Sensors - September 2001 - A Self-Calibrating Miniature Hall Effect Solution to Gear Tooth Speed Sensing

A Self-Calibrating
Miniature Hall Effect Solution to
Gear Tooth Speed Sensing

Advances in BiCMOS Hall effect sensor fabrication, circuit design, and packaging are satisfying automotive gear tooth speed sensing requirements.

Today’s automotive position and speed sensing applications demand high operational air gaps, tight switching timing accuracy, and excellent overall reliability. Significant challenges to Hall effect sensing are posed by systems possessing high-density (more than 48 teeth) and small-diameter targets with fine tooth pitches. Such target designs are prevalent in, but are not limited to, automatic braking system (ABS) and automatic transmission applications. The compact target geometry results in small signals that are difficult to resolve with conventional sensor technology. The extreme ambient temperatures in these systems create further complications. The severe operating temperature requirements and the need to provide stable operation over the wide temperature range push the limits of established IC fabrication technology.

Recent developments in BiCMOS Hall effect fabrication, circuit design, and packaging technology are major advances toward meeting automotive gear tooth speed sensing application requirements. Self-calibration techniques, threshold detection circuitry, greater onchip protection circuitry, and smaller packaging have dramatically increased usable air gaps, switching accuracy, and overall reliability. These advances have been instrumental in meeting the needs of fine-pitched speed sensing applications where two-wire operation is required.

The Hall Effect

Figure 1. The Hall effect produces a signal proportional to the intensity of the normal magnetic flux density.

In 1879 Sir Edwin Hall described the principle that now bears his name: when a bias voltage is applied to a silicon plate via two current contacts, an electric field is created and a current is forced. If the plate is then exposed to a perpendicular magnetic induction, b, the Hall electric field gives rise to the appearance of the Hall voltage between the two sense contacts. This Hall voltage, V_h, is proportional to the amount of magnetic field applied normally to the plate (see Figure 1). This basic principle is the foundation for all Hall effect sensors.

Back Biasing for Gear Tooth Sensing

Figure 2. The permanent magnet provides a source of magnetic flux. The changing profile of the target or gear provides changing reluctance in the magnetic circuit. The single-IC sensor provides an output signal when processing the changing Hall-effect signal.

Because Hall effect ICs detect the strength of a magnetic field, sensing is accomplished by changing the magnetic field passing through the IC. This is commonly accomplished through linear or rotary motion of a multipole magnet, but it is often more practical and cost effective to use a simple target or gear made from a ferrous material such as low-carbon steel.

The position and rotation of ferrous gear teeth can be detected using the Hall effect by measuring the changes induced by the gear at the face of an opposing magnet (see Figure 2). The presence of the ferrous gear alters the reluctance of the maŞnetic circuit and creates a concentration effect at the magnet surface. These changes can be measured by using a Hall effect sensing element located on the magnet face.

Element Configuration
Two basic methods of sensing are possible in the back-biased package—single and differential sensor configurations. A single Hall sensor element supplies a voltage proportional to the absolute value of the magnetic field that is induced normal to the element. Amplification and conditioning of the single-element signal can provide a digital representation of the target profile. Alternatively, a differential Hall element pair provides a signal proportional to the slope of the incident magnetic field. The differential sensor’s output is typically zero when it is faced with either a tooth or valley; it produces a signal only when it is in the region of a tooth edge. Both of these solutions offer advantages but have serious problems when used in their basic configurations.

Single-Element Gear Tooth Sensing
When the single-element Hall sensor is biased with a standard dipole magnet, a large field is induced through the Hall plate when no target is present. With conventional sintered magnets that possess high-energy products, this baseline magnetic field resides in the range of 2000–3000 G. In comparison, the amplitude of the field induced by fine-pitched targets passing by the back-biased sensor can be <100 G (see Figure 3).

Figure 3. The back biasing magnet provides very high baseline magnetic fields when presented with a single-element Hall effect sensor. Note the large shift in the baseline field over air gap for the small target valley.

The discrimination of this low-amplitude signal is difficult and the necessary circuitry is typically AC coupled. Furthermore, the slope of the magnetic field varies tremendously with air gap, making accurate edge detection difficult and limiting switch point accuracy. Additionally, the back-biased field values may change due to concentration effects caused by varying valley widths and target eccentricities, resulting in a nonuniform baseline.

Differential Gear Tooth Sensing
The differential Hall element configuration eliminates the undesired effects of the back-biased field through the mathematical process of subtraction. Since each of the two Hall elements on the IC sees approximately the same back-biased field, the differential baseline field is close to 0 G. As the target moves by the sensor, the resultant signal is still relatively small, as it was in the single-element configuration. Because the background field is also very small and close to zero, however, classical threshold-crossing techniques can be used to generate a digital representation of the target (see Figure 4).

