Rotary and linear encoders are standard equipment in today's automated motion systems. Both types, generally, are optically based, relying on light passing through or reflecting off a scale and then interpreted by opto-electronics. For rotary sensing, this works well, but the same cannot be said of optical technology when it's used in linear motion applications.
One lingering problem stems from the inherent difficulty of sealing linear scales. Optical linear encoders, as a result, are susceptible to interference from dust, solvents, and debris, limiting the environments in which they can be used. Other constraints are imposed by the fragile nature of optical scales (especially over long distances), the speed at which they can measure position, and thermal mismatches between glass scales and the metal machine beds they often contact.
Although there is an alternative to optics — linear magnetic encoders — these too are hampered by limitations. Rather than glass scales, magnetic encoders use magnetically imprinted tapes, operating on a principle like that employed in VHS systems and floppy disks. The technology is generally less expensive, but also less accurate and less precise.
Assessing the challenge
Measuring linear motion in an industrial environment can be a difficult task. It requires, at minimum, the consideration of four factors — sealing, speed, connectivity, and environment.
Sealing: Dust, liquids, and debris are common in industrial environments, and they can wreak havoc on traditional optical linear sensors. Other dangers include corrosive solvents and lubricants, both of which can work their way into all but the most tightly sealed scales, obscuring critical light paths and introducing errors.
Methods to combat these threats include frequent cleaning and mounting schemes carefully arranged to protect the scale. Adding supplemental filtering for optics is another option. This takes care of minor anomalies such as dust, light oils, fingerprints, and scratches, but it cannot compensate for more significant interference. A more proactive approach, although it requires compressed air and a dedicated clean-air system, is to repel contaminants with positive air pressure. All these remedies, of course, add time and cost, creating a compromise between capability and expense.
Speed: Anyone who has been stuck behind a cement mixer on a long uphill grade knows that the slowest vehicle sets the pace. The same holds true in automation. Processes can run only as fast as the slowest device in the system, be it a sensor, actuator, or mechanical drive.
When traverse speeds increase, so does the potential for linear measurement errors. Here, physical limitations, particularly in the read head, are a common error source. The read head, which contains the sensing mechanism, usually mounts on some form of carriage that rides on bearings. This allows the sensor to move along the scale.
Optical and magnetic encoders require the gap between the scale and sensor to remain constant within a very tight tolerance. But as acceleration and traverse speeds increase, so do the forces exerted on the carriage and bearings, making it more difficult (and eventually impossible) to maintain this crucial gap. For some applications, this inability to hold the gap is the “cement mixer” slowing everything down.
Connectivity: Computer users often take for granted that their new hard drive or MP3 player will be able to “talk” to their PC instantaneously upon connecting the two together. This sort of connectivity is the result of a concerted effort and the setting of standards that electronics manufacturers universally follow. In industrial electronics, it's a different story.
The electrical interface between sensors and their host systems bears the mark of many manufacturers. Not surprisingly, there is no single “language” to be found among sensors, controllers, and other industrial electronic devices. This can make it difficult, if not impossible, to employ new technology with existing devices. It also places a premium on devices that are more universally compatible with others.
Environment: In automated manufacturing facilities, powerful machines expend huge amounts of energy, generating mechanical shock, vibration, and heat, which can quickly ruin overmatched sensors. Only the most rugged and dependable sensors can survive in today's high-productivity environments.
The shortcomings in traditional technology point to a need for a new type of linear position sensor, one better matched to the challenges posed by industrial environments. Years in the making, such a device now exists. It uses inductive sensing technology, applying electromagnetic fields and measuring the interaction with surrounding materials. By eliminating light transmission from the equation, inductive sensing solves many of the problems previously encountered when measuring linear motion.
The operating principle is simple in concept. It begins with a standard reference signal transmitted through a drive coil. Adjacent sensing coils monitor the resulting electromagnetic field, which is distorted in a predictable and repeatable way when exposed to certain metals. Think of it as a metal detector, except here, the metal is a scale with cyclic ferromagnetic properties and the detector is the read head.
The scale itself is practically indestructible, a 0.6-in. diameter stainless steel tube loaded with a column of precision 0.5-in. diameter nickel-chrome ball bearings. Each scale is calibrated at the time of manufacture using a laser interferometer as the standard. In addition, the pre-load on the column of balls is precisely adjusted to lock in an appropriate scale factor, and the tube is then welded shut.
