AMR magnetic field sensors were briefly introduced in the December 1998 issue of Sensors ("A New Perspective on Magnetic Field Sensing "). Now we take a closer look at the way they work and where they are used.

Michael J. Caruso and Tamara Bratland, Honeywell, SSEC
Carl H. Smith and Robert Schneider, Nonvolatile Electronics, Inc.

Magnetoresistive sensors are made of a nickel-iron (Permalloy) thin film deposited on a silicon wafer and patterned as a resistive strip [1­3]. The film's properties cause it to change resistance by 2%­3% in the presence of a magnetic field. In a typical configuration, four of these resistors are connected in a Wheatstone bridge (see Figure 1) to permit measurement of the magnitude of the magnetic field along the direction of the axis. The bandwidth is usually in the 1­5 MHz range. The reaction of the magnetoresistive effect is very fast and not limited by coils or excitation frequencies.

Figure 1. The basic AMR sensor circuit provides a true differential output    (sign and magnitude) of a magnetic field.

AMR properties are well behaved only when the film's magnetic domains are aligned in the same direction. This configuration ensures high sensitivity and good repeatability with minimal hysteresis. During fabrication, the film is deposited in a strong magnetic field that sets the preferred orientation, or easy axis, of the magnetization vector M in the Permalloy resistors (see Figure 2). The M vector is set parallel to the length of the resistor and can be set to point in either direction, left or right, in the film. Assume for a moment that there is a current in the film flowing at a 45º angle to the length of the film. This creates an angle, u, between the current flow and M vector. The electrical properties of the Permalloy film have a relationship between the M vector in the film and the current flowing through the film. The film resistance is greatest when the current flows parallel to the M vector.

Figure 2. The Permalloy film has a magnetization vector, M, that is influencedby the applied magnetic field being measured. The resistance of
the film changes as a function of the angle between the vector M and a current flowing through it. This change in resistance is known as the
magnetoresistive effect.

An external magnetic field applied normal to the side of the film causes the magnetization vector to rotate and change angle u. This in turn will cause the resistance value to vary (R/R) and produce a voltage output change in the Wheatstone bridge. This change in the Permalloy resistance is termed the magnetoresistive effect and is directly related to the angle of the current flow and the magnetization vector.

Note in Figure 3 that the R/R change in resistance is symmetric about the angle axis and that there is a linear region about the 45º angle. The method used to cause the current to flow at a 45º angle in the film is called barber pole biasing, and is accomplished through a layout technique that places low-resistance shorting bars across the film's width. The current, preferring to take the shortest path through the film, thus flows from one bar to the next at a 45º angle. Figure 4 illustrates this effect for all four resistors in a simple Wheatstone bridge.

Figure 3. This curve shows how the resistance of the Permalloy film changes with the angle q between the direction of the current and the
direction of the magnetization, as defined in Figure 2. When the current flow is at a 0º angle with the vector M, the resistance is at a maximum.
When the current flows at 90º to the vector M, the resistance is at a minimum. AMR sensors are usually biased at a 45º angle using a
technique called barber pole biasing.

Figure 4. The AMR bridge is made up of four Permalloy parallel strips. A crosshatch pattern of metal is overlaid onto the strips to form
shorting bars. The current then flows through the Permalloy, taking the shortest path, at a 45º angle from shorting bar to shorting bar. This
establishes the bias angle between the vector M of the film and the current.

In the presence of an applied magnetic field, the magnetoresistive characteristic of the Permalloy causes a resistance change in the bridge and a corresponding change in voltage output. The bridge sensitivity, S, is often expressed in mV/V/Oe. The middle term, V, of this unit refers to the bridge voltage, V_b. When V_b is set to 5 V, and S is 3 mV/V/Oe, then the output gain will be 15 mV/Oe. Through careful selection of a bridge amplifier, bridge output levels of
1 V can be detected. This output corresponds to a magnetic resolution of 67 Oe, or 1 part in 15,000/Oe.

If the bridge output is amplified by a gain of 67, the total output sensitivity would be 1 V/gauss (G) (= 67 · 15 mV/G). If a full-scale range of ±2 G is desired, this implies a 4 V output swing centered on the 2.5 V bridge center value-or a span of 0.5­4.5 V. This signal level is suitable for most A/D converters. Using an AMR sensor and amplifier, precise magnetic field data can be derived that provide field magnitude as well as directional information.

