This paper reports the initial control experiments performed on a small magnetostrictive wire clamp assembly. Linear, lumped parameter, dynamic models are formulated based on magnetostrictive constitutive relations. Reduced order models are utilized to understand the expected performance in terms of the system parameters (e.g., the magnetoelastic parameters). It is shown that proportional flux rate (phi) feedback is fairly ineffective if used alone. Experimental results of proportional current feedback control are given. The models reported are useful for control analysis and actuator/sensor design.
Magnetostrictive materials often rely on magnetic fields generated through the use of a solenoidal coil. However, the field-generating coil also acts as a source of heat causing thermally induced strains in the magnetostrictive drive element. To insure that the useful magnetostrictive strains are large in comparison with the thermally induced strains, the solenoid may be optimized. This paper presents a simple one dimensional (1-D) magnetic model useful for predicting the magnetic field inside the magnetostrictive drive rod. The advantage of this model is that it can be evaluated very quickly, making it well suited for use in optimization algorithms. A figure of merit is presented that weighs the energy stored in the coil against the power that must be dissipated to maintain the field. With the magnetic model and cost function, the solenoid may be sized to maximize the volume averaged field in the magnetostrictive element per unit of volume averaged dissipated heat in the solenoidal coil. While previous work addressed field/power optimization at the center of air-cored selenoids, the work presented here considers optimization of the average field along a rod of permeable magnetostrictive material. The results indicate that coil quality decreases rapidly if the coil is thinner than optimal, but decreases rather slowly for a thicker than optimal coil.
A magnetostrictive water pump using Terfenol-D has been designed and built, achieving a flow rate of 15 ml/sec at 5 psi, using 41 watts. This is a higher flow rate and lower pressure than previous magnetostrictive pumps. The pump is 6' long and 3.6' in diameter. A model of pump performance has been developed, including valve inertia which limits the drive frequency, and trapped air in the chamber, which can reduce the flow rate and make the pump noisy. Methods have been developed to eliminate trapped air. The pump uses a hydraulic stroke amplifier, which turned out much stiffer axially than it was designed to be. This has adversely affected pump performance, because of finite Terfenol compliance and finite housing compliance. With a stroke amplifier of optimal stiffness, and with better quality Terfenol, the pump should be able to achieve a flow rate of 30 ml/sec at 5 psi, consuming 25 to 35 watts. Although the power is more than would be needed by a piezoelectric pump of the same performance, a Terfenol-D actuator offers important advantages, including low voltage and no known fatigue mechanism. Furthermore, much of the modeling would be relevant to a piezoelectric pump as well.
A magnetostrictive wire-bonding clamp for use in semiconductor packaging applications has been developed by Mechatronic Technology Co. Semiconductor industry trends, requiring high process throughput on increasing lead count packaging, make the magnetostrictive material Terfenol-D a candidate for this application. To construct this small, lightweight device, small samples of Terfenol-D were prepared by ETREMA Products, Inc. This paper reports the initial design, mathematical modeling, and experiments related to this initial prototype.
Nonlinear modeling and control methods can be used to increase the usable range of operation of Terfenol-D. Presently, in dynamic applications the usable range of Terfenol-D is often limited to approximately 850ppm. This limitation is imposed by harmonic distortion, spurious vibration, and/or tracking error considerations. These nonlinear effects are due to large variations in the magnetoelastic parameters and hysteresis. The preliminary results of this program indicate that a large performance advantage may be gained through proper control of the nonlinearities. As an example, a recently designed reaction mass actuator that weighs 1.125lbm can produce peak forces as high as 125lbf. However, to limit the open-loop total harmonic distortion to less than 2 percent requires that peak forces be limited to roughly 65lbf. To determine the magnetoelastic parameters, quasi-static experiments were performed with a specially designed apparatus. The research included modeling and simulation based on the static nonlinear magnetoelastic equations. Under assumptions of quasi-static magnetoelastic behavior, a fourth-order linear model was extended with the static nonlinearities. The model is compared with preliminary experiments. These types of models will allow nonlinear control strategies to be developed for Terfenol-D based actuators, thus extending the harmonic-free operating range.
A magnetostrictive reaction mass actuator possessing a large force to weight ratio has recently been developed at SatCon Technology Corporation. Achieving such high performance requires adequate modeling of this multidisciplinary device. The developed models allow performance optimization to be accomplished through parameter selection and control design. At the 1996 smart materials conference, the modeling and design issues associated with this actuator were discussed. Since that time, additional experiments and optimization have been performed that validate the proposed modeling. These experiments include validation of the thermal modeling and dynamic model validation through control experimentation. In addition to these results, a discussion of the trade-offs in terms of eddy current, controller, and thermal requirements, will be presented.
A high force to volume ratio magnetostrictive reaction mass actuator has been designed and developed. The actuator operates as a resonant device allowing the stored strain energy to be utilized. A discussion of the design issues associated with this actuator are presented. In addition, preliminary data is presented for this actuator. This data includes a linear analysis, evidence of parameter variation, and preliminary small signal tests intended to explore this behavior.