Whenever the speed of a dc servo motor must be controlled with absolute accuracy, there’s no substitute for tachometer-based feedback combined with a fullblown servo loop. But in other applications, adequate steady-state precision and sometimes superior dynamic performance can be had by taking advantage of the built-in physical constants of the motor itself.
For example, the fact that every permanent magnet dc motor has a fixed relationship between RPM and armature back-EMF means that a fair job of constant speed operation can be accomplished merely by driving the motor from a well-regulated power supply. Even better speed regulation can be achieved, sometimes rivaling that offered by tachometer feedback, by adding a positive-feedback term to the drive voltage that’s proportional to armature current. If this term is adjusted so as to cancel armature resistance (approximately equal to rated-voltage/locked-rotorstall current), motor speed will remain nearly constant over a wide range of loads.
Similarly, a stable data-sheet-specified relationship exists in every PM motor between armature current and output torque and therefore between armature current and acceleration. It follows that a constant maximum-rate spin-up to a specified speed can be achieved by applying full supply voltage until the integrated charge passed by the armature equals a value proportional to the setpoint speed.
This is the control principle employed in the figure. When TTL-/CMOS-compatible input RUN goes high, comparators A1 and A3 enable power amplifier A4+Q5+Q6, turning on the drive to motor M1. Simultaneously, comparator A2 releases the reset on armature current integration capacitor C1. Armature current monitor A5+Q3 injects a current onto C1 that’s proportional to armature current as sensed by emitter resistors R2 and R3 with the constant of proportionality adjusted by R4. Comparator A6 compares the resulting C1 voltage ramp to the setpoint pot R1 (SPEED). So long as V(C1) < V(R1), A6 holds the motor drive Darlingtons Q5 and A6 in saturation, thus applying the full unregulated supply voltage to M1.
This full-throttle acceleration phase continues until C1’s integrated charge causes V(C1) = V(R1). A6 pin 13 then goes low, turning off Q1 and passing control of the Q5 and Q6 Darlingtons over to A4. A4 then strives to hold M1’s speed constant by applying the armature-voltage tricks described in the first paragraph. Feedback from M1 to A4 pin 3 provide basic voltage regulation, while the current-proportional signal from R5 and Q4 provide armature-resistance-canceling positive feedback. Properly adjusting R4 will result in overshoot-free initial spin-up, while careful trimming of R5 provides constant motor speed against changes in load friction. When the RUN control logic input returns low, Q7 effectively shorts across the motor armature to provide dynamic braking and ensure rapid M1 deceleration.
This circuit was designed for service in a magnetic resonance research application in which a liquid sample must be held stationary during excitation with RF pulses, and then spun up to 1200 to 3600 RPM as quickly as possible. The 24-V hollowrotor dc motor used for M1 achieves loaded acceleration rates approaching 1,000,000 RPM/sec. Spin-up times in the single-digit-millisecond range are therefore easily achieved. Deceleration performance is similarly enthusiastic.
The magnetic resonance application entailed only a relatively low duty factor for motor operation. Therefore, the simple power supply illustrated (augmented with buckets—about 7 J—of energy-storage capacitance to satisfy the big startup surge) was adequate.