A concern for any magnetic sensor made of ferromagnetic material is the exposure to a disturbing magnetic field. For AMR sensors, this field actually breaks down the magnetization alignment in the Permalloy film that is critical to the sensor operation. The direction and magnitude of vector M are essential for repeatable, low-noise, low-hysteresis output signals. Figure 5A shows the AMR film exposed to a disturbing magnetic field. The Permalloy strip is broken up into randomly oriented magnetic domains that degrade the sensor operation shown in Figure 2.

Figure 5. The Permalloy film can be thought of as clusters of magnetic domains with their north-south polarization represented by arrows.
When randomized (A), the film will not appear to have any magnetization vector M. If a strong magnetic field is applied in the direction of the
film (B), these domains will rotate and "stick" in a preferred state where the north and south poles face each other. Now the vector M has
been established.

To recover the magnetic state, a strong magnetic field must be applied along the length of the Permalloy film. Within tens of nanoseconds the random domains will line up along the easy axis as shown in Figure 5B. Now the M vector is restored and the predictable magnetoresistive effect will occur. The M vector will stay in this state for years as long as there is no magnetic disturbing field present.

These domains are commonly realigned by winding a coil around the Wheatstone bridge resistors. Switching a high-current pulse through the coil (see Figure 6) will create a large magnetic field of 60­100 G and restore the M vector [4]. This process, referred to as flipping the magnetic domains with a set pulse, will also take place for a pulse in the opposite direction through this external coil. In this case, the reset pulse, the domains will all point in the opposite direction along the easy axis.

Figure 6. The magnetization vector M can be "set" in one direction or "reset" in the other direction by pulsing a current through a coil coupled
to the Permalloy film. A strong current will produce a magnetic field large enough to rotate, or flip, all the magnetic domains in the film, thus
producing the vector M.

Offset Reduction in AMR Sensors

Before addressing specific applications it is useful to understand how to operate the AMR sensor. Specifically, undesirable effects inherent in the sensor may interfere with magnetic field sensing such as bridge offset voltages and temperature effects. We will now turn to these concerns and examine some ways to perform automatic gain adjustment and real-time offset cancellation.

Figure 7. An AMR sensor can produce 30 mV output swings for a 4 G change in the applied magnetic field. The bridge transfer curve will
have a positive slope when the vector M is in the set state and a negative slope when it is in the reset state. The intersection of the zero field
axis turns out to be the bridge offset voltage (V_OS).

The use of a set/reset pulse has benefits in addition to restoring the sensor's properties after exposure to a high magnetic field. Figure 7, the transfer curves for a sensor after it has been set and then reset, shows an inversion of the gain slope and a common crossover point on the bridge output axis. This crossover point is the zero field bridge offset voltage. For this sensor, the bridge offset is around ­3 mV; the reason is a resistor mismatch during manufacture. This offset voltage is usually not desirable and can be reduced or eliminated using one of the four following techniques:

Manual Offset Trim. The most straightforward technique for offset reduction is to add a parallel trim resistor across one leg of the bridge to force both outputs to the same voltage. This must be done in a zero magnetic field environment, usually in a zero gauss chamber. It is labor intensive since each sensor may require a different value trim resistor.

Offset Strap. Another way to remove the offset voltage is to use a coil to create a field in the sensitive-axis direction. Static current through this coil can be set to null the bridge offset by adding or subtracting a field equal to the offset voltage. (Honeywell's family of AMR sensors has a patented onchip offset strap to accomplish this.) Again, the offset current must be determined in a zero gauss environment and requires a constant DC source. In this article, subsequent references to the offset strap will imply either the onchip strap or an external coil.

Figure 8. If the AMR bridge is measuring an applied magnetic field, then alternating the set and reset pulses will cause the bridge output to
switch between: V_SET = (+_VAPPLIED) + V_OS, and V_RESET = (­V_APPLIED) + V_OS.

Set/Reset with Microprocessor. A third method of canceling the bridge offset, V_OS, is by numerical subtraction. To measure a field H_APPLIED, first activate a set pulse (see Figure 8). After it has settled, take a reading and store it as V_SET. Repeat these steps for a reset pulse and store the reading as V_RESET. The expressions for these two readings, and their differences, are:

V_SET = S · H_APPLIED + V_OS (1)

V_RESET = ­S · H_APPLIED + V_OS (2)

V_SET ­ V_RESET = 2 · S · H_APPLIED (3)

Note that in Equation (3) there is no V_OS term and the desired field, H_APPLIED, is doubled. The advantage of this method is that any temperature drift of the bridge offset, including the amplifier, is eliminated. This is a powerful technique and easy to implement if the readings are controlled by a microprocessor.

