High Accuracy Speed and Angular Position Detection by Dual Sensor

(1)

Juli 2018

High Accuracy Speed and Angular

Position Detection by Dual Sensor

(2)

Teknisk- naturvetenskaplig fakultet UTH-enheten Besöksadress: Ångströmlaboratoriet Lägerhyddsvägen 1 Hus 4, Plan 0 Postadress: Box 536 751 21 Uppsala Telefon: 018 – 471 30 03 Telefax: 018 – 471 30 00 Hemsida: http://www.teknat.uu.se/student

High Accuracy Speed and Angular Position Detection

by Dual Sensor

Johan Östling

For many decades there has been a need in many industries to measure speed and position of ferrous gears. This is commonly done by

converting passing gear teeth from trigger wheels to electrical impulses to calculate speed and angular position.

By using Hall effect sensors or Giant Magnetoresistance sensors (GMR), a zero speed detection of gear teeth is possible while at the same time be cheap to produce and durable for harsh environments.

A specially designed trigger-wheel (cogwheel created for measurements) with gear teeth in a specific pattern, exact position can be detected by using a dual sensor, even when no earlier information is available. The new design of trigger-wheel also makes this new method more accurate and universal compared to previous solutions.

This thesis demonstrates and argues for the advantages of using a dual sensor for speed and angular position detection on gear wheels. Were one sensor do quantitative measurements for pattern detection in the teeth arrangements and the other sensor do qualitative measurements for position detection.

Tryckt av: Uppsala

ISSN: 1401-5757, UPTEC F 18060 Examinator: Tomas Nyberg

Ämnesgranskare: Uwe Zimmermann

(3)

Under många årtionden har det varit ett stort behov i många industrier att mäta hur snabbt saker roterar och vilken vinkel det roterande objektet befinner sig i. Det kan i

lastbilsindustrin handla om att mäta hjulens rotationshastighet, men även många andra roterande delar på drivlinan är av intresse. Fallet som tas upp i den här avhandlingen är metoder för vinkel detektering och rotationshastighet på vissa kugghjul i växellådan.

Att använda sig utav sensorer för att mäta roterande kugghjul ger inte alltid perfekta resultat, det finns alltid avvikelser från det faktiska värdet. När teknologin i fordon blir mer allt mer avancerad ställs högre och högre krav på att ha så bra koll på systemet man bevakar som möjligt. Vilket skapar ett intresse att utforska nya mätmetoder med eventuella förbättringar.

En vanlig metod för att mäta vinkel och varvtal på ett kugghjul är att placera en sensor nära tänderna på hjulet. Sensorn skickar elektriska impulser till en dator för varje tand som passerar, signalen som skapas kan därefter bearbetas i datorn för att beräkna vinkel eller hastighet.

Det finns en rad olika teknologier som gör det möjligt att detektera kugghjulständer, vissa är mycket noggranna men också mycket dyra. Det gäller att hitta en balans mellan prestanda och pris. I den här rapporten används främst två teknologier, Giant

Magnetoresistance sensorer (GMR) och Hall effekt sensorer.

Ett simpelt kugghjul har tänderna jämnt fördelade på ett varv, det gör det inte möjligt att urskilja en tand från en annan. Det finns i det fallet ingen tydlig referens vart på hjulet man befinner sig. I det här arbetet har ett specialdesignat kugghjul framtagits för att göra det möjligt att använda sig utav mönsterigenkänning med hjälp av en dubbel sensor.

Den nya metoden med dubbel sensor är en noggrannare mätning, ger mer information om vinkel och har en mer universell lösning jämfört med dagens populäraste metoder. Denna avhandling demonstrerar och argumenterar för fördelarna med att använda en dubbel sensor för att mäta varvtal och vinkel på kugghjul.

(4)

List of Figures

1.1 Old camshaft wheel. . . 2

1.2 Illustration of dual sensor concept on new camshaft wheel . . . 3

2.1 Hall-effect basic principle . . . 8

2.2 Typical behavior of a linear hall sensor output when it becomes satu-rated . . . 9

2.3 Hall effect sensor on triggerwheel. . . 9

2.4 GMR sensor with no magnetic field applied in the axis of sensitivity. Electron spin for the two blue layers are in opposite directions. Low probability for electron tunneling. High resistance.. . . 10

2.5 GMR sensor with a very strong magnetic field applied in the axis of sensitivity. Electron spin for the two blue layers are in the same di-rection, hence increasing the probability of electron tunneling. Low resistance. . . 11

2.6 Configuration for how a dual sensor could be mounted on the fly-wheel. Both the pattern and position sensor only detects the planar component of the magnetic field. . . 11

3.1 Selected sensors that was tested for this thesis . . . 13

3.2 Selection of GMR and Hall effect sensors, see figure 3.1 . . . 14

3.3 The encoder is a 13-bit digital output sensor, mounted on the axis of rotation for the test rig. The encoder resolution is 213(bits) = 8192 points each revolution, provides an accurate reference signal for other measurements.. . . 14

3.4 Picture displaying the magnets that were tested in figure 4.1. The magnets have unknown magnetic field strengths. Magnet number 3 is analyzed in section 4.1.2 . . . 15

3.5 Model: ScopeCorder DL750 from YOKOGAWA. . . 15

3.6 Model: CPX200D Dual 180 Watt DC Power Supply from PowerFlex. . 16

3.7 Picture of the test rig, a camshaft wheel is mounted on the left side of the test rig and is in this project referred to as camshaft wheel version 2 (see figure 3.10) . . . 17

3.8 The sensor is soldered to a PCB, that it the glued on the white hard plastic. Behind the sensor is a hole that the magnet fits through, that way the magnet can be placed close to the sensor. . . 18

3.9 A modified camshaft wheel with 2 pair of teeth cut in half, with 1 or 2 teeth in between the pair creating 2 unique reference points. This wheel can be used for measurements until the new camshaft wheel version 2 design arrives, see figure 3.10 . . . 19

3.10 Camshaft wheel version 2, the new design of the camshaft wheel that was specially made for this project. . . 20

(7)

3.11 The flywheel has not been redesigned but modified. Two pairs of teeth has been sawed in half, making it a total of 4 missing teeth. The amount of teeth is 114 (counting the wide teeth as one). One modified pair has has a single tooth between and the other modified pair has two teeth in between. . . 21

3.12 Trigger wheel has 60 teeth if the missing ones are ignored, resulting in a 6◦tooth period on average. . . 22

3.13 Sensor 11 . . . 22

3.14 Dual sensor designed for camshaft wheel version 2, uses two entities of linear hall effect sensors (Sensor 3). . . 23

3.15 Dual sensor designed for camshaft wheel version 2, uses two differ-ential hall sensors (Sensor 8) . . . 23

3.16 Dual sensor designed for flywheel, uses two GMR sensors (Sensor 11) 24

3.17 . . . 24

3.18 A 12-bit gray code for encoder to illustrate the least significant bit (far right). The digital signal is read by looking at what color the red dots are above during rotation. Black is 1 and white is 0, the pattern is design so that only one bit is changing for each step (gray code), making it easy to remove errors when disturbances are merged with the signal. This is not the gray code for the 13-bit encoder used in section 3.2.3, its just to clarify how only one of the encoder outputs could be used as reference signal. . . 25

