Texas Instruments TIDA-060040 User manual

Design Guide: TIDA-060040
Absolute Angle Encoder Reference Design With Hall
Effect Sensors for Precise Motor Position Control
Description
This reference design demonstrates a method for
absolute angle encoding typical in precision motor
control applications such as robotic arm control.
Angle encoding can be achieved using various
magnetic sensing technologies. This is done by
detecting two magnetic flux density, B-field, vector
components which are naturally 90° out of phase. As
the magnet spins, the sensor inputs are sinusoidal
which allows for calculations using trigonometric
properties. However, due to electrical and mechanical
characteristics of the system, it is often necessary
to correct the final result to achieve the greatest
accuracy. Typically in most precision motor control
applications, the final target error is less than
0.1°. This design explores the process for selecting
a magnet, determining placement, and correcting
for system-level imperfections to achieve a highly-
accurate angle measurement.
Resources
TIDA-060040 Design Folder
TMAG5170 Product Folder
DRV5055 Product Folder
Ask our TI E2E™ support experts
Features
• Measure angle in on-axis, in-plane, and off-axis
alignments for flexible alignment and sensor
placement
• Calibration process for accuracy of 0.1°
• Selectable digital SPI communication or analog
output
• Design files with sensor placement guidance
Applications
•Mobile robot motor control
•Mobile robot sensing module
•Robotic lawn mower
•Robot servo drive
•Servo drive position sensor
•Position sensors
•Self-balancing personal transporter
Motor Supply
Voltage
C2000
LaunchPad™
Interface
Motor Driver Hall-Effect
Sensor
TIDA-060040
BLDC
Motor
3.3 V
SPI Bus
www.ti.com Description
TIDUC07 – MARCH 2022
Submit Document Feedback
Absolute Angle Encoder Reference Design With Hall-Effect Sensors for
Precise Motor Position Control
1
Copyright © 2022 Texas Instruments Incorporated

1 System Description
In any precision motor control application it becomes necessary to assess angular position of the motor shaft to
ensure overall control of the system matches expectation. Inaccurate position data may result with impaired user
safety, wider manufacturing tolerances and yield loss, navigation failures, or damaged equipment. As a result,
it becomes essential in many applications to constantly monitor and evaluate angular position. For example,
in autonomous mobile robots and robotic lawnmowers, the ability to match the angular rotation speed of each
wheel is critical for proper navigation.
Although various technologies exist for angle measurement, this design demonstrates the use of two standard
one dimensional (1D) linear Hall-effect sensors or a three dimensional (3D) linear Hall-effect sensor. Each
technique is subject to various challenges which must be addressed.
Table 1-1. Angle Measurement Methods
METHOD ADVANTAGES DISADVANTAGES
3D Hall-effect (Method
demonstrated in this design)
• Single sensor can capture entire magnetic field
• Sensor placement is flexible for compact
solutions
• Angle position is available at system power up
• Immune to dirty working conditions
• Depending on range and placement, magnetic
field may be non-ideal
• Magnet cost
1D Hall-effect (Method
demonstrated in this design)
• Inexpensive sensors with analog output
• Compact solution size
• Angle position available at system power up
• Immune to dirty working conditions
• Requires precise sensor placement for
accurate phase alignment
• Magnet cost
Hall-effect Incremental Encoding • Captures speed and direction of rotating
magnet
• Simple calculation for incremental angle
changes
• Immune to dirty working conditions
• Requires a multipole ring magnet
• Provides incremental angle position only and
position at power up is unknown
Inductive Sensed Angle Encoding • Immune to influence from nearby fixed
permanent magnets
• Immune to dirty working conditions
• Requires precise design of sense coil and
metal target
Optical Encoding • Provides highest resolution data • Solution size tends to be bulky
• Must operate in clean conditions
Stepper Motor Pulse counting • Simple implementation
• Precision control can be achieved using geared
configurations
• Step size jitter provides uncertainty in absolute
position
• Start position is unknown
Sensorless Motor Control • Does not require additional sensing
components
• Does not detect
Motor position when stopped
• Does not work well at low speeds
• Can be difficult to manage at very high speeds
• Requires complex calculations
Not all solutions are able to use optical encoding due to contaminants such as dust, dirt, and grime. Optical
solutions tend to become bulky to create sealed environments for the sensor, which do not fit well into compact
designs.
For inductive and linear Hall-effect based solutions, the premise of the angle calculation uses sinusoidal outputs
which are 90° out of phase from each other.
System Description www.ti.com
2Absolute Angle Encoder Reference Design With Hall-Effect Sensors for
Precise Motor Position Control
TIDUC07 – MARCH 2022
Submit Document Feedback
Copyright © 2022 Texas Instruments Incorporated

