Specialists in Robotics and Motion

Section Links: Demonstration video General Information Advantages that apply to all embodiments Advantages that apply to high-accuracy embodiments Exploded view Functional description Alternate configurations Reducing costs for less demanding applications

DEMO       We have an animated demo available which we recommend. It will quickly introduce you to the concept.  It shows the system in action and quickly shows the advantages in terms of speed, profile, and travel.       The demo is an a movie file (AVI) that requires a fairly recent player and at least a Pentium-level computer.  Microsoft Windows Media Player versions 6.2 and above will work fine.       You can either download the compressed file to your disk, decompress it and play it from there; or play the movie (AVI) file directly in your browser.
Download compressed (ZIP) version (2.3 Mb):  Zip file 10.7 min at 28.8K, 48 sec at 384K	Play the movie (AVI) file (4.0 Mb):  Avi file 18.4 min at 28.8K, 1.4 min at 384K
If you need to update your Microsoft Windows Media player, you can download the latest version for free at.... Microsoft Windows Media Player

The subject positioning system, or Stage, is expected to fill an important need in the high-precision and/or high-speed positioning markets.  Most immediate need will be in the flat panel display and semiconductor manufacturing and test fields.  However, there are many other likely markets such as precision machining and medical inspection.

There are good reasons to expect this particular X-Y-Theta stage to surpass all existing systems in both accuracy and speed.  As an added by-product of the novel approach the system will cost much less than existing systems, be easily made vacuum-compatible, and occupy minimum space.

For the high-accuracy embodiment, we expect that it will achieve absolute accuracies in the range of 10 nanometers with maximum travel speeds of 50 inches per second.  We know these are astonishing numbers, but this document will point out why we have such high expectations.

The total manufacturing cost for such a stage will be much less than competing systems due to the simplicity of the components required.  This also applies to lower-accuracy versions because the system is highly configurable with a wide range of components.

Once a Theta offset is chosen, the stage will adjust the travel in its X and Y directions to match that Theta offset.  Thus the coordinates of motion are rotated to adjust the X and Y travel to be orthogonal with respect to Theta.  Common applications will require only a few degrees of rotation, and the configuration described herein covers that.  The patent describes an embodiment that permits 360 degrees rotation, but it is not described herein because it is not expected to be often needed and it adds some cost.  If Theta is not required, one actuator can be eliminated to save some cost.

The following is a brief outline the technical reasons why the stage should be able to achieve the stated speeds, accuracy, and cost goals.

It is easier to make this design vacuum-compatible than the conventional designs.  This is due to the thermal considerations discussed below.  If vacuum-compatibility is not required, air bearings can be used to increase the accuracy.

Minimum footprint for most stage configurations (including this one) is two times the stroke in both the X and Y unless the X and Y axes are independent.  Some other configurations achieve a reduced footprint by splitting the axes, but splitting the axes generally reduces accuracy and increases cost.  Some embodiments of this stage require only the minimum footprint whereas others require slightly larger than minimum footprint.  Also, some embodiments of this stage can be built in an open-frame configuration for applications where the platform must be accessed from both above and below.

Profile is lower than that of a stacked stage because one axis does not ride on the other.  The axes overlap, but the lower axis and upper axis share some of the same vertical height.

This stage is not stacked.  In other words one axis does not carry the other as is the case with most stages. Even the Theta-axis is not stacked.  Therefore the accumulated mechanical coupling errors that are transmitted from one axis to the other as the stage moves are also gone.  Since the platform moves over a flat surface, there is no overhang to cause pitch and roll problems.

Even greater advantage in terms of accuracy is achieved by the fact that position detection occurs directly beneath the point of work.  There are several methods of position detection that will achieve that result with this technology.  What is commonly done instead is to detect position from the edges of the mechanism and depend upon mechanical construction for accuracy at the point of work.  Detection directly at the point of work makes the system completely immune to Abbe and offset errors.

The mass to be moved can be of very low inertia, and the drive can be done with high speed motors of many types although it is particularly adaptable to linear motors.

Bearings are of a very simple design with few moving parts and can be manufactured very inexpensively.

Since servo technology is used to nullify a lot of mechanical errors, machine mechanical tolerances are very forgiving.  Parallel actuators do not need to be perfectly parallel, and orthogonal actuators do not need to be perfectly orthogonal.

The system will not get as hot as competing approaches.  The thermal conductivity to the outside world where heat is dissipated is much greater than the conductivity to the work area.  Additionally, the fact that all the motor coils are stationary makes it is easy to mount them directly to the base and (if necessary) apply any of several cooling methods to them. This is particularly helpful in a vacuum environment. Direct cooling of the motor coils in the carried axis is not possible with other systems without installing flexible coolant tubes.  The more efficient cooling greatly increases the overall system utility.

1. The laser interferometer usually required to achieve this degree of precision is not required.  Besides saving the cost of the laser itself, the carefully ground mirrors and all the ancillary hardware needed to stabilize the environment for the interferometer are not required.
2. Precision will be accomplished electronically and optically rather than mechanically.  This dramatically reduces the need for precision mechanical components and precision machining.  Although this is true for all embodiments, the advantage takes on another dimension with the high-accuracy embodiment.

This concept provides the ideal platform for taking advantage of the relatively new technology of holographic grids.  This new technology has shown a lot of promise but did not have a good means of implementation until now.  The technology consists of a feedback sensor in the form of an X-Y precision hologram or diffraction grating scanned by laser diode read heads.  Such a grids are manufactured with the same resolution as a laser interferometer.  The grid has two distinct advantages in accuracy over the laser interferometer:

1. Immunity to atmospheric variations.
2. Does not use a reflecting mirror and is therefore immune to the manufacturing tolerances and costs of machined mirrors.

