Machine Design

Sensors get smarter

Compact circuitry lets brainy sensors pack processing wallop.

By Dave Edeal
MTS Systems Corp.,
Sensors Div.
Cary, N.C.

Edited by Miles Budimir

Smaller, more powerful processors eliminate the need for external converter modules, such as this AEC-100 made by CMC Sencon, Inc. The circuitry formerly housed in an enclosure roughly 2 X 5 X 7 in. now fits into an output module roughly 0.5 1.5 2 in. This fits inside a standard sensor housing like the Temposonics R Series AQB quadrature interface sensor from MTS Sensors Div.

Typical quadrature signals are spaced 90° apart. The Z channel is used for referencing the zero-count measurement.

The principle of magnetostriction (a change in dimension of a ferromagnetic material due to a magnetic field) is the basis for position measurement. An electronic pulse travels along a waveguide from a reference head to a magnet. The time for a return pulse produced by twisting the element at the external magnet is measured and converted to distance.

Sensors with traditional analog outputs are still widely used for displacement feedback. However, demand for digital smart sensors continues upward. Constant pressure to increase productivity and reduce installation and maintenance costs has led to a need for smart plug-and-play components and subsystems. The move toward industrial fieldbus networks as a basis for machine control is proof of this trend.

Many manufacturers have integrated controllers and software into new products to meet the need for smarter components. A recent example of the integration is magnetostrictive-displacement sensors with directquadrature interface.

In the past, programming parameters, encoding outputs, and uploading programs to sensors could only be done at the control location using a discrete converter interface. Advances in digital microelectronics now let sensor manufacturers provide more sophisticated, digitally based products that are fully integrated, thereby eliminating the cost and complexity of separate converter interfaces. Lower costs and added functions also make the digital products a more attractive design option. In the case of magnetostrictive-displacement and level sensors, adding digitalsignal conditioning and signal-translation circuitry makes for highly optimized sensor packages. What's more, the additions are made without increasing the size of the standard sensor housing.

The fundamental principles behind magnetostrictive position sensing have been refined quite a bit since first brought to the industrial sensor marketplace.

Improvements have resulted in a fiftyfold increase in raw position resolution and a factor-of-five improvement in output linearity versus first-generation magnetostrictive sensors.

Improvements in manufacturing and materials processing have led to better performance. And more powerful digital components have enabled theoretical and signal-processing enhancements.

The basis of magnetostrictive measurement is time. What the sensor actually measures is the time it takes for a sonic pulse to travel a measured distance. This type of measurement takes advantage of the effect in magnetostrictive materials, where magnetic fields can produce mechanical strains and vice versa. A magnetic field produced in the magnetostrictive material of the sensor's waveguide interacts with the perpendicular magnetic field from a position-sensing permanent magnet, producing a torsional strain pulse. This is called the Wiedemann effect. The strain or return pulse travels mechanically away from the position magnet toward an electromagnetic pickup at the speed of sound in the waveguide. The measured position is simply the time it takes for the sonic wave to travel from the position magnet to the pickup, multiplied by the speed of sound in the waveguide material.

The resolution of the sensor is defined as:

R = 1/(G F)

where R = resolution, mm; G = gradient or the reciprocal of the speed of sound in the waveguide (around 0.35 sec/mm); and F = counter frequency, Hz.

The ability to sense extremely small slices of sonic-wave travel time makes for high resolution. This comes from using a high-speed counter, which activates after energizing or interrogating the waveguide with its magnetizing current. In the past, slower 28-MHz counters gave raw position resolutions of approximately 0.1 mm. The only way to achieve higher resolution values was to perform repeated back-to-back interrogations (recirculations) with each recirculation improving the sensor resolution by the reciprocal of the number of repeated interrogations.

The problem with this approach is that each recirculation adds to the update time. The use of high-speed 4-GHz counters yields resolutions of 2 microns with no increase in update time or concern for tracking errors due to position-magnet motion during recirculations. This is a key requirement for sensors that produce both position and velocity or multiple magnet-position outputs.

The high-end magnetostrictive sensors provide more precise measurements under a wider range of noise and vibration conditions. One reason for this is the use of a socalled Vallari transformer as part of the sonic pulse pickup. This small, flat piece of magnetostrictive material is attached to the waveguide perpendicular to its axis. The tape, passing through a coil, becomes magnetized by a small permanent-bias magnet. A torsional strain pulse acting on the tape alters the magnetic field. A voltage pulse is then produced in the pickup coil. This pulse then triggers the counter to stop and begin computing the position output.

The Vallari transformer has several functions. In one, it increases the raw sensor output signal by a factor of 13 over conventional concentric waveguide and pickup-coil configurations. In another function, along with double EMI shielding, it increases the sensor's signal-to-noise ratio from values on the order of 3 to more than 250. Also, because the pickup is perpendicular to the waveguide, it's essentially decoupled from longitudinal motion. This makes the sensor immune to external shock and vibration.

