 |
Description  |
|
|
BACKGROUND
1. Field of the Invention
The present invention relates generally to sensing and control of electric
motors, and more particularly, to precise sensing and control of the
position and orientation of a planar linear motor incorporating a
monolithic planar alternating current (AC) magnetic sensor.
2. Description of Prior Art
Motors that can move in a straight line (linear motors) are well known in
the art. Motors that can move freely in the plane are less well known, but
several examples exist. For example, the planar linear motor (henceforth
referred to simply as a planar motor) due to Sawyer (U.S. Pat. No.
3,376,578) can provide linear motion in two mutually orthogonal directions
in the plane as well as a small rotation in the plane.
Such a planar motor generally combines four linear-motor sections into one
forcer assembly that is capable of producing forces and torques in the
plane. The forcer is magnetically attracted to a patterned iron platen
surface while being forced away from the surface by an air bearing film;
the equilibrium separation being typically 10 to 15 .mu.m. The motor
sections have fine teeth [typically 0.5 mm (0.020 in.) wide on a 1.0 mm
(0.040 in.) pitch] and the platen has a two-dimensional array of square
teeth of corresponding width and pitch. After chemical or physical
machining, the platen surface is planarized using epoxy to form the
air-bearing surface. The combined motor sections making up the forcer ride
above (or hang below) the platen (stator) surface, and typically operate
on a flux-steering principle in open-loop microstepping mode. That is, a
string of pulses from the control computer serves to increment counters
which set proportional currents in the drive coils which, in turn, move
the stable magnetic equilibrium point which, in turn, provides a force
which moves the motor forward. These developments are chiefly due to
Sawyer, and date from the late 1960s.
Planar motors have many desirable attributes. Commercial systems such as
RobotWorld (V. Scheinman, "RobotWorld: a multiple robot vision guided
assembly system," in Robotics Research, the Fourth International
Symposium, Santa Cruz Calif., 1987, pp. 23-27, and "RobotWorld--unrolled
motors turn assembly on its head," Industrial Robot, Vol. 20, No. 1, 1993,
pp. 28-31) use forcers carrying vertical and rotational axes and vision
cameras suspended from a platen ceiling for automated assembly. Similar
systems have been developed by AT&T (P. F. Lilienthal, et al., "A flexible
manufacturing workstation," AT&T Technical Journal, 1998, pp. 5-14) and
Megamation (Anon., "Speed and precision from novel assembly robot,"
Assembly Automation, Vol. 9, No. 2, 1989, pp. 85-87) for a wide variety of
automation applications such as the placement of surface-mount components
on circuit boards (B. D. Hoffman, "The use of 2-D linear motors in surface
mount technology," Proc. 5th Int'l SAMPE Electronics Conference, 1991, pp.
141-151).
While offering many benefits, current planar motion systems are severely
limited because of their open-loop stepping operation which prevents the
achievement of maximum potential performance. To help ensure against loss
of synchrony (missing steps), only two-thirds to three-fourths of the
available force margin is used, reducing the forcer's potential maximum
acceleration and velocity. Even so, the forcer motors remain susceptible
to loss of synchrony if large enough unanticipated external forces are
acting. Additionally, settling times after moves are longer than desirable
and there is no way to reject low-frequency external disturbances. The
forcer has only moderate stiffness requiring high power dissipation when
holding a position.
Many have recognized that these problems can be solved or considerably
reduced in severity by incorporating a suitable position sensor that can
accurately measure the relative displacements of forcer and platen at high
enough bandwidth to be used for servo control for greatly improved
performance. Among the possible sensing strategies are laser
interferometry, tracking from light sources attached to the forcer,
optical sensing of teeth in the platen, capacitive sensing of teeth, and
magnetic sensing of teeth.
Interferometric or other optical tracking techniques are expensive and run
into trouble when multiple forcers are used in a cluttered environment. On
the other hand, sensors which are self contained and can be mounted on or
incorporated into the forcer would appear to be the most desirable. Such
sensors could use either magnetic, capacitive or optical principles to
generate electrical output when the forcer is driven over the platen
array. The output signals, either pulses or continuous waveforms, would
correspond to the platen array dimensions. These could be used for
closed-loop coarse distance control by pulse counting and/or intra-tooth
fine control by interpolating the analog waveform.
Sawyer himself recognized the desirability of a platen tooth sensor and
patented a method based on magnetic induction (U.S. Pat. No. 3,735,231).
