InvenSense MPU-6050 Instruction Manual

AN-MPU-x000A-01
InvenSense reserves the right to change the detail
specifications as may be required to permit improvements
in the design of its products.
InvenSense Inc.
1745 Technology Drive, San Jose, CA 95110 U.S.A
[phone]
www.invensense.com
Document Number: AN-MPU-x000A-01
Revision: 1.1
Date: 06/12/2015
PCB Design Guidelines for InvenSense
MotionTracking™Devices

AN-MPU-x000A-01
Document Number: AN-MPU-x000A-01 Page 2 of 14
Revision: 1.1
Date: 06/12 /2015
TABLE OF CONTENTS
1. Introduction .......................................................................................................................................................................................3
2. General Description of InvenSense MotionTracking Devices ............................................................................................................3
3. Compass Design Guidelines ............................................................................................................................................................... 4
3.1. Magnetic Distortion ...................................................................................................................................................................... 4
3.2. Calibration of Hard and Soft Iron Distortion................................................................................................................................. 5
3.3. Electric Current Noise ................................................................................................................................................................... 6
3.4. Heading Calculation for Compass Applications ............................................................................................................................ 6
3.5. Evaluation of magnetic sensor performance................................................................................................................................ 6
4. Accelerometer and Gyroscope Design Guidelines .............................................................................................................................7
4.1. Package Stress .............................................................................................................................................................................. 7
4.2. Thermal Requirements ................................................................................................................................................................. 8
4.3. Exposed Pad Requirements .......................................................................................................................................................... 8
4.4. PCB Trace Layout .......................................................................................................................................................................... 9
4.5. Noise Sources................................................................................................................................................................................ 9
5. Analyzing Sensor Data Issues Due to Sensor Placement .................................................................................................................11
5.1. Overview..................................................................................................................................................................................... 11
5.2. Analyzing Sensor Data ................................................................................................................................................................ 11
6. Quick Reference...............................................................................................................................................................................12
Revision History ....................................................................................................................................................................................... 13
TABLE OF TABLES
Figure 1. Change in Magnetic Field Caused by Hard Iron Distortion .........................................................................................................4
Figure 2. Change in Magnetic Field Caused by Soft Iron Distortion..........................................................................................................4
Figure 3. Typical Handset Components Causing Hard/Soft Iron Effects and Magnetometer Location .....................................................5
Figure 4. Example of Single Axis Helmholtz Coil ........................................................................................................................................7
Figure 5. Recommended Distance from Panelized PCB Bridges ................................................................................................................7
Figure 6. Avoid Connectors Directly Behind the Board..............................................................................................................................8
Figure 7. Exposed Pad (E-PAD) Requirements ...........................................................................................................................................8
Figure 8. Exposed Pad (EP) Dimensions and Sizing Recommended for MPU-6050 ................................................................................... 9
Figure 9. Speaker and Tactile Vibrations Can be Interpreted as Noise by the MPU .................................................................................. 9
Figure 10. Switched-Mode Power Supply Circuitry to Avoid ................................................................................................................... 10

AN-MPU-x000A-01
Document Number: AN-MPU-x000A-01 Page 3 of 14
Revision: 1.1
Date: 06/12 /2015
1. INTRODUCTION
This document provides high-level placement and layout guidelines for InvenSense MotionTracking™ devices, which may
incorporate a combination of gyroscope, accelerometer, and magnetometer sensors. Each sensor has specific requirements in order
to ensure the highest performance in a finished product. For a layout assessment of your design, including placement and estimated
magnetic disturbances, please contact InvenSense.
2. GENERAL DESCRIPTION OF INVENSENSE MOTIONTRACKING DEVICES
InvenSense MotionTracking Devices, such as the MPU-6050, MPU-6500, MPU-9150, or MPU-9250 consist of combinations of 3-axis
MEMS gyroscopes, 3-axis MEMS accelerometers, and 3-axis magnetometers as well as a Digital Motion Processor™ (DMP™)
hardware accelerator engine.

