Optical or magnetic linear encoders handle high-res needs in dirty jobs.
SIKO Products Inc.
Linear encoders help minimize errors in motion transmission such as backlash, hysteresis, abbe errors, ball-screw thermal expansion, and other problems. They also find use as motionfeedback sensors in high-speed positioning systems using linear servomotors.
There is a wide assortment of linear-encoder technologies from which to choose. Two of the more widely used choices today are optical and magnetic encoders.
Optical-encoder technology uses a media such as glass or steel tape with fine graduations. When illuminated by a light source such as a laser or infrared LED, these graduations either reflect or pass light to a detector. The size of the graduations determines the basic signal period. Typically, graduation sizes vary from hundreds of microns to hundreds of nanometers. Additional interpolation of the basic signal produces resolutions from tens of microns to tens of nanometers.
Magnetic technology can take several different forms. One uses a magnetized plastic media laminated to a steel tape strip. A noncontact sensor picks up the magnetic polarity changes embedded in the magnetized plastic. Magnetic induction-type encoders use a steel strip or bar that contains material voids or grooves. The system detects the change in reluctance caused by the gaps in material. Another version incorporates precision ball bearings under compression, again detecting the change in reluctance as the magnetic sensor passes the ball bearing. The magnetic pole width, strip-void index, or ballbearing size typically forms the basic signal period. For these sensors, periods vary from the tens of millimeters to the tens of microns. Signal interpolation boosts resolution from millimeters to the submicron range.
Optical-glass linear encoders offer high resolution, accuracy, and repeatability. The low coefficient of expansion of glass-media scales makes them a good choice where high "dead-length" accuracy is important, as in semiconductorfabrication and high-end metrology machines. Glass scales typically have smaller manufacturing and installation errors than tape scales. However, glass scales offer little immunity to dirt and contamination by fingerprints and dust.
Optical read heads with larger scanning windows, filtering optics, and averaging techniques can work where there are larger contaminants; but only for accumulation of debris smaller than the scanning window — typically a few millimeters. Alignment tolerances for pitch, roll, yaw, ride height, and runout are typically a product of scale pitch. The finer pitch of glass scales over magnetic scales usually makes glassscale installation more troublesome and time consuming.
One method that improves dirt immunity and reduces alignment difficulties is to enclose the glass scale in an aluminum extrusion with lip seals and a mechanical alignment carriage. An optional pressurized-air purge of the enclosed area helps keep contaminants out. However, enclosed scales are contact devices prone to wear. The enclosures add hysteresis and expense and do not work well for high-speed linear servo applications. Glass scales also have limited scale length and are fragile. New optical-scale materials entering the marketplace exhibit lower expansion coefficients and are less fragile than glass; but they're still impractical to ship and install in long lengths.
On the other hand, opticaltapescale encoders provide high resolution and accuracy. Tape scales are usually less expensive than glass scales. Ordered by the reel, tape scales offer cut-and-stick installation that is quick and often less expensive glass scales. Tape scales do not possess the same thermal properties as glass scales. However, special installation techniques let tape scales render thermal mastering to the substrate on which it is mounted. Unfortunately, dirt affects optical-tape scales the same as it does glass scales; and enclosing the scale eliminates the mechanical benefits of a true noncontact system.
Although usually generating coarser signal periods than optical systems, magnetic-encoder technology offers good resolutions, accuracy, and repeatability. The coarse scale pitch allows much larger installation tolerances.
Magnetic-scale accuracy today rivals optical-tape scales. Some magnetic encoders have accuracy in the single-digit micron range over meters of scale length with nanometer resolution. However, those do require a mechanical alignment carriage to maintain the higher installation tolerances needed for the micron-sized pole pitches.
Magnetic encoders usually have low magnetic attraction; therefore, they do not attract large amounts of ferrous material. Any buildup is easily brushed aside by the sensor. Magnetic sensors today are virtually unaffected by external magnetic fields making them suitable for linear-servomotor applications. In addition, magnetic encoders are immune to large dirt particles and submersion while remaining true noncontact devices. With higher shock, vibration, and temperature-range limits compared to optical systems, the difference in resolution and accuracy over optical is a small compromise in system design.
Though they do have benefits, high-resolution quadrature encoders may create system compatibility issues. These types of encoders need much higher bandwidth than standard encoders — sometimes in excess of 40 MHz. So when there is no alternative to a high-resolution encoder, the receiving electronics requires more thought. Enhanced resolutions must have a higher frequency response for the same linear velocity as well as a counter safety factor to prevent dropped counts.
The following equation computes real-time quadrature clock speeds:
S = R/V,
where S = edge separation in μsec, R = resolution in μm, and V = velocity in m/sec. The frequency of the required bandwidth is calculated as the reciprocal of the edge separation:
where FB= bandwidth frequency in MHz.
Experts recommend a safety factor for the counterclock frequency of between two to four. This practice allows for unsynchronized clocks, jitter, noise, and interpolation errors. Thus, the minimum counterclock frequency is given by the equation:
FCC= 2 X FB
where FCC= minimum counterclock frequency and FBthe required bandwidth frequency.
Many manufacturers already take output-clocking errors in mind when specifying sensor bandwidth. If that is the case, there's no need for the clock safety factor calculation.
An issue that arises frequently with encoder applications involves the calculation of FCC based only on programmed linear velocity. The maximum velocity programmed may be quite slow. However, system vibration, dither, and jitter may actually exceed the calculated FCC. Failure to account for these factors may result in encoder overspeed errors.
Some high-resolution encoders incorporate special timing circuits to minimize vibration and jitter errors on their outputs. The retimed outputs allow mating with a slower receiver; but retiming delays can create performance problems within a control loop. Real-time outputs avoid those performance issues.
Sinusoidal output voltages vary significantly with ride height and runout. Receiving electronics may not be able to correctly read the resulting signals. Even properly installed encoders may put out signals too small for the receiving electronics to read. For example, a properly installed encoder with sinusoidal output voltages spec'd from 0.6 to 1.2 VPP causes faults when coupled to a receiver expecting signals of 0.8 to 1.2 VPP. Automatic gain controls help stabilize output voltage levels to assure full compatibility.
Absolute linear encoders are making a comeback in the automation world. Absolute encoders eliminate homing routines, end-of-travel switches, and power outage concerns. But one drawback is that many absolute encoders generate outputs that follow proprietary protocols. These protocols limit the choices of compatible receiving networks. The best policy is to use open, nonproprietary protocols such as SSI or BiSS. One high-performance absolute system that also offers high speeds combines the SSI protocol with an additional analog output. Use of both outputs provides both absolute position and the real-time performance of analog outputs.
All in all, practicality is the rule when determining resolution requirements for a linear-encoder system.