Wireless Power and Data Transfer in Industrial
Nutrunners
SIMON CARLSSON
KTH ROYAL INSTITUTE OF TECHNOLOGY
SCHOOL OF ELECTRICAL ENGINEERING AND COMPUTER SCIENCE
Transfer in Industrial Nutrunners
SIMON CARLSSON
Master Programme in Embedded Systems Date: June 24, 2020
Supervisor: Mark Smith
Examiner: Carl-Mikael Zetterling
School of Electrical Engineering and Computer Science Host company: Atlas Copco Industrial Technique AB
Swedish title: Trådlös effekt och Dataöverföring i Industriella
Mutterdragare
Abstract
Wireless power and data transfer systems are experiencing an ever-growing consumer and industrial adoption. Its use in common devices has made the technology more accessible to people, but seldomly does it replace a physical connection inside a product.
When two parts of an assembly are to be electrically connected, the so- lution has traditionally been connectors or wires. However, typical connec- tion methods using physical connectors between two devices can be fragile and sensitive to dust and debris. In this degree project, a wireless power and data connection between two parts of an industrial nutrunner are evaluated.
A very compact nutrunner encasement calls for a minimal wireless interface with high efficiency. Additional complications are met when the nutrunner body is made of metal, which introduces losses. Electromagnetic simulations of a flexible PCB transformer with ferrite film backing are performed in the simulation software Finite Element Method Magnetics (FEMM), and electri- cal performance is evaluated in the circuit simulator LT-Spice. From the best performing solution, a physical model is constructed and evaluated.
The final implementation uses a flyback converter for power transfer, and
amplitude modulated data for bi-directional data transfer. Results indicate the
potential for excellent performance with 1 W power transfer with more than
50 % efficiency whilst simultaneously transferring data at a rate greater than
1 Mbit/s.
Sammanfattning
Trådlös effekt och dataöverförings system ser allt fler tillämpningar i konsu- ment och industriella sektorn. Dess implementation i vardagliga enheter har gjort teknologin mer tillgänglig för människor, men den ersätter sällan fysiska kontakter inuti produkter.
När två delar av en produkt skall sammankopplas elektriskt har den tradi- tionella lösningen varit kontakter eller sladdar. Däremot så kan typiska anslut- ningsmetoder med fysiska kontakter vara ömtåliga samt känsliga för damm och smuts. I detta examensarbete undersöks ett trådlöst effekt och data gräns- snitt mellan två delar av en industriell mutterdragare. En väldigt kompakt mut- terdragrinkapsling medför ett väldigt kompakt trådlöst gränssnitt med hög verk- ningsgrad. Ytterligare komplikationer uppstår när mutterdragaren är gjord av metall, vilket medför förluster. Elektromagnetiska simuleringar av en flexibel kretskorts transformator utförs i simuleringsprogrammet Finite Element Met- hod Magnetics (FEMM) och elektrisk prestanda undersöks i kretssimulatorn LT-Spice. En prototyp konstrueras från den lösning som presterat bäst och denna utvärderas.
Den slutgiltiga implementationen använder en flyback omvandlare för ef-
fektöverföring och amplitud modulerade data för dubbelriktad datakommuni-
kation. Resultat indikerar god prestanda med 1 W effekt överfört vid mer än
50 % verkningsgrad samtidigt som data överförs med en hastighet mer än 1
Mbit/s.
I would like to express my great appreciation for the people who was involved in this thesis. Firstly I would like to thank Erik Baker and Atlas Copco for pro- viding a fantastic foundation for the thesis, without whom this project would not have been possible. I would also like to thank my supervisor at Atlas Copco, Jonas Millinger, for always being there and providing excellent feed- back throughout the entire project. The entire mechatronics team was also very friendly and helpful in the project. A special thanks go to Guillermo Bossi, who was always happy to help out and come with useful feedback. Finally, I would like to thank my academic supervisor and examiner for making this fantastic degree project possible.
v
HMI Human Machine Interface
BHMI Basic Human Machine Interface EHMI Extended Human Machine Interface WPT Wireless Power Transfer
WDT Wireless Data Transfer ASK Amplitude Shift Keying FSK Frequency Shift Keying PSK Phase Shift Keying OOK On Off Keying RF Radio Frequency PCB Printed Circuit Board
vi
1 Introduction 1
1.1 Industrial Nutrunners . . . . 1
1.2 The Human Machine Interface . . . . 2
1.3 The Future of HMI . . . . 3
2 Background & Theory 5 2.1 Wireless Power Transfer . . . . 5
2.1.1 Coupled Inductors . . . . 5
2.1.2 Skin Effect . . . . 8
2.1.3 Proximity Effect . . . . 9
2.1.4 Driving Topologies . . . . 11
2.2 Wireless Data Transfer . . . . 14
2.2.1 Digital Modulation Schemes . . . . 14
2.3 Commercial WPT and WDT Systems . . . . 15
2.3.1 The Qi Standard . . . . 16
2.3.2 The RFID Standard(s) . . . . 18
3 Transformer Design 21 3.1 Structure . . . . 21
3.2 Finite Element Analysis . . . . 24
3.2.1 Measuring Parameters . . . . 27
3.2.2 Scripting and Analysis . . . . 27
3.2.3 Simulation Results . . . . 28
3.2.4 FEMM Conclusions . . . . 36
4 Electronics Design 37 4.1 Transformer Modelling . . . . 37
4.2 Power Driving Circuits . . . . 39
4.2.1 Flyback Converter . . . . 39
4.2.2 H-bridge Converter . . . . 43
vii
4.2.3 Sensitivity Analysis . . . . 47
4.3 Data Transfer Circuits . . . . 51
4.3.1 Zero Modulation . . . . 51
4.3.2 OOK Modulation . . . . 51
4.4 Combining WPT & WDT . . . . 53
4.4.1 Single Transformer Design . . . . 53
4.4.2 Dual Transformer Design . . . . 54
5 Experimental Evaluation 57 5.1 Design . . . . 57
5.1.1 Specifications . . . . 57
5.1.2 Schematic Design & Component Selection . . . . 58
5.1.3 Transformer Design . . . . 60
5.2 Measurements . . . . 61
5.2.1 Transformer Measurements . . . . 62
5.2.2 WPT Measurements . . . . 63
5.2.3 WDT Measurements . . . . 67
6 Discussion 69 6.1 Comments on Power Performance . . . . 69
6.2 Comments on the Transformer . . . . 69
6.3 Comments on Data Transmission . . . . 70
6.4 Comments on Trace Width Variation . . . . 71
7 Conclusions 73 7.1 Summary . . . . 73
7.2 Future Work . . . . 74
7.2.1 Pseudo Litz Wire . . . . 74
7.2.2 Implementing Radio Techniques . . . . 74
7.2.3 Orientation Sensing . . . . 74
Bibliography 77
A WPT & WDT PCB’s with schematic 79
Introduction
This chapter gives a brief introduction to the topic of industrial nutrunners. In addition to this, the Human Machine Interface will be presented, that is the basis for this degree project.
