5G Turntable Test Rig

(1)

IN

DEGREE PROJECT DESIGN AND PRODUCT REALISATION, SECOND CYCLE, 30 CREDITS

STOCKHOLM SWEDEN 2018,

5G Turntable Test Rig

A Two Degree of Freedom System for Antenna Characterisation

MARTIN SJÖGREN

KTH ROYAL INSTITUTE OF TECHNOLOGY

(2)

(3)

Sammanfattning

Detta examensarbete har undersökt vilka komponenter och vilken kontrollstrategi som ska användas i realiseringen av ett två-frihetsgradigt roteringsbord för antenn- karaktärisering.

De viktigaste komponenterna i roteringsbordet var de två motorerna som sköter varsin rotation. För att säkerhetsställa EMC i roteringsbordet valdes borstlösa DC motorer. Regleringen av dessa motorer kontrolleras av en regelbaserad online- inställd PID regulator.

Simuleringar av den nya designen av roteringsbordet utfördes i Simulink och Matlab och visade upp lovande resultat för både positioneringen och styrningen av elektromagnetiska störningar. Från ett momentant steg-input, vilket i en vanlig kontroller genererar hög EMI, så kunde den regelbaserade kontrollern istället skapa en mjuk, och därmed EMC-vänlig signal till motorerna.

(4)

(5)

Abstract

This Master’s thesis has investigated which components and what control strategy shall be used to develop a two-degree-of-freedom turntable for antenna characterisations.

The mechanical design of the turntable was firstly implemented, on which ba- sis the electrical components were chosen. The most important components in the turntable are the two motors, each of which handles one degree of freedom. To avoid electromagnetic interference, the brushless DC motor type was chosen. These motors were thereafter designed to be controlled by a rule-based online tuned PID controller.

Simulations of the new turntable design showed promising results, in both positioning accuracy and controllability of electromagnetic interference. Under a step input of the desired position, a conventional controller would generate an instantaneous and aggressive control signal (voltage demand), i.e. high electromagnetic interference. By contrast the rule-based controller could create a smooth inclining voltage demand, thus reducing the EMI from the motors.

(6)

(7)

Acknowledgments

I would like to thank both my academic supervisor Binbin Lian, and my industrial supervisor Erik Hansson for keeping up a continuous support throughout this thesis work.

Also, I would like to thank Michael Lindahl for the mechanical aspects of realising the turntable design, as well as Venkata Sai Krishna Varma Rudraraju who helped immensely with the finishing touches upon the CAD models.

Martin Sjögren October 12, 2018

(8)

(9)

List of Figures

1.1 What a radio signal looks like and how it propagates on a 2D plane. . . . 2

1.2 An illustration of the geometry of the radio unit to be rotationally positioned. The left figure shows the radio unit from the front, and the right figure shows it from the side. . . 3

2.1 The light duty 1 DOF system from MVG. Source: Microwave Vision Group, Electromagnetic compatibility. . . 12

2.2 The light duty 2 DOF system from MVG. Source: Microwave Vision Group, Electromagnetic compatibility. . . 12

2.3 1 DOF turntable system from ETS-Lindgren. . . 13

2.4 1 DOF turntable system from Maturo. . . 14

2.5 The reference tracking idea. r_s is the step input reference, and r is the adjusted reference signal. Source: KTH course, Dynamics and Motion control. . . 23

2.6 Standard block diagram for model following. Source: KTH course, Dy- namics and Motion control. . . 24

2.7 Rearranged block diagram for model following. Source: KTH course, Dy- namics and Motion control. . . 24

2.8 The structure of a "fuzzy PID controller". . . 26

2.9 The possible propagations of EMI, from source to victim. Source: Wikipedia, Electromagnetic compatibility [1]. . . 28

2.10 Vias on a PCB connecting to ground. Source: STMicroelectronics [2]. . . 29

3.1 The 3D model of the mechanical design of the turntable. . . 34

3.2 Left: The ULN2803A transistor array used. Right: A typical transistor and its pin setup. . . 39

3.3 The encoder A and B pulses. . . 40

3.4 Final 3D model of the mechanical design of the turntable. . . 42

3.5 An illustration of the plant model. . . 43

3.6 Velocity step response on the BLDC motor with neither load nor a gearbox. 45 3.7 The closed loop structure with a general PID controller in an error feedback fashion. . . 47