3.19 Illustration of teeth profile for camshaft wheel version 1. . . 26

3.20 A common version of a camshaft wheel with missing teeth.. . . 27

3.21 Section of the flywheel as seen from the side/radial direction, the left column of teeth is the part that is being measured in this project. The right column with slightly darker teeth is referred to as the "back wheel" 28

3.22 Illustration of the flywheel teeth and back wheel teeth as seen from the radial direction of the wheel (see figure 3.21 for real picture). The sensor placement is defined for axial position 0mm and 8mm. . . 28

3.23 Illustration shows three different scenarios that the Hall sensor can experience when the flywheel is spinning. Position 1 should give the same output as position 3. But since it’s almost impossible to place or retain perfect alignment, one of the hall elements inside will always be slightly before the other one, causing a small disturbance. Position 2 should give a large output that can be easy to detect . . . 29

3.24 Illustration of teeth profile for camshaft wheel version 2 . . . 30

3.25 Illustration of flywheel teeth used for defining axial position for pat-tern & position sensor. . . 31

4.1 Table that displays the results when magnets was tested on different sensors on camshaft wheel version 1. . . 34

4.2 Magnetic field strength as seen from the front and right side of magnet 3. It is clear that the magnet has been demagnetized . . . 35

4.3 Magnet 3 as seen from the top and bottom. Diameter of circles are proportional to magnetic field strength in orthogonal direction. The different sizes indicate that the magnet has in fact been demagnetized 35

4.4 Sensor 3 (linear Hall sensor) at 1mm and 2mm air gap at +2mm radial distance. . . 37

4.5 Sensor 8 (differential Hall sensor) at 1mm and 2mm air gap at +2mm radial distance . . . 37

(8)

4.6 Sensor 4 (digital Hall sensor) at 1mm and 2mm air gap at +2mm radial distance. . . 38

4.7 Sensor 11 (GMR sensor) at 1mm and 2mm air gap distance at +2mm radial distance . . . 38

4.8 Differential Hall sensor (sensor 8) at different radial positions at 1mm airgap . . . 39

4.9 Differential Hall sensor (sensor 8) at different radial positions at 2mm airgap . . . 40

4.10 Airgap set to 1mm. GMR sensor (sensor 11) measurements between the flywheel teeth and the backwheel for different axial positions . . . 41

4.11 Airgap set to 2mm. GMR sensor (sensor 11) measurements between the flywheel teeth and the backwheel for different axial positions . . . 42

4.12 Differential sensor (sensor 8) detecting a missing tooth for different airgaps and axial positions. . . 43

4.13 Sensor 8 used as pattern sensor on flywheel at different pitch angles. The signals inside the red rectangle are plotted in figure 4.14. . . 44

4.14 Figure 4.13 zoomed in to get a close look at how the disturbance in-creases for each degree of rotation in the pitch direction . . . 45

4.15 Sensor 8 used as pattern sensor on flywheel at different yaw angles. The signals inside the red rectangle are plotted in figure 4.16 . . . 46

4.16 Figure 4.15 zoomed in to get a close look at how the disturbance in-creases for each degree of rotation in the yaw direction . . . 47

4.17 Simultaneous measurement of a dual GMR sensor on flywheel at 1mm airgap. . . 48

4.20 Simultaneous measurement of a differential Hall sensor and a GMR sensor on the flywheel at 1mm airgap. . . 49

4.21 Simultaneous measurement of a differential Hall sensor and a GMR sensor on the flywheel at 2mm airgap. . . 50

4.22 Dual sensor measurement of camshaft wheel version 2 at 1mm airgap distance using sensor 3 . . . 51

4.23 Dual sensor measurement of camshaft wheel version 2 at 2mm airgap distance using sensor 3 . . . 51

4.24 Dual sensor with simultaneous measurements of the pattern and po-sition sensor at 1mm airgap, both being of the differential Hall sensor type (sensor 8). . . 52

4.25 Dual sensor using with simultaneous measurements of the pattern and position sensor at 2mm airgap, both being of the differential Hall sensor type (sensor 8). . . 52

4.26 Top graph is the positions sensor from figure 4.24 with added trigger points. Middle graph is the pattern sensor from the same figure and with triggers for every deep gap. The black dc signal in the middle graph is the sector indicator signal. The bottom graph is a teeth index tracking, useful for extracting data from individual teeth. The red square is stretched out in figure 4.27 . . . 54

4.27 Close look at when a new sector is detected and when the sector de-tection signal is set to a new value. . . 55

(9)

4.28 The use of a accurate reference signal removes the system error given by the test rig. . . 56

4.29 . . . 56

4.30 Distribution of all teeth periods obtained during 83 revolutions of camshaft wheel version 2 are shown in the blue histogram. Tooth in-dex number 15 are displayed in the yellow histogram with a slightly smaller average teeth width (0.04oless than the rest) . . . 57

4.31 Mean tooth width in degrees for teeth index 1:90 . . . 58

4.32 Mean teeth width for the old camshaft wheel using using a differential Hall sensor (sensor 8). The missing teeth are interpolated using linear interpolation. . . 58

4.33 Teeth periods for old camshaft wheel during 80 revolutions. The red samples are the missing teeth, yellow the teeth before and after the missing teeth. The blue samples are all evenly distributed teeth . . . . 59

(10)

List of Abbreviations and

Definitions

SNR Signal (to) Noise Ratio

RPM Rounds Per Minute

GMR Giant Magneto Resistance

DC Direct Current

BLDC BrushLess Direct Current

DAQ Data AcQuisition

PCB Printed Circuit Board

PWM Pulse Width Modulation

BLDC BrushLess Direct Current

Trigger Wheel Referred to as either a camshaft- or flywheel. Wheel designed to only be used for

measurements. Looks like a cogwheel.

Position Sensor Sensor placed in the region of evenly distributed teeth.

Pattern Sensor Sensor placed in the region of missing teeth or deeper cuts.

Trigger-Wheel Region Referring to either position region or pattern region.

Positive Measurement Direction Always left to right oriented when standing on front of test rig.

Sector Divides the trigger-wheel into a set number of sectors (compare to pie charts).

Backwheel The flywheel used in this project consists of two merged wheels, this is the wheel at the back of the flywheel and is not used for measurements.

Trigger Level Refers to the voltage level from where data is collected and stored as information. Trigger level at 0V indicates that whenever the signal passes 0V, the current time is saved.

(11)

(12)

Chapter 1 Introduction

1.1 Background

The demands placed on modern engines to provide maximum performance and fuel economy coupled with the drive to reduce emissions, require an increasing accuracy and specific functionality of sensors to monitor and help manage engine operation.

Despite a large variety of proposed sensors, it is not always possible to fulfill all requirements to very specific sensor functionality suited to new engine design. Some automotive manufacturers will insist on their own more general solutions that cover wider market.

Thus in some areas it is necessary to build a prototype or setup experiment se-ries to investigate in detail which sensor on the market is closer to fulfill the require-ments. Afterwards sensors can be designed and manufactured to exacting specifica-tions on a custom basis1.