The read head, which works in tandem with the scale, incorporates drive as well as sensing coils. The drive coil is a single winding that wraps around the sensing coil. The sensing coil is a bit more involved. It consists of six sets of windings, each of which is one ball diameter in length and is further divided into four identical coils. The four coils, designated A, B, C, and D, are evenly spaced in a repeating sequence so that, at any given instant, each phase coil aligns with the same point on each of the six balls within the read head's range.
In essence, the coils and balls in the inductive sensor form a magnetic circuit. The drive coil conducts ac current, generating an alternating magnetic field whose flux path is completed through the scale's steel balls. What's actually changing as the read head moves relative to the column of balls is the magnetic reluctance of the flux path. In other words, the varying reluctance presented by the balls moving through the flux path modulates the amplitude of the voltages induced on the pick-up coils.
The magnitude of this modulation effect is a function of the position of each coil relative to the steel balls in the scale. The result is a unique ratio of signal levels among the A, B, C, and D coil waveforms with respect to the reference or drive current waveform. Analyzing and comparing these signals gives the exact position of the read head over a particular ball. By tracking the number of balls traversed and adding the position of the ball currently in the flux path, the inductive sensor can get an accurate and repeatable position measurement.
Making the grade
An apples to apples comparison shows how inductive sensing measures up to other linear sensing technologies in industrial applications.
Sealing: Because the read head in an inductive sensor does not have to contact or physically “see” marks on a scale, but senses them through a stainless steel tube wall, the scale can be completely environmentally sealed to a rating of IP67. What's more, the read head itself can be filled with an epoxy or other encapsulant, resulting in a solid block, impervious to outside elements. Inductive sensors can even operate while submerged because there are no “lines-of-sight” that could become obscured by dirt or debris.
Speed: Inductive linear encoders are not subject to the speed limits imposed on their optical counterparts. Instead of a bearing-supported carriage riding on a track, inductive encoders rely on read heads that completely surround the scale. This eliminates the possibility of critical gaps varying with traverse speeds. With appropriate electronics, inductive encoders can measure position accurately at traverse rates of up to 20 m/sec.
Connectivity: To process their many complex signals, inductive encoders include onboard digital signal processors. These speedy chips analyze amplitude differences and calculate the read head's absolute position. They also translate position data into a series of digital square waves representing total movement. Programmable logic controllers, data acquisition systems, and other devices can easily read these digital waves to precisely monitor and control linear position.
Inductive encoders are not limited to the basic 5 Vdc output signal. Equipped with either line-driver or open-collector outputs, they can comfortably operate at voltages anywhere between 5 and 28 Vdc in industry standard quadrature format.
Environment: A sensor can't do much better than enclose itself in stainless steel, which is the case with inductive encoders. These hard-headed sensors can stand up to vibration, shock, temperature extremes, and chemical exposure better than their optical and magnetic tape cousins. Using steel scales instead of glass also reduces breakage and lost productivity. What's more, the temperature coefficient of a stainless steel scale closely matches that of most machine components and workpieces.
Repeatibility: Highly uniform ball bearings in the scale and ample a/d processing power make inductive encoders repeatable to within ±1 count. And with error mapping and appropriate correction factors or “teach points,” this repeatability can be leveraged to achieve system accuracies greater than the encoder's out-of-box accuracy.
Resolution and accuracy
Resolution is the smallest increment an encoder can express.
A linear encoder's accuracy is relative to scale length, the measurement's length, and characteristics of the part of the encoder involved in measurement. Accuracy is usually expressed as ±X µm over any one meter of scale length, where X is the maximum allowable error deviation from the mean over the given length.
Comparing linear encoders
At one time, designers had few options in linear encoders. Now, there are three distinct technologies from which to choose — optical, magnetic, and inductive. How they differ determines where each is likely to work best.
Optical: For the last several decades, linear encoders have been almost exclusively optically based. Optical encoders operate on the principle of light passing through or reflecting off a glass or metallic scale. A sensor collects the light and interprets position by measuring incident angles.