1.1 Industrial Nutrunners
A nutrunner is a collective name for a family of tools that have their sole pur- pose of tightening (and in some cases, un-tightening) nuts and bolts. In as- sembly type work, these bolts must be tightened quickly and always to spec- ification. The operation speed is especially important in line-type assembly work, where seconds of assembly time matters. The quality of the tightening is a measure of how well the tightening follows the specification. This can be a requirement on a specific torque or a specific angle of the bolt. At the end of the day, consistent clamping force is the most relevant aspect. Further- more, in many cases it is not enough to tighten these bolts to specification, the manufacturer typically wants some sort of tracking system to later verify and control that all bolts are tightened to the desired specification. Usually, this is accomplished by these industrial nutrunners by reporting back the measured data during a tightening to some controller or server; this allows the manufac- turer to track the entire assembly in regards to the fasteners, and find anomalies and correlations that might have otherwise gone unnoticed.
Atlas Copco produces these nutrunner solutions. They have tools ranging from small handheld battery-operated tools with sub 1 Nm torque, to large fixtured tools capable of several thousand Nm of torque. An example of a rather typical, angle, battery-operated nutrunner can be seen in figure 1.1.
1
Figure 1.1: A typical battery operated industrial nutrunner produced by Atlas Copco [1].
1.2 The Human Machine Interface
The Human Machine Interface, or HMI, is the component that forms the trans- mission link between the tool and the user. It can incorporate visual, tactile, and audible feedback to the user to signify status and modes of the tool. The HMI of a nutrunner is the immediate feedback a user has to the intent of the tool. The system has its core functionality in signaling the user if the current operation was performed successfully. If not, the user must get notified of this so he or she can redo the operation until it is up to standard.
A typical HMI of nutrunners produced by Atlas Copco features LED’s that
signify status based on color and position of the lights. This type of HMI is
called the Basic HMI (or BHMI). Typically, a BHMI incorporates a reversing
ring that allows the operator to put the tool in reversing mode, to redo a faulty
operation. It might also include a speaker and vibrator for audible and tactile feedback.
However, the HMI system can be far more sophisticated than this; displays may be used to visualize the intended operation more clearly, showing target torque, system status, or can even allow the user to change some settings. This type of HMI is called Extended HMI (or EHMI). An example of this can be seen in figure 1.2. An EHMI may also incorporate other features that may be necessary for particular assembly work, such as scanners for confirmation and unit tracking and location systems, to force the tool to behave differently if it is to be used outside the dedicated area of interest.
Figure 1.2: An EHMI module with integrated barcode scanner [1].
1.3 The Future of HMI
When trying to figure out how the future of HMI might be, one must consider the ultimate goal of the entire nutrunner system. What is most relevant to the operator and what is redundant.
In a nutrunner, the most important job is to tighten bolts and to convey the
result quickly and efficiently to the operator. Not a single bolt can afford to
be tightened to an insufficient specification. It is for this reason essential that
the nutrunner can convey the information to the operator efficiently enough to
always grab his or her attention, regardless of whether the operator is looking
away or has attention focused elsewhere. Investigations into HMI placement
in a pair of glasses has been done in prior thesis work at Atlas Copco [2].
However, this requires the operator to wear something that is linked to the nutrunner.
Another possibility is to move the HMI to a more optimal location while still keeping it on the tool. Through market analysis, this turns out to be the actual fastener itself. Placing the HMI on the front of the tool, close to the fastener, the operator will inevitably have the HMI feedback in the field of vision. However, future advancements in the nutrunners produced by Atlas Copco makes this implementation trickier. The front part (as it is called) de- picted in figure 1.1, is rigidly mounted to the rest of the nutrunner body. There are plans to make this operator interchangeable; this would make placing an HMI in the front of the tool a challenge. One obvious solution would be to use spring-loaded connector pins that connect the removable front part with the main nutrunner body. While this could be a feasible solution, the reliabil- ity of such an interconnection system is of great concern. Dust, dirt, or other particles can interfere with the connection to the point of failure and cannot be contained since this would not be a sealed system.
Another aspect would be to look into wireless connectivity between the front part and the main body. This interface would likely be more complicated but offer the advantage of zero mechanical components and probably a more robust connection over time. This system would require a wireless transfer of power in one direction, and transfer of data in both directions. A system like this has not been investigated for this application before. It would require tiny electronics to fit within the extremely compact nutrunners that Atlas Copco produces. It would also need to be inexpensive to manufacture with minimum manual labor required for assembly. In addition to the potential HMI benefits, an interface like this could enable other features in the nutrunner to also be implemented in the removable front part.
For these reasons, wireless power and data transfer system are investigated
and researched in this degree project.
Background & Theory
Wireless power and data transfer is nothing new. Plenty of research and well- established products and solutions already exist on the market. In this chap- ter, some of these will be presented, along with some fundamental theories of wireless power transfer and its limitations. In addition to wireless power transfer, wireless data transfer will also be given its introduction, and ways to modulate the signal will also be presented.