Figure 4. The differential Hall effect sensor eliminates high baseline field levels. The resulting signal use of classic threshold switching is centered about zero gauss, facilitating the techniques.

Gear Tooth Speed Sensing
The challenges in developing a high-performance differential gear tooth speed sensor lie in eliminating the false switching and large duty cycle variations associated with classic threshold sensors, and reducing the switch point drift over temperature. Although a conventional peak-detecting scheme could resolve these issues, this technique requires an external capacitor for peak holding. The capacitor represents additional cost, reduced system robustness, and high-temperature performance limitations duý to capacitance roll-off. Furthermore, the capacitor’s need for a third pin precludes the use of a conventional peak detector in a two-wire system. To meet all cost and performance requirements, and allow the use of a two-wire interface, an advanced differential device is needed.

IC Fabrication
Integrated circuit design complexity has increased dramatically over the past decade. Component count in sensor ICs has risen from 50 in 1980 to more than 7000 today. Semiconductors have progressed from standard bipolar and CMOS to BiCMOS and BCD (bipolar/CMOS/DMOS) merged technology processes. One of these, the DABIC4 (digital analog BiCMOS version 4), provides precision analog signal processing and complex logic functions in a fully integrated monolithic silicon IC. Bipolar components allow the design of low offset amplifiers, and CMOS components provide efficient A/D converter design. The increased component density provided by the DABIC4 process supports the development of sophisticated algorithms.

Using DABIC4, continuous operating temperatures of –40°C to 150°C can be accommodated. Surges beyond this range can be withstood up to a device junction temperature of 190°C. DABIC4 also provides reduced temperature-induced switch point shift compared to previous technology, resulting in a more consistent device over the full operating temperature range.

DABIC4 also permits design features such as reverse power supply protection, transient protections, wide operating voltage range, and output short-circuit protection.

The combined advances in analog and digital circuitry provide both low- and high-speed capability. Operation from zero speed to frequencies of at least 15 kHz is attainable.

Self-Calibration Amplification
The timing accuracy of typical sensors is determined by the variation in the slope of the magnetic signal with air gap. A large gain is typically required to generate a suitable signal at large air gaps. At close air gaps this gain makes the signal exceeć the dynamic range of the internal operational amplifier and results in signal clipping. The resulting variation in the slopes of the signals over the operational air gap range causes large timing accuracy and duty cycle shifts.

Figure 5. Due to the changing magnetic reluctance, the magnetic flux density decreases exponentially with air gap. The output resulting from fixed threshold switching techniques would provide poor accuracy and possibly missed edges.

Figure 6. Automatic gain control (AGC) normalizes the signal amplitude to provide consistent switching performance over air gap.

Improved timing accuracy and duty cycle performance are achieved with self-calibration circuitry provided by an automatic gain control (AGC) technique. This circuitry is engaged at device powerup and measures signal amplitude to normalize the device gain. Figure 5 shows peak-to-peak signals over air gap for a device without AGC; Figure 6 shows similar data for a device with AGC. Note the consistent switch point that results over air gap with the signals in the AGC circuit.

Since sequential targets do not have a signature region, they can be “learned” by the sensor within a single set of output transitions. To prevent the sensor from tracking errant signals created by transient events that occur after startup, AGC is disabled after the sensor has acquired enough data to learn the target. An example of such an event is a metal filing that momentarily attaches itself to a gear tooth and then, after several revolutions, falls off. AGC is enabled each time the device is powered on and disabled when the criterion discussed above is met.

Though the sensor should reject transient effects, it needs to discern steady-state changes in the magnetic system. To allow it to adapt to long-term changes such as a permanent alteration in installation air gap due to severe vehicle system impacts, an updating scheme is included in the device. The update algorithm allows the normalized signals to change if a variation is seen for a minimum number of consecutive output transitions. If the change is not present for at least this duration, then no change is made to the signal. If the change meets this minimum duration, the signal is allowed to change in the direction of the induced shift.

Disabling the AGC allows consistent and accurate switching despite the presence of rapid transients, while the update algorithm accommodates slower transients and steady-state shifts.

Digital Threshold Sensing
Dynamic thresholds eliminate the poor vibration performance associated with fixed-threshold or zero-crossing switching. To ensure that the switch points always occur within the dynamic range of the normalized signal, the thresholds are established as a percentage of the peak-to-peak signal. Threshold sensing close to the zero crossing also provides improved timing accuracy, and, therefore, duty cycle over the installation air gap (see Figure 7).