Optical encoders shine when it comes to measurements made under ideal conditions. Owing to the precision with which gratings can be applied to glass scales, optical encoders typically achieve high resolutions and accuracies. There is a downside, however. These finely crafted scales are susceptible to damage from shock and vibration, and the optics themselves are vulnerable to contamination, especially in an industrial environment.
Another consideration when using optical linear encoders is what's known as the “critical gap” — the distance between the optical sensor and scale. The angles of light coming off the scale determine position, so it's imperative that the scale-to-sensor distance remains constant. As traverse speeds increase, however, maintaining this critical gap becomes increasingly difficult.
Installation requirements for optical linear encoders vary. Some systems use reflective scales bonded to adhesive tape. Other styles use rigid scales encapsulated in larger guides or sheaths. If contamination is an issue, it may be necessary to pressurize the encoder, using an air compressor and appropriate filters and tubing.
Magnetic: Magnetic encoders tend to look like their optical cousins. However, rather than manipulating light beams, magnetic linear encoders monitor magnetic charges along a flexible tape scale. As opposed to optical scales, which have physical gratings, magnetic scales are patterned with magnetic poles. The read head senses these variations as it passes over them and uses this information to determine position.
Without the need for light paths and optics, magnetic linear encoders are less vulnerable to contamination than optical linears. This increased ruggedness comes at a price, however. Magnetic scale pitch (or charge spacing) is usually greater than the pitch achievable on an optical scale, which in turn limits resolution and accuracy.
Like opticals, magnetic linears must maintain a critical gap between the sensor and scale. This is necessary because position measurements are based on the strength of the magnetic charge, which is a function of the distance between the read head and tape. The mechanical requirement of the critical gap limits traverse rates, just as it does with optical encoders.
Installing a magnetic linear scale requires a clean, dry mounting substrate. An adhesive on the underside of the flexible tape holds it down, while a cover strip protects it from above. Any irregularities such as bumps or kinks can cause measurement errors, as does misalignment. The read head must be perpendicular to the scale within a couple of degrees, and the ride height (or critical gap) must remain constant to within a fraction of a millimeter over the entire length of the scale.
Inductive: Inductive encoders, the newcomer, work by monitoring the effect of a metal scale on an electromagnetic field. Instead of a read head gliding a fraction of a millimeter above a reference plane, the scale itself (the reference) passes right through the sensor. The elimination of critical gaps translates to traverse rates an order of magnitude faster than other encoder types.
Inductive linears are also inherently robust. Their scales are made entirely of steel. And the lines of electromagnetic flux coming from the scales and the wire coils monitoring them are impervious to contamination. As for accuracy, the signals generated by the inductive encoding technique can be analyzed to such a degree as to rival the accuracy and resolution of optical encoders.
Installing an inductive encoder is basically a matter of mounting its scale. This is done by bolting a set of universal mounting brackets to the host machine via drilled and tapped holes. A dial indicator can verify that the scale — held in the brackets — is parallel to the axis of travel. Bracket height and alignment are adjustable, providing additional leeway during installation.
Inductive sensing's operating principle
Understanding the inner workings of an inductive sensing read head starts with an appreciation of how its sensing coils are positioned, and why. In the read head image, notice that each turn of coil A is positioned precisely one ball pitch apart, placing it at the same relative position for all six balls. The same holds true for coils B, C, and D. This configuration causes each of the four coils to sense each sphere's effect on the electromagnetic field from a different “angle.”
When the scale moves through the read head, its nickel-chrome balls interfere with the electromagnetic field generated by the drive coil. This interference varies the amplitude of sinusoidal signals induced in the sensing coils, causing phase imbalances. For example, A and C are 180° out of phase, as are D and B. A is 90° out of phase with both D and B, and so on.
Phase angle differences result from the sensing coils' orientation with respect to the ball bearings and are carefully orchestrated to allow trigonometric relationships to be used when calculating the read head's position. First, complementary signals are electronically summed: A with C and D with B. Then, an a/d converter digitizes each complement. Since A-C and D-B are sine and cosine waveforms, the ratio of sine to cosine, or tangent, represents the read head's absolute position over one sphere. To find the total distance traveled, a linear encoder's on-board processor adds the read head's current position to the number of spheres traversed since the last sample.