One embodiment of the sensor in U.S. Pat. No. 3,735,231 includes a four
pole magnetic member having a pair of sense windings which can be in the
form of a printed circuit board disposed on non-adjacent poles at the
exposed end of the poles. The pair of windings provide outputs which are a
periodic function of the head relative to the platen along a single axis.
The patent does not teach control of the motor from the sensor signals.
U.S. Pat. No. 3,857,078 to Sawyer discloses a closed loop planar motor
using magnetic sensing. For detection along each axis, two pickoff
assemblies are utilized. Each pickoff includes two magnetic cores joined
by a magnetic cross piece having a drive coil wrapped around it. Each
magnetic core has two poles, each with three teeth. The two poles of one
core are spaced in a phase quadrature relationship with the two poles of
the other core. The flux in each core varies with the linear positioning
of the pickoff relative to the platen and the fluxes in the two cores are
in a quadrature relationship with each other. A sense coil is wound around
an upper horizontal portion of each core member. The two sense coils
provide quadrature related output signals having periodic relationships in
accordance with the actual displacement of the head along the platen. This
sensor, however, suffers from the disadvantage that since the magnetic
path is not symmetrical on both sides of the drive coil the outputs have a
large common mode (bias field) component which is not cancelled.
A magnetic sensing technique is disclosed by Brennemann, et al., ("Magnetic
sensor for 2-D linear stepper motor," IBM Technical Disclosure Bulletin,
Vol. 35, No. 1B, June 1992). This sensor is an AC magnetic sensor based on
self inductance of coils integrated with a planar motor. The sensor
includes four linearly shaped poles, each having a plurality of teeth. Two
poles on the left are separated from the two poles on the right by a
magnetic spacer. A sense coil (L1-L4) is wound around each of the poles.
The sensor for each axis consists of eight coils wound on eight poles.
Four poles are positioned in one quadrant of the forcer and four are
positioned in the diagonally opposite quadrant. The inductance of a first
sense coil L1 is at a maximum when the inductance of a second coil L2 is
at a minimum and vice-versa. The sense coils L3 and L4 are in phase
quadrature with the coils L1 and L2. Each four-pole sensor produces
quadrature related output voltages which vary sinusoidally with forcer
displacement along a single axis of the platen. The sensors can be used to
measure displacement along one of two axes and rotation about the z axis.
The above sensors are, however, relatively complex to make and use,
expensive to manufacture, difficult to shield from unwanted external
fields and have a relatively small signal. There is no discussion of motor
control based on such signals.
U.S. Pat. No. 5,434,504 to Hollis, et al. discloses a position sensor for
planar motors using inductive coupling to the platen teeth through planar
drive and planar sense coils. In one embodiment, the sensor includes first
and second magnetic members having teeth disposed relative to the teeth on
the platen. A single turn planar drive winding is disposed around at least
one of the teeth of the first and second magnetic members for producing a
first and a second drive flux within each of the magnetic members. A
single turn planar sense winding is disposed around at least one of the
teeth of the first and second magnetic members for generating first and
second outputs which are a periodic function of the position of the sensor
relative to the platen. In another embodiment, the sensor includes two
magnetic members each having four pole pieces. A drive winding is disposed
on each member for establishing first and second fluxes in each member
which are symmetrical about a center of each member. A sense winding is
wound around the center poles of each member for measuring the relative
flux therein and producing an output which is a periodic function of the
position relative to the platen. A disadvantage is the lack of an ability
to measure rotation in the plane with a single sensor substrate. Another
disadvantage is the difficulty during manufacturing of aligning multiple
sensor substrates in a single forcer. The patent does not incorporate a
control element for control of the planar motor from the derived signals.
A position sensing technique based on sensing capacitance between patterned
electrodes and the platen teeth is described in U.S. Pat. No. 4,893,071 to
Miller. In one embodiment, chevron-shaped electrodes couple
electrostatically to the platen teeth, from which position signals for
control can be derived. A disadvantage of this approach is capacitance
changes due to changing humidity and contamination between the electrodes
and the platen surface.
An optical sensing technique using colored stripes similar to that used by
an optical mouse device was patented by Hoffman and Pollack (U.S. Pat. No.
4,823,062). In this technique, optical filters are used to differentiate
between stripes in orthogonal directions. A disadvantage of this approach
is the need to interpose a pattern of colored stripes in the very thin air
bearing that exists between the platen and the forcer, leading to reduced
magnetic fields (or reduced air gap, requiring tighter manufacturing
tolerances). Additionally, there are manufacturing challenges in producing
large areas of precision made stripes and their subsequent bonding to the
platen surface.
