Noise and Noise Filtering

In high-precision measurement systems, the achievable position stability is not determined by sensor resolution alone. The noise behavior of the entire signal chain is equally important. An ideal sensor signal would be completely constant at standstill; however, in real measurement systems, small statistical fluctuations always occur. These fluctuations are referred to as noise and result in slightly different position values being output even when the measured object is stationary.

In optical encoders such as METIRIO®, analog sine and cosine signals are generated, from which the position is determined by means of a phase calculation. Fluctuations in these signals are therefore transferred directly to the calculated position. The resulting position noise depends not only on the readhead itself, but also on the downstream evaluation electronics, digitization, position calculation, and possible filter stages.

The RMS value is often used to describe noise. It describes the effective strength of the statistical fluctuations and is well suited for comparing different measurement or filter configurations. In addition, a peak-to-peak value is occasionally specified. However, this does not describe a fixed noise limit, but depends, among other things, on measurement duration, bandwidth, and sampling rate.

Frequency Components of Noise

Noise consists of signal components in different frequency ranges. In addition to low-frequency components, which can arise, for example, from mechanical drift or environmental influences, higher-frequency components also occur. The spectral noise density is therefore often considered in order to describe the noise behavior of a system more precisely.

Figure 1 schematically shows how strongly certain frequency components contribute to the overall noise. The RMS value depends on which frequency components are contained in the measurement signal. The larger the bandwidth considered, the more noise components can contribute to the resulting RMS noise. This makes it clear why the bandwidth of a measurement system is crucial for the observed noise: If a larger frequency range is captured, more noise components can also contribute to the measured value. If the bandwidth is reduced, the measurable noise decreases in many cases.

Figure 1: Schematic representation of the spectral noise density. A reduced bandwidth decreases the frequency range that contributes to the measured RMS noise.

Filters and Their Effect

Filters are used to selectively reduce certain frequency components of a signal. Low-pass filters are particularly relevant for many position measurements. They allow slow signal changes to pass and attenuate high-frequency components. This can significantly reduce position noise at standstill or during slow movements.

However, a filter does not act only on the noise, but on the entire measurement signal. If the filter bandwidth is selected too low, fast movements, accelerations, or vibrations can be attenuated or displayed with a delay. The measured position is then quieter, but may no longer fully represent the real movement.

The choice of filtering is therefore always a compromise between lower noise and sufficient dynamics. Stronger filtering can be advantageous for static measurements or slow positioning tasks. For fast movements, highly dynamic control systems, or vibration analyses, however, a larger bandwidth must be maintained.

The Signal Chain as an Overall System

In a high-resolution measurement system, the observed position noise results from the interaction of several components. These include the sensor, analog input stages, digital processing, position calculation, and, where applicable, additional filters.

Figure 2 shows a simplified signal chain of an optical position measurement system. The resulting position noise is not determined by the readhead alone, but also by optional analog filtering, the downstream electronics, and the position output. Depending on the integration, different controllers, interface modules, or additional filter stages may be used.

Figure 2: Simplified signal chain of an optical position measurement system. The resulting position noise is influenced by the METIRIO® readhead, optional analog filtering, the downstream electronics, and the position output.

It is therefore important to always consider noise values in the context of the respective configuration. A single numerical value does not automatically describe the sensor’s performance in every application. Rather, it indicates what noise level was achieved under specific measurement conditions, with a specific bandwidth and specific evaluation electronics.

This system-level view is particularly relevant when sensors are integrated into closed-loop control systems. A strongly filtered position signal can appear very stable, but at the same time it influences the dynamic behavior of the control system. Depending on the application, noise, latency, bandwidth, and motion profile must therefore be optimized together.

Application-Specific Noise Filtering

The optimum filtering depends on the respective measurement task. With a stationary axis, the best possible position stability is often the main priority. In this case, a reduced bandwidth can help to decrease high-frequency noise components. During continuous motion, however, the filter bandwidth must be high enough to capture the position change without significant attenuation.

Particular care is required for vibrating or accelerating systems. Filters can not only attenuate certain frequency components; depending on the structure of the signal processing, they can also influence the dynamic behavior. Filter settings should therefore always be checked based on the real application, especially when the measurement signal is used for active control.

Figure 3 shows the filter effect in the time domain. With a stationary axis, the unfiltered position signal fluctuates around its mean value. A low-pass filter attenuates fast signal components, making the filtered signal appear more stable. The illustration deliberately shows the effect of noise reduction. In dynamic applications, it must also be checked which bandwidth has to be maintained for motion, control, or vibration analysis.

Figure 3: Schematic representation of the filter effect in the time domain. Low-pass filtering reduces fast fluctuations in the position signal and thereby decreases the visible position noise. The appropriate filter bandwidth depends on the dynamics of the application.

SmarAct Metrology therefore does not consider noise and filtering in isolation, but as part of the entire measurement architecture. The goal is to coordinate sensors, evaluation electronics, and signal processing in such a way that the best possible combination of resolution, stability, and dynamics is achieved for the respective application.

METIRIO® and High-Precision Position Feedback

METIRIO® is a compact optical encoder for high-resolution position measurements.

For an application-oriented integration solution, SmarAct offers low-pass filters with different cutoff frequencies, allowing the noise filtering to be adapted to static measurements, slow positioning tasks, or applications with higher dynamics.

This allows the position measurement to be specifically tailored to the respective application — from stable standstill measurements to highly dynamic control processes.