High-Tech Sensors Bring New Precision To Machining

To get more output, machines are being pushed to limits that ordinarily would make them generate a lot of scrap. New optical position sensors provide submicron accuracy to meet these needs, and can take a pounding to boot.

Machining operations today are being driven to hit what were once unheard-of accuracies, while at the same time produce fast and deep cuts to get jobs out quicker. Discussions on the shop floor once focused on units of accuracy per meter. But today's deliberations are more likely to center on submicron accuracies.

Much of the responsibility for guaranteeing accuracy in the submicron range falls to the encoder used to gauge machine motion. Unfortunately, the need for high accuracy takes place in an environment that is downright inhospitable to precision encoders. Typical of this sort of environment, for example, are 15-g vibration levels aggravated by high heat.

A new generation of high-tech optical encoders (Photo 1) address these needs, making it possible to build machines that hit submicron accuracies while keeping costs down. The new way of thinking is to boost accuracy through electronics. In other words, precision can come from sophisticated feedback and control systems rather than through extraordinary mechanical rigidity and precision manufacturing and assembly.

Photo 1: HEIDENHAIN Offers A Variety Of Linear Encoders

This approach is increasingly important as machine buyers push builders into a lot of compromises. They are, for example, forcing builders to deal with higher vibration and harmonics because of such things as asking high-speed milling machines to operate with jig-bore accuracy, at a lower price than would be paid for milling.

 
Figure 1: Signal Generation Using Transmitted Light Method

Figure 2: Signal Processing Using Reflected Light Method

Measuring position

There are several methods of measuring position feedback in machine tools. Each has advantages. Magnetic encoders, for example, have a reputation for being less sensitive to contamination. Nevertheless, a magnetic encoder can still seize in the presence of heavy liquid and chip build up. By comparison, inductive encoders are known for ruggedness. Made of metal, they can withstand greater vibration and speed, and are perceived to be more reliable.

However, optical designs are the most practical choice for fulfilling requirements of submicron resolution and accuracy. For one thing, the problem of short-wave measuring errors, which can easily be overlooked in a superficial error analysis, argues in favor of optical encoders. Short-wave measuring errors result from distortion of the encoder output signals, resulting in interpolation errors. Interpolation increases the effective resolution of the encoder by calculating output values for distances that fall between two scale gratings or scanning signals. The smaller the distance between encoder gratings (i.e., the smaller the output signal period), the smaller the interpolation error. Output signal periods are typically 2 mm with inductive systems, 200 microns with magnetic systems, and 20 microns or finer with optical systems. Imperfections in the scanning signals produce measuring errors on the order of ± 1% to ± 2%.

In contrast to short-wave errors, long-wave errors come from scale graduation nonlinearity and scale deformations such as bending. Today's NC controls can usually compensate for long-wave errors.

Newer optical encoder designs also are as reliable as inductive and magnetic designs over temperature extremes, vibration, and in the midst of contaminants. Photoelectric NC encoders combine high accuracy and a thermal expansion coefficient approaching that of steel with high reliability and high resistance to vibration. The LS 106 sealed linear encoder for machine tools, for example, has a high precision glass scale with a grating period of 20 microns. The scale is secured to an extruded aluminum housing by an elastic adhesive. Flexible lips seal the housing from below and keep out dirt, swarf, oil, cutting fluid, and other contaminants.

Photoelectric encoders contain a scale which is specially optimized to ensure stability during changes in temperature. A scanning unit typically travels along the scale. Within the scanning unit, a semiconductor light source and condenser lens generate a light beam. The beam projects through the grating of what is called a scanning reticle and onto the scale. Photosensors sitting behind the scale detect the changes in light intensity as the machine moves, caused by interactions between the scale grating and the reticle grating alternately blocking and passing the light. The sensors produce sinusoidal signals corresponding to the changes in light intensity. Figure 1 shows a diagram of sinsusoidal signal generation using transmitted light method.

Because the scale directly bears on the accuracy of the position measurement, encoder manufacturers take numerous precautions to keep the scale stable. As the scanning housing warms up, for example, the elastic adhesive allows it to expand without affecting the length of the scale. The housing is connected to the machine with special fasteners to prevent thermal constraining forces from arising which could shift the scale position and thus change the measured position with reference to the machine. In addition, the thermal expansion coefficient of the scale nearly matches that of a steel workpiece.

High resistance to vibration comes from judicious specification of the fasteners and the use of a wide support surface for the scale within the aluminum housing. And the mechanical and optical design surrounding the semiconductor light source results in a device that is insensitive to fluids and condensation.