Another optical technique based on reflected light was developed by
Nicolson, et al., ("Optical sensing for closed-loop control of linear
stepper motors," Proc. Int'l Conf. on Advanced Mechatronics, Tokyo, Japan,
August 1993). This technique uses light produced by light emitting diodes
(LEDs) shining through slit-shaped masks and viewed by photodiode
detectors to produce quadrature position signals from the platen teeth. A
disadvantage is noisiness of the signals derived from the reflected light
due to random scratches, dirt, and corrosion on the top surfaces of the
platen teeth.
U.S. Pat. No. 5,324,934 to Clark discloses the use of fiber optics to
determine the position, velocity, and direction of movement of a planar
motor. There are a pair of channels, each one of which has two bundles of
optical fiber. A first end of the optical fiber of each of the bundles is
disposed within a narrow elongated slit. One of the bundles conveys light
directed upon the opposite end of the bundles to the platen surface
adjacent to the slit. The remaining bundles convey light reflected from
the surface to a photodetector. The slits are spaced appropriately
relative to the platen tooth pitch spacing so that both position and
direction of motion can be ascertained. A disadvantage is complexity of
manufacture and susceptibility to scratches and contamination.
U.S. Pat. No. 5,818,039 to Lampson describes an optical reflectance sensor
which uses a charge-coupled device (CCD) detector to sense motion in the
plane. A plurality of detectors is mounted onto a forcer to sense motion
along a particular direction, with the detector being insensitive to
motion along an orthogonal direction. Disadvantages include the complexity
of optically coupling the CCD detectors through optical beam splitters and
cylindrical lenses with its attendant bulkiness and cost. As with other
optical methods, there is susceptibility to corrosion, dirt, and
contamination.
Another optical technique was proposed by Brennemann, et al., ("An optical
means of sensing position of a Sawyer motor on a magnetic grid surface,"
IBM Technical Disclosure Bulletin, vol. 37, pp. 375-378, May, 1994). This
method is an improvement on conventional optical sensing of platen teeth.
It uses a fluorescent dye embedded in the epoxy backfill between the
platen teeth. When the dye is illuminated through slits with light of
wavelength .lambda..sub.1, it re-radiates light at a wavelength
.lambda..sub.2 >.lambda..sub.1. By using optical filters to remove the
reflected component .lambda..sub.1, only the component .lambda..sub.2 is
detected, eliminating variations in reflectance due to scratches and
contamination. Brennemann and Hollis published a paper ("Magnetic and
optical-fluorescence position sensing for planar linear motors," Int'l
Conf. on Intelligent Robots and Systems, Vol. III, August, 1995, pp.
101-107) making detailed comparisons between optical and magnetic methods.
All of the sensors and control methods for planar motors heretofore
described suffer from a number of disadvantages:
(a) Sensors based on optical reflectance techniques are susceptible to
scratches, dirt, and corrosion on the top surface of the platen. These
sensors measure the perceived optical position of the platen teeth which
may or may not represent the true position of the teeth. As a result,
position signals for use by a control system are corrupted by spatial
noise, i.e. as the planar motor carrying the sensing head moves across the
regular teeth of the platen surface, the perceived position signals will
be irregular to the extent that contamination effecting the optical signal
is present.
(b) Sensors based on optical fluorescence techniques are bulky and
difficult to integrate within a planar motor structure since the necessary
optical components must occupy a three-dimensional extent. Additionally,
these sensors cannot be used on conventional platen surfaces which lack a
fluorescent dye. Incorporation of dye into the epoxy backfill of a platen
surface requires an additional manufacturing step. Also, it is difficult
to provide a uniform concentration of dye, and failing to do so introduces
unwanted spatial noise in the position signals.
(c) Sensors based on capacitance techniques require incorporating shielded
electrode arrays into the planar motor structure. These electrode arrays
must be constructed in ways to eliminate common mode signals resulting
from changes of flying height of the planar motor. The capacitance signal
is subject to changes in humidity in the environment and to contamination,
e.g dielectric films on the platen surface. These considerations make it
difficult to provide reliable sensing and control.
(d) Sensors based on magnetic techniques using large three-dimensional core
structures have the advantage that they are measuring the planar motor
position with respect to intrinsically magnetic objects, namely the
ferromagnetic teeth of the platen. On the other hand, most of the magnetic
sensors described in the prior art use structures that are bulky,
difficult to manufacture, and difficult to integrate with a planar motor.