2.1 Wireless Power Transfer
Wireless energy transfer has been of great interest ever since the first human- made electric circuit was created. However, practical and useful wireless power transfer systems have only become available to the public in the last couple of decades. More efficient and compact semiconductor devices in co-junction with tight tolerance components have all lead to an ever-growing wireless in- frastructure.
Currently, one of the more popular wireless power transfer standards is the Qi standard, designed to transfer power to a mobile, consumer device for convenient charging.
2.1.1 Coupled Inductors
Inductive wireless energy transfer is by far the most common. It provides medium distance transfers while not requiring any specific hardware or driver to function. Furthermore, it can be quite efficient, so battery-powered devices can incorporate these transmitters without wasting too much energy. Figure 2.1 illustrates a typical inductive wireless energy transfer system. It is con-
5
structed with an air-core transformer (L
p,L
s) for the wireless interface with their corresponding equivalent series resistance (R
p,R
s), and an AC voltage source (V
p) and load (Z
L). The load impedance (Z
L) is typically replaced by some rectifier and target circuit to obtain a DC source from the wireless inter- face.
For increased transfer efficiency, it is common to see these systems utilize a resonance parallel or series LC circuit. In figure 2.1, the parallel option is illustrated.
Ls Lp
Cp Cs ZL
Vp
Rp Rs
K
Figure 2.1: Wireless power transfer with parallel resonance capactiors.
The resonance capacitors (C
Pand C
S) are used to amplify voltage or cur- rent at a given frequency. The resonance frequency for both the series LC and parallel LC circuit can be found as:
f
0= 1 2π √
LC (2.1)
The series resonance configuration amplifies the voltage over the primary wind- ing while the parallel configuration amplifies the current. The level and range of amplification are determined by the quality factor, Q. A high Q factor pro- vides greater amplification at the cost of narrow resonance bandwidth.
An important part of the air-core transformer is the coupling coefficient of the primary and secondary side inductors. This is described by:
K = M
q
L
pL
s(2.2)
Where M is the mutual inductance and can be derived from Neumann’s for- mula [3]:
M = µ
04π
I I
dl
1· dl
2|r
2− r
1| (2.3)
K ranges from zero to one, where one corresponds to 100 % magnetic flux
coupling. In WPT systems, K is usually smaller than one but usually kept as
high as possible to ensure maximum efficiency.
Figure 2.2: Transformer T-equivalent transformation.
For calculations it is common to model the transformer as a T-equivalent circuit according to figure 2.2 [4]. Using Kirchoff’s current law the impedance can be calculated as:
Z
prim= R
1+ jω(L
1− M ) (2.4)
Z
sec= R
2+ jω(L
2− M ) (2.5)
Z
M= jω(M ) (2.6)
The total impedance seen from the primary side is then given by:
Z
th= (Z
sec+ R
3)Z
MZ
sec+ R
3+ Z
M+ Z
prim(2.7) Z
this a non-real impedance and requires a matching network to make it real; this is especially important if a Class-E topology is to be used as the driving circuitry.
Stray Capacitance
At higher frequencies, the effect of stray capacitance between the transform- ers’ windings becomes more prominent; this can reduce the overall bandwidth of the transformer and result in un-wanted behavior to specific transient sig- nals. Stray capacitance is highly dependant on the chosen geometry and can be mitigated to some extent but not completely.
Magnetic Cores
In WPT systems, the inductors tend to be loosely coupled with K < 1. This lower K is due to the lack of a shared magnetic core that links the primary side with the secondary using a high permeability material. However, these types of systems may still use high permeability materials elsewhere in the design.
These materials may be used to steer the magnetic field to better couple the
inductors or to shield the system from closely mounted electronics or nearby metals [5]. Whenever a high permeability magnetic material is introduced to a magnetic circuit, the concern of saturation becomes apparent.
If the magnetic flux density increases above the material’s saturation limit, the magnetic performance of that material diminishes, and the system may no longer perform as intended. To reduce the risk of saturation, one can reduce the number of turns, increase the physical size of the used magnetic material, or reduce its relative permeability.
2.1.2 Skin Effect
A conductor carrying a time-varying electric current produces a magnetic
field. This magnetic field induces electromotive forces on the environment
and itself. These forces create eddy currents in the environment and in the
conductor itself. Eddy currents in the conductor have opposite direction to the
net current flow in the center of the conductor, and hence combine to reduce in
amplitude. On the outer edges, the eddy currents are in the same direction as
the net current, so the currents amplify. The effective conductor area reduces
and results in ohmic losses. This phenomenon is known as the skin effect. An
illustration of the skin effect is depicted in figure 2.3. The current i(t) pro-
duces a magnetic field H(t) which creates eddy currents i
e(t) that reduce in
the center and combine towards the outer edge of the conductor.
J r)
H(t )
i t)
0 r
L(r )
ro
−ro
0 ro r
−ro
0 ro r
−ro
H(r ) iet)
Figure 2.3: Illustration of the skin effect [6].
The majority of the current flows on the outer edge δ distance towards the center of the conductor. δ is the skin depth and can be calculated as:
δ =
s
ρ
πµ
rµ
0f [m] (2.8)
Where ρ is the resistivity of the material, and f is the frequency. For the sake of simplicity, the skin depth for copper at room temperature for some relevant frequencies are displayed in table 2.1. For the geometries that are relevant for this application, frequencies above 100 kHz have relevant skin effect losses.