AN-MPU-x000A-01
Document Number: AN-MPU-x000A-01 Page 4 of 14
Revision: 1.1
Date: 06/12 /2015
3. COMPASS DESIGN GUIDELINES
3.1. MAGNETIC DISTORTION
Distortions in the magnetic field are often observed within electronic devices and affect the angle accuracy of magnetic data. There
are two basic types of magnetic field distortion.
Hard Iron Distortion: This is a scaling of the sensed magnetic field, caused by magnetized or rare earth magnetic materials. Since the
magnetic field is constant, it can be compensated for by subtracting the additive field, as long as the field does not cause the
compass to saturate. Common devices which cause Hard Iron distortion include speakers, motors, and auto-focus actuators. Fig. 1
shows the change in sensed magnetic field from a 30 µT field caused by Hard Iron distortion.
Figure 1. Change in Magnetic Field Caused by Hard Iron Distortion
Soft Iron Distortion: This is the result of the magnetic field bending around materials which interact with a magnetic field. When
placed near a sensor, the material may bend, or attenuate the field causing a reduction in sensitivity or angular accuracy of the
magnetometer. Common sources of soft iron distortion include iron, cobalt, nickel, and their alloys, ferritic and martensitic stainless
steels (SUS400, SUS 6000, etc.). These materials may be used in a wide variety of common components, including heat sinks,
fasteners, hinges, connectors, batteries, spring steel dome switches, chip capacitors, and some integrated circuits. Fig. 2 shows the
change in sensed magnetic field caused by Soft Iron distortion, which has the effect of transforming the typically spherical sensed
field into an ellipsoid.
Figure 2. Change in Magnetic Field Caused by Soft Iron Distortion
Ferritic, Martensitic (400 series), and work-hardened Austenitic (300 series) stainless steels are “soft iron” materials, and can affect
the sensed magnetic field. Stainless steels which may not have a significant interaction with magnetic fields in their annealed state,
may exhibit an increased interaction when cold-worked. This may occur during a stamping, forming, or bending process which
commonly occurs. Therefore, it is important to evaluate both the mechanical process, or to examine completed parts, as well as the
material to be used.
Materials which have no (or only very weak) interaction with magnetic fields include copper, aluminum, plastic, magnesium, brass
(Cu + Zn), SUS305, titanium, gold, silver, and some stainless steels.
In discussion, the term “Hard Iron” commonly refers to a rare earth magnetic material, which has a permanent magnetism. Similarly,
“Soft Iron” commonly refers to an iron (Fe) containing substance that may be magnetized. It is important to note that both Hard Iron
Distortion and Soft Iron Distortion may be exhibited by either material to varying degrees.
-40
-20
0
20
-20
-10
0
10
20
-20
-10
0
10
20
-20
0
20
-20
-10
0
10
20
-20
-10
0
10
20

AN-MPU-x000A-01
Document Number: AN-MPU-x000A-01 Page 5 of 14
Revision: 1.1
Date: 06/12 /2015
Magnetic fields decay quickly with distance. Typically, the best resolution to either hard or soft iron distortion will be achieved when
locating the compass as far away from sources of distortion as is practical, and/or substituting alternatives to “soft iron” materials in
the product. It is recommended to keep the magnetometer at least 10mm away from permanently magnetized materials in order to
reduce the likelihood of device saturation. Figure 3 provides an example of magnetometer location with respect to sources of hard
and soft iron distortion in a finished product.
Figure 3. Typical Handset Components Causing Hard/Soft Iron Effects and Magnetometer Location
Care should be taken in order to avoid introducing soft or hard iron effects to the magnetometer based on current device
configuration. Folding or rotating devices such as clamshell phones, dual-screen tablets, laptops, convertible notebooks; or critical
accessories, such as smart cases, port adapters, docking stations, should be analyzed for magnetic interference for all possible
orientations of the device. Magnetic or “soft iron" materials should not be placed near the magnetometer when the device in any
configuration as it will invalidate the current calibration and require an immediate recalibration for best performance.
3.2. CALIBRATION OF HARD AND SOFT IRON DISTORTION
Hard iron distortion can be corrected through a sensor bias estimation. The bias introduced by hard iron distortion may be
calculated by measuring the magnetometer at known orientations and estimating a fixed offset, or applying a spherical model to
magnetometer data and computing the center of magnetic sphere.
Compensation of soft iron distortion is more difficult, as the warping of the magnetic field causes a non-uniform reduction in angle
sensitivity. A distortion matrix, ellipsoid fitting, or other characterization of sensitivity vs. orientation must be employed.
Both hard and soft iron distortion may vary over time, as parts become magnetized, or introduced by the surroundings. InvenSense
utilizes “in-use” calibration algorithms to adjust for changes during run-time, providing best performance for a variety of
environments.
As of eMPL 5.1, InvenSense compass calibration algorithms include:
- Soft Iron Correction: A module which generates a mapping of spherical magnetometer data: tracking center shift, and
scaling axes according to sensed data. This correction reduces the effect of soft iron distortion introduced by device design,
and may be triggered to recalibrate based on device configuration changes.
- Hard Iron Correction: A compass vector calibration module continuously estimates the compass bias and radius during run
time.

