JoAnn P. Close, Lawrence M. DeVito, David E. Quinn
Analog Devices
Wilmington, MA
Copyright © 1990 IEEE. Reprinted from International Solid-State Circuits Conference Digest of Technical Papers, Vol. 33, February 14-16, 1990, pp. 240-241, 304.
This material is posted here with permission of the IEEE. Internal or personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution must be obtained from the IEEE by sending a blank email message to pubs-permissions@ieee.org.
By choosing to view this document, you agree to all provisions of the copyright laws protecting it.
Combining precision analog circuits with a Hall cell yields a monolithic sensor for detecting time-varying magnetic fields. The device withstands the rigorous environment of an automobile engine ignition control system where it detects teeth on a rotating ferrous wheel. The open-collector digital output responds to field changes of 15 ±1 Gauss over a -55° C to 150° C range. Yield is improved by laser trim of silicon chromium resistors, eliminating the process dependence of Hall cell sensitivity. An ac-coupling scheme using an external 1µF capacitor rejects system offsets of several hundred Gauss while allowing full sensitivity with signals as slow as 4Hz. Other features include controlled start-up and shutdown behavior and insensitivity to system noise. The 110-transistor, 10,700mil² device is packaged in a 4-pin SIP, and operates on a single 5 to 16 volt supply.
The chip is placed on end in a slotted pole of a permanent magnet. As gearteeth pass, the magnetic field around the chip is distorted, producing pulses as shown in Figure 1. The sensor triggers on magnetic field changes exceeding the ±15 Gauss operate and release points. This configuration of sensor and magnet eliminates runout problems, but at the expense of reduced signal amplitude. Consequently, this application requires greater control of magnetic sensitivity. Magnetic signal amplitude decreases as distance from the target to the sensor (airgap) increases. The range of airgap in the application requires the sensor to handle pulses from 40 Gauss p-p to 600 Gauss p-p. To meet airgap requirements and simultaneously avoid triggering on noise between signals, operate and release points are required to fall between ±10 and ±20 Gauss.
Figure 2 shows the sensor block diagram. The Hall cell output is amplified by differential preamp A1. The resulting signal, VHALL, drives the averaging circuit and one side of the output comparator, C1. When VHALL exceeds the averaging circuit output VAV, IUP is switched to charge CHOLD, the 1µF external capacitor. When VAV exceeds VHALL, IUP is switched to ground and IDOWN discharges CHOLD. VAV follows a slowly-moving VHALL voltage. Maximum error is set by the hysteresis of C2 (less than 2mV, corresponding to a 1 Gauss uncertainty). VAV drives the other side of C1. When the difference between the fast VHALL and the slow VAV exceeds the hysteresis of C1, the sensor output changes state. The low-frequency performance of the sensor depends on the magnitude of IUP and IDOWN and the value of CHOLD. This scheme accurately tracks VHALL DC value even with spurious leakages at the CHOLD node. AC coupling behavior is demonstrated in Figure 3a.
Target surface roughness causes much noise at small airgaps. The concomitant large signal amplitude is exploited to avoid false triggering on the noise between gearteeth signals. Diode Dl pulls down on VAV in the presence of large signals. The trigger points are shifted away from the noisy quiescent level of the signal, avoiding false triggering. (Figure 3b)
Figure 4 shows sensor signal path circuitry. Temperature variation of the Hall gain is compensated for by operating the cell with a current having -800ppm/°C temperature coefficient, derived from the bandgap reference. Process variations of Hall gain are compensated for by adjusting the room temperature voltage across the cell to 2V using laser trimmed resistor RH. The gain of A1 is also temperature and process invariant, as it depends on the ratio of thin film resistors. The hysteresis of C1 depends on the bandgap voltage and resistor ratios, to ensure constant hysteresis over temperature and process. Temperature variation of sensor gain is typically less than 100ppm/°C from -55°C to 150°C.
Figure 5 shows the averaging and start-up circuits. Emitter followers buffer CHOLD from both comparator inputs. Q38 base current, IDOWN, is a replica of Q8 base current. IUP, the tail current of differential switch Q41 and Q42 is set in a similar way. The PTAT currents driving the bases of Q7, Q8 and Q48 are produced by subtracting a 150mV PTAT voltage from the Vbe of Q54. The drop across R23 sets up a 500:1 current ratio between the collector current of Q54 (50µA PTAT) and collector currents of Q51, Q52, and Q53 (100nA PTAT). Upon power-up, CHOLD must charge to nominally 3 volts before the sensor functions. Excessive start-up time results if IUP alone ramps up the capacitor. The start-up quickly charges CHOLD to operating value and then switches itself off unless reactivated by a power-up/down sequence. C2 is used to sense when to stop pulling up CHOLD, eliminating possible overshoot and undershoot. The state of the output comparator is forced low by the start-up sequence. Upon application of supply voltage, the base of Q93 is held low due to CHOLD: this turns on Q71, allowing Q44 to pull up the CHOLD node via Q47. The base of Q36 will hang low relative to Q29, turning on Q72. As VAV rises above 2Vbe, Q93 turns on, enabling the latch formed by Q40 and Q43. Q72 is still on and Q44 continues pulling up on VAV. When VAV exceeds VHALL, the output of C2 changes state; reversing the Q40, Q43 latch. Q44 turns off, ceasing to pull up CHOLD, and Q62 turns on, eliminating further influence of Q72 and Q73. The Q40, Q43 latch remains set, keeping the startup circuit off.
The authors acknowledge T. Freitas, B. Surrette, F. Holden, and Mod B manufacturing organization contributions to this project.
This project is a joint development with Airpax Corporation of North American Philips.
FIGURE 1 - Geartooth sensing. Bn is the field component normal
to the sensor package. Straight up and down field registers as null.
FIGURE 2 - Geartooth sensor block diagram
FIGURE 3a - AC coupling behavior of sensor. VAV tracks a
slowly varying magnetic field and the sensor does not respond.
When VAV cannot keep up with quickly varying fields,
the sensor responds.
FIGURE 3b - Avoiding false triggering at small airgap.
Large signals cause VAV to be pulled down, shifting operate
and release points away from signal quiescent level.
The sensor responds only to large signals.
FIGURE 4 - Simplified sensor signal path circuitry
FIGURE 5 - Averaging circuit and start-up
FIGURE 6 - Chip micrograph