These sensors are also difficult to shield from unwanted magnetic
interference, stemming chiefly from electric currents in adjacent motor
windings. Signals from these sensors have not proved suitable for precise
control of position and orientation of planar motors.
(e) Sensors based on magnetic techniques using small planar core structures
have the advantage as in (d) that they are measuring the planar motor
position with respect to intrinsically magnetic objects, namely the
ferromagnetic teeth of the platen. On the other hand, they are difficult
to integrate with a planar motor and require difficult precise alignment
during manufacture.
What is needed is a platen sensor which is not susceptible to scratches,
dirt, or other film contamination and which can readily be integrated with
existing planar motor technology to provide precise lateral position as
well as rotational orientation on the platen surface coupled with
effective means of control so as to affect precise closed-loop planar
motion.
OBJECTS AND ADVANTAGES
Accordingly, several objects and advantages of the present invention are:
A precision closed-loop planar motor with features including:
(a) an integrated monolithic sensor that is insensitive to scratches which
may be present on the top surface of the platen teeth, corrosion of the
top surface of the teeth, or other contaminating films which may be
present and which would adversely effect an optical signal, and will
therefore provide high quality position signals for precise closed-loop
control;
(b) an integrated monolithic sensor that does not require the use of
special, non-standard platen surfaces;
(c) an integrated monolithic sensor that is unaffected by humidity or
dielectric film contaminants on the platen surface;
(d) an integrated monolithic sensor that is small enough and flat enough to
readily be integrable with existing planar motor designs, occupying only
space which is presently unused in such motors;
(e) an integrated monolithic sensor that is comprised of a single
monolithic planar part, easily manufacturable, and capable of measuring
all degrees of freedom in the plane (orthogonal directions of translation
as well as a small rotation), providing precise signals for closed-loop
control;
(f) an electronic processing unit of special design such that signals from
the integrated monolithic sensor can be measured with extreme (sub
micrometer) precision; and
(g) a controller of special design that utilizes the processed precise
position and orientation signals from the integrated sensor to provide
precisely controlled motion in the plane.
Further objects and advantages of our invention will become apparent from a
consideration of the drawings and ensuing description.
SUMMARY OF THE INVENTION
A closed-loop planar motor forcer comprising a monolithic integrated AC
magnetic position and orientation sensor, electronic signal processing,
and controller, enabling the forcer to execute precise motion in
translation in two orthogonal directions, and small rotations in the
plane.
A closed-loop planar motor forcer incorporating a monolithic integrated AC
magnetic position and orientation sensor of a physical form which allows
it to be integrated in otherwise unused space in the forcer.
A closed-loop planar motor forcer incorporating a monolithic integrated AC
magnetic position and orientation sensor comprised of a single planar
magnetic substrate incorporating planar drive and sense windings and
capable of providing precise position and orientation signals.
BRIEF DESCRIPTION OF THE DRAWINGS
The features of the invention believed to be novel and the elements
characteristic of the invention are set forth with particularity in the
appended claims. The invention itself, however, both as to organization
and method of operation, may best be understood by reference to the
detailed description which follows taken in conjunction with the
accompanying drawings in which:
FIG. 1A shows a bottom view of the closed-loop planar linear motor.
FIG. 1B shows an overhead view of the closed-loop planar linear motor.
FIG. 2 shows an exploded view of the integral magnetic position/orientation
sensor.
FIG. 3A shows the flexible printed circuit of region 44 of FIG. 2 using
parallel sense windings.
FIG. 3B shows the flexible printed circuit of region 44 of FIG. 2 using
series sense windings.
FIG. 4 is a cross-sectional view of one set of sensing elements of the
sensor shown in FIG. 2.
FIG. 5 is a cross-sectional view of one set of sensing elements after
planarization to form an air bearing.
FIG. 6 shows a block diagram of the AC magnetic position/orientation sensor
electronic circuit.
FIG. 7 shows a schematic diagram of digital timing block 104 shown in FIG.
6.
FIG. 8A shows the measured sensor position output vs. time for a number of
samples.
FIG. 8B shows a histogram of the data samples shown in FIG. 8A.
FIG. 9 shows a block diagram of the computer controller.
FIG. 10 shows an abstract geometric representation of the force resolution
block 92 of FIG. 9.
FIG. 11 shows the performance obtained with the closed-loop planar linear
motor.