2.1.3 Proximity Effect
Much like the skin effect, the proximity effect is an undesirable effect that
reduces the cross-sectional area of a conductor. The proximity effect occurs
whenever two adjacent conductors are carrying high-frequency signals. At
frequency, a conductor produces a magnetic field B axially around the con-
ductor. This field induces eddy currents in any adjacent conductors, resulting
Table 2.1: Skin depth in copper δ
Cuat different frequencies f . Frequency f Skin depth for copper δ
Cu1 kHz 2.1 mm
10 kHz 0.66 mm
20 kHz 0.47 mm
50 kHz 0.30 mm
100 kHz 0.21 mm
200 kHz 0.15 mm
500 kHz 93 µm
1 MHz 66 µm
2 MHz 47 µm
5 MHz 30 µm
10 MHz 21 µm
20 MHz 15 µm
50 MHz 9.3 µm
in loss and hindering their current carrying capacity. This phenomenon in- creases with proximity to the adjacent conductor. The current profile under the proximity effect is depicted in figure 2.4. Here we can see that current will either flow in closer proximity or further proximity to the adjacent conductor, depending on the relative current direction. In a transformer application, the proximity effect might appear between adjacent windings and hinder perfor- mance by increasing the real part of the impedance of the winding.
J
x
i i
i i
J
x
Figure 2.4: Proximity effect for two conductors carrying current in opposite
direction (left) and in the same direction (right) [6].
2.1.4 Driving Topologies
The induced magnetic field produced by an air-core inductor is directly pro- portional to the current that passes through it. Hence, to obtain a strong field sufficient to induce a desired current in the secondary side, a power driving stage is required.
Depending on the topology of the air-core transformer and load require- ments, the driving topology might differ. However, they usually share a few aspects:
• Voltage step up
• Current amplification
• High frequency capable
• High efficiency H-Bridge
A typical approach to this problem is with the use of a full bridge or half-bridge driver. The full-bridge driver or the H-bridge provides the best performance since it doubles the supply voltage, a schematic illustration of the H-bridge can be seen in figure 2.5.
M2 M1
Vdd
M3 M4
V
Figure 2.5: An H-bridge used as a driver for WPT.
The H-bridge requires that transistors M
1, M
3and M
2and M
4are driven
in pairs to obtain the doubling in voltage. The H-bridge can be very stable
and provide a high current (depending on transistor selection). However, it does not produce smooth transitions, which leads to spikes in voltage during switching. Furthermore, transistors must be driven hard to make sure that minimal energy is wasted since the H-Bridge does not provide zero voltage switching.
In wireless power applications, the driver would tend to switch an inductive load (primary winding), which might cause voltage runaway due to the sudden change in current, V = L
dIdt. Providentially, the H-bridge always provides a current return path, either through ground or V dd; this results in a controlled voltage over the inductor [7].
Flyback Converter
A flyback converter is one of the simplest forms of driving current through a transformer. The topology for the flyback converter can be seen in figure 2.6.
It consists of a driving transistor that switches the inductor directly. When current is let off from the switch, the voltage at the node between the inductor and driving transistor increases drastically, causing a great negative voltage differential over the inductor. This voltage is especially high if the secondary side of the transformer is poorly coupled. This high voltage appears since, unlike the H-bridge, the flyback does not provide any return current when the transistor is switched off. Yet this voltage can be beneficial since it means that we can obtain a higher secondary side voltage. However, it can also be harmful since typically integrated transistors may not be rated for these higher voltages. By adding a parallel resonance capacitor, the current has a return path, the voltage is limited, and the driver obtains an increase in current.
One drawback that the flyback converter shares with the H-bridge is the lack of sinusoidal voltage and current; this results in radiated emissions span- ning over a broader frequency spectrum; this produces losses. Like the H- bridge, the flyback does not provide zero voltage switching, which in turn means that significant power can be lost in the switching transistor. And fi- nally, unlike the H-bridge, the flyback converter does not produce symmetric output voltage over the secondary side, the output is biased toward one side.
This uneven bias might cause an uneven load of any potential core material
and, potentially, earlier saturation of it [8].
Figure 2.6: A flyback converter driving a WPT interface.
Class-E Amplifier
The class-E amplifier, developed by Nathan Sokal, is a self resonating ampli-
fier designed for RF applications. The class-E amplifier can be seen in figure
2.7. In many cases, it is very similar to the flyback converter, but solves many
of its issues. It has zero voltage switching, which means that very little to no
energy is wasted in the switching transistor. Furthermore, it incorporates a
shaping network consisting of a constant current inductor and series LC res-
onance circuit that provides sinusoidal current and voltage, which means that
very little energy is wasted in the out-of-band spectrum. Additionally, the
resonating nature of the circuit can provide a boost in both current and volt-
age, much higher compared to the H-Bridge. However, it requires carefully
chosen components, and if it’s implemented in a wireless power situation, it
also requires that the secondary side always has the same real to imaginary
impedance factor (power factor). Otherwise, the resonance frequency might
shift, and the amplifier is not as efficient [9].
L1
Q1 C1
C2
R1 L2
Vdd
Figure 2.7: The class-E amplifier driving a resistive load.
2.2 Wireless Data Transfer
Data transfer over an inductive interface introduces some complications in re- gards to bandwidth. The self-inductance in a transformer will act as a filter and attenuate any frequencies that are not in the pass-band. Thus, a close to constant data rate is required to make sure that the data is not distorted at the secondary side. It is because of this issue beneficial to utilize some type of modulation scheme. This modulation enables arbitrary data rates since the frequency is dictated only by the carrier wave. However, direct data transfer is still viable in some cases and is briefly discussed in section 4.3.1.
2.2.1 Digital Modulation Schemes
Much like their analog counterparts, digital modulation combines a signal with a carrier wave. How these signals are combined and later decoded depends on the modulation scheme.
Amplitude Shift Keying (ASK)
The simplest and most straightforward approach to signal modulation is Am-
plitude Shift Keying or ASK. Like AM, this modulation multiplies the base
signal with the carrier wave to produce the modulated signal. De-modulation
is later done with a level detector. ASK can use multiple levels of mixing to
provide more than two symbols. Nevertheless, more levels usually mean it is more susceptible to noise.
A particular case of ASK modulation appears when only two levels are used with 100 % mixing depth. I.e., carrier wave present represent logic 1, and no carrier wave present represents logic 0. This particular case is usually referred to as On-Off Keying or OOK and enables the construction of very simple transceivers. A drawback of OOK is the non-constant signal power and that the carrier wave has to be substantially higher in frequency than the data stream [10].