AN-MPU-x000A-01
Document Number: AN-MPU-x000A-01 Page 6 of 14
Revision: 1.1
Date: 06/12 /2015
3.3. ELECTRIC CURRENT NOISE
Magnetic fields can be created by both AC and DC electrical current, or by changes in the electric field. Electromagnetic noise will be
picked up by the magnetic sensor as noise on the magnetic field. The magnetic sensor should be located away from the source or
traces containing current fluctuations as detailed in Table 4.1.
Current Fluctuation
[mA]
Recommended Distance* from Trace
[mm]
2
0.2
10
1
50
5
100
10
200
20
(*) distance where fluctuation of magnetic field is below +/- 2µT
Table 4.1: Recommend Distance from Fluctuating Current Traces
3.4. HEADING CALCULATION FOR COMPASS APPLICATIONS
A simple calculation of the earth’s magnetic field can be obtained by calculating the angle of the X and Y axis magnetic fields as
sensed by the magnetometer:
  󰇛 󰇜.
For applications which require an accurate heading angle, this approach is not sufficient. Using this simplified method, the heading
angle is highly dependent on the orientation of the MotionTracking Device. As the Earth’s magnetic field is sensed as a 3-
dimensional vector, the simplified calculation is only accurate when the earth field vector is aligned with the magnetometer’s XY
plane. This is not typically the case, as the earth field has a significant magnetic inclination (or magnetic dip), which is geographically
dependent. For example, within the Continental United States, this inclination angle can vary between 55 and 75 degrees. Because
of the high inclination angle, the heading output will be extremely sensitive to small changes in pitch and roll when rotating the
device on a flat plane.
A much more accurate calculation includes rotating the sensed magnetic field to a horizontal plane prior to computing the heading
angle. For static measurements (when the device is stationary), an accelerometer reading is sufficient to estimate roll and pitch,
while during dynamic movements, the orientation output of Motion Fusion is recommended.
Quaternion rotation may be applied to the magnetometer data, or calculation using Euler roll / pitch angles, using the following
example:
󰆒   󰇛󰇜   󰇛󰇜 󰇛󰇜   󰇛󰇜 󰇛󰇜
󰆒   󰇛󰇜   󰇛󰇜
  󰇛󰆒 󰇜
3.5. EVALUATION OF MAGNETIC SENSOR PERFORMANCE
When installed within a device, the magnetic field can be evaluated by analysis of sensor angular accuracy and noise.
The standard deviation of magnetometer data (noise) should be less than 1 uT for raw data, so that digitally filtered data can be
below 0.5 uT and reported at a high data rate. A noise level of 0.5 uT corresponds to a typical angle deviation or jitter of 0.5 degrees.
Angle accuracy can be measured by analyzing the heading angle of the sensor relative to its orientation. At precise rotations, the
output and device orientation can be compared. Typical angle accuracy of +/- 2 degrees is reasonable for a calibrated magnetometer
in consumer device applications.
Alternatively, a Helmholtz coil can provide a reference magnetic field signal for angle accuracy measurements, rather than rotating
the device. A Helmholtz coil passes current through two coils of wire, placed at a precise distance, equal to the radius of the two
coils, as shown in figure 4. The magnetic field generated near the center of the coil has a uniform value over a significant volumetric
space. By using a multi-axis Helmholtz coil, the generated field can be rotated around a device precisely, independent of any
environmental sources.