As a result, the grid accuracy can be even better than that of a laser interferometer.  The grid has manufacturing imperfections, but they are generally predictable and can be software-compensated.

Besides eliminating Abbe and offset errors as described above, the high-accuracy version comes with the holographic grid attached directly to the moving platform.  Other detection systems tell you where the components are that move the platform surface, but ignore the fact that the platform may not be where the components say it is.  Thus the high-accuracy embodiment eliminates errors caused by:

If the stage is built with air bearings or with PTFE-based friction bearings, the final source of error is eliminated - Stiction.

There is only one error source that is not brought into the range of insignificance by the design, and that is thermal expansion of the workpiece relative to the holographic grid.  However the effect of that error will be much smaller than in competing designs.  Only the temperature changes in the platform and grid affect the result.  Temperature changes in the actuating mechanisms and base do not apply.  Also, this design easily lends itself to techniques for reducing the temperature error. For extreme accuracy, the platform that carries the X-Y grid and the workpiece can be made of material that has the same expansion coefficient as the workpiece.

The platform is guided from the edges by 6 edge bearings, which are moved by the 3 actuators.  The channel for the Y-axis actuator is not as deep as the channel for the X-axis actuators.  Therefore the Y-axis actuator overlaps the X-axis actuators, but there is no contact between them, and one axis does not "carry" the other as is the case with most stages.  The platform rests upon the base, and is held up by a either an air bearing, multiple roller bearings, or a zero-stiction friction bearing.  The first unit makes use of an air bearing so that the platform actually floats on the base.

The means of moving the actuators in their respective channels will vary from application to application as will the bearing mechanism for those actuators.  The drawing herein depicts a simplified rendition of linear servo motors, and does not specify the actuator guide bearings.  This was done to illustrate the concept as simply as possible.  Each linear motor consists of a forcer and the actuator bar itself.  The actuator bar contains magnets, and the coil is in the forcer.  Since the forcer coils are mounted directly to the base, cooling is very efficient.

Each actuator includes a fixed edge bearing on one end and a compliant edge bearing on the other so that the compliant bearing preloads the platform against the fixed bearing.  Thus the position of the platform is set by the 3 fixed bearings.  Theta (rotation) is accomplished by offsetting the two X-axis actuators with respect to each other.  This offset moves the platform to a slight angle.  Any such rotation will require some separation between the edge bearings of the Y-axis actuator to accommodate the apparent added length caused by the inverse cosine of theta.  Thus Theta motion creates a need for more compliance on the compliant Y-axis edge bearing, and motion in Theta will be limited by the design of the edge bearings.  The edge bearings in the first product are air bearings wherein the compliant side is piston-preloaded, but the edge bearings also can be of many types.

Since the platform moves on a flat surface, height variation, pitch, and roll are very well constrained.  There is some load shifting because the platform center of gravity will move with respect to each actuator.  At high speeds, this could cause some instability in yaw.  All stage designs have some load shifting, but in this case the load shift is very predictable and the controller will automatically adjust in those cases that require it.

Rotation can be eliminated when not needed. See "Reducing costs.." below.

The patent covers a configuration that allows 360 degrees of rotation without a loss in accuracy.  It adds some complexity and cost, but it has application.

The product can be made in an "open-frame" configuration which will permit access from below as well as from above.  This will be accomplished by either using two Y-axis actuators, or by offsetting the single Y-axis actuator and constructing stiff support mechanisms at each end of the Y-axis actuator to reach from the actuator's offset position to the center position of the base - thereby placing the Y-axis edge bearings where they would be if the actuator were not offset.

When footprint is at a premium, the base can be made smaller by positioning the X-axis actuators all the way to the outer edge of the base, connecting the two X-axis actuators by stiff rods that pivot at the end points of the actuators, then mounting the X-axis edge bearings on those connecting rods.  This eliminates the added footprint in the Y-direction caused by the separation between the two X-axis actuators.

The preloaded bearing can be on the same side of the actuator as the fixed bearing with a vacuum or magnetic preload.

There are many possibilities for the actuator drive mechanism besides the linear motors described herein.  Those possibilities include lead screw drives, and belt drives, and piezoelectric linear motors.  For less demanding applications, stepper motors will be used.  The first product uses piezoelectric linear motors and air bearings for the actuators.

The X-Y grid can be replaced by linear scales attached to each of the motor actuators.  This provides the same point-of-work detection advantages as the grid provides because the positions detected by the three sensors triangulate to the point of work.  Three sources of mechanical error come back into play: Compliance of the bearing supports, straightness of the platform edge guide, and orthogonality of the platform edges.  Errors caused by temperature changes within the actuators also come back into play.  However all other accuracy benefits remain, and absolute accuracies in the sub-micron range will still be achieved in many cases.

Inexpensive bearings could be used under the platform and at the motor contact points in place of air bearings or special low-stiction bearings.  This would reduce the absolute accuracy that the servo could achieve, but it will still be possible to achieve micron level accuracies.

If there is no requirement for rotation, one of the X-actuators can be eliminated.  In that case, the one actuator would support all 4 x-axis edge bearings

Beyond that, there are many other compromises that can be made to reduce cost when speed and accuracy can be traded off.  Motors can be rotary instead of linear.  Feedback can be eliminated entirely and stepper motors used instead.  Drive mechanism can be ball screw, acme screw, or even belt drive.  If flatness or heat dissipation is not critical, the base can be made of a variety of inexpensive materials with limited machining.