Angle (deg)
Magnetic Flux Density (mT)
0 45 90 135 180 225 270 315 360
-24
-20
-16
-12
-8
-4
0
4
8
12
16
20
24
Input 1
Input 2
Figure 1-1. Sine and Cosine Inputs
With outputs of this form, use Equation 1 through Equation 4 to describe the absolute angle.
Out1 = sin θ(1)
Out2 = cos θ(2)
tan θ=Out1
Out2(3)
θ= atan Out1
Out2(4)
As Equation 4 shows, determine the angle by calculating the arctangent of the ratio of the two outputs. To
simplify this calculation step in software, use the atan2() function available in many coding libraries. This function
automatically considers the sign of each input and applies adjustments to produce an output ranging from ±180°.
An additional option is to use a device with an integrated CORDIC calculator. CORDIC is an algorithm that
approximates a binary search by performing vector rotations and has been optimized for digital logic. Devices
such as TMAG5170 and TMAG5273 are capable of generating angle outputs using the device outputs with
minimal total system latency.
Linear Hall-effect solutions may be implemented in the following arrangements, which will be explored in more
detail in Sensor Location.
• 1D In-Plane
• 1D Off-Axis
• 3D In-Plane
• 3D Off-Axis
• 3D On-Axis
www.ti.com System Description
TIDUC07 – MARCH 2022
Submit Document Feedback
Absolute Angle Encoder Reference Design With Hall-Effect Sensors for
Precise Motor Position Control
3
Copyright © 2022 Texas Instruments Incorporated

2 System Overview
System Specifications
PARAMETER SPECIFICATION
Operating Voltage 3.0 V–5.5 V (DRV5055A4)
2.3 V–5.5 V (TMAG5170)
Magnetic Field Input ±85 mT (DRV5055A4)
±100 mT (TMAG5170)
Output format 0.2 V to Vcc – 0.2 V (DRV5055A4)
SPI up to 10 MHz
Best Angle Accuracy < 0.1°
2.1 Block Diagram
Motor Supply
Voltage
C2000
LaunchPad™
Interface
Motor Driver Hall-Effect
Sensor
TIDA-060040
BLDC
Motor
3.3 V
SPI Bus
Figure 2-1. TIDA-060040 Block Diagram
2.2 Highlighted Products
2.2.1 TMAG5170
The TMAG5170 is a high-precision linear 3D Hall-effect sensor designed for a wide range of industrial
applications. The high level of integration offers flexibility and accuracy in a variety of position sensing systems.
This device features 3 independent Hall sensors at X, Y, and Z axes. A precision signal-chain along with an
integrated 12- bit ADC enables high accuracy and low drift magnetic field measurements while supporting a
sampling of up to 20 kSPS. On-chip temperature sensor data is available for system-level drift compensation.
2.2.2 DRV5055A4
The DRV5055 is a linear Hall-effect sensor that responds proportionally to magnetic flux density. The device can
be used for accurate position sensing in a wide range of applications. The device operates from 3.3-V or 5-V
power supplies. When no magnetic field is present, the analog output drives half of VCC. The output changes
linearly with the applied magnetic flux density, and five sensitivity options enable maximal output voltage swing
based on the required sensing range. North and south magnetic poles produce unique voltages.
System Overview www.ti.com
4Absolute Angle Encoder Reference Design With Hall-Effect Sensors for
Precise Motor Position Control
TIDUC07 – MARCH 2022
Submit Document Feedback
Copyright © 2022 Texas Instruments Incorporated

2.3 Design Considerations
2.3.1 Magnet Selection
The magnet selected for testing was R834DIA, an N42 Grade diametric ring magnet. The magnet is specified
with an inner diameter of 4.75 mm (0.1875 in) and an outer diameter of 12.7 mm (0.5 in). The magnet thickness
is 6.35 mm (0.25 in).
Figure 2-2. Diametric Ring Magnet
2.3.2 Magnet Shape
A ring magnet format was selected for its ability for easily mount to a rotary shaft. Solid cylindrical magnets
without a bore hole in the center, must be mounted using a secondary fixture to ensure proper alignment and
centering
2.3.3 Magnet Rotation Speed
As the speed of the motorized system increases, this impacts a fixed-phase delay error from the sensor due
to the actual propagation delay in the system. The ability of the controller to manage this delay impacts device
settings and end performance.
For instance, TMAG5170 offers averaging modes which can help improve the observed input referred noise.
Minimizing this noise helps achieve the highest accuracy result, but proportionally increases the overall
integration time. This increase in integration time determines the resulting phase-related error.
AVERAGING MODE SAMPLING PATTERN TOTAL CONVERSION TIME
1 × XY 75 μs
32 × XY 1.625 ms
1 × XYZZYX 175 μs
32 × XYZZYX 4.825 ms
This integration delay can cause speed limitations for a sensor. In the case of the 32 × sampling using the
XYZZYX conversion pattern, the total integration time is 4.825 ms. At a motor speed of about 35 RPM, the
www.ti.com System Overview
TIDUC07 – MARCH 2022
Submit Document Feedback
Absolute Angle Encoder Reference Design With Hall-Effect Sensors for
Precise Motor Position Control
5
Copyright © 2022 Texas Instruments Incorporated

