Variable Off-Time Motor Current Control, Bhakti Vyas
The problem is that in motoring mode, electric machines cause winding current waveforms to be destabilized by the induced voltage. For magnitude control, we want to maintain a constant controlled average current. There are a number of ways of doing this, but the new idea variable-off-time control does it by using a prior knowledge of what the PWM off-time should be for a stable waveform. The on-time is controlled by the current control loop, while the off-time is generated by an off-time generator according to a derived equation of what it should be for stable current waveforms. This off-time function is nonlinear, but only has a range of x2, and can be somewhat accurately approximated by a linearized off-time generator. The off-time generator can be implemented using a simple 555 timer circuit, with timing current from a speed-to-current converter and appropriate offset. An embedded microcontroller PIC16F877 can be used to calculate the linearized and exact off-times, then the duty ratio and the frequency of a PWM generator on the microcontroller set to drive the motor bridge driver. The current ripple over the motor speed range was measured.
Field oriented stepper motor control,
(pdf 712 KB), Bhavin Shah
This is a project to control stepper motors using control theory rather than traditional stepping. This will obtain the maximum theoretical performance from the step motor, thus obtaining performance that is on par with more expensive servo motors. This project is being implemented with an Analog Devices DSP.
load control, (pdf 1.92MB), Srujan Kusumba
Dynamometers are electro-mechanical instruments used to place a controlled mechanical
load on torque-producing devices such as motors. They are used to characterize motor
torque as a function of speed. A dyno is a controlled, mechanical, rotational load.
It controls either speed or torque and measures both. With a dyno, the torque-speed
curves of motors can be plotted, and their motor-drives can be tested over the intended
operating range. Dynos are to motors and motor drives as oscilloscopes are to electronics
a basic test instrument.
A typical bench-top dyno costs $10,000 or more. The corresponding electronics test instrument, the digital storage oscilloscope, can be bought for around $1,000, a rather sizeable difference. Consequently, for small laboratories on limited budgets, it can make sense to build a dynamometer for testing the motors. The goal of this thesis is to build a cost efficient dynamometer for small motor performance testing.
Speed control is done using an H-bridge configuration of the MOSFETs to PWM the load current through Rload (load resistance). The speed at which the motor is running is measured by using an optical encoder attached on the shaft of the generator and the reference speed is set by the user by using a potentiometer. The error speed then is sent to the controller which changes the duty cycle of the PWM (switching of the MOSFETs) to change the load current.
Sensorless control of stepper motor using Kalman filtering, (pdf 2.88MB), Chirayu Shah
Despite model improvements and different control algorithms, much work remains to be done to attain maximum motor performance. This work attempts to achieve control and velocity tracking for a step motor using optimization techniques. The resulting system displays practical stabilization for velocity tracking of a voltage-fed permanent-magnet stepper motor. The control design is an output-feedback design that utilizes stator current and rotor position measurements. The goal of this work is to design a controller that is robust to load torques, cogging forces, and other disturbances satisfying certain bounds by estimating the speed and position of the motor using an extended Kalman filter. The controller and the estimator are implemented in a Microchip PIC16F877 microcontroller using embedded technology, and speed control is achieved using the estimated motor state. The PIC16F877 is interfaced to a PC through USART communication to send the real time data to the PC for analysis and comparison with Matlab results.
Radial basis function neuro controller for a PM stepper motor,(pdf 3.21MB), Saikiran Gumma
Changes in the environment, unmeasurable disturbances, changes in the system parameters, and component failures are some the characteristics of complex dynamic systems that necessitate intelligent control techniques. Traditionally plant dynamics are first modeled and verified through experiments, and then controllers are designed. However, such controllers are limited by the accuracy of the identified model and cannot accommodate large variations in parameters. While adaptive control is a natural choice to overcome parametric uncertainties, other major issues remain unsolved at this level. Although the region of operation is considerably increased compared to classical control systems as adaptive controllers tune themselves, they don't possess long term memory. Thus, adaptation must be repeated every time the system is confronted with changing operating conditions. To tackle such problems intelligent control techniques have been developed, neuro-control being one of those. In this work we use a specialized learning architecture with a radial basis neural network to develop an inverse dynamic model for a nonlinear permanent magnet stepper motor. This neuro-controller is initially trained offline using the bold driver gradient descent algorithm and is later used in feedback as a controller. Its performance is then compared with traditional PD controllers tuned for various trajectories and external disturbances. The effect of the number of neurons and initialization of the neural network weights on the performance of the controller is also studied.