected, measure the RMS AC voltage again. A successful fix will drop the RMS noise floor below 1 mV (or the inherent thermal noise floor of the resistor).
- Verify Bandwidth (The Step Response): Reconnect the signal source. Inject a 1 kHz square wave. If the RC filter is too aggressive, you will see rounded edges (exponential RC charging curves) instead of sharp transitions. Adjust the capacitor value if signal fidelity is compromised.
For deeper analysis on how probe capacitance interacts with high-impedance nodes, refer to the Keysight Application Note 5989-6162 on Oscilloscope Probe Loading.
Ground Termination Rules for Shielded High-Z Inputs
A common reflex when dealing with noisy high-impedance inputs is to wrap the cable in a braided shield. However, shielding advice is useless—and actively harmful—without strict ground-termination rules.
If you run a shielded cable to a 1 MΩ input, the parasitic capacitance between the shield and the inner conductor (typically 30-50 pF per foot) forms a low-pass filter with your high $Z_{in}$. A 10-foot cable can easily roll off your high-frequency response. Furthermore, if you ground the shield at both ends, you create a ground loop, turning your shield into a giant inductive pickup coil for magnetic noise.
The Mandatory Grounding Rules:
- Single-Point Grounding: For low-frequency analog signals (<100 kHz), the shield must be grounded at one end only (typically the receiver/system ground side) to prevent ground loop currents from flowing through the shield.
- No Pigtails: Never terminate a shield using a twisted "pigtail" wire to a ground pin. At frequencies above 1 MHz, the inductance of the pigtail renders the shield useless. You must use a 360-degree shield termination, such as a metal backshell or a shield-clamping gland, to maintain a continuous Faraday cage.
- The Driven Shield (Guard) Alternative: For ultra-high impedance circuits (like piezo sensors or pH probes), grounding the shield destroys bandwidth due to cable capacitance. Instead, use a driven shield. Buffer the input signal with a unity-gain op-amp and drive the shield with that buffered signal. Because the shield and the inner conductor are at the exact same AC potential, no current flows through the parasitic capacitance, effectively reducing the cable capacitance to zero. For a complete guide on implementing guard rings, see the Analog Devices MT-036 Tutorial on Op-Amp Input Impedance.
Ultimately, signal integrity is not about eliminating noise; it is about designing the input impedance so that the noise has nowhere to go. Terminate your floating nodes, match your RF lines, and always verify with a 10x probe.