1.2 Position detection on ferrous gears

1.2.1 Current measurement technique

When assembling hundreds of parts into a larger structure, space is always a con-cern. While there are existing sensors available today that can sense position and velocity very accurately, they are hard to install inside a gearbox and can create more problems than they solve. Therefore its common to use a sensor that is placed at the teeth of a cogwheel to detect when a tooth is passing by, then converting the detected magnetic field to electrical impulses that can be used to determine position. Due to he harsh environment inside gearboxes, the most common sensor choices are inductive and Hall sensors. GMR sensors (giant magnetoresistance) are growing in popularity but is a newer technology compared to Hall sensors. It has been shown that sensors that measures magnetic field have a high performance in harsh environ-ments. Because of the no contact functionality of magnetic field sensors they don’t get affected by dirt like a optical sensor would.

A common way for measuring position and angular velocity for a rotating sys-tem is to use a wheel similar to a cogwheel, referred to in this project as a trigger-wheel (see figure1.1). This type of wheel is common since the larger teeth work as reference points, which usually relates to when fuel injection of the motor cylinders are needed.

A sensor is placed close to the teeth with a back biasing magnet behind it, this will magnetize the teeth close to the sensor, hence making it possible for the mag-netic field sensor to detect the change magmag-netic field as the teeth from the wheel passes the sensor.

(13)

FIGURE1.1: Illustration of what typical sensor outputs (differential and direct field) might look like for trigger wheels that has larger

teeth.

The main reason for this project is to improve on the current measurement method. Identified measurement problems:

Uneven teeth width When a trigger wheel has different teeth width, the trigger points for when the sensor crosses a specific threshold voltage is not evenly distributed. That causes a need for interpolation during time frames when a "bigger" teeth is in front of the sensor. This gives a small deviation from the current position since larger teeth not only gives less trigger points each revo-lution but also gives less accurate triggers since the bigger teeth contributes to a larger magnetic field that shifts the zero crossing slightly to the side when big teeth are measured. This is true for all type of magnetic sensors and requires additional predictive filtering to correct for the shift in zero crossing.

Sector detection is not available Using the wheel design in figure 1.1, the wheel has four big teeth, dividing the wheel into four parts called sectors. There is no way of knowing what sector the current wheel position corresponds to since the big teeth are identical and cannot be distinguished. It is possible to know how many teeth have passed during a certain amount of time but it is not possible to know which tooth has passed.

No universality Motor cylinders need fuel injections when the trigger wheels are at specific angles/positions. Since the sensor is less accurate around larger teeth, timings for fuel injections can not be made around that area due to the lower accuracy. During the construction of trigger-wheels, the placement of big teeth together with the amount of motor cylinders needs to be taken into account so that injections can be made when the sensor is most accurate. This makes it common to do a redesign of the trigger wheel when using a different amount of motor cylinders.

(14)

1.3 Dual sensor concept

The idea of using two sensors is to make one sensor do quantitative measurements to determine position i.e. measuring evenly distributed teeth with high accuracy. The other sensor should do the qualitative measurements by measuring the pattern of deep cuts i.e. the timing differences between the deep cuts. In figure1.2the sensor is measuring the deeper cuts with different amounts of teeth in between each deep cut. By combining the "position sensor" and "pattern sensor", the amounts of teeth in between each deeper cut can be translated to the actual sector. The illustration below is just a basic concept and not the actual wheel used in this project. In this example, sector 1 is determined by having 2 teeth in between two trigger points from the pattern sensor during clockwise rotation of wheel. Sector 2 has 3 teeth, sector 3 has 4 teeth and sector 4 has 5 teeth and so on. Relation between isolated teeth and sector number changes when the wheel direction switches.

FIGURE1.2: Illustration showing basic concept of how a dual sensor

(the two black rectangles) can in theory determine the current sector and give higher accuracy due to the removal of the wider teeth that

the old camshaft wheel has (see figure1.1)

One of the things that will be looked into during this project is how close the two sensors be mounted in relation to one another. The goal is to make it possible to separate the pattern output from the position output without crosstalk.

The position sensor and the pattern sensor have a potential problem that needs to be investigated. How the created magnetic field around the deeper cuts will affect the signal on the position sensor assigned for evenly distributed teeth. This problem is also valid for the reversed scenario when unwanted magnetic field from the posi-tion region reaches the pattern sensor. This result will give a hint on how close the two sensors can be mounted in relation to each other.

(15)

1.4 Project description

The main goal is to build sensor prototypes to demonstrate the effect of using a dual sensor according to the principle laid out in section1.3. A literature study is done to learn how present solutions for ferrous gear measurements are made, us-ing that knowledge, a market analysis is made to find newly developed sensors that might have a better performance than current solutions. Combining new sen-sor with sensen-sors that have already been proven to work, gives a small collection of around 10 different sensors with varying sensitivities, measurements techniques (linear/differential) and technologies (GMR/Hall). The dual sensor cannot be used on current gears without modifications, therefore a new trigger-wheel design needs to be produced. The idea is to find how the current design can be modified or com-pletely changed to be compatible with the dual sensor concept. Designing a new trigger-wheel will be of high priority during the project’s first phase,. A modifica-tion is made on two already existing trigger wheels (camshaft and flywheel). This makes it possible to test the chosen sensors and experiment on geometry optimiza-tion and using different back bias magnets. This will eliminate sensors not suitable for this project and will limit the amount of measurements. The best suitable sen-sors are used to measure the flywheel and the new camshaft wheel to find optimal positions for the pattern sensor and position sensor (defined in figure1.2). After all single sensor measurements are done, different combinations can be made to design 2 or 3 different dual sensor prototypes. Software to analyze and demonstrate how the sector is detected is then written. The dual sensor prototypes are then compared to the old measurement technique to demonstrate the improvements.

1.5 Limitations

To avoid the project of becoming too large in comparison to the 20 weeks’ time frame for this project. Some scope limitations were set.

• No optimization using every possible solution/setup. If a good result is found, its moved to the pile of good results and no extensive optimization is made. • All testing is done on the rig assembly, no testing on real trucks. Environmental

effects will therefore not be investigated.

• Rotational speed of test rig is set to 50RPM for all measurements, sensor selec-tion is not made in regard to accuracy at different velocities.

• The dual sensor prototypes are not finished products, its simply two sensors placed at a calculated distance from each other at a specific position on the trigger-wheel. Both sensors are measured simultaneously and result is ana-lyzed and discussed.

• Software is limited to display principle of sector detection

• Advanced signal processing commonly used in automotive gear detection such as adaptive trigger levels, self-calibration such as automatic gain control (AGC), packaging options to move back bias magnet closer to sensor elements, self-calibration, predictive filtering and reconstruction of signal to a readable digi-tal signal for the control unit is merely touched upon. When demonstrating the principle of sector detection, final results are analysed and treated in such way

(16)

that demonstrates how the control unit could use the dual sensor information in a real time.

• No real time implementation is done on the software part, although the same techniques are used to make it easy to transform the code to a real time work-ing algorithm.

• Experimenting with different rotational speed of the test rig, no optimization is needed for this project. The rotational speed for all measurement are made at 50 RPM

• This project does not aim to build a complete product since the primary aim is to demonstrate the effect of using two sensors and to show that it is indeed possible.

• Direction detection is necessary since the relationship between isolated teeth and sector changes if the direction changes. This was not touched upon in this project but can easily be implemented by using sensors with a direction detection feature.