A variation is to add V_SET and V_RESET instead of subtracting them; the result is 2 · V_OS. This approach can be used to periodically check the offset voltage, say during the power-on cycle or once every 10 min. The V_OS can then be subtracted from all subsequent readings. Doing so will allow increased input signal bandwidth and help reduce power consumption.

Figure 9. A simple feedback loop around the bridge amplifier will remove the bridge offset (VOS) as well as the amplifier offset and any
temperature drift with these offsets. An additional amplifier can be used to provide a gain of +1 and ­1 synchronous with the set/reset
switching to demodulate the bridge amplifier to a DC level.

Electronic Feedback. The bridge offset can be eliminated electronically using a feedback amplifier (see Figure 9). The basis of operation is to modulate the sensor input signal to a higher frequency, remove the offset, and then demodulate it back to a DC voltage. This can be accomplished with the set/reset switching property shown in Figure 6. By using a square wave of 200 Hz to alternately create set and reset pulses, the bridge output voltage will switch between V_SET and V_RESET as described in Equations (1) and (2). This switching of V_OUT1 helps reduce the signal noise by modulating the low-frequency signals of interest to a higher band, away from the 1/f noise, and where the flatband noise is minimal. Before the output of amplifier #1 is connected, the intermediate signal, V_OUT1, is in the form of a square wave with amplitude related to 2 · H_APPLIED and an offset level of V_OS as shown in Figure 8.

Amplifier #1 is designed with a low-pass frequency response so that its output will not follow the 200 Hz square wave from the bridge. Instead, it will output a negative DC level corresponding to the V_OS of the bridge and any offset of amplifier #2. When this signal is connected to the (+) input of amplifier #2, it cancels these offsets. Now, the intermediate signal V_OUT1 is in the form of a square wave with amplitude related to 2 · HAPPLIED and centered around V_REF. By using a selectable ±1 gain block controlled by V_SET/RESET, the output signal, V_DEMOD, will be demodulated. This produces a DC level that is directly proportional to H_APPLIED. An additional low-pass filter (~10 Hz) should filter the V_DEMOD signal to eliminate any residual switching noise at 400 Hz out of the demodulator. This circuit has very low temperature drift since the bridge offset and temperature variations, as well as the offset and temperature effects of the bridge amplifier, are continuously being canceled. The magnetic signal bandwidth is limited to 10 Hz for this example.

Compensating for Hard Iron Effects

Any external magnetic field can be canceled by driving a defined current through the offset strap. This is useful for eliminating the effects of stray hard iron distortion of the Earth's magnetic field. One example is reducing the effects of a car body on the Earth's magnetic field in an automotive compass application. If the sensor is in a fixed position within the automobile, the effect of the car on the Earth's magnetic field can be approximated as a shift, or offset, field. If this shift can be determined, it can be compensated for by applying an equal and opposite field using the offset strap.

In-Circuit Gain Calibration

The offset strap can also be used to autocalibrate the AMR bridge during normal operation. This is useful for occasionally checking the bridge gain for that axis or making adjustments over a temperature drift. This can be done during power-up or anytime during normal operation. The concept is simple: take two points along a line and determine the slope of that line-the gain. When the bridge is measuring a steady applied magnetic field the output will remain constant. Record the reading for the steady field and call it H1. Now apply a known current through the offset strap and record that reading as H2. This can be as simple as switching a 1 k resistor in series with the offset strap using a microprocessor output. The current through the offset strap will cause a change in the field the AMR sensor measures-call that the delta applied field    (H_APPLIED). The AMR sensor gain is then computed as:

AMR_GAIN = (H2 ­ H1) / H_APPLIED (4)

Closed-Loop Circuit for Precision Measurements

The offset strap can be used as a feedback element in a closed-loop circuit (see Figure 10). Using the offset strap in a current feedback loop can produce desirable results for measuring magnetic fields. To do this, connect the output of the bridge amplifier to a low-pass filter driver connected to the offset strap. Using high gain and negative feedback in the loop, this technique creates a canceling, or offsetting, magnetic field that will drive the AMR bridge output to zero. The resultant current through the offset strap indicates the strength of the field being canceled. This current is measured using a resistor, R_SENSE, which generates an output voltage, V_SENSE. The results are extremely good linearity and temperature characteristics. The idea in this circuit is to always operate the AMR bridge in the balanced-resistance mode, i.e., whatever magnetic field is being measured, the current through the offset strap will cancel it out. The bridge always "sees" a zero field condition. The resultant offset current required to cancel the applied field is a direct measure of that field strength and can be translated into the field value.