(14)

3.8 Left: Step response of π [rad] of the closed loop system with the different PID controllers. Right: Their corresponding effort curves [V ]. . . 47 3.9 The set point signal next to the reference signal r with the initial parame-

ters. . . 50 3.10 Step response together with the final iteration of the Trajectory planner. . 51 3.11 The trajectory planner implementation in Simulink. . . 52 3.12 Left: The position, velocity and reference signal plot when given a set point

of π rad. Right: The controller and Model follower efforts. . . 52 3.13 Left: The position and reference signal plot when given a set point of π

rad at t = 0.1s. Right: The control effort. . . 54 3.14 The effort curves from the controller with and without the Model follower. 54 3.15 Left: The position and reference signal plot with different values of the

P-only controller. Right: The control efforts. . . 55 3.16 Left: The position and reference signal plot with the initial PI controller

and with changed integral part. Right: The control effort. . . 56 3.17 The position and reference signal plot with the initial PIDF controller and

with changed derivative part. . . 57 3.18 Left: The control efforts of each different PIDF controller. Right: A close-

up of the peak values of the different controllers’ effort curves. . . 58 3.19 Left: Two different controllers implemented with the Trajectory planner,

where one focuses on no overshoot, and the other on EMI. Right: Their corresponding control efforts. . . 60 3.20 Left: The position and reference signal plot with the final PIDF controller

with the Trajectory planner strategy. Right: Its corresponding control effort. . . 61 3.21 Left: The behaviour of the final PIDF controller next to the neighbouring

controllers changing P. Right: Their corresponding control efforts. . . 62 3.22 Left: The behaviour of the final PIDF controller next to the neighbouring

controllers changing I. Right: Their corresponding control efforts. . . 62 3.23 Left: The behaviour of the final PIDF controller next to the neighbouring

controllers changing D. Right: Their corresponding control efforts. . . 63 3.24 The feedback control loop with the online tuned PID controller. . . 64 3.25 Inside the subsystem Rules LUT. . . 64 3.26 Left: The position response and the reference point. Right: The controller

effort. . . 66 3.27 The gains of the PID controller throughout the duration of the simulation. 67 3.28 All communication and power supply cables for one entire motor package. 69 4.1 The positioning response when given a reference value of π rad for each

control strategy. . . 72

(15)

4.2 The effort response when given a reference value of π rad for each control strategy. . . 72 4.3 The control strategies’ rate of change in effort. . . 73 4.4 The built turntable with a mockup for the DUT. . . 74 4.5 The turntable from above and side, illustrating the set point positions. . . 76 4.6 The control effort plots of the rotations of a) 10^◦, b) 30^◦ and c) 60^◦. . . . 78 D.1 Pin map layout of the EM-316 BLDC motor driver from OEMmotor.

Source: OEMmotor. . . .

(16)

(17)

List of Tables

3.1 Rough parameters in the mechanical design. . . 36

3.2 Parameters of the motors. . . 37

3.3 Characteristics of the brake. . . 38

3.4 Renewed parameters of the mechanical design. . . 42

3.5 Chosen parameters in the PID designs. . . 47

3.6 The magnitudes of the conventional PID controllers’ control efforts. . 48

3.7 Initial parameters of the Trajectory planner. . . 49

3.8 Final parameters of the Trajectory planner. . . 50

3.9 The magnitudes of the PIDF controllers’ control efforts, with tuned D gains. . . 59

3.10 The two different parameter sets for the controllers that show how overshoot and EMI are trading off each other. . . 60

3.11 The parameters of the PIDF controller in the Trajectory planner simulation and a couple of neighbouring controllers. . . 61

3.12 Lookup table for the PID parameters for a certain error index. . . 65

4.1 Set points for the tests. . . 75

4.2 Verifying Motor 1’s positioning capabilities. . . 76

(18)

(19)

Nomenclature

Symbol Description

J_tot The total inertia attached to Motor 1 [kgm²] m_tot The total mass attached to Motor 1 [kg]

