Design of A Novel 6-DOF Planar Maglev System: Article in Press
Design of A Novel 6-DOF Planar Maglev System: Article in Press
Design of A Novel 6-DOF Planar Maglev System: Article in Press
Abstract
This paper describes a novel single-deck and six degree-of-freedom (DOF) planar maglev positioning system design. The proposed
design uses an array of solenoids to levitate and to hold a permanent magnet carrier in place. The solenoids are excited separately to
generate restoring forces when the permanent magnet carrier is displaced from its equilibrium position. The research uses the ANSOFT
finite element analysis simulation to analyze the solenoid restoring forces and to motivate a suitable permanent magnets arrangement.
Active control on the solenoid currents is then used to maintain the carrier position. The system identification is carried out by perturbing
the experimental set-up from its equilibrium position. The simulation results based on the identification models show that simple control
is effective for maintaining the carrier position. Initial implementation has also showed that the concept is feasible.
r 2006 Elsevier B.V. All rights reserved.
0304-8853/$ - see front matter r 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.jmmm.2006.01.154
ARTICLE IN PRESS
Y.-C. Lai, J.-Y. Yen / Journal of Magnetism and Magnetic Materials 304 (2006) e386–e390 e387
center of the solenoid. Fig. 2 shows that there maintains bottom solenoids will be excited with different currents to
enough buoyancy force to elevate the magnet. The force provide restoring forces to the yaw magnets. The roll
altitude relationship from the simulations does not exhibit control is similar as shown in Fig. 4e and pitch control as
a concave characteristic; therefore, some active control shown in Fig. 4f. Thus, one is able to control all 6-DOF of
may be necessary. the maglev carrier stage.
2.2. Design of the 6-DOF planar maglev system 3. System identification and control
Based on the simulation results, the proposed design Fig. 5 shows the prototype experimental system with six
assembly provides a restoring force, by placing the magnet sensors. There are four LED displacement sensors and two
between two solenoids, to move the carrier back from any eddy current displacement sensors to detect the 6-DOF
position deviations and a levitation force, by placing the movements. The system is thus a six-input and six-output
magnet upon the solenoid, to counteract the carrier system. Because of the very complicated system configura-
gravity. From the analysis, a novel single-deck planar tion, the servo design adopts a straight forward approach
maglev system is proposed as shown in Fig. 3. The maglev to treat each channel as independent single-input single-
platform, 140 140 mm2, has 29 circular magnets with 37 output sub-system.
solenoids to form the array. Fig. 4a shows the five This research carries out the system identification by
horizontal magnets designed to provide buoyancy force separately perturbing the coil currents that control the
and to control the carrier altitude. The analysis results motion axes and measuring off the movements of the
suggested the design in Figs. 4(b) and (c) to achieve x and y corresponding axes signals. The system ID procedure is
direction motion. Respectively, there are eight horizontal based upon the autoregressive with exogenous input
magnets controlled by 12 solenoids (4 rows) to exert
restoring forces to the carrier. The corresponding forces
generated by the magnets are also shown in the figures.
When yaw disturbance occurs as in Fig. 4d, the four
Fig. 1. A 3D mimic framework of the circular magnet and solenoid. Fig. 3. Design structure of the novel planar maglev system.
Fig. 2. X-axis and Z-axis force of the magnet at different air gaps.
ARTICLE IN PRESS
e388 Y.-C. Lai, J.-Y. Yen / Journal of Magnetism and Magnetic Materials 304 (2006) e386–e390
Fig. 4. A 6-DOF driver of the novel planar maglev system: (a) Z-direction, (b) X-direction, (c) Y-direction, (d) Z-axis rotation, (e) X-axis rotation, and (f)
Y-axis rotation.
(ARX) model and the identification procedure resulted in 6 results. The responses in all three directions are all highly
second-order system models for the axes motions. The oscillatory. There is an oscillation of magnitude reaching
dynamic system simulations are carried out based on the 1.5 mm in the z direction. There are still a lot of rooms for
identification models. Unlike what was expected, the further servo tuning; however, these results clearly indicate
models consist of rather simple structure, and simple PI- that the controller is able to hold the stage in place. Due to
control seem to suffice. Fig. 6 shows the resulted system space limit, the pitch, roll and yaw responses are not
responses on x, y and z direction. The system identification shown.
data was taken at 1 kHz sampling rate. The control
responses do not show any overshoot, and the signals 4. Conclusions
converge quickly. This indicates that the active feedback
control would be effective and achieve stable positioning By investigating the FEA analysis results, it is possible to
on the carrier. Fig. 7 shows the initial implementation construct a solenoid-controlled pattern in the planar
ARTICLE IN PRESS
e390 Y.-C. Lai, J.-Y. Yen / Journal of Magnetism and Magnetic Materials 304 (2006) e386–e390