In cases where there is a lot of vibration and temperatures are high, the encoders is a glass scale which is cemented over almost its entire width onto a steel carrier which is bolted directly to the machine bed. This connection is very rigid and results in a natural frequency which is sufficiently above the high-energy frequencies of machine tools. An example of an encoder using this method is the LS 103.

The optical scanning procedure used in such devices is slightly different than typical configurations (Figure 2). It consists of bouncing a light beam off the scale at an angle. The light reflects off the back of the scale, passes through the scale graduation into the scanning reticle and into a photosensor. In addition, a high precision coupling compensates for measuring errors caused by deformations in the scale.

New techniques

There is a limit to the resolution available from ordinary optical encoders. The limit is basically set by the smallest practical distance that can be resolved between the graduations on the scale. To avoid interference from diffraction effects at the graduation lines, the grating period can be no smaller than 10 microns.

However, even finer resolution and accuracy is possible using the interferential scanning principle. Interferential scanning, which is also an optical technique, can resolve graduations finer than 4 microns. Diffraction effects generate the scanning signals. The phase grating of the scale is applied to a reflective surface of glass or steel and likewise consists of reflecting steps. The height of a step is one quarter the wavelength of the light source.

Interferential methods can also be employed in the use of phase grating for two-dimensional measurements. This makes it possible to build small measuring systems for comparators in two Cartesian coordinates.

Encoder manufacturers have also developed new measuring techniques based on magnetic rather than optical scanning. Magnetic encoders have some inherent advantages in terms of insensitivity to contamination. One recently developed device uses a magnetoresistive principle to gauge distances. Here, the scale consists of permanently magnetic material with north and south poles alternating every 200 microns. The magnetic field permeates magnetoresistive tracks and changes its resistance. The relatively small changes in resistance are detected by a bridge circuit.

Scale movements generate two 90 degree phase-shifted sinusoidal (quadrature) signals. An interface circuit converts these into signals which resemble those of photoelectric encoders. Suitable pulse-shaping circuits can resolve 1 micron.

The scale and scanning reticle material in these encoders are each only about one millimeter thick, allowing for a compact design. A magnetic linear encoder of this type can be integrated into a linear guideway. The insensitivity to oil and cutting fluids, combined with a mechanical seal against swarf, results in a rugged design.

Interpolation magic

Encoder manufacturers often process output signals through a special kind of circuitry known as interpolation electronics. This circuitry takes the two sinusoidal signals put out by the encoder, which are 90 degrees out of phase with each other, and from them produces output signals of a higher frequency. These outputs can measure distance traveled with more resolution than is available from non-interpolated encoder outputs.

For example, one type of interpolation circuit, which uses an array of resistors, can produce signals corresponding to one-twentieth, one-fortieth or even one-hundredth of the grating period. The idea behind the resistive technique is to separate the two encoder signals into 10 signals spaced from 0 to 162 degrees. These signals are then combined to produce two output signals having a frequency at least five times higher than that of the encoder output. The resulting signals are square waves, and subsequent electronics use the time between two successive pulse edges for one measuring step.

As an example, with five-fold interpolation and the usual four-fold evaluation of the square-wave pulse trains in the subsequent electronics, a linear encoder with a grating period of 20 µm can provide a measuring step of 1 µm (0.00005 in.).

There are other interpolation methods that provide even higher frequencies. One digital method digitizes the incoming encoder signals, then uses the resulting information to repeatedly pull entries out of a table in computer memory. These entries correspond to encoder position. Circuitry uses this information to generate square-wave output signals.

Another digital method is called arctangent calculation. Here two digitized encoder signals feed a processor which calculates trigonometric relationships between the two to produce position values within one signal period. While this is happening, the two encoder signals are converted to square waves and counted. The system derives the actual position value from the counted value and the calculated angle. In addition, correction values are read from a memory table and combined with the position value to accommodate systematic errors.

The result of this process is a codes value that can be transmitted over a serial or parallel interface to a control, computer or display device.

Interferential measuring principle

Figure 3: Photoelectronic Scanning using Interferential
Measuring Principle Encoders

High-accuracy applications such as those in the electronics industry frequently make use of encoders with interferential scanning. Linear encoders such as the LIP 101A employ a reflection-type diffraction grating applied to a carrier of solid steel, glass, or glass ceramic (Figure 3). Scale grating periods are 4 or 8 microns. Distances can be measured in increments from 1 µm to +0.005 microns, depending on the model, with accuracy grades to ± 0.1 micron. A large gap between the scale and scanning reticle simplifies installation and makes the encoder more forgiving of contamination.