Permanent magnet direct current (PMDC) motor drives must deliver fast and accurate velocity tracking, yet they remain difficult to control because the terminal voltage and armature current can readily exceed the operating limits imposed by the converter and machine, leading to winding damage and converter degradation. Conventional cascade controllers cannot systematically enforce these physical limits during rapid transients without sacrificing tracking bandwidth, and they confront saturation only after the control command has been generated. This study presents a single-loop constrained finite-set model predictive control (FS-MPC) strategy for high-performance PMDC speed regulation and benchmarks it against standard cascade proportional-integral (PI) and anti-windup cascade PI controllers under identical operating conditions. A discrete-time PMDC model derived from the coupled armature-current and rotor-speed dynamics predicts the future trajectories, and at each sampling instant a finite set of admissible terminal-voltage candidates is evaluated through a cost function that jointly weighs speed-tracking error, control effort, and a soft armature-current limit. The proposed controller reaches the reference speed in approximately 0.02 s with negligible overshoot, whereas the anti-windup and standard cascade PI controllers settle in roughly 0.15 s and 0.18–0.20 s, the latter with pronounced overshoot. This transient gain is obtained at the cost of a higher short-duration current and torque excursion alongside a small finite-set steady-state ripple. The findings indicate that embedding voltage limitation and current-limit penalization within a single predictive decision can curtail loop-interaction and windup effects, offering a practical route to faster, constraint-aware speed control for compact industrial PMDC servo drives.