Magnetic Ball Suspension System (MBSS)

• MBSS is a nonlinear, contactless electromechanical system that uses magnetic forces to levitate metallic balls with high precision.
• Its modeling integrates nonlinear state-space representations and MIMO configurations to capture dynamic behaviors and determine equilibrium points.
• Robust control strategies—including state feedback, LQR, observer-based, and fuzzy logic methods—ensure stable operation amid inherent nonlinearities.

A Magnetic Ball Suspension System (MBSS) is a class of inherently nonlinear and open-loop unstable electromechanical systems designed to levitate a steel or magnetic ball using electromagnetic forces with no physical contact. MBSSs are foundational in precision engineering, magnetic bearings, and contactless actuation, and extensively serve as benchmark platforms for advanced control theory validation. They manifest in various configurations, including classical single-axis electromagnetic suspension, multi-floating-body arrays, and rotational analogs analogous to Levitron systems (Nwachukwu, 22 Jan 2026, Doff et al., 29 Jun 2025, Abuelenin, 2015).

1. Physical Principles and System Modeling

The MBSS operates by counterbalancing gravitational force with an attractive magnetic force generated by an electromagnet or magnet assembly. The vertical position $y(t)$ of a ball of mass $m$ is modeled by the net force: $m\ddot{y} = m g - F_\mathrm{em}(i, y),$ where $F_\mathrm{em}(i, y) = K i^2 / y^2$ is the magnetic force, with $K$ dependent on actuator geometry (Nwachukwu, 22 Jan 2026). The actuating electromagnetic coil is modeled as a resistive-inductive (R–L) circuit with position-dependent inductance $L(y) = L_0 / y$: $u(t) = R i(t) + L(y) \dot{i}(t),$ with $u(t)$ as input voltage. The state variables typically are position $y$, velocity $\dot{y}$, and coil current $i$.

This model generalizes to Multi-Input Multi-Output (MIMO) settings with several balls (floaters) and actuators. For each channel, magnetic force satisfies

$F_m(i, z) = \frac{\mu_0 N^2 A}{2 z^2} i^2,$

where $N$ is coil turns, $A$ is cross-sectional area, $\mu_0$ is permeability, and $z$ is vertical gap (Abuelenin, 2015).

Alternative MBSSs employ permanent magnet dynamics, as in the “dynamical Levitron” configuration, leveraging off-axis rotation and spatial arrangement rather than current control to achieve levitation (Doff et al., 29 Jun 2025).

2. State-Space Representation and Equilibrium Analysis

A canonical MBSS can be cast into nonlinear state-space form: $$\begin{align*} \dot{x}_1 &= x_2, \ \dot{x}_2 &= g - \frac{K x_3^2}{m x_1^2}, \ \dot{x}_3 &= \frac{x_1}{L_0} \left( u - R x_3 \right), \end{align*}$$ where $x_1 = y$, $x_2 = \dot{y}$, $x_3 = i$, and input $u(t)$ is the electromagnet coil voltage (Nwachukwu, 22 Jan 2026).

At equilibrium, equating derivatives to zero yields: $$m g = \frac{K i_0^2}{x_{10}^2} \implies x_{10} = \sqrt{\frac{K}{mg}} i_0.$$ This defines the bias current and nominal suspension point for given parameters. Linearization via Jacobian analysis around this equilibrium gives explicit matrices $A$, $B$ suitable for classical control design. Representative parameter values are $m = 0.2\,\text{kg}$, $g = 9.8\,\text{m/s}^2$, $R = 10\,\Omega$, $L_0 = 0.5\,\text{H·m}$, $K = 0.01\,\text{N·m}^2/\text{A}^2$, and an equilibrium point $x_0 = [0.06,\,0,\,0.8]^\top$, $u_0 = 8\,\text{V}$ (Nwachukwu, 22 Jan 2026).

3. Controllability, Observability, and System Properties

Controllability and observability are central to MBSS control feasibility. The canonical (linearized) model's controllability matrix $\mathcal{C}$ and observability matrix $\mathcal{O}$ are constructed as: $$\mathcal{C} = [B,\, AB,\, A^2B], \quad \mathcal{O} = \begin{bmatrix} C \ CA \ CA^2 \end{bmatrix}.$$ Full rank ($3$ in single-axis MBSS) establishes the ability to achieve arbitrary state regulation and state estimation using output feedback, even in the presence of unmeasured plant states (Nwachukwu, 22 Jan 2026, Abuelenin, 2015).

Multi-floater MIMO settings, with local plant blocks coupled via global feedback, retain this property under decoupling assumptions, subject to channel symmetry and cross-coupling avoidance (Abuelenin, 2015).

4. Control Architectures and Algorithms

MBSS stabilization and regulation demand sophisticated nonlinear and robust control methodologies:

4.1 Pole-Placement State Feedback

For state measurement availability, pole-placement (e.g., via Ackermann’s formula or \texttt{MATLAB place}) directly assigns closed-loop eigenvalues. Typical targets are fast, non-oscillatory roots (e.g., $\{-5, -10, -20\}$) yielding control law

$u = -K x,$

with $K$ calculated to match desired pole locations (Nwachukwu, 22 Jan 2026).