(17)

(18)

Chapter 2 Hardware Fundamentals

2.1 Gearbox

Transmission provides a controlled application of the power output given by the engine. It delivers power to the wheels by performing torque and speed conversions using gears and gear trains from the rotating power source given by the engine,

The camshaft is used for operating poppet valves in reciprocating engines. The shaft is placed underneath the cylindrical bank and have protruding oblong lobes (called cam), one for each valve. A camshaft wheel is simply a wheel connected to that rod. Measurements on the camshaft wheel gives information of the camshafts current state and is important information since it gives the fuel injection timings for each cylinder.

The flywheels purpose is to store rotational energy, made possible by having a large moment of inertia that can resist changes in rotational speed. Commonly used in reciprocating engines to smoothening the power output, given by the intermittent active torque from the individual pistons. Since the flywheel is rotating proportional to the camshaft it can therefore be used to measure the fuel injection timings. Fly-wheel turns two times for every camshaft Fly-wheel turn.

2.2 Measurements techniques for trigger wheels

2.2.1 Inductive sensor

A commonly used sensor for position detection on gears is the inductive sensor, based on Faraday’s law of induction (eq. 2.1). Its main advantage is that it has a high accuracy at higher rpm, requires no supply voltage and is cheap and easy to build. It’s drawback is not being able to handle detection of zero speed. Reason being that the generated current from the coil inside the sensor is proportional to the rate of change in the magnetic field i.e. the rotational velocity.

For a spool of wire with N turns, the induced voltage V between the two ends of the circuit is directly proportional to the temporal variation of magnetic fluxΦ that the spool is exposed to. The following equation describes the basic workings of the inductive sensor.

V= NdΦ

dt (2.1)

It then follows why the zero or close to zero speed detection is not possible, if the magnetic flux is zero, the induced voltage e is also zero. It is therefore difficult to know if the object that is being measured is standing still or moving slowly, since the sensor gives close to the same result for both scenarios.

(19)

2.2.2 Hall Effect Sensor

A Hall effect sensor2 is a solid state device and is good at sensing position, velocity or directional movement of a material that is magnetized. They are cheap to make, no contact and do not wear out, low maintenance and can handle vibrations and are very robust. Their basic function is that they deliver a voltage signal directly proportional to the magnetic field that it measures.

The hall element is built by doping a thin layer of semiconductor material with impurities, usually p-type. A directional current with constant value is flowing through the material. When an external magnetic field is present, with field lines perpendicular to the surface, the electrons and holes in the material will be pushed in opposite directions producing a measurable electric field (see figure2.1).

FIGURE2.1: Main principle of the hall effect sensor, a magnetic field

creating a electric field3

Hall effect is defined as the generation of a measurable voltage by using a mag-netic field. How magmag-netic fields interact with moving charged particles is described by the Lorentz force:

F=qE+qv×B (2.2)

where F is the force the particles experience, q is the charge of the particle, E is the electric field the particle is traveling through, v is the speed of the particle and B is the magnetic field the charge is traveling through. The buildup of electrons and holes at the sides of the hall element is the measurable voltage Vhin figure2.14.

VH = RH

I t ·B

(2.3) The hall element is said to be saturated when all the electron and holes have been pushed to opposite direction and the electric field has reached its maximum voltage.

(20)

The relationship between the magnetic flux density and the created voltage from the hall effect is a linear relationship if the sensor is said to be proportional. That linear relation ends when the hall element becomes saturated (see figure2.2), increasing the magnetic flux density no longer increases the voltage.

FIGURE2.2: Typical behavior of a linear hall sensor output when it

becomes saturated

Gearteeth sensing with Hall effect sensor

If a hall-effect sensor is to be used to detect a ferromagnetic5cogwheel/trigger-wheel its teeth have to be magnetized first. Figure2.3 is illustrating then main principle. When a permanent magnet of sufficient strength is placed near the teeth, with its axis of magnetization in the radial direction. The magnet gets attracted to the tooth due to the lower reluctance path in the ferromagnetic material. The teeth become magne-tized and the total magnetic flux density between the magnet and the teeth increases. A hall element placed in-between these two parts can detect the increase/decrease of magnetic flux density as teeth pass by. This is why zero-speed detection can be achieved since it only measures the amplitude of the field, not the rate of change as the common inductive sensor does.

FIGURE2.3: Basic concept of how a hall effect sensor can detect

(21)

2.2.3 Giant Magnetoresistance Sensor

The GMR sensor can get very large differences in electrical resistance for very small changes in magnetic field strength. A popular use for the GMR technology is to give sensitive read-out head for compact hard drives, and was in fact awarded the Nobel prize in physics 20076. In recent years it has become more common for the automotive industry to make use of the GMR sensor for various applications such as speed and angular position detection for gearwheels7. GMR was considered one of the first real applications for the field in nanotechnology when it was discovered. Figure2.4shows the basic structure of a GMR sensor when no magnetic field is applied. The two blue layers are ferromagnetic alloys with a very thin conducting and nonmagnetic layer (red layer). One of the blue layers are pinned i.e has a fixed magnetization, the other layer is free to move due to its anti-ferromagnetic coupling8

and will in its natural state be in opposite direction as the other blue layer. Since the electron spin will line up with the magnetic field from the pinned layer. The resistance between the blue layers is large when no magnetic field is present.

If a magnetic field is applied in the axis of sensitivity, the green arrow in figure

2.5. The electron spins in the blue layers will start to line up with green arrow pro-portional to the magnetic field strength. Since the red layer is so thin, the effect of quantum tunneling is present. The odds of electron tunneling between the blue lay-ers increases linearly with the applied magnetic field strength. The flow of electrons between the blue layers can be understood analogically be interpreting it as current flow. If the current flow increases with a constant potential, the resistance is chang-ing. This is of course a simplified explanation for the workings of a GMR sensor but is sufficient for performing this project.9

10

FIGURE2.4: GMR sensor with no magnetic field applied in the axis

of sensitivity. Electron spin for the two blue layers are in opposite directions. Low probability for electron tunneling. High resistance.

(22)

FIGURE2.5: GMR sensor with a very strong magnetic field applied in the axis of sensitivity. Electron spin for the two blue layers are in the same direction, hence increasing the probability of electron tunneling.

Low resistance.

A example for how GMR sensors can be used in the dual sensor design is dis-played in figure2.6. Since the GMR sensor only measures the magnetic field in the axis of sensitivity, it can be oriented to only measure planar components from the gear teeth. Planar components illustrated with red arrows. This makes it possible for the pattern sensor to ignore all teeth except for the missing teeth.

FIGURE2.6: Configuration for how a dual sensor could be mounted

on the flywheel. Both the pattern and position sensor only detects the planar component of the magnetic field.

(23)

2.2.4 Differences between GMR and Hall-sensors

Sensitivity GMR sensors have 50-100 times larger raw output signal compared to silicon Hall effect sensors. The reason they appear to be similar in sensitivity is be-cause of the internal electronics of the hall sensor has that amplifies the signal. GMR sensors can handle larger airgaps in gear-tooth applications.

Precision Because of the larger raw output signal from the GMR sensor, the precision is several times larger over a wide range of supply voltages and temperatures.

Power The high impedance of GMR sensors allows for a very low power consumption compared to the Hall effect that is never switched off.

