Teck Seng Low

A nonlinear speed observer for permanent-magnet synchronous motors

The application of vector control techniques in AC motor drives demands accurate position and velocity feedback information for the current control and servo control loops. The authors describe a speed observer system suitable for use with permanent magnet synchronous motors as a software transducer. The observer is developed from the dq model of the machine. Design considerations for the observer are discussed. The nonlinearities in the machine model present a problem to the observer design, so a state detection technique is used to achieve global stability and consistent convergence of the observer system. The simulations show that the performance of the observer is robust against noise and parameter uncertainties. >

Servo performance of a BLDC drive with instantaneous torque control

Brushless DC (BLDC) drives with permanent-magnet motors are suitable as servo-motors if properly controlled. Their performance can be superior to that of conventional DC servos if NdFeB or rare-earth cobalt magnets are used for motor excitation and novel control techniques are implemented for its servo performance. The availability of inexpensive but high-performance digital signal processors allows the implementation of a novel instantaneous torque control algorithm for the BLDC drive applications. Described are the servo performances of a BLDC drive with instantaneous torque control. To assess the quality of this controller, its performance is compared with those obtained using conventional current controllers. Details of the control strategies are also described.<<ETX>>

A methodology for neural network training for control of drives with nonlinearities

The learning process of a multilayered feedforward neural network involves extracting a desired function from the training data presented through an appropriate training algorithm. To achieve the desired function, the generation of good training data is necessary. A closed-loop methodology for neural network training for control of drives with nonlinearities is presented. Problems associated with the more common open-loop training scheme, and how these are addressed by the proposed closed-loop method, are discussed. An inverse nonlinear control using a neural network (INC/NN), a control strategy which incorporates the neural network for control of nonlinear systems, is described and used to demonstrate the effectiveness of the closed-loop training scheme. Simulation studies and experimental results are presented to verify the improvement achieved by the closed-loop training methodology. >

An optimal speed controller for permanent-magnet synchronous motor drives

An optimal control system synthesis method which can achieve vector and speed control simultaneously for permanent-magnet synchronous motor (PMSM) drives is proposed in this paper. A state-space multiple-input-multiple-output (MIMO) model for PMSM is first developed and the compensation for the nonlinearities in this model is discussed. A pseudo-linearized PMSM model is dynamically constructed through the state detection, and subsequently an optimal speed controller is developed based on this linearized model. The integral control technique is incorporated to eliminate possible speed offsets. A speed observer is further developed to eliminate the speed sensor from the drive. >

Effects of unbalanced magnetic pull in spindle motors

This paper analyzes the effects of the unbalanced-magnetic-pull developed in spindle motors for hard disk drives. An analytical technique is used in the investigation which is based on magnetic circuit and field theory. Computational results show that the UMP development is closely related to the magnetic structure, armature winding and armature current of the spindle motors.

Steam Laser Cleaning of Plasma-Etch-Induced Polymers from Via Holes.

As the wafer industry enters into submicron processes and below, the demand for new cleaning technology after plasma etching increases. The cleanliness of via holes becomes very crucial for the success of low-resistance interconnecting high-density ultra-large-scale integrated devices. In this study, a relatively new approach in removing the sidewall and bottom polymers resulting from reactive ion etching of via holes, using dry and steam laser cleaning techniques is investigated. The presence of a layer of isopropanol (IPA) film on the wafer surface during excimer laser irradiation is found to improve the removal efficiency greatly even at fluences as low as 80–100 mJ cm-2–much lower than the damage threshold of the underlying Al–Cu metal film with titanium nitride anti-reflective coating of 250–280 mJ cm-2. The explosive evaporation of IPA and the creation of bubbles at the liquid-substrate interface were proposed to be the reason for the improvement. Experimental results show however that the presence of a layer of acetone film does not improve but even impedes the laser cleaning process. An explanation is offered for this phenomenon in terms of the difference in the absorbance of the two liquids at the laser wavelength.

Wavelength Effects in the Laser Cleaning Process

The wavelength effects in the laser cleaning process were studied both theoretically and experimentally. As the reflectivity and absorption coefficient depend on laser wavelength, different wavelengths will result in different temperature distributions in a particle or a substrate so that various cleaning forces and efficiencies are induced in the laser cleaning process. For laser-induced removal of copper and aluminum particles from quartz surfaces, the cleaning efficiency is higher and the threshold fluence is smaller for a shorter wavelength under the same cleaning parameters. This is because laser irradiation at a shorter wavelength causes a higher temperature rise and induces a greater cleaning force in a copper or aluminum particle, which results in higher cleaning efficiency and lower threshold fluence. The theoretical analysis can successfully predict and explain the experimental results.

A Cleaning Model for Removal of Particles due to Laser-Induced Thermal Expansion of Substrate Surface

Taking Van der Waals force and cleaning force due to fast thermal expansion of substrate surface induced by pulsed laser irradiation into account, a cleaning model was established for removal of tiny particles from substrate surfaces. The cleaning condition and cleaning threshold can be obtained from this model. Theoretical predictions have been verified by the experimental results for removing quartz particles from silicon substrate. For laser-induced removal of quartz particles from silicon substrate surfaces, the cleaning threshold is about 135 mJ/cm2. Cleaning efficiency increases with increasing laser fluence, and large particles can be removed more easily than small ones.

An intelligent indirect dynamic torque sensor for permanent magnet brushless DC drives

In this paper, we present a technique for indirect sensing of the dynamic torque feedback signal which is applicable to permanent magnet BLDC drives. This technique is based upon a judicious use of the motor equations, and on parameter estimation methods. It is intelligent in the sense that the technique provides self-calibration at start-up, and it is also possible to extend the technique to overcome effects of parameter drifts by incorporating real-time on-line parameter estimation. The effectiveness of the proposed intelligent indirect dynamic torque sensor has been demonstrated in typical speed and position control experiments where the performance obtained is shown to be superior to that obtained with conventional sinusoidal current controllers.

Effect of magnetic anisotropy distribution in longitudinal thin film media

The effects of anisotropy axes distribution on the properties of longitudinal recording media are simulated using the micromagnetic modeling. The angle between the anisotropy axis and the preferred direction is assumed to follow the Gaussian distribution. Hysteresis loops are simulated with the field applied in the preferred direction and perpendicular to it. Orientation ratios (OR) are calculated for different anisotropy distribution. OR (coercivity) increased from 1.4 to 79.2 when standard deviation of anisotropy distribution σ decreased from 0.7 to 0.05. The effect of the anisotropy distribution on the recording performance is evaluated, too. The results show that narrowing the anisotropy distribution will increase the OR and decrease the transition noise power and transition length. It was found that a critical point occurred at σ=0.15, below which the OR can be maintained at a sufficiently high value and yet the integrated noise power can be kept low.