J_a The inertia of the antenna module [kgm²] m_a The mass of the antenna module [kg]

J_r The inertia of a motor’s rotor [kgm²] T Torque [N m]

T_e Electromagnetic torque [N m] of a DC motor

ϕ₁ Rotation angle of the turntable on the horizontal plane [rad]

ϕ₂ Rotation angle of the turntable on the vertical plane [rad]

˙

ϕ Angular velocity [rad/s]

¨

ϕ Angular acceleration [rad/s²] t Time [s]

k_T Torque constant of a DC motor [N m/A]

k_emf Back-emf constant of a DC motor [V_rms/krpm]

R_m Line-to-line resistance of a BLDC motor [Ω]

L_m Line-to-line inductance of a BLDC motor [mH]

n Gearbox ratio [−]

d_r Rotor dissipation constant for a DC motor [−]

d_l Load dissipation constant [−]

s The continuous complex frequency Laplace variable [irad/s]

z The discrete complex frequency Laplace variable [irad/s]

τ_m The mechanical time constant of a DC motor [s]

τ_e The electrical time constant of a DC motor [s]

P Proportional gain of a PID controller [−]

I Integral gain of a PID controller [−]

D Derivative gain of a PID controller [−]

(20)

Symbol Description

Tf Filter coefficient of a PID controller [−]

r_s Angular set point reference [rad]

r Angular reference [rad]

σ Transfer function pole [irad/s]

ω_n Natural frequency pole [irad/s]

ξ Damping coefficient [−]

e Angular position error [rad]

e_i Error index [−]

u Control effort [V ] Ts Sampling time [s]

Abbreviation Definition

AC Alternating current BLDC Brushless DC motor

CAD Computer-Aided Design CPR Cycles Per Revolution

DC Direct current DOF Degrees Of Freedom DUT Device Under Test

EMC Electromagnetic Compatibility EMF Electromagnetic Field

EMI Electromagnetic Interference

KTH Kungliga Tekniska Högskolan (The Royal Institute of Technology) MCU Micro Controller Unit

OTA Over The Air

PCB Printed Circuit Board

PID Proportional-Integral-Derivative (Controller) PMSM Permanent Magnet Synchronous Motor

rpm Revolutions per minute SOTA State Of The Art

(21)

Chapter 1 Introduction

In this first chapter, the reader is presented with the background for this thesis, the research problem and how the implementation can be solved.

1.1 Background

At Ericsson, the development of the 5G network, the 5th generation of cellular mobile communication technology, is well under way to make new opportunities available.

The new generation of networks will undoubtedly ensure growth for innovations in multiple areas, such as IoT, and Ericsson wants to be first with the development.

"5G technology will encompass an evolution of today’s LTE technology (4G) with the addition of new radio access technologies, often in higher frequencies." - Ericsson

5G will bring to the world faster network speeds, and most of all, low latencies.

The low latency will be useful in time critical applications, e.g. autonomous driving.

There are multiple new problems emerging with the development of 5G. The particular problem addressed by this Master’s thesis is the required technology for accurate control of the direction of transmitted or received radio signals. Because of the higher frequencies compared to 4G/LTE (and older generations), the 5G radio waves will get increasingly more difficult to manage because of its more fragile characteristics, and this is the fundamental reason for turning this problem into a thesis. The countermeasure for this adversity of weaker signals would be to, in the pre-development phase at Ericsson, make it possible to aim the signals, thus opening up the possibilities to:

• Focus signals to a specific direction

(22)

• Create a 3D illustration of the propagation of the radio signal

• Characterise antenna/radio patterns

The first and second points suggest a need for the system to be able to rotate a radio unit/antenna in both the horizontal and vertical direction, thus making it possible to paint up a 3D picture of the transmitted radio signal, see Figure 1.1 for an illustration of a radio signal on a 2D plane.

Figure 1.1: What a radio signal looks like and how it propagates on a 2D plane.

The third point suggests a need for the process to be automatic which enables fast, repeatable and reliable measurements, as many cases require high angular accuracy. The accuracy is even more evident when discussing far field measurements, where the lobes are growing thinner at increasing distances.