magnet will rotate 1° over the duration of a single conversion. By contrast, comparing the fastest integration time
of 75 μs, the magnet would have to spin 64 × faster to similarly rotate by 1° during one conversion.
For any motor, especially when loaded, there must be a deceleration period as the shaft approaches the target
position. It is necessary, therefore, to understand the phase delay to predict when to begin the deceleration. As
the shaft speed is increased or decreased, the system must account for this delay to accurately understand the
actual motor position.
While increased averaging reduces the impact of noise on the measurement, Figure 2-3 demonstrates that the
increased sample time results in a greater discrepancy between actual motor shaft angle and the collected data
from the sensor.
Speed (RPM)
Angle Phase Error (deg)
1 2 3 4 5 7 10 20 30 50 100 200 500 1000 5000
0.0002
0.0005
0.001
0.002
0.005
0.01
0.02
0.05
0.1
0.2
0.5
1
2
5
10
20
50
1x Averaging
2x Averaging
4x Averaging
8x Averaging
16x Averaging
32x Averaging
Figure 2-3. Angle Phase Error vs Motor Speed
2.3.4 Sensor Location
Simulations of the magnetic field were performed to evaluate the potential locations for on-axis, in-plane, and
off-axis alignments.
With a diametric magnet, the largest magnitude magnetic field vector at any air gap distance occurs when
evaluating the field aligned to the polarization of the magnet. In this case, the first target location to examine is
centered vertically with the magnet along a line directed radially outward from the center of the magnet.
Figure 2-4. In-Plane Sensor Alignment
At this location, it is worthwhile to examine the peak magnetic flux density as well as the quality of the magnetic
field for each component of the B-field over one full rotation. Figure 2-5 demonstrates the peak magnitude which
is observed as the horizontal distance of the sensor changes as depicted in Figure 2-4.
System Overview www.ti.com
6Absolute Angle Encoder Reference Design With Hall-Effect Sensors for
Precise Motor Position Control
TIDUC07 – MARCH 2022
Submit Document Feedback
Copyright © 2022 Texas Instruments Incorporated

Air Gap Distance (mm)
Magnetic Flux Density (mT)
0 3 6 9 12 15 18 21 24
0
60
120
180
240
300
360
420
480
540
600
660
720
780
Figure 2-5. In-Plane Magnitude vs Airgap Distance
The subsequent plots depict the magnetic field vector components at a few select locations to demonstrate the
nature of the magnetic field. In each case the sensor location is perfectly aligned with the center of the magnet,
and therefore there is no Z component in the vector.
2-mm Airgap Magnetic Field Inputs is an example of a sensor placed too close to the magnet source. The input
on the (X-axis) has a peak value of nearly 150 mT while the (Y-axis) input has a peak value over 300 mT, which
exceeds the linear input range for TMAG5170.
Angle (deg)
Magnetic Flux Density (mT)
0 45 90 135 180 225 270 315 360
-400
-350
-300
-250
-200
-150
-100
-50
0
50
100
150
200
250
300
350
400
450
Bx
By
Figure 2-6. 2-mm Airgap Magnetic Field Inputs
Additionally, there is some minor harmonic distortion to the sinusoidal inputs. Ultimately any distortion to the
magnetic field inputs results in some periodic angle error. For all magnets, the sinusoidal nature typically
improves with range.
5.9-mm Airgap Magnetic Field Inputs shows a case where the field is too far from the sensor. While the X and Y
components have an ideal sinusoidal form, the 4.5 mT peak input signal is too small to be used in high-precision
applications. With such a small input signal, the output noise could cause significant errors in angle calculations.
www.ti.com System Overview
TIDUC07 – MARCH 2022
Submit Document Feedback
Absolute Angle Encoder Reference Design With Hall-Effect Sensors for
Precise Motor Position Control
7
Copyright © 2022 Texas Instruments Incorporated