Frequency Shift Keying (FSK)
FSK modulates the signal based on frequency; this has some benefits com- pared to ASK. Firstly, it is more immune to noise pollution since small changes in amplitude on the receiver doesn’t change the decoding. This immunity to variations means that the modulation depth is constant with distance and can be detected easier with weaker signals. But, FSK has more significant spec- trum pollution compared to ASK, which might interfere with other devices.
Construction of FSK transceivers is trickier than ASK, mainly since it requires carrier synchronization on the receiver, making it more complex [10].
Phase Shift Keying (PSK)
The final modulation scheme provides all the benefits of FSK with less spec- tral pollution since data is encoded in the carrier phase. With PSK, more sym- bols can be encoded on the same carrier without increasing spectral pollution.
Quadrature Phase Shift Keying or QPSK encodes two bits with four symbols of different phases; this enables higher data throughput at the same carrier fre- quency. Unfortunately, just like FSK, PSK requires carrier synchronization and, in many cases, need a higher degree of accuracy in this compared to FSK [10].
2.3 Commercial WPT and WDT Systems
WPT systems are not new inventions; they have existed ever since Nikola
Tesla’s time. However, in recent years, mass adoption of the technology has
taken place in the form of commercial products, mainly in the consumer elec-
tronics sector. Wirelessly charging mobile phones, wireless payment, and au-
thentication methods, and wireless charging of vehicles are a few of the current
implementations of WPT systems. Studying these well established WPT and WDT systems helps in researching new implementations.
2.3.1 The Qi Standard
Overview
The association, Wireless Power Consortium, develops wireless standards, which includes the popular Qi standard. The Qi standard defines an induc- tive power transfer system for handheld mobile devices, such as smartphones.
Consider figure 2.8; a transmitter contained in some enclosure that includes the power transmitting coil and driving electronics. The receiver, containing the receiving coil and rectifying circuitry, which, when placed over the trans- mit coil, starts the transfer process. Power ranges from 5 W to 15 W in the latest version.
PTx coil Power cable
PRx coil
Figure 2.8: WPT in Qi products illustrated with charging pad and smartphone with integrated Qi receiver [11].
The operating frequency of the system typically is between 87 kHz to 205 kHz, but frequencies outside this range may also comply with the standard.
Typically the frequency would be used to regulate the power, with the lower range for highest power transfer.
To negotiate at what frequency and what power level the system should
operate, Qi uses a data protocol. The Qi standard dictates an FSK type mod-
ulation of the power carrier wave to transfer this information, among other
details that might be useful for the transmitter or receiver [11]. The data is overlayed on the power signal with the use of backscatter modulation.
Circuit Topology
The Qi standard dictates a two capacitor resonant circuit, as shown in figure 2.9. Capacitor C
Sis the main resonant capacitor, while C
d, with its series switch, is used in detecting resonance.
CS
Cd
LS
Figure 2.9: A dual resonance capacitor system used to detect resonance at the receiver [12].
The resonance circuit increases the efficiency of the power transfer; this is achieved when the primary and the secondary side both have the same resonant frequency. The resonance circuit has some Q value. This Q value dictates how narrow the resonance is in the frequency domain. The quality factor of the circuit, shown in figure 2.9 (with C
ddisconnected) can be described as:
Q = 2πf
SL
SR
S(2.9)
Where R
Sis the effective resistance of the coil, and f
Sis the resonance fre- quency. A high Q value provides the highest efficiency while a lower value might yield a more stable system that is more tolerable to variations.
The complete (simplified) transmission model can be seen in figure 2.10.
Z
Lis the load impedance, signified by a separate cell. However, this will usu- ally include a full bridge rectifier and smoothing circuitry as well as a matched load.
Communication Protocol
The transmitter and receiver may require some negotiation regarding the de-
sired frequency or power requirements. For this, an FSK type modulation
Lp Cp
Rp
Ls Cs
Rs kop
is
Power Transfer Interface
PTx PRx
ZL iL
uL
Load Circuit uop
fop ip
Figure 2.10: Simplified Qi, WPT circuit with all relevant abrevieations [12].
scheme is used over a backscatter type communications link. Backscatter mod- ulation implies that the receiver and transmitter have a means of modulating the load and transmitted power, respectively. Then, each side must also be able to detect the variation in power to retrieve the data. In other words, this forms an AM modulated signal with, in this case, a shallow AM depth. The data is, in turn, modulated using FSK modulation that the corresponding side has to decode. This is a half-duplex communications link, meaning that only one side can transmit at a time while the other is listening. The data rate is also relatively slow. Exact numbers are not given, but the backscatter modulation has to be significantly slower than the primary WPT wave, which could be as low as 87 kHz [12].
2.3.2 The RFID Standard(s)
The RFID standards are a set of standards for wirelessly powered and commu- nicating identification tags. Frequencies, modulation schemes, and protocols vary between the different types of standards. However, the functional princi- ple is similar to many of the systems.
In figure 2.11, one can see the functional block diagram of a typical RFID system. The transmitter, or in this case, the tag reader, outputs a carrier wave at 13.56 MHz; this is then picked up by a receiving coil located in a transponder.
A parallel resonance capacitor circuit is used on both the transmitting and
receiving coil to increase the power transfer efficiency.
∼
BP
Chip Magnetic field H
Reader
Binary code signal
Transponder Ri
DEMOD
frdr fs
C1 C2
Figure 2.11: Block diagram of a typical RFID system. Card reader (left) and transponder/card (right) [13].
Typically, the signal is rectified using a single diode and capacitor. When sufficient charge is built up in the capacitor, the transponder regulates a load to communicate back to the reader. However, some variations of RFID use full- duplex transmission, then the RX and TX communication is split in frequency rather than in time.
