Hall-effect speed sensors for reliable operation

For high-contamination environments and those subject to temperature extremes, Hall-effect devices provide rugged, reliable and cost-effective speed sensing. They may be implemented with ring magnets, vanes and gear-tooth configurations. The speed of a rotating shaft is a common measurement made across a wide variety of applications. Optical encoders are typically used where high angular resolution or update rates are required, but in some cases they are overkill. For many industrial, consumer, and automotive designs, a few pulses per shaft revolution are more than adequate.



Hall-effect speed sensors provide rugged, low-cost solutions to shaft-speed measurement problems. They operate on the principle of sensing magnetic fields, so they are essentially immune to dust, oil and other sources of contamination that can cause severe malfunctions in optical sensors. In addition, because strong magnetic fields (greater than 100Gauss) are not commonly found in nature, magnetic speed sensors are relatively immune to accidental actuation and other forms of interference.



The three most common speed-sensing schemes that can be implemented with Hall-effect-based speed sensors are ring-magnet detection, vane detection and gear-tooth detection. Each of these methods requires the addition of a special target to the shaft being monitored, and requires a particular type of sensor to detect that target. This text will look at the characteristics of both the targets and the sensors used to implement each of the above sensing schemes.

HALL EFFECT SENSING

One of the major reasons for choosing Hall-effect sensing technology over competing technologies is that silicon-based Hall transducers can be fabricated on standard bipolar and CMOS integrated-circuit processes. This means that significant amounts of signal-processing circuitry can be incorporated on the same die as the transducer. It also means that the Hall sensors can be manufactured relatively inexpensively.



The basis of the Hall effect is that moving charge carriers in an electrical current are deflected at right angles to both their original trajectory and an externally imposed magnetic field. In metals, this effect is very small and difficult to measure. In semiconductors, such as silicon and gallium arsenide, while still small, the Hall effect is sufficiently pronounced to be useful for making magnetic sensors to measure fields in the 1-10,000Gauss range. Because the sensitivities of semiconductor Hall transducers are still very low –in the 10 to 100µV/Gauss range – additional signal conditioning is usually required for any practical application. Fortunately, many semiconductor manufacturers provide a preamplifier and a threshold detector integral to their sensor ICs.



A ring magnet is a disk- or toroid-shaped magnet onto which an alternating pattern of north and south poles has been magnetized. Because the boundaries of the poles usually are not marked in any way, the number, placement, and size of the individual poles are not always obvious. It is possible to visualize the pole pattern using magnetic view-film, which when placed over a ring magnet clearly indicates the outlines of the poles.



Conceptually, a ring magnet is the simplest type of a Hall-effect speed sensor. The ring magnet is mounted on a shaft and spun past a suitable magnetic pick-up. The sensor element in the magnetic pick-up only provides microvolt-level signals in response to the field provided by the ring magnet, so most commercial models included on-board signal-processing and interface electronics, providing a TTL-compatible logic output. Most commercially available offerings operate in one of two modes: switched (the output is activated in response to the presence of a particular pole, and deactivates when the south pole is removed) or latched (the output is activated in response to the presence of one pole and remains activated until deactivated by the presence of the opposite pole).



The advantage of the latched mode of operation over the switched mode is that it provides a more uniform duty-cycle output and allows increased spacing from the ring magnet to the sensor. In either case, the sensor provides one pulse per revolution for each north-south pair of gear-tooth poles. A 10-pole ring magnet, for example, would generate only five pulses per revolution.



Ring magnets offer an easy-to-implement speed-sensing method, but the cost of a suitable magnet can be significant. For price-sensitive, high-volume applications, the best scheme is often vane detection. A vane detector works by using a thin ferrous target to shield a magnetic sensor from a magnetic field generated by a bias magnet. When the ferrous target is absent, a magnetic sensing element can detect the presence of the bias magnet. When the target is placed between the magnet and the sensor, it shunts the field lines away from the sensor. In order to effectively shunt the field, the area of the target must be comparable to that of the magnet and to the air gap between the magnet and the sensor.



An example of a vane detector is the Cherry VN101501. This device is enclosed in a package similar to that of many opto-interrupters. The bias magnet resides in one of the towers, while the sensor element resides in the other. The sensor incorporates integral signal conditioning, and provides a switched digital output. Because the magnet, sensor and housing all have been designed to work together for a wide variety of targets, the need for magnetic engineering on the part of the user has been largely eliminated.



Two problems are commonly encountered when using vane detectors, and they usually result when one tries to migrate an opto-interrupter design to a magnetic vane design. The first is the matter of material selection. For a magnetic vane to function, it must be able to shunt the field. This immediately rules out the use of materials such as plastic, aluminum and most stainless steels, which are not ferromagnetic. Even appropriate materials, when too thin, will fail to reliably actuate a magnetic vane detector. The second problem is that the bias magnet will exert a significant mechanical force on the target, and attempt to pull it into the detection gap. If the shaft provides significant torque, such as the output of a motor, this is usually not a big problem. But if there is only a limited amount of torque, such as in a flow meter, a ring-magnet-sensing scheme is usually a better choice.



GEAR-TOOTH DETECTION

Ordinary ferrous gears present the most challenging type of target for a speed sensor. They are also, unfortunately, a preferred choice of target, as they often are already present in a mechanical system as power-transmission components. Besides gears, gear-tooth sensors can also be used to detect gear-tooth-like objects, such as bolt heads, roller chains, and metal stampings. Because gears are not normally magnetized, they must be detected by the perturbations they cause in a magnetic field generated by the sensor assembly. There are many different schemes for implementing gear-tooth sensors with a wide variety of technologies. We will limit our discussion to just a few of the most common types that can be implemented with Hall-effect technology.



The simplest gear-tooth sensing architecture is shown in Figure 1. The sensor element is placed on the face of a magnet. When a gear tooth passes in front of this assembly, it causes an increase in the magnetic fields seen by the sensor element. When a valley passes in front of the sensor, the magnetic field drops. By setting a suitable threshold level, the presence of a gear tooth may be discerned. The key problem with this type of sensor, however, is in determining what constitutes a suitable threshold. Sensors using this approach often will incorporate circuitry that dynamically adjusts the threshold value in response to the magnetic fields that are actually detected. Figure 2 shows examples of the signals obtained from this type of sensor.



Another type of gear-tooth sensor is the gradient detector. For many applications, such as automobile ignition timing, it is important to know exactly when the edge of a target passes the sensor. The single element gear-tooth approach previously discussed often does not provide very good edge-detection consistency, particularly if there is any variation in the spacing between the target and the sensor. Measuring the gradient at the face of the sensor, however, provides an excellent indication of where the edge of a tooth is. It is difficult to measure the actual gradient, but a good approximation can be obtained by subtracting the outputs of two sensors placed close together, as shown in Figure 3.



The signals obtained from the individual sensor elements are similar in appearance to those that come from the sensor in the single-element scheme, but are skewed as they each “see” a different part of the target: one signal leads the other. Subtracting the two signals yields a resultant gradient signal that clearly indicates where the edges of the target teeth are (Figure 4). To obtain a high-quality gradient signal requires good matching of the sensitivities and offsets of the individual sensor elements.



Again, one of the key implementation issues is where to set the thresholds to discriminate between leading and trailing tooth edges. For demanding applications, a variety of signal-conditioning schemes have been developed, ranging from simple analogue threshold detection to elaborate digital signal-processing systems. For many applications, however, a simple signal-conditioning circuit will suffice.