Integrating Active Damping with Shaping-Filtered Reset Tracking Control for Piezo-Actuated Nanopositioning

Piezoelectric nanopositioning systems are often limited by lightly damped structural resonances and the gain–phase constraints of linear feedback, which restrict achievable bandwidth and tracking performance. This paper presents a dual-loop architecture that combines an inner-loop non-minimum-phase resonant controller (NRC) for active damping with an outer-loop tracking controller augmented by a constant-gain, lead-in-phase (CgLp) reset element to provide phase lead at the targeted crossover without increasing loop gain. We show that aggressively tuned CgLp designs with larger phase lead can introduce pronounced higher-order harmonics, degrading error sensitivity in specific frequency bands and causing multiple-reset behavior. To address this, a shaping filter is introduced in the reset-trigger path to regulate the reset action and suppress harmonic-induced effects while preserving the desired crossover-phase recovery. The proposed controllers are implemented in real time on an industrial piezo nanopositioner, demonstrating an experimental open-loop crossover increase of approximately 55Hz and a closed-loop bandwidth improvement of about 34Hz relative to a well-tuned linear baseline.

💡 Research Summary

**
The paper addresses the bandwidth limitation of piezo‑actuated nanopositioning stages caused by lightly damped structural resonances (first mode at 710 Hz, second at 1150 Hz). Conventional proportional‑integral (PI) control cannot push the closed‑loop crossover frequency beyond a fraction of the first resonance without sacrificing phase margin or amplifying noise. To overcome this, the authors propose a dual‑loop architecture.

The inner loop employs a non‑minimum‑phase resonant controller (NRC) for active damping. NRC provides gain‑phase decoupling, allowing aggressive attenuation of the dominant resonance without adding phase lag. Its parameters are set as (k = \gamma |G(0)|^{-1}) and (\omega_a = n \omega_n), where (\omega_n) is the first resonant frequency.

The outer loop handles trajectory tracking. It consists of a conventional PI controller, notch filters for higher‑order modes, a low‑pass filter, and, crucially, a constant‑gain lead‑in‑phase (CgLp) reset element placed in series with the tracking controller. CgLp is built from a generalized first‑order reset element (GFORE) combined with a linear lead‑lag filter. The design yields an approximately flat magnitude response while delivering a selectable phase lead (5°–20°) at the desired crossover frequency, thereby restoring phase margin that was sacrificed by aggressive linear tuning.

A key observation is that larger phase‑lead settings increase the nonlinearity of the reset action, which excites higher‑order harmonics. Using higher‑order sinusoidal‑input describing functions (HOSIDFs), the authors show that the dominant third harmonic can cause gain fluctuations of –1 to +2 dB around the crossover and, more importantly, creates multiple zero‑crossings in the reset signal. This “multiple‑reset” phenomenon invalidates the standard HOSIDF assumption (two resets per period) and leads to degraded error‑sensitivity and tracking performance, especially for the 15° and 20° phase‑lead cases.

To mitigate these adverse effects, a shaping filter (C_s(s)) is inserted in the reset‑trigger path. The filter processes the error signal before it reaches the reset condition, effectively limiting the frequency content that can cause spurious resets. Consequently, the higher‑order harmonics are suppressed, the reset signal exhibits only the intended two‑reset‑per‑period behavior, and the sensitivity peaks are reduced.

Experimental validation is performed on a commercial PIHera single‑axis nanopositioner (100 µm travel) with a NI CompactRIO‑FPGA platform running at 33.33 kHz. System identification yields a plant model (G(s)) with the aforementioned resonances. Seven cases are tested:

• Case 1 – well‑tuned linear controller (baseline).
• Cases 2–5 – same linear controllers but augmented with CgLp providing 5°, 10°, 15°, and 20° phase lead, respectively.
• Cases 6–7 – Cases 5 and 4 plus the shaping filter in the reset path.

Results show that the baseline achieves an open‑loop crossover around 210 Hz and a closed‑loop bandwidth of ~120 Hz. Adding CgLp shifts the crossover up to 260 Hz and bandwidth to ~170 Hz, but the 15°/20° configurations suffer from pronounced third‑harmonic peaks and increased tracking error. Incorporating the shaping filter restores the intended phase lead while suppressing the harmonic peaks, pushing the crossover to ~275 Hz (≈ 55 Hz improvement over baseline) and the closed‑loop bandwidth to ~154 Hz (≈ 34 Hz gain). Sensitivity functions become smoother, and tracking RMS error is reduced by roughly 20 % compared with the linear baseline.

The paper’s contributions can be summarized as follows:

1. Integration of NRC and CgLp – Demonstrates that active damping of resonances and nonlinear phase‑lead compensation can be combined effectively in a dual‑loop scheme.
2. Quantitative HOSIDF analysis – Provides insight into how higher‑order harmonics arise from aggressive reset designs and how they affect sensitivity and stability.
3. Shaping‑filter solution – Introduces a simple yet powerful filter in the reset‑trigger path that mitigates harmonic‑induced multiple resets without compromising the desired phase lead.
4. Experimental verification – Validates the approach on real hardware, showing substantial bandwidth expansion (≈ 55 Hz open‑loop, ≈ 34 Hz closed‑loop) and improved tracking performance.

Overall, the work advances the state‑of‑the‑art in precision motion control by showing that carefully engineered nonlinear reset elements, when paired with appropriate shaping, can break the traditional water‑bed limitations of linear controllers and deliver higher speed without sacrificing robustness. The methodology is applicable not only to piezo‑based nanopositioners but also to any precision electromechanical system where lightly damped structural modes limit performance.