Range of use Because of the GMR’s extra feature of only measuring the planar component of a magnetic fields in a single direction, called the axis of sensitivity, it can do measurements that is either harder or impossible for the Hall sensor. 11

Cost& availability GMR sensors are a newer technology compared to Hall effect sensors and is therefore not as present in the current market, they have many advantages of Hall sensors and are expected to become more common in the years to come. Today the price for GMR sensors is larger than Hall effect sensors, this differ-ence will most likely decrease when GMR sensors become more common.

2.3 Sensor types

Bipolar sensor A latching sensor that changes state when the polarity is reversed, always a digital sensor.

Unipolar sensor A unipolar sensor can only detect one polarity of a magnetic field.

Omnipolar sensor Does not care about polarity of magnetic field, it only responds to the absolute value of the magnetic field. It will detect increase and decrease of absolute strength but it can not determine polarity

Differential sensor This sensor is built by combining 2 or more sensor elements (no specific po-larity) into a single sensor, separated by a certain distance. The output is the difference between the elements inside the differential sensor

(24)

Chapter 3 Implementation

3.1 Components

3.1.1 Sensors

Hall-effect and GMR sensors are targeted when picking sensors for this project. A previous master thesis12at Scania used these types of sensors to measure flywheels. Therefore it was a good time saver to analyse the results from that project and pick sensors that gave the best results. Five sensors were selected and another six sensors were ordered. All GMR sensors that were picked had similar saturation so that the same magnet could be used in the experiment. The sensors had different character-istics and were built for different purposes. Some of the differences that the sensors had is related to them being omnipolar or bipolar, linear vs non-linear output, built in filters, saturation levels, differential measurement and digital vs analogue out-puts. Here is the complete list of chosen sensors, although some turned out to be unfit for this project, see figure3.1.

(25)

FIGURE3.2: Selection of GMR and Hall effect sensors, see figure3.1

FIGURE3.3: The encoder is a 13-bit digital output sensor, mounted on

the axis of rotation for the test rig. The encoder resolution is 213(bits)₌

8192 points each revolution, provides an accurate reference signal for other measurements.

3.1.2 Magnets

To minimize the amount of measurements, and since most of the sensors had around the same sensitivity and saturation, it was decided to use only one magnet for all measurements. To figure out what magnet to use without wasting too much time, the sensors were powered on and connected to the oscilloscope. The sensors were

(26)

then dragged across the teeth of the trigger wheel by hand in different types of com-binations of magnets (see figure4.1) and distances. The oscilloscope was only used to see if the sensor could pick up the difference in magnetic field for the teeth and the gaps in the camshaft trigger wheel. colors were used to describe what type of result a given distance for a specific magnet was giving. It could either get saturated, be too big to detect the small gear teeth, having too much noise, or sometimes miss detection of a tooth, see section4.1.1.

FIGURE3.4: Picture displaying the magnets that were tested in figure

4.1. The magnets have unknown magnetic field strengths. Magnet number 3 is analyzed in section4.1.2

3.1.3 Oscilloscope

The only point worth noting about the used oscilloscope is that a low pass filter was used for most measurements, with a cut off frequency set to 40MHz. The sampling frequency for most measurements was between 20kHz and 200kHz depending on how much time the measurement took. Manual for the oscilloscope is found here.13

(27)

3.1.4 Generator

A generator (see figure3.6) was used for supplying voltage to all sensors that were tested, including a encoder that was partially used as a reference signal. The supply voltage is always 5 V unless otherwise is noted.

FIGURE3.6: Model: CPX200D Dual 180 Watt DC Power Supply from

PowerFlex.

3.1.5 Test rig

The speed of a BLDC motor is controlled from a micro controller using PWM signals. Engine or gearbox parts that needs to be investigated during rotation in a controlled environment, can be mounted on the test rig (see figure3.7). That way the efficiency of measurements becomes higher, compared to testing it on real trucks where the setup is more problematic.

A micro controller (Arduino) is sending PWM signals to a BLDC motor that drives a shaft. Parts from a engine or gearbox that rotates, like camshaft wheels or flywheels, can be mounted on the shaft. The main use of the test rig is to measure the rotating parts using sensors in a controlled environmen The steps for controlling the test rig

1. The setup for test rig are defined in the LabVIEW interface, such as rotation speed, direction and acceleration

2. The commands from LabVIEW are parsed by the Arduino, converting to PWM signals structured in such a way that the motor servo drive is compatible 3. The output from the Arduino gets sent to the motor servo drive which powers

the BLDC motor

4. The motor servo drive controls the speed of the BLDC motor using the PWM signals and feedback from a encoder

(28)

FIGURE3.7: Picture of the test rig, a camshaft wheel is mounted on the left side of the test rig and is in this project referred to as camshaft

wheel version 2 (see figure3.10)

LabVIEW

LabVIEW stands for "Laboratory Virtual Instrument Engineering Workbench" uses a visual programming language designed in a development environment, mainly used for system-design. The micro controller that controls the test rig is programmed using LabVIEW.

Placement station

To be able to measure the trigger-wheels (section3.1.6) on the test rig using the sen-sors in section3.1.1, the sensors are mounted onto something referred to as a place-ment station, figure3.8shows the placement station. The placement station is fixed to the table by having magnets that can be activated, by lowering the magnets onto

(29)

the ferromagnetic surface on the table the station is stuck to the table. The planar position can be oriented with high precision using the wheels on the side.

FIGURE3.8: The sensor is soldered to a PCB, that it the glued on the

white hard plastic. Behind the sensor is a hole that the magnet fits through, that way the magnet can be placed close to the sensor.

3.1.6 Trigger wheels

Camshaft wheel version 1

Since the dual sensor required a new design of the camshaft wheel, that version had to be constructed at the workshop. To make use of the time before the camshaft wheel version 2 arrived (see figure 3.10), a simplified version was constructed by just cutting off teeth on an existing camshaft wheel that would have the same teeth width as version 2 would have. It should be noted that the wider teeth (can be seen in figure3.9) show up in all measurements and should be ignored. All measurements of version 1 will stop when version 2 is constructed and ready for use.

(30)

FIGURE3.9: A modified camshaft wheel with 2 pair of teeth cut in half, with 1 or 2 teeth in between the pair creating 2 unique refer-ence points. This wheel can be used for measurements until the new

camshaft wheel version 2 design arrives, see figure3.10 Camshaft wheel version 2

Figure3.10shows the new design of camshaft wheel, made specially for the dual sensor. The new design is best suited for axial measurements when using a dual sensor. This wheel has 6 pairs of gaps making it possible to divide the wheel in up to 6 sectors.The camshaft wheel has 90 teeth, making each tooth period 4◦ on average. The term sector is illustrated in figure1.2.

(31)

FIGURE 3.10: Camshaft wheel version 2, the new design of the camshaft wheel that was specially made for this project.

Flywheel

No new design of a flywheel was tested in this project. The flywheel is larger and will always give more accurate measurements than the camshaft wheel. The larger amount of teeth on the flywheel provides more trigger points per wheel revolution, that fact together with the larger diameter results in better position accuracy. If the principle works on the camshaft wheel it will work on the flywheel. Although a similar modification as camshaft wheel version 1 (figure3.9) was made by cutting of teeth on the flywheel with different amount of teeth in between. This was only done to 4 teeth so only 2 sectors are possible to detect.

