Gene F. Franklin

Supervisory control of a rapid thermal multiprocessor

An application of supervisory control theory to a piece of semiconductor manufacturing equipment is presented. The approach allows the flexible design and reliable update of processing recipes to accommodate frequently changing manufacturing requirements. An input-output interpretation of supervisory control theory is given. This interpretation leads to a generic implementation scheme for manufacturing systems. A synthesis fixpoint algorithm implementation using binary decision diagrams enables the design of supervisors of realistic size. A sample synthesis for an oxide growth recipe is performed on a state space of the order of 10/sup 6/ states. The actual implementation of the logic sequencing control software for the application under investigation is described. >

Control systems engineering education

This comprehensive article deals with the important field of control systems engineering education. Efforts have been made to present some historical perspectives, major concepts and thoughts on a practical curriculum when this field is viewed as a discipline. Also discussed are curricular issues including typical laboratory systems with emphasis on the role of simulation, logic and sequencing, and real-time simulation. An elaborate section is devoted to CACE software, its role in teaching and learning, potential shortfalls, and trends in software development and use. Views from industry are sought in terms of desirable skills in the practicing engineer and continuing education needs. A survey of a few academic programs and a complete list of textbooks in control over the past three decades appear in the appendices.

A brief history of disk drive control

The purpose of this article is to provide a history of control in disk drives for rigid magnetic disks (often called hard disks). It begins with a short general history of disk drives. It then considers the feedback loop and trace the evolution of the individual components. Finally, we will focus on the control system itself, viewed in the context of overall history and the component history.

Model reduction via balanced realizations: an extension and frequency weighting techniques

Two model-reduction methods for discrete systems related to balanced realizations are described. The first is a technique which utilizes the least controllable and observable subsystem in deriving a balanced discrete reduced-order model. For this technique as L/sup infinity / norm bound on the reduction error is given. The second method is a frequency-weighting technique for discrete- and continuous-time systems where the input-normal or output-normal realizations are modified to include a simple frequency weighting. For this technique, L/sup infinity / norm bounds on the weighted reduction errors are obtained. >

An error bound for a discrete reduced order model of a linear multivariable system

The design of feasible controllers for high dimension multivariable systems can be greatly aided by a method of model reduction. In order for the design based on the order reduction to include a guarantee of stability, it is sufficient to have a bound on the model error. Previous work has provided such a bound for continuous-time systems for algorithms based on balancing. In this note an Linftybound is derived for model error for a method of order reduction of discrete linear multivariable systems based on balancing.

A new optimal multirate control of linear periodic and time-invariant systems

The optimal multirate design of linear, continuous-time, periodic and time-invariant systems is considered. It is based on solving the continuous linear quadratic regulation (LQR) problem with the control being constrained to a certain piecewise constant feedback. Necessary and sufficient conditions for the asymptotic stability of the resulting closed-loop system are given. An explicit multirate feedback law that requires the solution of an algebraic discrete Riccati equation is presented. Such control is simple and can be easily implemented by digital computers. When applied to linear time-invariant systems, multirate optimal feedback optimal control provides a satisfactory response even if the state is sampled relatively slowly. Compared to the classical single-rate sampled-data feedback in which the state is always sampled at the same rate, the multirate system can provide a better response with a considerable reduction in the optimal cost. In general, the multirate scheme offers more flexibility in choosing the sampling rates. >

Design of ripple-free multivariable robust servomechanisms

The design of multivariable control systems to provide zero steady-state error in the presence of non-decaying reference and disturbance signals and in spite of perturbations to system parameters is referred to as the robust servomechanism problem. The application of standard robust servomechanism theory to sampled data systems can guarantee asymptotic tracking at sampling instants only. Between sampling instants, the output will normally contain a component referred to as ripple. In this paper, the robust servomechanism theory is extended to sampled-data systems and a technique is proposed for ripple-free tracking of nondecaying inputs including sinusoids and polynomials. It is shown that a continuous internal model is necessary and sufficient, to provide ripple-free response. All the poles of the system can be assigned to the origin to yield a continuous ripple-free deadbeat response. The integrators of the hold devices are used as part of the continuous internal model for constant inputs. An example is provided to illustrate the design technique.

Disk drive pivot nonlinearity modeling. I. Frequency domain

This paper describes studies done at HP Labs on the actuator pivot bearing nonlinearity of a small disk drive. The nonlinear frictional behavior of the pivot bearing varies with actuator position, from drive to drive, and with time and temperature. This nonlinear behavior has made the traditional linear disk drive models inadequate. The paper shows how the swept-sine/describing function method was used to develop a nonlinear model of the pivot. This model departs from the classical friction models, but does a good job of matching laboratory frequency domain measurements.

On the stability of adaptive pole-placement controllers with a saturating actuator

The authors examine how existing pole-placement algorithms work in the case of a saturating input. The stability of such algorithms and the modifications needed to make them work in the case of saturation are examined. The net result is that the analysis of G.C. Goodwin and K.S. Sin (1984) can be slightly modified to prove the stability of adaptive pole-placement, but only for the case where the plant is stable. It is shown that, for unstable plants, it is possible to push the plant into a state from which it cannot be returned to the origin (with a saturating input). Thus, it is fairly easy to generate examples of instability in the adaptive control system that would not exist with unbounded control authority. >

Experiments in load-adaptive control of a very flexible one-link manipulator

Abstract An adaptive control algorithm based on the Self-Tuning Regulator concept has been experimentally demonstrated on a very flexible one-link robotic manipulator. The very lightly-damped structural resonances of the manipulator coupled with the use of a primary sensor separated from the actuator by flexible structure (non-collocated sensor and actuator) to give good accuracy in tip positioning make high-performance controllers very sensitive to modeling error. The adaptive controller is able to maintain precise tip position control despite wide variations in end effector load. An identification algorithm is employed which estimates the mass of the tip load. It makes use of a simple parameterization of system transfer functions which is a linear-fractional expression in the mass of the load. The Linear Quadratic Gaussian synthesis procedure is used for control design, a polynomial interpolation to a precalculated table of controller gains is made on-line to implement a controller appropriate to the identified value of load mass. Rate of convergence of the identification algorithm is such that the system is able to adapt to a 40% change in moment of inertia during a single commanded step change in position.