The actual power transfer is much lower than that of the previously men-
tioned Qi standard. However, its use of backscatter modulation for transferring
mainly data makes it interesting. Again, the RFID system is typically a half-
duplex system, i.e., only the transponder or the reader may transmit at one
given time. Data modulation can be of ASK, FSK, or PSK (depending on
standard), and the data transfer rate can be of 500 kbit/s [13].
Transformer Design
The transformer design in this degree project is quite different from other WPT and WDT systems. It utilizes a flexible PCB coil rather than wire-wound turns of a conventional air-core transformer. The structure of this solution is pre- sented along with how this transformer is simulated in a finite element anal- ysis software, FEMM. Results from the FEMM simulations are presented in section 3.2.3.
3.1 Structure
The WPT and WDT systems that are researched in this degree project are to be implemented in a commercial nutrunner product. Please consider figure 3.1.
Figure 3.1: The intended nutrunner structure. Main nutrunner body (left), and removable part (right). Orange section on the removable part illustrates the secondary side coil placement.
The nutrunner in question is composed of two components: The main body that houses the majority of the electronics in the nutrunner (left in the
21
figure), and The removable section that houses electronics for the HMI and other peripherals (right in the figure). These two components are, to this day, mechanically connected, but not electrically connected. The structure in question aims to provide power and data connectivity between these two components, without disrupting the mechanical structure. A cross-section of the power transfer section of figure 3.1 is depicted in figure 3.2.
Figure 3.2: Cross section of the cylindrical transformer. The inner part is the secondary side of the transformer (removable part), and the outer part is the transmitter, i.e., the primary side of the transformer. The dotted line shows the axis of symmetry. Figure not to scale.
This is an air-core transformer that is housed between the two components.
The clearance in this structure is at the maximum of 1.9 mm. This thin struc- ture means that the transformers’ windings are required to be sufficiently thin to fit both the primary and secondary sides within this gap whilst also not in- terfering with the mechanical structure. For this, a flex PCB structure for the primary and secondary windings is proposed. A flex PCB can be thinner than a conventional wire-wound transformer and has the benefit of possibly re-using the same manufacturing process that is already implemented in the nutrunner.
An illustration of flex PCB coil can be seen in figure 3.3. The PCB flex coil
only has a determined height and width (set by the mechanics of the system),
all other geometries of the coil are modifiable to an almost infinite extents.
Figure 3.3: An illustration of the proposed flex PCB coil (not to scale).
The nutrunner housing is constructed with a mix of ferrous and non-ferrous materials. Placing a magnetic system close to these materials may result in considerable losses in the system due to eddy currents. This is an apparent non- desirable effect of placing an AC magnetic circuit near these materials. Some ideas to re-direct these losses include ferrite and iron cores that contain the flux inside the core. The primary coil would then be placed on one half of this core, and the secondary would be placed on the other half. There would then have to be an air gap in between these two halves to provide free movement.
However, this solution would be fairly complex to manufacture. It would also require a sufficient air gap between the core halves to not interfere with the mechanical structure, hindering the magnetic advantage. It is for this reason that a thin ferrite sheet is instead explored.
The ferrite sheet is depicted in figure 3.4. The ferrite sheet is placed behind the flex PCB coil to re-direct most of the magnetic flux induced by the coil away from the metal structure. Unlike a ferrite or iron core, this film would not cover the coil ends but instead allow for some flux leakage in these areas.
This construction would also be less complex to manufacture since no unique
3D geometry is required.
Figure 3.4: Ferrite adhesive films from Wurth Elektronik [14].
The ferrite sheet can be ordered with different relative permeability and varying thickness. The film is flexible enough to be wound around the axle and can be cut to any shape. One potential drawback of the ferrite film is the relatively small cross-sectional area that might cause lower magnetization flux densities than a solid ferrite core.
3.2 Finite Element Analysis
For verifying the proposed solution’s magnetics performance, a finite element
analysis software is used. The chosen software is called FEMM (Finite Ele-
ment Method Magnetics). This software was chosen due to its simplicity in
the way it is operated and for its scriptability. FEMM is a 2D simulations pro-
gram that can be used for linear or nonlinear magnetostatic problems, linear
or nonlinear time-harmonic magnetic problems, linear electrostatic problems,
and steady-state heat flow problems [15]. Due to its 2D simulation topology, it
is only suitable for 2.5D, and axis-symmetrical geometries, where 2.5D would
be a 2D outline with some depth and axis-symmetrical is a cross-section ge-
ometry that is revolved around a center axis. The proposed solution is a trans-
former with two inductors, wound around a center axis; this lends itself well for an axis-symmetrical simulation.
In figure 3.2 is the illustration of the cross-section of the proposed trans- former solution. The structure is modeled in FEMM by scripting commands in MATLAB. This setup allows for parameter-driven modeling where differ- ent parameters of the structure can be stepped about each other. Parameters that are of interest include:
• Number of primary side turns
• Number of secondary side turns
• Primary and secondary side coil radius
• Track spacing
• Track width
• Track height
• Multi-layer track stack up
• Excitation frequency
The geometry is seen in figure 3.2 was cut on the symmetry line, and the right-hand side was used for simulations in FEMM. The resulting topology from FEMM, for a set of parameters, can be seen in figure 3.5. This figure shows the two flex PCB’s with ferrite film backing placed at 1.5 mm distance.
The same structure with attached measurements can be seen in figure 3.6.
This figure shows all relevant parameters altered during the simulation, and
the dimensions mentioned will be referencing this figure.
Figure 3.5: Model of the transformer as simulated in FEMM. Primary side (right) and secondary side (Left) with the same amount of turns and geometry.
Both coils are placed on a ferrite material.
RSecondary RPrimary
Delta carrier
Trace clearance Trace width
TFerriteTPCB Symmetry line
HFerrite
PCB offset
X
X
X
X X
X
X
X
NSecondary NPrimary
Figure 3.6: Illustration of the flexible coil transformer with all relevant mea-
surements.
