In this chapter we will analyse the results of the magnetic levitation system, we designed.
First we will give some comments about the sensor. After that we will talk about the current amplifier and finally we will discuss some control strategies. As we stated it is possible to improve the results that we obtained. Therefore, we will give some recommendations at the end of this chapter.
6.1 Conclusions
6.1.1 The sensor
In this report we have designed an inductive sensor. This technique should make it possible to sense and actuate on the same cail. However, it was of major importance for the inductive sensor to be able to sustain a sinusoidal current with constant amplitude to the coil together with the actuating currents. SO far now we did not completely succeed in reaching this goal. Moreover, sensing and actuating on the same cail appeared to be impossible when large actuating currents (> 2.5[A])occurred, as the coil became saturated. Despite the limitations mentioned above, the inductive sensor appeared to be accurate and reliable when no actuating currents appeared. Therefore we can say we succeeded in developing an appropriate sensor, although we did not yet succeed in sensing and actuating on the same cail.
6.1.2 The current amplifier
We developed a current amplifier that was able to supply a bandlimited current to the coil.
The current, "delivered" by the amplifier very precisely followed the voltage input. This was not easy as a cail is a large inductive load, and thus trying to suppress a change in current.
However, as mentioned in the earlier section, we still did not succeed in also supplying the sensing current very accurate. In the next section we will try to deliver a solution by giving a few recommendations.
6.1.3 The control system
After we had designed an appropriate cail, a sensor and a current amplifier, we tried to develop a controller that could be realised in practice and would be able to stabilize the balI under the actllating cail. We succeeded in levitating and stabilizing the balI, hy using a
43
simple PD-controller. HQwever, simulations of the system suggested much better results to be possible. The simulations showed it should be possible to keep the balI very accurate and steady in our wished equilibrium point. In practice the balI kept moving. These differences of praetiee and simulations can be explained by the following remarks:
• The model of the levitation system is not completed. Nonlinear effects of the eurrent amplifier and the sensor should be taken into account, as well as the influence of the second cail, that was used to sense the ball's position.
• In our realisation of the controller it was very hard to actuate the equilibrium eurrent exactly. Furthermore it was hard to exactly adjust the gain.
• The magnetic system is poorly defined and thus no exact knowledge of the forces is available. Furthermore the infiuence of temperature and saturation have been neglected.
Summarizing, we may conclude that it is very hopeful that we are able to levitate the ball. However, it is still not possible to keep the ball steady and to sense and actuate on the same eoil. Therefore a lot of things must and can be improved. In the next section we will give some recommendations on how these improvements can be reached.
6.2 Recommendations
First some suggestions are given about the design of the coil. After that we will say something about improving the sensor. Next we will analyse if the current amplifier ean be improved.
Finally we will talk about an optimal way of eontrolling thesystem.
6.2.1 Design of the coil
Although it is very hard to design an optimal coil, it might be an improvement to make the coil longer (increase I) and give it a bigger erosssectional area (increase A). This would probably avoid saturation effeets, without increasing the resistanee of the coil too much. This could make it possible to sense and aetuate on the same cail.
6.2.2 Improving the sensor
The limitations of the sensor are mainly the linearity of the output and the output noise.
The restrietion of the linear part is a consequence of the restricted area of operation of the multiplier used to realize (3.10). The noise is introduced by aJI electronic components. As a matter of fact, improving the sensor is a question of using better electronic devices.
6.2.3 The current amplifier
The current amplifier is very accurate, if we look at supplying the bandlimited actuating currents. However it is not possible yet to sustain the sensing current very accurately as weIl.
A suggestion for a solution of this problem is given in appendix D.
CHAPTER 6. CONCLUSIONS AND RECOMMENDATIONS 45
6.2.4 The control system
To he ahle to control (not only levitate) the position of the ball, it wil! he necessary to use a more accurate controller, for example a PID-controller. To realize this in practice it will be necessary to do the control by means of the Digital Signa] Processor (DSP). Both controlling and sensing (Look-up-table) can he improved by using the DSP. Moreover it deserves attention to try out the technique of exact linearization or Feed Forward control to optimize the control system.
[I] BUrT-Brown IC DATA BOOK- Linear products, 1994.
[2] Harris Power MOSFETs, 1994.
[3] F. Auer. Combined Electromagnetic Suspension and Propulsion for Positioning with sub-Micrometer Accuracy. Delft University Press, 1995.
[4] P.B. Houtkamp. Preparatorystudy of a two axis mirror deflection system with magnetic hearings. University of Technology Eindhoven, 1996.
[5] T. Sugie and M. Kawanishi. Design of a magnetic levitation system considering parameter variations. Proceedings of the Asian Contol Conference 1994, pages 159-162, 1994.
[6] T. Sugie, K. Simizu, and J. Imura. hoo control with exact linearization and its application to rnagnetic levitation systerns. IFAC, Vol.
4,
pages 363-366, 1993.[7] M.L. Thijssen. A 2-dimensional magnetic levitation system. Master's thesis, University of Technology Eindhoven, 1996.
[8] D. Vischer and H. Bleuier. SeIf-sensing active magnetic levitation. IEEE Transaction on magnetics, Vol. 23, No. 2, pages 1276-1281, 1993.
[9] T. Wong. Design of a magnetic Ievitation control system - an undergraduate project.
IEEE Transaction on education, vol. 29, no. 11, pages 196-200, 1986.
46