AN-MPU-x000A-01
Document Number: AN-MPU-x000A-01 Page 7 of 14
Revision: 1.1
Date: 06/12 /2015
Figure 4. Example of Single Axis Helmholtz Coil
4. ACCELEROMETER AND GYROSCOPE DESIGN GUIDELINES
4.1. PACKAGE STRESS
MEMS accelerometer and gyroscope motion processors are mechanical devices and are affected by package stress. Bending in the
PCB caused by mounting locations, screw holes, or misalignment will transfer board stress to the package, and can alter the output
of the MPU, or in extreme cases may even damage the MEMS structure.
The MPU should be placed in a location where there will be minimal board stress. Typically, this is away from any fixed mounting
location, screw hole, or large insertion components, such as buttons, shielding boxes, connectors, etc. During the design phase, the
estimated misalignment, mounting method, and board geometry may be used to determine the areas which have the least internal
stress, through static or finite element analysis.
Package stress can also be introduced from thermal sources during soldering or reflow processes. Uneven thermal expansion and
cooling during the assembly process introduces this stress. It is recommended not to exceed the conditions in the reflow diagram
provided within the device’s Product Specification document. This diagram represents maximum conditions required for component
reliability testing. Typical lead-free reflow solder processing is conducted between 235°C and 260°C.
It is recommended not to hand solder the MPU, as the uneven application of heat during soldering may introduce an undesired bias
offset in the part. Do not place any component pads or vias within 2mm of the package land area, in order to ensure even cooling
and minimal mechanical coupling between the MPU and adjacent devices.
Any epoxy-sealed parts on the board should be placed away from the MPU such that the epoxy resin does not come in contact with
the package. Curing shrinkage or uneven thermal expansion may introduce package stress and adversely affect the sensor output.
Keep the distance from the MPU more than 15mm away from bridges for PCB separation by router (fig. 5). Deflection from a routing
drill or saw can damage the MEMS device. Similarly dull router bits and saw blades can cause excessive mechanical vibration; which
should be avoided. Do not snap apart panelized boards, since snapping apart the PCB boards may introduce severe bending forces
and mechanical shock which may damage the MPU.
Figure 5. Recommended Distance from Panelized PCB Bridges
Do not place connectors or test points for Pogo pins on the PCB surface below the MPU location, as in figure 6. Deflection and shock
from snapping the connectors and pressure from the Pogo pin during functional test on a production line may damage the MEMS
part.

AN-MPU-x000A-01
Document Number: AN-MPU-x000A-01 Page 8 of 14
Revision: 1.1
Date: 06/12 /2015
Figure 6. Avoid Connectors Directly Behind the Board
Additional considerations to avoid MEMS damage in component placement:
Avoid any source of external point load directly below or above the MEMS device as this can cause MEMS
breakage (Note: Device is qualified to withstand up to 10Kgee shock)
Avoid shock or impact in manufacturing flow or phone assembly process
4.2. THERMAL REQUIREMENTS
The internal measurement of the MEMS sensor is dependent on temperature. For InvenSense MPU devices, software based
temperature compensation is available, however, variations in device temperature may cause changes in sensor accuracy and
should be avoided. Care should be taken for placement of the MPU relative to heat sources, which may include processors, power
management circuitry, or high current devices. The temperature gradient across the MPU should be minimized for best results.
4.3. EXPOSED PAD REQUIREMENTS
PCB land patterns are defined within the Product Specification document, and should be followed closely. The center Exposed Pad
(EP), for MPU devices is a No Connect (NC) pad. To avoid package stress, do not solder the EP to the PCB. The EP is not required for
heat sinking, and should not be soldered to the PCB. There is no electrical connection between the EP and the CMOS.
It is also strongly recommended to define a keep out layer beneath the MPU, and not place any trace, fill, or via on the top layer
under the exposed pad, described in figure 7.
Figure 7. Exposed Pad (E-PAD) Requirements
Except for the EP, a solder mask opening is required for all pin footprints. All pins should be soldered to the board to reduce uneven
assembly stress, and the solder paste stencil should not have an opening for the exposed pad to prevent stress and pitch
misalignment.