The stakeholder for this project, Ericsson, wants a dedicated system that is adapted for their usage of 5G to ensure accurate test results of high value. The side lobes are an example of interesting characteristics to measure, as they for some cases are a direct measure of losses in the radio transmission.

(23)

1.2 Problem Description

There are three subjects which are considered as this thesis’ problems. They are the mechanical design, electronic component choices and the control strategy. The reasons are presented below.

Existing turntables lack the mechanical ability to rotate a device around a centre apart from the centre of the turntable itself. This is the main aspect leading to the need of a new mechanical design. If the device’s sender (or receiver) is not in the centre of rotation, then it would not be possible to create a 3D illustration of the radio signal.

The radio units which Ericsson will mount on to this turntable are weighing up to 40 kg. They are rectangular cuboids and their mass centres are outside of the geometrical centre, and the sender or receiver units are as well, see Figure 1.2 for an illustration. Since the senders and receivers must be in the centre of rotation, it creates a mechanical design problem.

The exact designs of the radios are protected by Ericsson under an NDA (Non Disclosure Agreement), so they are not shown in this report. However, the important detail is that they are non symmetrical and are quite thick, which entails only more non-symmetry when tilting the radio unit.

Figure 1.2: An illustration of the geometry of the radio unit to be rotationally positioned. The left figure shows the radio unit from the front, and the right figure shows

it from the side.

Apart from the mechanical aspects, it becomes evident that thorough research has to be conducted for the new/additional components, in order to achieve a rota-

(24)

tion symmetrical around the senders/receivers of the radio unit. Since the devices which will be used for tests by Ericsson are easily disturbed by Electro-Magnetic Interference (EMI), all components must either have no sources of EMI at all, or have controllable EMI.

A 2 DOF turntable must naturally have two rotational units (actuators). Their angular positions are controlled with an external microprocessor and a power supply.

Whenever controllers are applied to a system, there are electrical signals which may become sources of heavy EMI, and since the actuators have to be in close vicinity of the radio unit, or the device under test (DUT), these cannot be allowed to emit too much disturbances.

A control signal that both stabilises the turntable and reduces EMI is therefore critical. Apart from being Electro-Magnetic Compatible (EMC), the controller have to realise the timing and accuracy requirements. The current research lacks knowledge about how EMI is caused by a controller. This particular subject is therefore studied in this thesis, in both theory and experiments.

In measurements where the transmitted signal has to travel longer distances (e.g. five or more meters), the thinner the lobe of the signal will become. Thus, high angular precision is needed in both rotations, in the order of fractions of a degree according to employees at Ericsson. It would then give a clear depiction of the entire lobe distribution, and not an otherwise low resolution picture where characteristics could easily be missed.

The radio unit’s signals may be easily affected by EMI, both from the outside and from internal components. Attenuation of these disturbances, and consequent testing depend on the EMC of the entire system. EN 55022 [3] is the standard for IT equipment electro-magnetic compatibility, and strictly followed by Ericsson in this project as well.

1.2.1 Requirements

“The system” refers to “the 2 DOF rotating table”.

What the system must be able to do:

• The system must rotate in such a way that the DUT’s sender piece will always be in the rotational centre.

• The system must have high EMC and the induced EMI must be minimum and not affect the test results.

(25)

• The system must be able to receive certain input from an operator, such as the target rotational angle.

• The system must be able to rotate 360^◦ around the vertical axis, with an accuracy of 0.1^◦.

• The system must be able to rotate 30^◦ around the horizontal axis, with an accuracy of 0.1^◦.

• The system must be able to work properly with a total weight of 100 kg on top of the turntable.

• The settling time, i.e. the time from command to the time when the system reaches the correct position, must be within 20 seconds for all cases.

• The steady state error, i.e. the error between desired and the actual angles of the radio unit, must be within 0.1^◦.

• The system must be repeatable, i.e. it must give the same result each time it gets the same command, within the specified allowed steady state error.

• The system must be robust in the sense of being able to handle different antenna modules with e.g. different weights, without deteriorating the perfor- mance.

• Regardless of the magnitude of the individual components’ effects on the total EMI, they must all be reduced as much as possible, for increased robustness.

• The two rotations must be individually controllable, thus they can be two single input, single output (SISO) systems, rather than one multiple input, multiple output (MIMO) system.

