Compact, High-Dynamics Motor Design Breaks New Ground for Industrial Robotics
In an era where industrial automation is rapidly evolving, the demand for compact, high-performance servo motors has never been greater. Traditional surface-mounted permanent magnet synchronous motors (SPMSMs), while reliable and widely adopted, face inherent limitations in meeting the stringent requirements of modern industrial robots—particularly in terms of size, weight, and dynamic responsiveness. A newly proposed motor architecture promises to overcome these barriers, offering a compelling alternative that could reshape the landscape of robotic actuation.
Engineers at Shandong University and Shandong Jingchuang Magnetoelectric Industry Institute Co. Ltd have developed a novel Disk-type Transverse Flux Permanent Magnet Brushless Motor (DTFM) specifically tailored for industrial robotics applications. This design eliminates axial winding overhangs, leverages a low-inertia disk-shaped rotor, and integrates soft magnetic composite (SMC) materials to achieve a striking balance between miniaturization, torque density, and dynamic performance. The results, detailed in a recent study published in Transactions of China Electrotechnical Society, demonstrate significant improvements over conventional motors in both physical footprint and operational agility.
The core innovation of the DTFM lies in its structural reimagining. Unlike radial-flux motors, where magnetic flux travels parallel to the rotational axis, transverse flux machines route magnetic flux perpendicular to the axis, effectively decoupling the electrical and magnetic circuits. This architectural freedom allows for more efficient use of internal volume. By adopting a disk-shaped configuration—where stator modules are arranged on both sides of a thin rotor disc—the team further reduced axial length while maintaining high torque output.
“Industrial robots need to start and stop rapidly, often hundreds or thousands of times per shift,” explained Zhang Wenjing, lead author and doctoral researcher at Shandong University’s School of Electrical Engineering. “Every gram of rotor mass and every millimeter of axial length matters. Our DTFM design directly addresses these constraints.”
One of the most persistent challenges in transverse flux motor design has been high cogging torque—unwanted torque ripple caused by interactions between permanent magnets and stator teeth during rotation. Excessive cogging torque leads to vibration, noise, and reduced positioning accuracy, all of which are unacceptable in precision robotic tasks. The research team tackled this issue through a sophisticated multi-objective optimization strategy that combined backpropagation (BP) neural networks with a multi-objective particle swarm optimization (PSO) algorithm.
Rather than relying solely on computationally intensive finite element analysis (FEA) for every design iteration, the team first generated a training dataset using an L25 orthogonal experimental design. This dataset mapped four key geometric parameters—namely, the angular spans of inner and outer permanent magnet arrays (α_in and α_out), the stator pole shoe angle (β_s), and the SMC pole shoe thickness (h_SMC)—to two critical performance metrics: average output torque under rated current (12.5 A) and peak-to-peak cogging torque under no-load conditions.
A BP neural network, enhanced with genetic algorithm-based training to avoid local minima and improve generalization, was then trained to predict motor performance from these geometric inputs. This surrogate model enabled rapid evaluation of thousands of design candidates within the PSO framework, ultimately yielding a Pareto-optimal solution set that balanced torque maximization against cogging torque minimization.
The selected optimal configuration featured α_out = 31.5°, α_in = 28.5°, β_s = 25.3°, and h_SMC = 4.1 mm. Three-dimensional FEA validation confirmed the model’s accuracy: the optimized DTFM achieved an average output torque of 20.06 N·m—virtually unchanged from the initial design—while slashing cogging torque from 2.07 N·m to just 0.24 N·m, a reduction of over 88%.
To validate these findings in the physical world, the team constructed a full-scale prototype. Experimental tests at 3,000 rpm revealed a clean, sinusoidal back-electromotive force (back-EMF) waveform, confirming the motor’s suitability for field-oriented control strategies commonly used in high-performance servo systems. Dynamic response tests showed that the motor could accelerate to and decelerate from 3,000 rpm in approximately 2 seconds under no-load conditions—a testament to its low rotational inertia.