Figure 10. A magnetic field from a controlled DC current can be coupled to the AMR sensor to oppose the applied field being measured. The
strength of this controlled current is a direct measure of the applied field being nulled, or bucked, out.

There are many uses for the offset strap in addition to those described here. The key point is that the ambient field and the offset field simply add to each other and are measured by the AMR sensor as a single field.

AMR Sensor Applications

The AMR sensors on today's market do an excellent job of sensing magnetic fields within the Earth's field-below 1 G. They are used to detect ferrous objects such as planes, trains, and automobiles that disturb the Earth's field. Other applications include magnetic compassing, rotational sensing, current sensing, underground drilling navigation, linear position sensing, yaw rate sensing, and head tracking for virtual reality.

Vehicle Detection. The Earth's field provides a uniform magnetic field over a wide area, say, several square kilometers. A ferrous object, a car in this case, creates a local disturbance in this field whether the object is moving or standing still. AMR magnetic sensors can detect the change in the Earth's field caused by this disturbance.

A single-axis sensor can detect the presence or absence of a vehicle up to 15 m away, depending on the target's ferrous content. Drivers entering parking garages could use this information to find the most satisfactory available parking space. Another application is to control railroad crossing gates based on the detection of approaching trains. Here, two sensors could be used to detect presence, direction of travel, and speed.

Magnetic disturbances can be used to classify vehicles for toll road applications. A 3-axis AMR magnetometer placed in the lane of traffic will provide a rich signal output for vehicles passing over it. Figure 11 is a magnetometer output for three vehicles driving over it at roughly 1 s, 3 s, and 5 s on the time axis. The type of vehicle (e.g., car, truck, bus) can be classified through pattern recognition and matching algorithms.

Figure 11. AMR sensors can detect Earth field anomalies as three cars drive by, as shown here. Three sensors provide the X, Y, and Z axes
and show unique signatures for the three different types of vehicles.

Electronic Compass Using AMR Sensors [5]. The Earth's magnetic field intensity is ~0.5­0.6 G and has a component parallel to the Earth's surface that always points toward magnetic north. This is the basis for all magnetic compasses.

Figure 12. The earth's magnetic field can be thought of as a simple dipole magnet. The horizontal, or level, componet will always point toward magnetic north, which is often different from true, or grid, north.

The Earth's magnetic field can be approximated with the dipole model shown in Figure 12, which illustrates that the Earth's field points down toward north in the northern hemisphere, is horizontal and points north at the equator, and points up toward north in the southern hemisphere. In all cases, the direction of the Earth's field is always pointing to magnetic north. It is the components of this field that are parallel to the Earth's surface that are used to determine compass direction. The vertical portion of the Earth's magnetic field is ignored.

To achieve a compass accurate to 1º requires a magnetic sensor that can reliably resolve angular changes to 0.1º. The sensors must also exhibit low hysteresis (<0.05% F.S.), a high degree of linearity (<0.5% F.S. error), and be repeatable. The magnetic fields in the X and Y plane will typically be in the 200­300 mG range-more at the equator, less at the poles. The required magnetometer resolution can be estimated by using the relationship:

azimuth = arc-tan (Y/X) (5)

To resolve a 0.18º change in a 200 mG field would require a magnetic sensitivity of better than 0.35 mG. Solid-state AMR sensors are available today that reliably resolve 0.07 mG signals giving a 5 3 margin of detection sensitivity.

Compasses are not usually confined to a flat and level plane but rather are often handheld units, attached to an aircraft, or mounted on a vehicle traveling uneven terrain. It is therefore more difficult to determine the azimuth, or heading direction, since the compass is not always horizontal to the Earth's surface. Errors introduced by tilt can be quite large, depending on the amount of the dip angle. Compass tilt is typically corrected with an inclinometer, or tilt sensor, that determines the roll and pitch angles [6]. In aviation terminology, roll refers to the rotation around the X, or forward direction, and pitch to rotation around the Y, or left-right, direction as in shown in Figure 13.

Figure 13. The electronic compass uses three AMR sensors as X, Y, and Z axes to measure the Earth's magnetic field. The compass tilt
angles-roll and pitch-are referenced to a level plane surface.