There was not an explicit requirement on the rise time of the system, which is usually an interesting term in control theory. The reason for the lack of requirement on that part was because of the fact that only the whole duration of a movement is important for repeated testing. Ericsson had agreed that as long as the response times are feasible, they are content, which in this case led to a response time of less than 20 seconds at least. Furthermore, this meant that when the simulations takes place, the more detailed requirements for rise time, settling time, overshoot, etc. can be formulated according to some logical sense in the perspective of what is feasible for this particular structure.

What the system should be able to do:

(26)

• The system should be controlled through an external unit, a GUI, from outside of the test environment.

• The system should be made modular, as to make changing of components easier.

• The system should be built to handle multiple arguments as input, e.g. a sequence of rotations with options to choose rotational speed and timings of the sequential rotations.

(27)

1.3 Research Questions

The Research questions for this thesis are formulated as follows:

• Which control strategy can be used so that the EMI is controllable and held to a minimum, while a sufficient¹ precision for radio characterisation can still be reached?

• How can the EMC control system be built with the required accuracy according to EN 55022?

1.3.1 Research and Implementation Methodology

The outline for this thesis is to gather knowledge about available technologies and components to realise this EMC focused 2 DOF turntable design. The gathering of information comes foremost from a State Of The Art (SOTA) analysis.

SOTA

The goal of the SOTA is to have enough information from existing EMC turntables from well-performing suppliers. It is therefore held as a quantitative research method which will result in a summary of the most used components in existing turntables with evaluations of their applicability to this thesis’ design. This makes the later evaluation steps involve thoroughly supported claims.

An exploratory literature review and qualitative in-depth study are also performed. It involves research of the needed components for this turntable design, and most importantly, different control strategies. Books in control theory, papers and academic lectures, as well as other projects of EMC turntables, lays foundation to the research. When studying electro-mechanical components, such as motors, they can be placed in categories of "EMC" or "not EMC", simply because of their electrical nature.

The research in this thesis shall therefore be able to produce an answer to the question of what components are needed for the realisation of the new turntable design. It shall also prove what control strategy would be the best choice of implementation, relying on their controllablity of EMI. However, if it does not give a clear answer on itself, this shall be evaluated later in simulations.

1Sufficient in the way that testing can be done multiple times with concise result.

(28)

Implementation

Apart from the research, the structure of the thesis follows the V-model [4] in its way of conducting the product development. Thus, there are continuous verification steps, to make sure that the design is always applicable and that it satisfies the requirements.

After the research and SOTA analysis, simulations are conducted to answer the first research question. A designed simulation model of the physical turntable and its dynamics, all done with Matlab and Simulink [5], is developed. A 3D model of the entire system is modelled with Creo Parametric [6], used by Ericsson.

The procedure for simulating a physical model follows some of the material from the course Dynamics and Motion Control [7] given by KTH (Royal Institute of Technology). This ensures correct and robust modelling strategies.

Making the simulation of the plant, i.e. the turntable, in Simulink visualises how some of the physical properties of the turntable change during the rotational sequences. This is important to refer to when the design of the controller takes place, so that limiting factors, such as available voltage, maximum torque rating, etc., are not violated.

After an accurate plant is designed in Simulink, the chosen control strategies are implemented in a Simulink model, depicting the whole closed loop with positional feedback, all in continuous time. When designing and simulating different controllers, dynamic behaviours such as rise time, overshoot, steady state error, and similar, are presented and compared to each other. It is also important to see if or how the EMI would be affected by the behaviour of the different controllers, since this and the accuracy are the main concerns of the control implementation stage.

To control a real turntable, the controller has to be discretised as a final step in order to be implemented on a microcontroller. Ericsson has many times used the development board Arduino Uno [8] for their purposes, and is therefore used in this thesis as well. The method for discretising a controller follows the methods of the Dynamics and Motion Control course as well as Embedded systems 2 [9], an additional course from KTH.