Under load, the prototype delivered a rated torque of 16.5 N·m at 3,000 rpm, with a peak efficiency of 90% and an efficiency of approximately 82% at nominal operating conditions (including inverter losses). While the measured torque was slightly lower than simulated predictions—attributed to minor manufacturing tolerances such as chamfering on silicon steel laminations and small air gaps between SMC pole shoes and the stator core—the overall performance remained highly competitive.
Most notably, when compared directly to a commercially available radial-flux permanent magnet servo motor of equivalent power rating, the DTFM prototype demonstrated dramatic advantages in form factor and dynamics. Its axial length was reduced by 40.8% (190 mm vs. 308 mm), total mass decreased by 21.9% (22.8 kg vs. 28 kg), and—perhaps most critically—rotational inertia dropped by 54.2% (34.76 kg·cm² vs. 76 kg·cm²). These reductions translate directly into faster acceleration, improved energy efficiency, and enhanced control precision in robotic joints.
The strategic use of SMC materials played a pivotal role in enabling this performance. Unlike traditional laminated silicon steel, which exhibits anisotropic magnetic properties and higher core losses at high frequencies, SMC offers near-isotropic behavior, lower eddy current losses, and superior formability for complex 3D geometries. In the DTFM, SMC was used exclusively for the stator pole shoes—the components that interface directly with the air gap and shape the magnetic flux path—while the stator yoke and core limbs retained silicon steel for its higher permeability. This hybrid approach optimized both magnetic performance and manufacturability.
From an engineering perspective, the DTFM also simplifies assembly. Earlier C-core transverse flux designs required intricate stacking and alignment of monolithic stator segments. By splitting the C-core into two U-shaped halves mounted on opposite sides of the rotor disc, the new architecture allows for modular assembly and easier integration of ancillary components—such as electromagnetic brakes and encoders—within the central bore of the stator. This design foresight ensures compatibility with standard industrial robot joint housings without compromising axial compactness.
The implications extend beyond robotics. Any application demanding high torque density in a constrained axial envelope—such as aerospace actuators, medical robotics, or electric vehicle wheel drives—could benefit from this architecture. Moreover, the optimization methodology itself represents a significant methodological contribution, showcasing how machine learning can accelerate electromagnetic design while maintaining physical fidelity.
Critically, the research adheres to rigorous scientific standards. All simulations were validated against physical prototypes, and performance claims are grounded in empirical data from controlled experiments, including no-load back-EMF measurements, dynamic response tracking, torque-current characterization, and efficiency mapping. The comparative analysis against a market-available motor further strengthens the credibility of the findings by providing real-world context.
Looking ahead, the team plans to explore advanced cooling strategies to further increase power density and investigate the integration of the DTFM with next-generation wide-bandgap semiconductor inverters for even higher switching frequencies and efficiency. They also aim to refine the manufacturing process to minimize performance deviations between simulation and reality—particularly by optimizing the interface between SMC and silicon steel components.
This work underscores a broader trend in electromechanical engineering: the convergence of novel topologies, advanced materials, and intelligent design algorithms to solve longstanding performance bottlenecks. As industrial robots continue to penetrate new domains—from micro-assembly in electronics to collaborative tasks alongside human workers—the need for smarter, smaller, and more responsive actuators will only intensify. The DTFM represents a significant step toward meeting that need.
Authors: Zhang Wenjing¹, Xu Yanliang¹, Li Shucai²
Affiliations:
¹ School of Electrical Engineering, Shandong University, Jinan 250061, China
² Shandong Jingchuang Magnetoelectric Industry Institute Co. Ltd, Linyi 276000, China
Journal: Transactions of China Electrotechnical Society, Vol. 36, No. 14, July 2021
DOI: 10.19595/j.cnki.1000-6753.tces.200401