Access Official Technical PDF

Share Summary:LinkedInEmail

In industrial automation, a sensor signal may be small long before it reaches a controller, data-acquisition system, or converter. Noise introduced by the sensor interface and signal-conditioning electronics can obscure useful information and limit measurement resolution. Effective low-noise system design therefore starts with the complete measurement chain: the sensor, its source impedance, the amplifier, the signal bandwidth, and the measurement method.

The goal is not to eliminate every fluctuation. Fundamental noise arises from the physical behavior of components and cannot be removed entirely. Instead, engineers identify the important noise sources, estimate their combined effect, and shape the system so that unwanted noise contributes as little as practical while the required signal remains intact.

Distinguish interference from fundamental noise

Unwanted disturbances can come from outside the circuit or from the components themselves. External interference may couple into a system from power wiring, radio transmitters, nearby circuits, or mechanical vibration. Depending on the problem, shielding, filtering, improved layout, or a change in measurement setup may reduce these effects.

Fundamental, or intrinsic, noise is different. It is generated by the electrical devices and materials in the system. Thermal noise in a resistance is one example. Shot noise and low-frequency noise are other mechanisms relevant to electronic design. Unlike external pickup, intrinsic noise cannot be completely removed by relocating or shielding the circuit. Its effects must be managed through device and circuit choices.

This distinction helps guide troubleshooting. If shielding or changes to the measurement arrangement reduce the disturbance, external coupling may be significant. If not, engineers should consider the noise generated by the sensor, resistors, and active devices themselves.

Thermal noise: a baseline for circuit design

Thermal noise results from the random motion of charge carriers in a conductor. Its voltage depends on resistance, absolute temperature, and the bandwidth over which noise is measured. For a resistor, the root-mean-square thermal noise voltage is expressed as E = √(4kTRΔf), where k is Boltzmann’s constant, T is temperature in kelvins, R is resistance, and Δf is noise bandwidth.

The relationship has practical consequences. Increasing resistance or measurement bandwidth increases the resistor’s noise voltage; reducing either reduces it. However, narrowing the bandwidth is useful only when the system can still pass the signal changes it needs to measure. Similarly, a low source resistance may reduce thermal noise, but the sensor’s electrical characteristics and interface requirements also matter.

Thermal noise is often described as white noise because its power is distributed evenly across frequency over the range where the model applies. The total noise measured by a system depends not just on the noise level per unit bandwidth, but on how much frequency content the system accepts.

Account for noise bandwidth, not just cutoff frequency

A filter’s conventional −3 dB bandwidth and its noise bandwidth are not necessarily the same. The −3 dB point describes where signal power has fallen to half its reference level. Noise bandwidth, by contrast, is the width of an ideal rectangular response that would pass the same total noise power as the actual frequency response.

Because real circuits have shaped frequency responses, using only the −3 dB bandwidth can misstate the noise accumulated at the output. For simple filter arrangements, noise bandwidth can be calculated from the frequency response. For more complex systems, numerical analysis or circuit simulation can help estimate the integrated noise over the relevant frequency range.

In practice, bandwidth should be set deliberately. A measurement chain that accepts more frequency content than the application requires may include additional noise. But filtering too aggressively can also remove useful signal information. The design task is to balance signal response and noise performance.

Model the whole sensor and amplifier interface

An amplifier should be evaluated in relation to the source it must measure. A sensor brings its own signal and noise characteristics, as well as an output impedance. The amplifier adds noise too. A useful analysis represents amplifier noise with equivalent input noise voltage and current sources, then considers their effects alongside the sensor and other circuit elements.

This approach supports more informed decisions about active devices and circuit topology. It also helps engineers locate the dominant contribution: the sensor, the amplifier, or another part of the signal path. If one source is already much larger than the others, reducing a smaller contribution may have little effect on total system noise. Noise contributions from independent sources are combined by their mean-square values, rather than by simply adding their root-mean-square amplitudes.

Feedback, passive components, power supplies, and voltage references also belong in a system-level assessment. A quiet amplifier alone does not guarantee a quiet measurement if other circuit elements add significant noise or interference.

Use simulation and measurement together

SPICE-based tools can support noise analysis by representing circuit elements and their noise sources, examining frequency-dependent behavior, and estimating noise over a specified band. Simulation is most useful when its models and assumptions reflect the circuit being designed. It can compare design alternatives, but it does not replace checking the assembled system.

Measurement provides a practical check on the noise model. The result depends on the instrument, measurement bandwidth, and observation time. The test setup itself must not introduce significant interference or alter the system response. Comparing measured and modeled results can help reveal whether the noise assumptions, circuit implementation, or test conditions need closer examination.

Design implications for automation systems

For engineers developing sensor interfaces and low-level analog front ends, a disciplined workflow can make noise decisions clearer:

  • Define the measurement need: Identify the signal range and frequency content that must be preserved.
  • Characterize the source: Include the sensor’s output impedance and noise in the analysis.
  • Identify major contributors: Compare sensor, amplifier, and component noise rather than focusing on a single device specification.
  • Choose bandwidth intentionally: Limit unwanted frequency content without suppressing useful signal information.
  • Verify the design: Use circuit analysis and simulation, then measure with appropriate bandwidth and instrumentation.

These steps apply across instrumentation and control systems, from sensor preamplifiers to analog interfaces feeding converters. They help teams distinguish an electronics limitation from a system-level interference problem and focus design effort where it can matter most.

Conclusion

Low-noise system design is a system problem, not simply an amplifier-selection exercise. Thermal noise establishes a baseline, while source impedance, amplifier characteristics, circuit components, and noise bandwidth determine how that baseline affects a measurement. By modeling the sensor-to-converter path, controlling bandwidth, and validating assumptions through measurement, automation teams can make sounder decisions about signal conditioning and achievable measurement performance.

Access Official Technical PDF

About The Author