Verification

Lastly, this thesis verifies the model design with a real-world turntable. A full scale turntable is therefore built, based on the designed 3D model. One crucial aspect is

(29)

to analyse for EMC when it is in motion, as well as when it is at standstill between operations. Ericsson’s radio measurements will all be conducted when the turntable is at steady state. All EMI sources must therefore be located and dealt with. The accuracy requirements of the system are also verified with the real system, under multiple scenarios to ensure the robustness and repeatability. Finally, all specific actions that are made to strengthen the EMC of the system, are shown if they have any significant effect on the accuracy or other aspects of the system.

1.3.2 Ethical aspects

5G technology will satisfy our current need for telecommunications. It will bring opportunities such as autonomous driving in a mass scale, and this alone suffers from critically ethical aspects. In a near future, there will be roads full of autonomous vehicles, sending and receiving each other’s information continuously. It is extremely critical that the information sent will be received, and not lost and forgotten. In such a case, the worst scenario could lead to a collision between multiple vehicles.

Therefore, it is important that the systems handling the communications are robust, well defined, reliable, and so on. All parameters of a reliable system shall be predictable to ensure safety. It is also important to have as low latency as possible, so the received information is not old, and thus maybe too late to prevent a collision.

What this project delivers is the foundation of radio parameter defining, since it will be used by Ericsson to conduct antenna characterisations. The technology inside of the testbed system will not necessarily touch on any ethical aspects, but the results of the system will. The outline of this paragraph is to push the important message even further about the need of robustness. If the calibrations are wrong, the future development will produce errors, and the possibility of doing harm would exist.

(30)

1.4 Scope and Limitations

The scope of this project involves finding components to realise the turntable testbed system. The following list is therefore important to consider as limitations to its scope:

• Only well known components tested by others are used, to ensure that no unknown errors or disturbances to test data may occur.

• Only components without heavy EMI are considered.

• Only sensors without both heavy EMI and moving mechanical parts are considered.

• Tests are performed in Ericsson’s lab environments. No other tests outside of their facilities are conducted.

(31)

Chapter 2 SOTA analysis

In the process of developing a turntable with high EMC and low EMI characteristics, a thorough study has to be done prior to the design of the product. This ensures smarter design choices based on the State-of-the-Art.

The two terms EMC and EMI are important to fully understand. Knowing how electro-magnetic energy can disturb and damage radio signals and electrical components is important for building a turntable used for radio signal characterisation.

2.1 SOTA of Turntables

Various companies that design and manufacture turntables with focus on EMC are presented in this section. Their products have been researched for the purpose of getting an overall idea of what shall be produced from this thesis. Many of their products explicitly state that they are for antenna construction and thus they put focus on EMC. Not all of the found manufacturers are presented here, but can instead be found in the reference list as references [10] and [11].

2.1.1 MVG

A good point of reference in turntable design, given by colleagues from Ericsson, is the company MVG [12]. They manufacture high quality turntables and their customers can choose between articles according to their requirements on force and precision. Their smallest 1 DOF turntable, see Figure 2.1, can hold as much as 910 kg on top of it’s platform [12], which is significantly heavier than what this project has a need for. This implies that the mechanical requirements and high-end functionality of the turntables from MVG can be considered above the scope of this project, but still seen as a source of inspiration regarding design.

(32)

Figure 2.1: The light duty 1 DOF system from MVG. Source: Microwave Vision Group, Electromagnetic compatibility.

The 2 DOF turntables that MVG have uses a second motor that lets the entire table tilt up and down, see Figure 2.2 below. The axis of rotation is thus far beneath the surface of the table, of which the DUT (Device Under Test) will be placed upon [13]. Therefore, for Ericsson’s needs where the antenna modules have to be on the same level as the rotating axis, this mechanical design is regarded infeasible.

Figure 2.2: The light duty 2 DOF system from MVG. Source: Microwave Vision Group, Electromagnetic compatibility.

Looking more into the details of their turntable designs, the type of motor that MVG use could be found, written in one of their datasheets [14] provided at their website [12]. They state that they use a DC motor with a gear reducer and an

(33)

incremental encoder, thus making it a closed loop servo control. With their controller attached in the loop, this makes for an excellent angular precision system, gaining an accuracy of ± 0.03^◦. It is however not stated more thoroughly about what type of DC motor they have. The 1 DOF turntable which can hold 910 kg, has a motor which can generate a torque of 12 Nm and run at approximately 2.6 rpm (revolutions per minute). These numbers are good to know and relate to when the design of this thesis’ turntable will take place.