Figure 7. This graph demonstrates that AGC and low-hysteresis dynamic switching thresholds provide consistent switching over a wide air gap range.

System offset is a limiting factor in conventional threshold detection. Large teeth or valleys can produce large offsets due to installation tilt or nonuniform back-biasing magnets. In these undesirable scenarios, a threshold can be crossed and create a timing shift in the position of an edge by as much as one tooth. Since threshold sensing is less immune to system offsets than is peak detecting, the technique is best suited to targets that produce sinusoidal signals. Since ABS and transmission applications meet this criterion, and because vibration is prevalent in these environments, threshold sensing with gain adjustment is the optimum solution.

Figure 8 shows actual oscilloscope traces of the output of two sensors that have been subjected to mechanical vibration in an ABS application simulated in a laboratory environment.

Figure 8. AGC and dynamic zero-crossing thresholds provide hysteresis that is independent of air gap, resulting in excellent vibration immunity. Disabling of AGC after the calibration phase also provides insensitivity to high-amplitude, short-duration (impact) events.

The device on the top is a differential Hall sensor optimized with AGC and dynamic thresholds; the device on the bottom is a variable reluctance (VR) sensor. Note the chatter of the output on the VR and the completely clean signal of the Hall.

Two-Wire Operation
Conventional sensors commonly use an open-collector output structure that necessitates a third signal pin in addition to the supply and ground pins. The space and cost requirements of today’s ABS and transmission applications call for two-wire sensors, which provide an output signal by modulating the total current demand of the device. A sense resistor in the control module allows for the interpretation of the output using a simple comparator (see Figure 9).

Figure 9. Two-wire operation is easily achieved through the use of a sense resistor and comparator. An integrated voltage regulator and circuit protection from advanced fabrication technology allows for a wide range in supply voltage.

Packaging

Figure 10. The SG package provides a completely integrated, back-biased Hall effect sensor. The optimized sensor/magnet system requires minimal user expertise in magnetic circuit design.

Figure 11. The SB package provides all of the required Hall effect sensor elements in a mechanically assembled module.

Recent innovations in packaging help the device achieve high air gap performance and meet restrictive space requirements. The SG package allows the magnet to sit closer to the IC than it does in conventional packaging. This geometric advantage allows the device to work with a higher air gap and a smaller magnet than possible with standard packages. The resultant small package fits easily in the tight spacing dictated by ABS applications (see Figure 10).

Additional benefits are realized in the SG package through the single-step molding operation. Sensors are typically manufactured with successive mechanical assembly steps (see Figure 11), and the clearances required by this technique result in voids throughout the interior of the sensor unit. The SG has no such air gaps, translating into improved heat dissipation and elimination of the air entrapment that can occur during subsequent potting operations (see Figure 12).

A further improvement in heat dissipation is realized through the reduced heat conduction path of the SG package. Single-step molding

Figure 12. In this cross-sectional view of the SB package (A) and the SG package (B), note the compact SG assembly that results in a small-size, intimate magnet contact for IC heat sinking and the elimination of air pockets to allow robust postprocessing.

eliminates the layer of plastic normally placed between the IC’s lead frame and the magnet. The increased thermal conductivity signifies the magnet’s greater heat sinking capacity, permitting operation at higher ambient temperatures.

The lead configuration of the SG makes for easy PCB surface mounting and simple attachment of a bypass capacitor. The spacing provided between the two leads allows an axial leaded capacitor to be welded across them. Attachment to the sensor can be made with a lead frame to preclude the use of a costly PCB.

Conclusion
New advances in Hall effect sensor technology are providing solutions to the increasing demands of speed-sensing applications. Advances in IC fabrication permit increased component density, which supports the needs of sophisticated algorithms. These circuits address the problems created by applications with increased air gaps, mounting-induced magnetic offsets, and high-vibration environments.

The advanced circuitry and differential sensor configuration eliminate certain problems associated with conventional sensors. In addition, the resulting sensor output signal is independent of both speed and supply voltage and is very insensitive to mechaüical system vibrations. IC advances have also led to more robust components capable of operation at higher temperatures and voltages. Packaging advances reduce sensor size without compromising performance and contribute to an expanded operating temperature range.

For Further Reading
Milano, Shaun, and Ravi Vig. 1997. “Self-Calibrating Hall Effect Gear Tooth Sensing Technology for Digital Powertrain Speed and Position Measurement,” Proc 30th Annual Intl. Symp on Automotive Technology & Automation, Vol. 2, ISATA.