high accuracy slides


Thread Starter

Bill Sturm

Has anyone on the list tried to build an industrial 2 axis slide system with accuracy and repeatability in the range of 0.1 to 0.5 micron? We have been quoting such a system that started out as plus or minus 0.5 micron at around six inches of travel. The accuracy would need to be maintained for less than 2 inches of stroke.

We have planned on using a high end motion controller with sine encoder feedback on the servo motors for the velocity loop and high resolution linear scales (1 microinch) for the position loop. We plan to use zero backlash ground ballscrews and extremely low friction slides.

I realize that the slides must be machined very accurately. We also know that the mechanics must be very stiff. Linear motors are probably out due
to environmental concerns and also because one of the axis is vertical.

Has this been done? I believe that it is possible, but I wonder how difficult.

Thanks for any input,
Bill Sturm
Applied Grinding Technologies

Jim Griffin - Applied Motion Systems

This motion certainly is possible, but not trivial. It is often done with linear motors, or piezo motors, and in mechanically isolated environments, but using ballscrews may not be completely unrealistic.

.1 micron repeatability is not that tough, though it is perhaps at the limit for the best screw drives. At that level "zero backlash" really just means pre-loaded. There will be some hysteresis. Compensation from a linear scale will be required. Compensation for temperature is likely required as well, unless the environment is very constant.

I have a two-axis stage on my desk that can do 2 inches travel with +/- .1um repeatability - but not absolute accuracy. To maintain flatness, straightness, roll, pitch and yaw errors this small is a trick to say the least. I would have to look at air-bearing table specs to see what the best quoted specs actually are. {You could look to see what Dover has as a benchmark}.

I am assuming that you would like to try to use just cross roller bearings. Over a distance of even 2 inches, 100nm "accuracy" (a true path in space) is very tight and will require laser measurement of the assembly to verify the specs. Obviously .5um is easier, but I believe mapping the system is required to cost effectively get what you seek.

My first questions: What are the loads? Is this a one off or an application that will repeat? What are the speed, duty cycle, and life requirements?

I assume from the name of the company that there is to be some kind of grinding dust and probably metallic particulate (hence the concern about a linear motor). {Ballscrews, bearings, and encoders don't like that stuff much either}.

If the loads are light enough, Nanomotion Piezo motors are a good solution that costs less than ground ballscrews and really has no backlash. They are self-locking for use in verticle applications, but their force is limited to only a few pounds of thrust. Only one high-resolution linear encoder is required per axis, as opposed to a screw drive running with separated position and velocity feedback. With interpolation, linear encoders are fairly cost effective and available to 10nm or better.

In terms of actual accuracy in the sub-micron range, the encoder itself is an issue as well, and systems are often mapped to allow compensation for abbe errors and actual encoder scale deviation. The specs for the encoders often show accuracy to 'x'um/m, but there is probably no guarantee that the error is linear over a given distance.

There are several companies that can deliver stages complete with motors and encoders, and can have the laser measurment /mapping completed prior to shipment. Some high-end controllers, such as Delta Tau, can apply both on-axis and cross-axis compensation for measured deviations from the desired path. We sell, and integrate, some of these products.

Please let us know if we can help you.


Jim Griffin
Applied Motion Systems
12000 NE 60th Way
Vancouver, WA 98682
[email protected]

John Bundschuh

Dear Bill:

It sounds like you have the right part. A cross roller bearing stage driven by a ball screw and with a high resolution linear encoder for feedback will work fine. You could use a linear motor if the z-axis has a counter balance.

If you use a ball screw you may want to have a rotary encoder on the motor for improved velocity control.

Your repeatability will be less than 0.1 micron. Your accuracy will depend on the scale and the slide.

New Way Bearings also has dove tail style of airbearing slide that would provide improved smoothness and straightness over cross roller bearings. We have tested the New Way slide with a 40 nanometer resolution encoder and have only 2-3 counts of error during a move and less when in position.

Regards, John
John Bundschuh
Cymatix, Inc.
Here is a link application note on the SPii+ multiaxis controller showing a .8 nanometer position error - motor enabled with no movement.


This type of accuracy is obtained using a 500 line/mm Sin/Cos encoder with the SPii+ controller. The SPii+ has a built-in capacity to multiply encoder inputs on up to eight axes and still close the servo loop at 50 microseconds.

Error mapping tables compensate for mechanical issues.

Bob Lea
Applications Engineer
ACS-Tech80, Inc.