4.2 Observer-Based Control

When direct state measurement is impractical, full-order Luenberger observers reconstruct state vectors: $$\dot{\hat{x}} = A \hat{x} + B u + L(y - C \hat{x}),$$ where observer gain $L$ is placed for eigenvalues significantly faster than the control loop (e.g., $\{-20, -30, -40\}$), ensuring rapid convergence of the estimation error dynamics (Nwachukwu, 22 Jan 2026).

4.3 Optimal and Robust Control

Linear Quadratic Regulator (LQR) control solves: $J = \int_0^\infty (x^\top Q x + u^\top R u)\,dt,$ yielding the optimal feedback gain via the algebraic Riccati equation. Proper selection of weighting matrices $Q, R$ ensures a trade-off between performance (speed, accuracy) and control effort (Nwachukwu, 22 Jan 2026).

4.4 Fuzzy Logic and Hybrid Control

MIMO MBSSs frequently employ a hierarchical fuzzy logic architecture:

• Main Fuzzy Logic Controllers (FLCs) for each channel, mapping error and error rate to control action via fuzzy rules and membership functions.
• Supervisory FLCs that adapt gains based on error magnitude, extending dynamic range for large displacements.
• Global plane-leveling PD controller to enforce coplanarity and suppress cross-axis sway (Abuelenin, 2015).

Such hybrid arrangements substantially improve stability under strong nonlinearities, actuator saturation, and parameter variation.

5. Dynamical Levitron and Geometric Analogs

A distinct MBSS incarnation, the “dynamical Levitron,” utilizes the interplay of geometric offset ($\delta_R$) and rotation ($\omega_R$) between two Neodymium permanent magnets. Levitation does not require precise rotor (floater) spin but is stabilized by lateral displacement and high rotational speed of the rotor magnet. System equilibrium and stability are set not by $\mu/m$ (magnetic moment/mass) ratio alone, but by geometric parameters: $m_f/\mu_f \propto \delta_R,$ with a critical offset $\delta_R^c \sim 0.3\,\text{mm}$ ensuring stability for typical separation $r_f \sim 10\,\text{mm}$. This paradigm offers a complementary mechanism to current-driven MBSS, circumventing some constraints of classical Levitron systems (Doff et al., 29 Jun 2025).

6. Performance Metrics and Comparative Simulation

MBSS assessment necessitates quantitative metrics:

• Overshoot ($M_p$, %)
• Rise time ($t_r$, s)
• Settling time ($t_s$, s)
• Integrated control effort ($\int_0^T \|u(t)\|^2 dt$)

Simulation campaigns demonstrate that:

• Linear models display minimal oscillations and rapid stabilization.
• Nonlinear simulations incur larger overshoots and longer settling times due to strong nonlinearity and input constraints.
• LQR-based controllers minimize control effort and offer robustness, though their closed-loop performance is not universally superior to well-tuned state feedback.
• Observer-based output feedback tracks the performance of full-state controllers where direct measurement is infeasible (Nwachukwu, 22 Jan 2026, Abuelenin, 2015).

Method	$M_p$	$t_r$	$t_s$	$\int \|u\|^2 dt$
State-Feedback LTI	2%	0.5	2.0	10.0
State-Feedback NL	5%	1.0	5.0	15.0
Observer-LTI	3%	0.6	2.5	12.0
LQR-LTI	1%	0.4	1.8	8.0
LQR-NL	4%	0.9	4.8	11.0

Representative results confirm that nonlinearity substantially affects transient response and energy consumption.

7. Challenges, Design Considerations, and Extensions

MBSS practical implementation requires mitigation of sensor noise, parameter uncertainty (notably in $K$ and $L(y)$), and actuator saturation. Avoidance of friction requires rigorous levitation margin preserving sufficient clearance at equilibrium points (Nwachukwu, 22 Jan 2026). Tuning of control and observer gains—especially LQR weighting and fuzzy logic scaling factors—is critical for robust and stable operation, particularly across varying payload mass or actuator non-idealities.

The supervisory fuzzy structure is extendable to more complex MBSS topologies, including ring or multi-ball arrangements, via modular expansion of local and global controllers (Abuelenin, 2015). The “dynamical Levitron” approach offers distinct design trade-offs, leveraging geometric placement rather than electromagnetic feedback, and suggests new avenues for passive stabilization in scaled systems (Doff et al., 29 Jun 2025).

---

References

(Nwachukwu, 22 Jan 2026): "Design, Modelling, and Control of Magnetic Ball Suspension System" (Doff et al., 29 Jun 2025): "Magnetic levitation by rotation described by a new type of Levitron" (Abuelenin, 2015): "Design and simulation of a hybrid controller for a multi-input multi-output magnetic suspension system"