The phenomenal improvement in motion-control electronics is both a blessing and a curse for designers of control systems. It enables the designers to develop more sophisticated motion-control systems with higher levels of performance. But, it also expands the number of potential solutions that must be evaluated. However, a software design package that simulates the control system operation can both improve its performance and do it with less effort.
At a major elevator-door manufacturer in Connecticut, engineers wanted to lower development costs, raise product quality, optimize motion-control electronics, and speed the time to market of their door systems, Figure 1. Using a software simulation and design package, called VisSim, from Visual Solutions, Inc., Westford, Mass., they designed, prototyped, and optimized a door system in roughly half the time and cost of the previous trial-and-error method. And, this was only their first time using the program.
The elevator-door system consisted of the following five components, of which the first three are similar to those used in previous designs. Only the digital controller and encoder (last two items) were new:
• A 1/4-hp, armature-controlled dc motor with a rated speed of 85 rad/sec at 110 V. The motor drive controller regulates the armature voltage and current and provides a fixed dc supply for the field windings. This constant field excitation produces a fixed-field flux so the motor torque depends only on the armature current. Modulating the voltage to the armature controls the speed and the voltage polarity determines the direction of rotation.
• A 10:1 ratio gearbox to reduce the motor speed applied to the door. A standard bevel-gear arrangement was found to minimize system backlash at a reasonable cost. The gearbox is used in conjunction with a belt drive (see following item) to provide a combined speed-reduction ratio of 23:1.
• A linkage mechanism to transfer rotary to linear motion. The mechanism consists of a cogged belt drive and connecting rods between gearbox and door. Designers modeled the door and transfer mechanism as a mass-spring-damper combination. They used a nonlinear relationship between rotary and linear motion of the mechanism to convert beltsheave rotation to an equivalent door displacement.
• An incremental encoder, having a resolution of 1,024 parts per revolution, to sense the belt sheave rotation (up to 180 deg). An incremental encoder was selected, rather than an absolute type, because of its lower cost. The software converted the quadrature output signal from the encoder to a relative angle. And, the encoder was initialized when the door reached either the “full open” or “full closed” position.
• A microprocessor-based digital controller. Engineers first modeled this controller, which contains the control algorithm, as an analog computer that sends a continuous signal (zero time delay between signals). They also evaluated Intel 80386, 80286, and 8088 microprocessor chips, each at its respective operating speed. DSP chips were ruled out because of higher cost.
Performance specifications for the elevator door system were:
• Maximum door open and close times = 1.3 sec.
• Maximum motor armature current = 6 A.
• No perceptible door vibration allowed.
• One feedback sensor available, a digital encoder on the motor output shaft.
• At 1.3 sec, the door must be within 0.1 in. of full-open or full-closed position.
Simulating the door system
Because the cost of microelectronics varies widely, the designers concentrated on evaluating different options in this area.
The simulation software enabled the designers to create the door operating system, beginning with a system block diagram, Figure 2. Each of the major component blocks contains a simulation model. Standard models for an encoder, dc motor, gearbox, and digital controller were used to accelerate the design process. The software, running on a PC with Windows, enabled the design team to make changes and see the results interactively.
Initially, the simulation was used to operate the motor at its rated speed while time, position, and speed data for a door-open cycle were recorded. Then the digital controller algorithm (model) used the recorded data to translate position error into a motor-velocity command. In effect, the controller compares two inputs — a door-position command and sensed door position — and sends one voltage output, a velocity command to the motor. The maximum speed recorded was 15.5 in./sec, which corresponds to a 110-V command signal, or 100% of the motor voltage.
Continue on Page 2
After several design iterations, an ideal control system — one that provides continuous (analog) signals — satisfied the performance specifications, Figure 3. The software’s interactive design environment, with plots and displays, enabled the designers to diagnose and solve problems within each component, then see the effect at the system level.
The next step was to determine what type of controller microprocessor — 80386, 80286, or 8088 — would meet the performance requirements. Each of these devices performs calculations within a different increment of time, called the update time. Microprocessors with longer update times cost less.
To determine the maximum update time required, engineers modified the ideal continuous-control algorithm to include a discrete-time approximation. After several design iterations, the team selected two update times for evaluation: 20 msec and 50 msec. Figure 4 presents the door response at the full-open position for the ideal, 20-msec, and 50-msec controllers.
The 20-msec update time of an 80386 was chosen based on its performance in the simulation. At this point the preliminary controller design was complete and ready to be evaluated.
The software can be used for design verification by substituting hardware for simulated components in the control-system diagram. Thus, engineers replaced models of the motor, gearbox, linkage, and door with actual components and interfaced them to the simulated controller, using an Advantech 812PG Input/ Output card to form a hardwarein- the-loop simulation. This simulation, also called rapid prototyping, enabled the engineers to test the prototype system before finalizing the controller design.
The curves in Figure 5 compare the results of a door open-close cycle for both the simulation and the real equipment. The simulated system (top) contains the controller and door-system model. On the bottom, the door-system model has been replaced by the actual equipment.
An add-on software module, called Vis- Sim/RT, was used to control the equipment via signals between encoder, controller, and motor.
In the test system a door-position sensor was not available, so the position was approximated from the encoder signal. As shown in Figure 5, the simulated controller provided insufficient door displacement when applied to the real equipment.
Cut the prototypes
Rather than building a prototype motion- control system only to find that it didn’t provide enough door displacement, the design team used the software to modify the 20-msec controller design until it met the specifications. This was done by manipulating certain control variables, called anticipation parameters, which enabled the designers to predict performance of the door system and adjust the control algorithm accordingly to achieve desired performance. Thus, the software enabled the designers to move from component enhancement to system re-simulation to test-hardware verification, thereby, greatly reducing the design time, and increasing the doorsystem performance.
By fine-tuning these parameters even further, the designers eventually found that a slower, less expensive, 80286 microprocessor, operating at a 50-msec interval, was satisfactory.
Tuning the anticipation parameters often leads to system instability that could damage actual equipment. Thus, the simulation provided a safe environment for optimizing the system without risking equipment damage.
Ric A. Kolk is a professor of electrical engineering at the University of Hartford in West Hartford, Conn.