(32)

FIGURE 3.11: The flywheel has not been redesigned but modified. Two pairs of teeth has been sawed in half, making it a total of 4 miss-ing teeth. The amount of teeth is 114 (countmiss-ing the wide teeth as one). One modified pair has has a single tooth between and the other

mod-ified pair has two teeth in between.

Common camshaft wheel

Measurements on the old camshaft wheel are necessary to compare the potential improvements that the dual sensor provides. Using larger teeth is decreasing the ac-curacy of the sensor output, the reason being the increased density of ferromagnetic material in the large teeth. A uneven distribution of magnetized material that the sensor detects during a revolution, causes a varying amplitude in the output signal, making it harder to do accurate zero crossings. A large teeth will appear to be wider than it actually is because of the increase in sensor output amplitude. The illus-tration in figure1.1shows how the voltage output from the sensor increases when crossing over to large teeth and also demonstrates how the version with large teeth could look like.

This camshaft wheel, shown in figure3.12, does not have big teeth, but it does have a lower density of ferromagnetic material at the parts where teeth are cut off. This will in effect cause the same problems in zero crossing detection as the large teeth version. The gaps will appear smaller than what they actually are, but the decrease in accuracy of sensor output will be the same as the large teeth version.

Ideally a trigger wheel with wider teeth should have been measured instead of the missing teeth version since the former is more common. But this idea was im-plemented at the end of the project and there was no available version of wide teeth trigger wheels that could be mounted on the test rig.

(33)

FIGURE3.12: Trigger wheel has 60 teeth if the missing ones are

ig-nored, resulting in a 6◦tooth period on average.

3.2 Measurement setup

3.2.1 Single sensor soldering

All sensors where tested individually and were soldered onto separate PCB’s. The selected sensors had different types of packaging, each package required a specific type of PCB. The figure below shows how sensor 11 looked like when soldered onto a PCB that fits a SOIC8 package.

FIGURE3.13: Sensor 11

3.2.2 Dual sensor combinations

Some examples of how the dual sensor prototypes looked like after the sensor type and orientation had been decided, some sensor packages where easy to put on the

(34)

same PCB’s, others had very tiny pins and were hard to solder by hand. The specific PCB design required for the dual sensor was not possible to build in time frame for this project and a more makeshift solution was presented. Three different types of dual sensor configurations are presented below.

• Figure3.14shows two sensors that are soldered to the same PCB. Both sensors had simple packaging

• Figure 3.15 shows the solution for when the sensor packages required ma-chine made PCB’s, there was no need to order specific PCB designs since the makeshift version is sufficient enough for measurements. This way also makes it easier to adjust the distance between the sensors between measurements. • Figure3.16shows a design when a large distance between sensors are required,

this design made it very easy to adjust the distance between the sensors.

FIGURE 3.14: Dual sensor designed for camshaft wheel version 2,

uses two entities of linear hall effect sensors (Sensor 3)

FIGURE 3.15: Dual sensor designed for camshaft wheel version 2,

(35)

FIGURE3.16: Dual sensor designed for flywheel, uses two GMR sen-sors (Sensor 11)

3.2.3 Reference signal

FIGURE3.17

The 13-bit encoder required an DAC converter to be used with a oscilloscope, the analogue DAC converter did not work as intended. So the next idea was to con-nect the encoder to LabVIEW using a DAQ with digital inputs, then writing a gray code converter and do a digital to analog conversion in the computer. When this finally started to work, the DAQ used to import the encoder signal did not have suf-ficient performance to get the high sampling frequency that was necessary to meet the Nyquist criteria.14

The solution that ended up working was to use only 1-bit from the encoders gray code, see figure3.18. This bit could be sent to a oscilloscope to measure a square wave with 1024 periods per revolution instead of the 13-bit 8192 points. The 1-bit that was used is the layer to the far right in figure3.18. The encoder can usually give a position in angle, but since only one bit is used. That calculation had to be

(36)

done by writing a software program in MATLAB. The resulting vector could then be interpolated using cubic spline to give an accurate relation between time, index and degrees.

FIGURE 3.18: A 12-bit gray code for encoder to illustrate the least

significant bit (far right). The digital signal is read by looking at what color the red dots are above during rotation. Black is 1 and white is 0, the pattern is design so that only one bit is changing for each step (gray code), making it easy to remove errors when disturbances are merged with the signal. This is not the gray code for the 13-bit encoder used in section3.2.3, its just to clarify how only one of the

encoder outputs could be used as reference signal.

3.3 Single sensor measurements

3.3.1 Camshaft wheel version 1

Crossover between teeth and gaps

The smaller the dual sensor the better, therefor one should strive to place the position sensor and the pattern sensor as close as possible to each other. A problem that can occur is when the sensor is placed in the position section of the wheel, it may also detect the pattern of gaps that the pattern sensor is supposed to measure. This will shift the zero crossing of the position sensor witch gives a less accurate position result. The crossover between pattern and position sector are measured using all the chosen sensors, this will make it possible to eliminate sensors that have a lot of crosstalk.

For each sensor, measurements are made at 50 rpm in the position region with even distributed teeth, the sensor then captures one cycle of the camshaft wheel. The next measurement moves 2mm closer to the pattern region and another cycle is measured. This repeats until the sensor has moved all the way from the position region to the pattern region. Combining the measurements will create a picture of how the neighboring region is interfering with the measurement of the present region, depending on the radial position of the sensor (in case of camshaft wheel).

Sensor placement and definition of radial direction is illustrated in figure 3.19. The left half of the teeth in the figure has 2 sawed off teeth, a pattern the sensor can detect to determine sector. The right half of the teeth only measure position. When

(37)

the center of the sensor is on top of the sawed off tooth edge, it is defined to be at radial position 0mm, where the positive direction is moving away from the center of the wheel (to the left in the figure above)

FIGURE3.19: Illustration of teeth profile for camshaft wheel version

1.

3.3.2 Camshaft wheel without evenly distributed teeth

Camshaft wheel with missing teeth, this wheel is used to demonstrate the decrease in accuracy when not using evenly distributed teeth. This version causes the same problems as a wheel with big teeth, like the wheel illustrated in figure1.1.

The measurement on this wheel (see figure3.20) is done for 83 revolutions at 50 rpm together with a reference signal, then each tooth width is noted. The missing teeth are interpolated using a simple linear interpolation. The expected result is shorter missing teeth, and wider teeth before and after the missing teeth.

(38)

FIGURE3.20: A common version of a camshaft wheel with missing teeth.

3.3.3 Flywheel

Exploring potential problems with crosstalk

The flywheel assembly consists of two wheels, only one is used for measurements and is referred to as flywheel teeth (left column of teeth in figure3.21). To test how the "back wheel’s" teeth (right column of teeth) interfere with the measurements on the flywheel teeth, a GMR sensor is used to measure one complete cycle of the wheel For every revolution the sensor is moved 1mm closer to the back wheel. This is done until the front teeth are clearly not visible in the measurement. This gives a picture of how the crossover looks and can be used to decide if it’s good to use a GMR sensor to measure the flywheel or if the back wheel interfere too much.