3.2.1 Measuring Parameters
The coils in figure 3.5 are constructed using two separate circuits. One for the primary and one for the secondary side. Both are defined as series circuits, which means that each turn of the coil is in series. The desired outputs of the FEMM analysis are:
• Coupling coefficient
• Self-inductance
• Impedance
• Q factor
• Flux density
Of these only the coupling coefficient can not be directly (or indirectly) ex- tracted within FEMM. For this, a workaround, described in [15] and in a thread [16] is used. This workaround requires three simulation runs to be performed.
The first is with driving the primary side with current I and obtaining the pri- mary side inductance, L
1. Also, the flux induced by the primary coil on itself is stored as Φ
11. Secondly, the primary side is disconnected, and the secondary side is driven with a current I, the secondary side inductance is measured and stored as L
2. And thirdly, both the primary and secondary sides are driven with current I, and the total flux is measured and stored as Φ
21. The resulting values can be used in equation 3.1 and 3.2 to calculate the mutual inductance and coupling coefficient, respectively.
M =
Φ
21− Φ
11I
(3.1)
K = M
√ L
1L
2(3.2)
3.2.2 Scripting and Analysis
FEMM has scripting possibilities for creating structures, performing simula-
tions, and extracting outputs. A toolkit called OctaveFEMM taps into these
commands by providing the appropriate toolbox in MATLAB [17]. Octave-
FEMM is officially supported by FEMM and is installed with FEMM 4.2. The
script written in MATLAB constructs the geometry, which is defined by a set
of global parameters. These parameters define crucial parts of the structure
that are of interest to modify. By keeping these global, the structure can be rebuilt multiple times with slightly different attributes. This enables plotting of the change in critical parameters versus changes in geometry.
From the same script that generates the geometry, the simulation condi- tions are also set. These include boundaries, material properties, meshing, ini- tial conditions, and circuits. The script saves the desired outputs from FEMM in variables in MATLAB to be manipulated or plotted. The simulation flow can be seen in figure 3.7.
Draw structure Set up inital conditions and global
parameters
Draw boundry
Dene materials
and circuits
Create mesh and simulate
Save output Loop
Modify circuit
2X
Main script
Modify global parameter Plot output
Figure 3.7: Simulation flow when solving for parameters in FEMM and MAT- LAB.
3.2.3 Simulation Results
The structure generated in FEMM can be seen in figure 3.5. This base struc-
ture was used throughout the simulation in FEMM. Some parameters were
modified and swept to obtain dependencies and optimum geometry.
Figure 3.8: Transformer structure in FEMM with plotted magnetic flux density
|B| in Tesla.
Primary and Secondary Turns Variation
Firstly, the effect of the turns ratio on the primary and secondary side was
investigated. In this simulation, a model consisting of single-sided coils on
both the transmitting and receiving end was used. The plot of the turns ratio
vs. coupling coefficient can be seen in figure 3.9. We can see that the highest
coupling is achieved on a linear, diagonal line, increasing with the number of
turns. The highest coupling coefficient is reached at N 1 = N 2 = 10, and the
lowest is measured at the extremes when one side only has a single turn, and the
other has multiple turns. What is not as easily visible in this figure is that the
coupling coefficient starts to decay at turns higher than 10. A higher number
of turns is not plotted for single-sided coils since the compact geometry did
not allow for it. The simulation was performed at frequency f = 100 kHz to
reduce the skin effect’s impact on the result.
Figure 3.9: Coupling coefficent with different windings on primary and sec- ondary side.
Other Parameters VS Number of Turns
An additional simulation was performed to see what the affect the number of turns has on other parameters. The simulation was done with N 1 = N 2 = N that shows the dependency vs. Re(Z), inductance, coupling coefficient, and quality factor, the plot of which can be seen in figure 3.10. Again, this simulation was done at frequency f = 100 kHz to minimize the skin effect.
Here we can see the almost linear dependency of resistance and number of
turns. We can also see the exponential dependency of inductance. The quality
factor is seen as a decaying function that settles at a quite low value due to the
low frequency and high resistance to inductance ratio.
0 2 4 6 8 10 Number of turns (N)
0 1 2 3 4
Resistance (Ohm)
re(Impedance)vs NO. turns
0 2 4 6 8 10
Number of turns (N) 0
0.5 1 1.5
Inductance (H)
10-5 Inductance vs NO. turns
0 2 4 6 8 10
Number of turns (N) 0.65
0.7 0.75 0.8 0.85 0.9
Coupling coefficent (K)
Coupling coefficent vs NO. turns
0 2 4 6 8 10
Number of turns (N) 0
1 2 3
Quality factor (Q)
Quality factor vs NO. turns
Figure 3.10: Parameters versus number of turns with N
P rimary= N
Secondary. To study the effects of frequency, the same parameters as in figure 3.10 were analyzed at three different frequencies. This time the coils have been constructed using double-sided windings, which produce higher Q factors for any given frequency. The results can be seen in figure 3.11. Here, the number of turns is represented per side; each side is in series, so the total number of turns is the same as twice the annotated values. We can see that skin effect becomes a large contributor to the coil’s resistance at higher frequencies. We can also see minimal deviation for both the inductance and coupling coefficient plots. However, the quality factor improves greatly with higher frequencies.
One can also see the inverse relationship between the number of turns and
frequency in the quality factor plot. For a given rise in frequency, the peak of
the quality factor is reached at a lower number of turns.
0 2 4 6 8 10 Number of Turns per side (N) 100
102
Resistance (Ohm)
re(Impedance) vs NO. turns 10 MHz
1 MHz 100 kHz
0 2 4 6 8 10
Number of Turns per side (N) 0
2 4 6
Inductance (H)
10-5 Inductance vs NO. turns 10 MHz
1 MHz 100 kHz
0 2 4 6 8 10
Number of Turns per side (N) 0.65
0.7 0.75 0.8 0.85
Coupling coefficent (K)
Coupling coefficent vs NO. turns
10 MHz 1 MHz 100 kHz
0 2 4 6 8 10
Number of Turns per side (N) 0
20 40 60 80
Quality factor (Q)
Quality factor vs NO. turns 10 MHz 1 MHz 100 kHz
Figure 3.11: Parameters versus number of turns per layer (using 2-layer wind- ings) and at multiple frequencies. N
P rimary= N
Secondary.