DETAILED DESCRIPTION OF THE INVENTION--FIGS. 1A, 1B, 2, 3A, 3B, 4-7, AND 9
Referring to FIGS. 1A and 1B, there is an overall view of the closed-loop
planar motor with integral AC magnetic position/orientation sensor.
FIG. 1A shows a bottom view of the hardware part of a preferred embodiment
showing a nonmagnetic housing 20 whereon an electrical connector 21 is
mounted. Flexible cable (or tether) 22 attached to connector 21 serves to
conduct electrical signals to and from a computer, along with a supply of
air for an air bearing surface comprised jointly of the bottom surfaces of
housing 20 and elements 23-26 and 28. Toothed linear motor segments 23-26
provide electromotive force along directions parallel to the sides of
housing 20. All of the elements 20-26 are well known in the art and
together constitute what is commonly known as the moving part of a "planar
linear motor," or "forcer" or "Sawyer" motor, after its inventor. We refer
to the combination of a forcer and its platen (stator) simply as a planar
motor. Planar motor forcers operate on a flat iron surface, or "platen" in
which an array of small square posts has been formed on a regular grid.
Typically, the post dimensions are 0.020 in..times.0.020 in., on a pitch
spacing of 0.040 in..times.0.040 in. Spaces between the posts are filled
with a non-magnetic material. To the planar motor heretofore described,
integral AC magnetic position/orientation sensor 28 is incorporated into
housing 20 at a central location.
FIG. 1B shows an overhead view of the hardware part of the closed-loop
planar motor forcer showing the upper extent of the motor segments 23-26
and integral AC magnetic position/orientation sensor 28. Motor current
wiring from 23-26 (not shown) collect at connector 21 and are directed
through tether 22 to conventional electrical pulse width modulation (PWM)
drive circuitry located elsewhere. Electronic circuit board 29, preferably
contained within housing 20, connects with sensor 28. Output wires from
board 29 (not shown) are connected to the computer by way of connector 21
and thence through tether 22.
FIG. 2 shows a detailed and exploded view of the AC magnetic position
sensor 28 of FIGS. 1A and 1B. Magnetically soft substrate (preferably of
manganese-zinc-ferrite or similar material) 30 has a plurality of raised
linear teeth formed on its surface in patterns corresponding to and in
juxtaposition with the square posts of the platen surface. Shown in the
figure are a preferable arrangement of four groups of six teeth each,
comprising four narrow teeth 35-38 and two wide teeth 32, 33. Substrate 30
can be formed by any number of manufacturing processes such as grinding,
etching, machining, etc. Substrate 30 is mounted on nonmagnetic base 50,
preferably a ceramic of matching temperature coefficient, by means of
adhesive 51. Rigid circuit board 53, containing a plurality of electrical
connectors 54 is, in turn, mounted on base 50 by means of adhesive 52.
Flexible printed circuit boards 40, (one of which is shown in FIG. 2)
extending from connectors 54 of rigid circuit board 53, and corresponding
to each group of raised teeth on substrate 30, containing multiple slots
42, each of which corresponds to an individual tooth of substrate 30, are
adhered to substrate 30 by means of adhesive 41, thus forming a sensor
group 44. Magnetically soft shield 60, preferably of .mu.-metal, formed in
the shape of a box, is mounted, in turn, onto rigid circuit board 53 by
means of adhesive 62. Electrical ground pin 55 is soldered to shield 60 at
hole 61.
FIG. 3A shows a closeup view of flexible printed circuit board 40 shown in
FIG. 2. In another embodiment, a single flexible circuit board can be used
with four wing-shaped tabs such as that shown in the figure. In particular
are shown slots 42 formed in the insulating substrate of the flexible
printed circuit board, single-turn drive winding 43 which encircles the
two wide slots 42, and single-turn sense windings 45-48 encircling the
four narrow slots 42 forming two independent circuits, each having their
windings in parallel opposition. That is, windings 45 and 46 encircle
their respective slots in opposite directions and are connected in
parallel. Similarly, windings 47 and 48 encircle their respective slots in
opposite directions and are connected in parallel. Drive and sense
windings are preferably formed on opposite sides of 40, as indicated by
dashed and solid lines in the figure. To facilitate connections, several
vias 49 serve to connect top side wiring by bridging under other top side
wires without contact.
FIG. 3B shows a closeup view of one of four alternatively wired flexible
printed circuit boards 40 shown in FIG. 2. Here, single-turn sense
windings 45-48 encircle the four narrow slots 42 | | |