(39)

FIGURE 3.21: Section of the flywheel as seen from the side/radial direction, the left column of teeth is the part that is being measured in this project. The right column with slightly darker teeth is referred

to as the "back wheel"

FIGURE3.22: Illustration of the flywheel teeth and back wheel teeth

as seen from the radial direction of the wheel (see figure3.21for real picture). The sensor placement is defined for axial position 0mm and

(40)

Using Hall sensor to detect cut off teeth on flywheel

The idea is to use a differential hall sensor with two hall elements inside with suffi-ciently large distance between them. The sensor is then placed right on the edge of a sawed off tooth, so that one hall element is in the gap, and the other is on top of the tooth. One of the hall elements (the right square at position 2 in figure3.23) will be above a teeth, and the other hall element will be at the gap. That way the sensor should in theory give a large pulse output at every missing teeth.

FIGURE 3.23: Illustration shows three different scenarios that the

Hall sensor can experience when the flywheel is spinning. Position 1 should give the same output as position 3. But since it’s almost im-possible to place or retain perfect alignment, one of the hall elements inside will always be slightly before the other one, causing a small disturbance. Position 2 should give a large output that can be easy to

detect

The expected result is that at close distance the output will be very good, but the output from position 2 will be smaller very fast when airgap between sensor and gear tooth increases.

How the sensitivity of the output changes depends on the sensor placement needs to be explored. One test is to check how exact the sensor needs to be placed on the edge of the tooth. It may give a good output right on the tooth edge, but change drastically if moved in any direction which is bad.

To test the sideways direction according to the illustration above, the sensor will be placed at the sawed of tooth while the wheel is stationary. A placement station is used to move the sensor very accurately in various direction. So for this experiment it was only moved sideways, referring to figure3.23above. The maximum output is noted, and after the sensor is moved 0.5mm to the side and a new measurement is written down manually. By moving from position 2 all the way to the right and all the way to the left, and writing down the sensor output every 0.5mm. A plot can be made to show how the Peak to peak output of position 2 changes depending on the sideways placement of the sensor.

(41)

Placement sensitivity regarding pitch and yaw angle

The measurements on the flywheel are made from the radial direction, and therefore the pitch is defined as the tilting of the sensor elements so that 2 hall elements in the differential sensor are passing tooth edges at different times. This misalignment will cause an unwanted output that needs to be analyzed to understand its potential disturbance to the wanted output. Yaw is using the axis of rotation of the flywheel as reference. Changing the yaw angle of the differential sensor will cause the hall elements inside the sensor to be at different distances to the teeth. They will in effect have different airgap distances and might give a small output when going between teeth and gaps.

3.3.4 Camshaft wheel version 2

Figure 3.24 defines radial position for the camshaft wheel version 2. The 0mm position corresponds to when the sensor is perfectly aligned with the outer tooth, positive direction is towards the center of the wheel for measurement convenience. When the sensor is placed at the outer radius of the camshaft wheel, the sensor is right on top of a tooth, and this position is defined as 0mm. Then the radial distance from the center of the wheel to the sensor is decreased by increments of 1mm so that the whole range of position region, pattern region and the cross section between them is captured. The crossover between the two regions of the camshaft wheel can therefore be mapped. Whereby a calculated decision of where each sensor should be placed is made, that would avoid unwanted disturbance from neighboring re-gion but also give the sensor some freedom in radial movement without changing the characteristics of the output. It’s good if there is some degree of freedom in sen-sor position so that the result is similar even if small changes in airgap and radial position is made. This will minimize the disturbance of the gaps when measuring teeth’s, and the disturbance of teeth’s when measuring gaps.

FIGURE3.24: Illustration of teeth profile for camshaft wheel version

2

3.3.5 Magnet analysis

To minimize the amount of measurements and since most of the sensors had around the same sensitivity and saturation, it was decided to use only one magnet for all

(42)

measurements. To figure out what magnet to use without wasting too much time, the sensors were powered on and connected to the oscilloscope. The sensors were then dragged across the teeth of the trigger wheel by hand in different types of com-binations of magnets and distances. The oscilloscope was only used to see if the sensor could pick up the difference in magnetic field for the teeth and the gaps in the camshaft trigger wheel. colors were used to describe what type of result a given distance for a specific magnet was giving. It could either get saturated, be too big to detect the small gear teeth, having too much noise, or sometimes miss detection of a tooth, see section4.1.1.

3.4 Dual sensor measurements

3.4.1 Flywheel - dual sensor

The placement of the dual sensor on the flywheel is defined in figure 3.25. The axial position of the pattern & position sensor is defined to 0 mm when the sensor is aligned to the edge of the sawed of tooth with the positive direction defined as left to right. The red arrows illustrate the magnetic field that the GMR sensors are detecting. The blue square is the back biasing magnet seen from above.

FIGURE 3.25: Illustration of flywheel teeth used for defining axial

position for pattern & position sensor.

3.4.2 Camshaft wheel version 2 - dual sensor

By measuring the crossover between the part of the trigger wheel that detects section i.e. cut off teeth (Fly-wheel/Camshaft version 1) or extra deep cuts (camshaft version 2) and the part that detects position (region of the wheel with evenly distributed teeth), the placement of the position & pattern sensor can be determined to ensure

(43)

that interference is not a problem between the two sections of measurement. This measurement is only done for analogue sensors that gave the best result when the single sensor measurement was done (see section4.2).

(44)

Chapter 4 Results

& Discussion

4.1 Magnets

4.1.1 Picking Magnets

The result from section 3.3.5 is displayed in table 4.1. The colors describing the results for the different sensors and magnets are explained below:

No detection Simply means the change in magnetic field is not strong enough to show a change in output. The SNR is very low

Low SNR A change in signal is present but the noise level is so high it might be problematic to get a smooth curve even after filtering.

Saturated The magnet is too strong for the sensor, its output is the same as its input with no variation whatsoever. The semiconductor has a high current passing thru it. A larger separation of electrons and holes are no longer possible.

Good result A smooth, very noticeable signal with high SNR

Partial detection Only applies to digital sensors where the trigger level inside the IC is either too low or to high so that even when a teeth is passing by, the difference in field is not big enough for output signal to change from its current state in some situations i.e. partial detection.

Bipolar sensors was removed after this test was done, it was made clear that bipolar sensors is of no use when measuring magnetic fields from gear teeth. To trigger a bipolar sensor, the magnetic field polarity need to switch sign, this never happens when using a back biasing magnet.

The planar component of the magnetic field in camshaft wheel version 1 was problematic to detect using GMR sensors due to the small teeth width, therefore only Hall sensors were used in this experiment. GMR sensors also require a more precise placement compared to Hall sensors, doing the measurement by hand was therefore not very productive.

All magnets had similar magnetic field strength, therefore magnet 3 was picked since the larger size makes it easier to mount on the placement station.