Liquid-filled tilt sensors, resembling a glass thimble, incorporate electrodes that monitor the fluid movement as the sensor changes angles. Newer solid-state accelerometer-type tilt sensors measure the Earth's gravitational field by means of an electromechanical circuit [7]. These devices output an electrical signal equivalent to the angle of tilt.

To compensate a compass for tilt, knowing the roll and pitch is only half the battle. The magnetometer must now rely on all three magnetic axes so that the Earth's field can be fully rotated back to a horizontal orientation. In Figure 13, a compass is shown with roll () and pitch () tilt angles referenced to the right and forward level directions of the observer or vehicle. The X, Y, and Z magnetic readings can be transformed back to the horizontal plane (XH,YH) by applying the rotational equations:

XH = X · cos() + Y · sin() · sin() + Z · cos() · sin()               (6)

YH = Y · cos() ­ Z · sin() (7)

Once the X and Y magnetic readings are in the horizontal plane, Equations (6) and (7) can be used to determine the azimuth:

azimuth = arc-tan (YH / XH) (8)

For speed in processing the rotational operations, a sine and cosine lookup table can be stored in program memory to minimize computation time. A block diagram for a tilt-compensated compass with a serial bus interface is shown in Figure 14. After the azimuth is determined, the declination correction can be applied to find true north according to the geographic region of operation.

Overview

The magnetoresistive effect was first described by William Thompson, Lord Kelvin, in 1856. Only after the development of thin film technology-100 years later-could anisotropic magnetoresistive sensors enter the commercial arena. These devices are widely used today for position detection in high-density read heads for tape and hard-disk drives, sensing vehicular wheel speed and crankshaft position, and measuring linear and angular position and

Figure 14. To determine an accurate magnetic north heading, the five parameters-magnetic vector (X,Y,Z) and tilt (roll, pitch)-must be
measured. The signals are often converted to digital form and sent to a microprocessor for heading calculation and display.

displacement in the Earth's magnetic field. Their low cost, high sensitivity, small size, noise immunity, and reliability give them an advantage over mechanical or other electrical alternatives. Highly adaptable and easily assembled, AMR sensors are solving a variety of problems in custom applications.

Unit Conversions from SI to Gaussian

79.6 amps/meter = 1 oersted

100 microtesla = 1 gauss

1 gauss = 1 oersted (in free air)

1 gauss = 10­4 tesla = 105 gamma

1 nanotesla = 10 microgauss = 1 gamma

References

1. P. Ciureanu and S. Middelhoek. 1992. Thin Film Resistive Sensors, New York, Institute of Physics Publishing.

2. B.B. Pant. Fall 1987. "Magnetoresistive Sensors," Scientific Honeyweller, Vol. 8, No. 1:29-34.

3. J.E. Lenz et al. 1992. "A Highly Sensitive Magnetoresistive Sensor," Proc Solid State Sensors and Actuator Workshop.

4. Magnetic Sensors International Data Book. 1995. Munich, Siemens Aktiengesellschaft.

5. M.J. Caruso. Feb. 1997. "Application of Magnetoresistive Sensors in Navigation Systems," Sensors and Actuators 1997, SAE SP-1220:15-21.

6. Gregory J. Olson et al. June 1994. "Nongimbaled Solid-State Compass," Proc Solid State Sensor and Actuator Workshop.

7. M. Horton and C. Kitchin. Apr. 1996. "A Dual Axis Tilt Sensor Based on Micromachined Accelerometers," Sensors.

Michael J. Caruso is an Applications Engineer and Tamara Bratland is Magnetic Product Line Manager, Solid State Electronics Center, Honeywell Inc., MN14-3B35, 12001 State Hwy. 55, Plymouth, MN 55441-4799; 612-954-2198, fax 612-954-2257, mike.caruso@corp.honeywell.com or www.ssec.honey well.com

Dr. Carl H. Smith is Senior Physicist and Robert Schneider is Director of Marketing, Nonvolatile Electronics, Inc., 11409 Valley View Rd., Eden Prairie, MN 55344; 973-635-7576 (Smith), 800-467-7141, fax 612-996-1600 (Schneider), www.nve.com

Home | Contact Us | Advertise © 2009 Questex Media Group, Inc.. All rights reserved. Reproduction in whole or in part is prohibited. Please send any technical comments or questions to our webmaster.