Precision feedback with magnetostrictive sensors requires accurately triggering counters from the pickup-coil signal. Slicing time more accurately also produces better output repeatability and linearity. Triggering off of a noisy signal results in increased measurement uncertainty. Sensors that use a concentric pickup coil design must rely on amplification and filtering of the raw signal to improve noise and vibration immunity. The drawback to this type of conditioning is that it typically adds phase lag to the sensor output, resulting in larger position errors, especially at higher velocities.

Quadrature is a common feedback technique in industrial control. It's most often generated by encoders for both linear and rotary motion. The signal is a square-wave pulse train that is communicated at a 5-Vdc, TTL level. As is typical of quadrature outputs, the AQB interface provides both polarities for each format (A, A, B, B, Z, Z) to reduce interference from external electrical noise.

For encoders, resolutions and pulse widths are usually defined by the fixed physical characteristics of the hardware. For example, the number of physical lines per inch etched in a glass scale defines the resolution or number of pulses transmitted in a defined displacement increment, usually as counts/in. (cpi). As the speed varies, the width of the quadrature pulses change accordingly. Slower speeds mean longer pulse widths, and higher speeds result in shorter pulse widths so sensor speed can be computed directly by the controller. Direction of motion comes by monitoring the channel (A or B) that leads the other by a half pulse width. One problem with this method is that at low speeds (near zero), the pulse widths increase significantly so that updating the speed to the correct value takes longer, producing a choppy speed output.

The AQB interface differs from other feedback schemes in several ways. The fundamental difference is that quadrature-pulse outputs are not a result of a fixed hardware configuration such as an encoder grid. The AQB sensor provides a pulse output with a fixed width, and a frequency independent of position-magnet speed. Although speed is not a direct result of the output, it is easily computed as a function of counts (position change) for a given time interval. At even the slowest speeds, position and velocity information is regularly updated at the user-specified pulse frequency.

Another important difference: sensor resolution and fixed-pulse frequencies are user programmable. This gives the sensor a range of applicability for a given stroke length. The resolution can be varied from 50 to 12,500 cpi (0.02 to 0.00008 in./count), regardless of stroke. The fixed-pulse frequency has a user-selected range of 8 kHz to 1 MHz. Other adjustable parameters include sensing polarity, zero-reference (Z channel) pulse location, and width and burst mode.

The basic function of the AQB output module is to keep track of absolute-position changes and convert this information into quadrature signals. Each new absolute-position count subtracts from the previous one, resulting in an incremental output. When the result is a positive number, the channel A signal will lead channel B by 90° (one-half pulse width), and vice versa when the result is negative. Update times are cut by the simultaneous execution of sensor interrogations, signal conversions, and output generation. Typical output update times are less than 1.5 msec for stroke lengths up to 100 in.

Because it's a magnetostrictive sensor, the AQB version can produce inherently absolute position information at any time. Absolute-position information is required after a power loss to the axis controller or quadrature interface. When power is restored, controllers can't determine displaced position without an absolute reading or some sort of rehoming algorithm. Incremental encoders that produce quadrature output cannot produce an absolute output.

The AQB sensor solves this problem by generating a send-all or burst output. This is a continuous stream of quadrature pulses at the prescribed pulse frequency corresponding to the absolute position of the position magnet. The AQB sensor provides this output either at power-up after a user-specified delay, or at any time during normal operation using a switched power input signal. The burst-at-start-up feature can be disabled or programmed with a delay from immediate to 30 sec.

The digital electronics approach has also been applied to a number of other devices including fieldbus sensors with CANbus, DeviceNet, Profibus DP, Interbus-S and Modbus protocols. Advantages include the ability to provide diagnostics, upload sensor parameters, or download process information. These capabilities minimize startup times as well as maintenance and retrofit efforts. And because the protocols allow varying degrees of programming flexibility, configurations or recipes within a given standard can be customized for a specific application.

Other smart sensors feature built-in Synchronous Serial Interface (SSI). The programmable SSI sensor generates absolute-position-encoder output. This interface is rapidly becoming the industry standard for providing the highest resolution (2 micron) and serial data throughput (up to 7,500 measurements/sec) for servosystem feedback sensors. Sensors with built-in position controllers provide external serial parameter programming and controller inputs as well as direct-control outputs suitable for most servovalves. This eliminates the need for an external controller and driver interface.

All of these sensors use the same electronics for raw position sensing. As system and sensor requirements change, this modular design approach will enable relatively simple integration of newer interfaces such as Ethernet or ControlNet into a wider variety of sensor application housings.

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.