Effects of Trace Widths
The dependencies on trace width were also investigated. The idea is that trace width lowers the resistance until the width is equal to the skin depth, at which point it will roll-off. We can see this for 100 kHz in figure 3.12 and for 1 MHz in figure 3.13. The trace width was modified while keeping the size of the coil the same. This means that the clearance between traces decreases with in- creasing trace widths. For 100 kHz, we can see that the coupling coefficient is quite independent of trace width. We can also conclude that the skin depth has minimal effect on the resistance since it falls off quite linearly with increasing trace width.
Interestingly, the inductance decreases with trace width. For 1 MHz, we
can see that 0.22 mm trace width produces the lowest resistance, while the
peak Q is at 0.2 mm. At this frequency, the skin depth is quite prominent in
the resistance of the system. However, instead of rolling off, the resistance
increases for greater trace widths than 0.22 mm. This is further discussed in
section 6.4.
0.15 0.2 0.25 0.3 Trace width (mm)
2 2.5 3 3.5 4
Resistance (Ohm)
re(Impedance)vs trace width
0.15 0.2 0.25 0.3
Trace width (mm) 0.8
0.9 1 1.1 1.2 1.3
Inductance (H)
10-5 Inductance vs trace width
0.15 0.2 0.25 0.3
Trace width (mm) 0.84
0.845 0.85 0.855 0.86 0.865
Coupling coefficent (K)
Coupling coefficent vs trace width
0.15 0.2 0.25 0.3
Trace width (mm) 1.8
2 2.2 2.4 2.6 2.8
Quality factor (Q)
Quality factor vs trace width
Figure 3.12: Parameters versus tracewidth at 100 kHz.
0.15 0.2 0.25 0.3
Trace width (mm) 3.6
3.8 4 4.2 4.4
Resistance (Ohm)
re(Impedance)vs trace width
0.15 0.2 0.25 0.3
Trace width (mm) 0.8
0.9 1 1.1 1.2 1.3
Inductance (H)
10-5 Inductance vs trace width
0.15 0.2 0.25 0.3
Trace width (mm) 0.84
0.845 0.85 0.855 0.86 0.865
Coupling coefficent (K)
Coupling coefficent vs trace width
0.15 0.2 0.25 0.3
Trace width (mm) 12
14 16 18 20
Quality factor (Q)
Quality factor vs trace width
Figure 3.13: Parameters versus tracewidth at 1 MHz.
Effects of Trace Clearance
To further investigate the effects of trace clearance, an additional sweep of
trace clearance at 1 MHz was performed. The results from the simulation can
be seen in figure 3.14. Measurements were performed with a fixed position
of the outmost turn; this means that the inner windings get smaller radius as spacing increases. In figure 3.14, we can see that resistance has a minimum dependency on the clearance and that we obtain a peak in coupling at 0.25 mm clearance.
0.05 0.1 0.15 0.2 0.25 0.3
Trace clearance (mm) 5
5.02 5.04 5.06 5.08 5.1
Resistance (Ohm)
re(Impedance) vs trace clearance
0.05 0.1 0.15 0.2 0.25 0.3
Trace clearance (mm) 1.4
1.6 1.8 2 2.2
Inductance (H)
10-5Inductance vs trace clearance
0.05 0.1 0.15 0.2 0.25 0.3
Trace clearance (mm) 0.8
0.805 0.81 0.815 0.82
Coupling coefficent (K)
Coupling coefficent vs trace clearance
0.05 0.1 0.15 0.2 0.25 0.3
Trace clearance (mm) 18
20 22 24 26
Quality factor (Q)
Quality factor vs trace clearance
Figure 3.14: Parameters versus trace clearance at 1 MHz.
Axial Play
The application is so that the coils cannot be guaranteed a perfect alignment
between primary and secondary coils. This is mainly in terms of axial play
were play in the mechanical linkage allows for some amount of vertical move-
ment. A screengrab from FEMM where the maximum axial play of 0.75 mm
can be seen in figure 3.15. In reality, the maximum play in the linkage would
be smaller than this. The resulting plots from this simulation are depicted in
figure 3.16. Most notably from these plots is the square drop in the coupling
between the two coils.
Density Plot: |B|, Tesla 3.844e-002 : >4.046e-002 3.641e-002 : 3.844e-002 3.439e-002 : 3.641e-002 3.237e-002 : 3.439e-002 3.034e-002 : 3.237e-002 2.832e-002 : 3.034e-002 2.630e-002 : 2.832e-002 2.428e-002 : 2.630e-002 2.225e-002 : 2.428e-002 2.023e-002 : 2.225e-002 1.821e-002 : 2.023e-002 1.618e-002 : 1.821e-002 1.416e-002 : 1.618e-002 1.214e-002 : 1.416e-002 1.011e-002 : 1.214e-002 8.092e-003 : 1.011e-002 6.069e-003 : 8.092e-003 4.046e-003 : 6.069e-003 2.023e-003 : 4.046e-003
<0.000e+000 : 2.023e-003
Figure 3.15: Image of the structure in FEMM with 0.75 mm difference in axial position.
-1 -0.5 0 0.5 1
Axial offset (mm) 4.9
5 5.1 5.2 5.3
Resistance (Ohm)
re(Impedance) vs Axial offset
-1 -0.5 0 0.5 1
Axial offset (mm) 1.71
1.715 1.72 1.725 1.73
Inductance (H)
10-5 Inductance vs Axial offset
-1 -0.5 0 0.5 1
Axial offset (mm) 0.74
0.76 0.78 0.8 0.82
Coupling coefficent (K)
Coupling coefficent vs Axial offset
-1 -0.5 0 0.5 1
Axial offset (mm) 20
20.5 21 21.5 22 22.5
Quality factor (Q)
Quality factor vs Axial offset