2.1.2 ETS-Lindgren

Another company who is known for their turntables with high EMC is ETS-Lindgren [15]. Looking at some of their products, see [16] for reference and Figure 2.3, they use fibre optics for their signals I/O between the motor base and controller. This takes the design further towards making it both EMC and decreasing EMI, since this ensures that it will not interfere with the test data, which conventional electrical cables would otherwise entail. Additionally, their control unit and encoder are placed in shielded enclosures, which further improves the EMC. This company is in other words a good source of how to lower EMI sources.

Figure 2.3: 1 DOF turntable system from ETS-Lindgren.

One prominent aspect of the turntable designs that they have is the flatness of the products. They state on their website that it minimises the customer’s struggle when placing the product in the intended environment. There is no need of exca- vating the ground for this product to stay as close to the floor as possible, if the height where to be limited. They also state that their products are "... virtually Radio Frequency (RF) invisible, so site attenuation characteristics are virtually un-

(34)

changed." [16].

In the datasheet [17] for the turntable presented in the figure above from ETS- Lindgren, the type of motor is not specified other than that it is a variable-speed motor with voltage supply of 208/230 VAC, i.e. the standard voltage outlet level in many countries. The top speed is however clearly specified at 2.5 rpm, and the nominal load to 272 kg.

Reflecting this design onto the one which shall be manufactured for this thesis, the velocity 2.5 rpm can be translated to 15^◦/s, which points towards it being able to conduct multiple tests within a reasonable amount of time. The weight requirements will not be as high for this turntable, just as the turntable from MVG.

It is nonetheless a good example to follow in the area of EMC, since they show that they have put much effort into reducing the EMI and shielded all components.

ETS-Lindgren could nevertheless not present any feasible 2 DOF turntables.

2.1.3 Maturo

A third company which manufactures positioning equipment (among them are turntables specifically with EMC in mind) is Maturo [18]. They have several turntables for different loads and uses, see Figure 2.4 for one example of their products. This is the most applicable one for this thesis, which has a load capacity of 100 kg [19].

Unfortunately, Maturo does neither have any applicable 2 DOF turntables.

Figure 2.4: 1 DOF turntable system from Maturo.

The presented turntable from Maturo also uses fibre optic lines for I/O signals, thus reducing their EMI. From the datasheet [20], the motor they use is presented to be a brushless DC (BLDC) motor with 200 W of power, and has a velocity range of 0.5 to 3.0 rpm. This could therefore be a good reference for this thesis’ design, since it handles similar weight and velocity constraints, while at the same time reaching good EMC. The positioning accuracy they can achieve is also presented in the datasheet, and is stated to be within ± 0.5^◦.

(35)

2.1.4 Summary of the turntable SOTA

It is shown that there exists many different turntable designs, but as the SOTA proves, there is no design which can place a device on an axis of rotation which would lead to a symmetrical rotation somewhere on the device itself.

The presented suppliers have however been able to show what components they have chosen to implement to acquire good accuracy, as well as having an EMC environment. Most of them have chosen DC motors to rotate their systems, which is something that this thesis takes into consideration in the coming research segment about the electronic components. Moreover, encoders were used by these companies to feedback the angular position.

2.2 Electronic components

A most pressing research area for the design and implementation of an EMC turntable is naturally the choice of motors. The other electronic components (motor driver, position feedback sensor, microcontroller) naturally follows the choice of motor. The type of motor that should be used in the final product depends highly on their characteristics in their EMC properties, while it can still maintain a high accuracy. The motors that the presented turntable manufacturers are using are of the DC type, but it is still arguably important that a quantitative research about different types of motors should be done before making a final decision.

Without going too far into the details of describing different motor types, these following aspects are sought after to know for this thesis:

• is the type of motor prone to induce EMI?

• what is the motor’s ability to be used as a positioning actuator?

• can this type of motor handle the load requirement of this thesis?

There are roughly speaking two types of motors: electrical and non-electrical.