(45)

No detection Low SNR Saturated Good result Partial detection 1 TLE4997 0mm 1mm 2mm 0mm 1mm 2mm 0mm 1mm 2mm 2 SS466A 0mm 1mm 2mm 0mm 1mm 2mm 0mm 1mm 2mm 3 SS495A1 0mm 1mm 2mm 0mm 1mm 2mm 0mm 1mm 2mm 4 TLE4941 0mm 1mm 2mm 0mm 1mm 2mm 0mm 1mm 2mm 6 A1395 0mm 1mm 2mm 0mm 1mm 2mm 0mm 1mm 2mm 7 TLE49215 0mm 1mm 2mm 0mm 1mm 2mm 0mm 1mm 2mm 8 iC-MZ (D) 0mm 1mm 2mm 0mm 1mm 2mm 0mm 1mm 2mm Airgap sensor-geartooth

Airgap between magnet and pcb board Mag_rec Mag_disc Mag_weak 1mm

Airgap sensor-geartoot

h

Airgap between magnet and pcb board Mag_rec Mag_disc Mag_weak

2mm 1mm 2mm 1mm Airgap sensor-geartoot h

Airgap between magnet and pcb board Mag_rec Mag_disc Mag_weak 2mm

1mm

Airgap sensor-geartooth

Airgap between magnet and pcb board Mag_rec Mag_disc Mag_weak 2mm 2mm 1mm Airgap sensor-geartooth

Airgap between magnet and pcb board Mag_rec Mag_disc Mag_weak Mag_rec Mag_disc Mag_weak

Airgap between magnet and pcb board

2mm 1mm

Airgap between magnet and pcb board Mag_rec Mag_disc Mag_weak

Triggerwheel for camshaft

Direction of movement MAGNET PCB board Sensor 2mm 1mm

FIGURE4.1: Table that displays the results when magnets was tested

(46)

4.1.2 Demagnetization analysis of magnet 3

It turned out that GMR sensor measurements had a very low repeatability since the result could vary by a very large amount, this was shown to be because of a demag-netization in the back biasing magnet. Magnet 3 that was used for all measurements, was originally in a box with other magnets, some of them being a lot stronger. Plac-ing a weak magnet next to a strong magnet will likely demagnetize it15. The fol-lowing plots shows the magnetic field strength of magnet 3 from the sides (figure

4.2)and from the top and bottom (figure 4.3). This was done manually by placing a direct field hall sensor with known sensitivity, orthogonal to a few points on the magnet. -10 -5 0 5 mm -15 -10 -5 0 5 10 15 20 mm

Magnetic field seen from the front

-10 -5 0 5 10 mm -15 -10 -5 0 5 10 15 mm

Magnetic field seen from the right side

FIGURE4.2: Magnetic field strength as seen from the front and right

side of magnet 3. It is clear that the magnet has been demagnetized

-4 -2 0 2 4 mm -4 -2 0 2 4 mm

Magnetic field strength (top)

-4 -2 0 2 4 mm -4 -2 0 2 4 mm

Magnetic field strength (bottom)

FIGURE4.3: Magnet 3 as seen from the top and bottom. Diameter

of circles are proportional to magnetic field strength in orthogonal direction. The different sizes indicate that the magnet has in fact been

(47)

4.2 Single sensor measurements

4.2.1 Camshaft wheel version 1 measurements

This section displays the result after testing a variety of Hall-effect and GMR sensors, using different radial positions and air gaps. To limit the amount of figures due to the approximate 500 measurements that were made, a few cherry picked sensors are selected at a specific radial distance of +2mm (defined in figure3.19). The supply voltage is 5 V and the trigger-wheel is spinning at 50 rpm for all measurements.

The following guidelines were used when picking results:

Vpp- Peak to peak voltage The signal should be around 100mV or larger in its peak to peak value

Airgap The sensor needs to have good readings at 1mm & 2mm airgap distance.

SNR Larger SNR values for signals are of higher priority i.e. sensors with high noise compared to Vppwas discarded.

No focus on digital sensors Digital signals were decided not to be used due to the limitations in analysis

Sensor polarity Bipolar sensors were discovered not to be usable in this project since the triggerwheel always have the same magnetic field polarity. Bipolar sensors require change in polarity to be turned on/off.

Best results

The graphs below shows one revolution of camshaft wheel version 1 at 50 rpm with supply voltage 5 V at radial position +2mm.

Figure 4.4 shows the result for a linear Hall sensor, smooth measurements at both 1mm & 2mm airgap. V pp > 200mV at 1mm airgap & V pp > 50mV at 2mm airgap The offset voltage has a similar pattern for each revolution of the triggerwheel indicating that the disturbance is from the test rig or triggerwheel.

Figure4.5shows the result for a differential Hall sensor. Output has a constant offset of around 1.7V regardless of supply voltage for sensor. This was the only differential sensor that was tested, was picked mainly for that reason. The variation in peak to peak voltage is similar for each revolution indicating that the disturbance is from the test rig or triggerwheel.V pp>400mV at 1mm airgap & V pp >120mV at 2mm airgap

(48)

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 s 4.2 4.4 4.6 4.8 V

Sensor 3 (Hall) 1mm airgap

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 s 4.2 4.4 4.6 4.8 V

Sensor 3 (Hall) 2mm airgap

FIGURE4.4: Sensor 3 (linear Hall sensor) at 1mm and 2mm air gap at

+2mm radial distance 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 s 1.4 1.6 1.8 2 V

Sensor 8 (diff Hall) 1mm airgap

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 s 1.4 1.6 1.8 2 V

Sensor 8 (diff Hall) 2mm airgap

FIGURE4.5: Sensor 8 (differential Hall sensor) at 1mm and 2mm air

gap at +2mm radial distance

Discarded sensor results

The graphs below shows one revolution of camshaft wheel version 1 at 50 rpm with supply voltage 5V at radial position +2mm.

Figure4.6shows the result for a digital Hall sensor. At +2mm air gap distance, the sensor output seems to be good, but at 1mm air gap distance, the sensor stops triggering correctly (most noticeable at around 0.55 s & 0.75 s). It is not possible to analyze waveform from the actual signal when using digital sensors. Therefore all digital sensors are avoided for the rest of the measurements.

(49)

Figure4.7shows the result for a GMR sensor. This is a example of when a sensor is discarded for having to low signal to noise ration (SNR) in the output. Since a MATLAB script is written to trigger every time the signal passes a certain threshold voltage. It’s hard to get accurate trigger points if there is a lot of noise in the signal, even when using adaptive trigger levels.

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 s 0.4 0.6 0.8 1 V

Sensor 4 (digital Hall) 1mm airgap

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 s 0.4 0.6 0.8 1 V

Sensor 4 (digital Hall) 2mm airgap

FIGURE4.6: Sensor 4 (digital Hall sensor) at 1mm and 2mm air gap at

+2mm radial distance 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 s 2.1 2.15 2.2 2.25 V Sensor 11 (GMR) 1mm airgap 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 s 2.1 2.15 2.2 2.25 V Sensor 11 (GMR) 2mm airgap

FIGURE4.7: Sensor 11 (GMR sensor) at 1mm and 2mm air gap

High Accuracy Speed and Angular Position Detection by Dual Sensor

Juli 2018

High Accuracy Speed and Angular

Position Detection by Dual Sensor

High Accuracy Speed and Angular Position Detection

by Dual Sensor

Johan Östling

Contents

List of Figures

List of Abbreviations and

Definitions

Chapter 1

Introduction

1.1

Background

1.2

Position detection on ferrous gears

1.3

Dual sensor concept

1.4

Project description

1.5

Limitations

Chapter 2

Hardware Fundamentals

2.1

Gearbox

2.2

Measurements techniques for trigger wheels

2.3

Sensor types

Chapter 3

Implementation

3.1

Components

3.2

Measurement setup

3.3

Single sensor measurements

3.4

Dual sensor measurements

Chapter 4

Results

& Discussion

4.1

Magnets

4.2

Single sensor measurements