Angle (deg)
Magnetic Flux Density (mT)
0 45 90 135 180 225 270 315 360
-10
-8
-6
-4
-2
0
2
4
6
8
10
Bx
By
Figure 2-7. 20-mm Airgap Magnetic Field Inputs
The goal is to target an input near to the maximum of the linear magnetic sensing range of the target sensor.
With target devices such as TMAG5170 and DRV5055, it is necessary to allow for system tolerances while
selecting a sensor location to maximize SNR. Variations in device sensitivity, installation alignment, and magnet
quality can all have an impact on the observed field magnitude.
Sensitivity options for these two devices are shown in Table 2-1.
Table 2-1. Linear Hall-Effect Sensor Input Ranges
DEVICE LINEAR MAGNETIC SENSING RANGE (BL)
TMAG5170A1 ±25 mT , ±50 mT, ±100 mT
TMAG5170A2 ±75 mT , ±150 mT, ±300 mT
DRV5055A1 DRV5055Z1 ±22 mT
DRV5055A2, DRV5055Z2 ±44mT
DRV5055A3, DRV5055Z3 ±88 mT
DRV5055A4, DRV5055Z4 ±176 mT
To allow for either device, a horizontal sensor range of 4.65 mm (DRV5055) and 5.9 mm (TMAG5170) from the
magnet face were selected to ensure at all possible positions that the input field would be within the BL of the
TMAG5170A1 using the ±100 mT setting and DRV5055A4.
At this range, the expected input magnetic field components are shown in 5.9-mm Airgap Magnetic Field Inputs.
Angle (deg)
Magnetic Flux Density (mT)
0 45 90 135 180 225 270 315 360
-10
-8
-6
-4
-2
0
2
4
6
8
10
Bx
By
Figure 2-8. 5.9-mm Airgap Magnetic Field Inputs
System Overview www.ti.com
8Absolute Angle Encoder Reference Design With Hall-Effect Sensors for
Precise Motor Position Control
TIDUC07 – MARCH 2022
Submit Document Feedback
Copyright © 2022 Texas Instruments Incorporated

It is evident that the sensor inputs at this range are appropriate for TMAG5170, but there is significant amplitude
mismatch between the X and Y axes. If these inputs were used directly, the resulting angle calculation would
effectively map an elliptical input. The arctangent is a circular function, and so this contributes significant
measurement error if not corrected. Mapping the X and Y magnitudes into a coordinate plane reveals this
relationship.
Bx (mT)
By (mT)
-100 -80 -60 -40 -20 0 20 40 60 80 100
-100
-90
-80
-70
-60
-50
-40
-30
-20
-10
0
10
20
30
40
50
60
70
80
90
100
Natural By
Corrected By
Figure 2-9. Benefit of Amplitude Matching in Angle Calculations
The effective angle error is significant. To counter this effect, attenuate the (X-axis) results or the (Y-axis) result
requires a gain increase. Either transformation results in normalization of the inputs for the angle calculation and
resolves this error to the expected circular format.
5.9-mm Airgap Magnetic Field Inputs shows that the in-plane alignment is not ideal for DRV5055 in the SOT-23
package option due to the directionality of the magnetic field vector. When perfectly coplanar to the center of the
rotating magnet, there is no vertical component. If using the SOT-23 package option for this device, the sensor
detects the Z-component of the field vector which in this alignment is 0 mT for all angles. Instead, it is more ideal
to use the LPG through-hole package in this location. TIDA-060040 includes an optional component footprint
for this package variant adjacent to the target location for the SOT-23 package. Make sure the sensors have
the same vertical spacing from the PCB to ensure each observes an equivalent input from the rotating magnet,
and that the sensor is installed orthogonal to the surface of the PCB. If using the DRV5055 solution, remove the
unused package option to allow for a more ideal PCB layout.
www.ti.com System Overview
TIDUC07 – MARCH 2022
Submit Document Feedback
Absolute Angle Encoder Reference Design With Hall-Effect Sensors for
Precise Motor Position Control
9
Copyright © 2022 Texas Instruments Incorporated

PCBPCB
N
S
NS
Figure 2-10. DRV5055 Sensitivity
An alternative to setting tight controls for the mechanical assembly using the TO-92 package is to provide
vertical offset for the sensor location which results with a component of the magnetic field vector directed
vertically.
Figure 2-11. Off-Axis Sensor Alignment
The result is an alignment which is no longer coplanar to the magnet center. This placement is referred to as
off-axis (or out-of-plane). In effect, all placements which are not coplanar to the magnet or centered on the
axis of rotation fall into this category. The key characteristic is that the magnetic field vector has a sinusoidal
component in all three axes. As a result, there is a wide range of locations suitable for the SOT-23 package
option.
System Overview www.ti.com
10 Absolute Angle Encoder Reference Design With Hall-Effect Sensors for
Precise Motor Position Control
TIDUC07 – MARCH 2022
Submit Document Feedback
Copyright © 2022 Texas Instruments Incorporated