AN-MPU-x000A-01
Document Number: AN-MPU-x000A-01 Page 9 of 14
Revision: 1.1
Date: 06/12 /2015
4.4. PCB TRACE LAYOUT
Traces connected to pads should be as symmetrical as possible. Symmetry and balance for pad traces will improve component self-
alignment and lead to better control of solder paste reduction after the reflow process.
For high speed interfaces, such as I2C and SPI, all clock and data traces should be routed with the same length, and away from the
serial bus or other high speed traces. Power traces should also be routed away from high speed signals, and kept 10mil or thicker for
a 0.5oz copper PCB. Provide a solid ground return path, with traces 10mil or thicker for a 0.5oz copper PCB.
The PCB Layout Diagram and recommended pad size is provided within the MPU device Product Specification documents. Figure 8
below provides an example of a PCB Layout Diagram from PS-MPU-6000A-00 (revision 3.3). Please use the most recent revision of
the product specification for the device that you are working with.
Figure 8. Exposed Pad (EP) Dimensions and Sizing Recommended for MPU-6050
Place vias outside of the solder area and near the pad, as they can cause elevation changes. Do not place vias within the pad outline;
as vias and their related plating materials can contribute to an orientation offset and non-uniform mechanical package stress.
Eight NC (No Connect) pins can be soldered to the board for mechanical stability, but those pads on the board should not be
connected electrically.
4.5. NOISE SOURCES
Physical noise sources can cause unnecessary vibration and contaminate the desired measurement. The MPU should be mounted in
a rigid location, which will have minimal external vibration.
Moving parts which cause vibration and are not intended to be measured, such as speakers, vibration/haptic motors, buttons, etc.
(figure 9), should be mechanically isolated from the MPU.
Figure 9. Speaker and Tactile Vibrations Can be Interpreted as Noise by the MPU
Active signals may harmonically couple with the gyro MEMS devices, compromising gyro response. InvenSense MPU gyroscopic
sensors operate at drive frequencies: X = 33+/-3 kHz, Y = 30+/-3 kHz, and Z = 27+/-3 kHz. To avoid harmonic coupling don’t route
active signals directly below or near the package. For best performance, design a ground plane under the EP to reduce PCB signal
noise. If the MPU device is stacked under an adjacent PCB board, design a ground plane to shield the MPU from the adjacent PCB.
Electrical sources, such as a switched-mode power supply (SMPS) shown in figure 10, can cause high frequency vibration. SMPS with
switching noise below 150 kHz (including Harmonics) can reduce device performance.
Powered devices that may generate acceleration or vibration to the MEMS structures can cause damage to MEMS devices. Examples
of such components are inductors, capacitors, PMIC, haptic motors, speakers, etc. If there is a source of acceleration or vibration in
PCB Layout Diagram
SYMBOLS
DIMENSIONS IN MILLIMETERS
NOM
Nominal Package I/O Pad Dimensions
e
Pad Pitch
0.50
b
Pad Width
0.25
L
Pad Length
0.35
L1
Pad Length
0.40
D
Package Width
4.00
E
Package Length
4.00
D2
Exposed Pad Width
2.70
E2
Exposed Pad Length
2.60
I/O Land Design Dimensions (Guidelines )
D3
I/O Pad Extent Width
4.80
E3
I/O Pad Extent Length
4.80
c
Land Width
0.35
Tout
Outward Extension
0.40
Tin
Inward Extension
0.05
L2
Land Length
0.80
L3
Land Length
0.85
PCB Dimensions Table (for PCB Lay-out Diagram)

AN-MPU-x000A-01
Document Number: AN-MPU-x000A-01 Page 10 of 14
Revision: 1.1
Date: 06/12 /2015
the vicinity of the MEMS device, we recommend the customer to test on the PCB with a vibrometer to confirm that the MEMS
device is not being excited by any resonance frequency.
The recommendation is to place any acceleration or vibration sources as far away as possible from MEMS devices. However, in the
situation that the customer has limited board space and placement options are limited, the recommended distance is ≥5mm away
from the MEMS device. If placement is uncertain, customer should consult with local FAE to provide a more detailed analysis.
Figure 10. Switched-Mode Power Supply Circuitry to Avoid