Under the electrical category, there are DC and AC motors, which can both be split into more subcategories if desired. Non-electrical motors are e.g. hydraulic or pneumatic motors, i.e. using pressured fluids or air to move a rotor instead of currents powering either coils or electro-magnets [21] [22].

(36)

2.2.1 DC motors

The DC motor type is, relative to AC motors, generally known to be easier to both understand and control. There are a number of ways to categorise them according to their working conditions. DC motors can e.g. either be brushed or brushless, which corresponds to how the electrical commutation works.

Brushed DC motors are much simpler, but easily wear out due to the fact that there are physical metal brushes on the stator touching the surface of the rotor. The stator is a permanent magnet, and the rotor is an electromagnet. This mechanical interface reduces the motor’s life expectancy by a substantial amount and can even create sparks in the motor, which in itself can become a large source of unwanted EMI.

A brushless DC motors (BLDC) instead has a permanent magnet as its rotor, and a number of winding-pairs around the stator [21]. Current runs through certain windings for certain rotor positions, creating a rotating magnetic field to ensure a continuous rotation of the permanent magnet rotor.

The BLDC motor does however require the knowledge of exactly where the rotor’s north and south poles are, so the correct coils can be energised. There are two ways to know the position of the rotor: either with a sensor, or via sensorless detection. The sensor is usually a Hall effect sensor [21]. If the BLDC motor uses sensorless techniques it then uses the Back-EMF effect [23].

Another way to distinguish different types of DC motors is through the winding principle in the stator. It can either be in Shunt (parallel) or in Series. A Series DC motor has a high starting torque but decreases rapidly with increasing load. Shunt has a lower starting torque than a Series DC motor, but it does not degrade much with increasing load [24].

It is moreover quite easy to find a suitable DC motor, since there exists many in different shapes and sizes. Gearboxes are additionally used to increase the torque, at the cost of decreasing the speed.

2.2.2 AC motors

AC motors are both in their design and implementation more complex than DC motors. They do however present an easy relationship between the speed of the rotor and the AC power supply frequency. They are proportional, which means that speed control is effortlessly achieved by using an AC motor [25]. There is unfortunately no such relationship to the position.

(37)

There is research being done on sensorless AC positioning motors [26] [27], where the position can be known, with the knowledge of the electrical quantities at particular time instants, much like the back-EMF technique from the BLDC motors.

Using AC motors for position control, with e.g. induction motors or permanent magnet synchronous motors (PMSM) [22], tends to be difficult, and requires an advanced control algorithm to become reliable and accurate. One paper proposes the usage of a combination of a Sliding Mode Controller and a Linear Quadratic Reg- ulator [28] in order to get their optimal controller. Another utilises a Fuzzy-sliding controller [29] to control their system. Other research projects implement an Opti- mised adaptive sliding-mode position control [30], and another chooses an Adaptive fuzzy backstepping with input saturation position control method [31]. There are however possible implementations with traditional control strategies, see e.g. [32]

for a successful PD controller used in position control of an induction motor. The method achieves an accuracy between 0.002 rad and 0.004 rad in angular position.

STMicroelectronics [2] states an important aspect about AC motors. They generate harmonic signals at the power input and output, and thus create EMI. This has been proven to affect surrounding electrical devices and also power networks for this reason. AC drivers can both cause and be affected by this, which could only worsen the test signal data if they would be operating when radio measurements were made. It is quite clear that this is one of the reasons for the lack of AC motors in turntable design with EMC in focus.

2.2.3 Non-electrical motors

The second major type of motors are the non-electrical motors: pneumatic motor and hydraulic motor. As explained, they are driven by air pressure or fluid respec- tively. They are generally not as efficient in power as the electrical motors can be, but they make up for in power density. A pneumatic motor would be much smaller than an electrical motor with the same power output, while also producing less heat and no ability to generate sparks [33]. Since the non-electrical motor does not use electrical components, there are simply no EMI sources, other than electrical signals running through cables.

Pneumatic stepper motors have shown to have higher precision and controllability than normal pneumatic motors [34], and could therefore be used in a positioning system.

These types of motors would however need a constant supply of pressurised air or fluid. This reason alone makes these types of motors cumbersome in relation to

(38)

how little of an effort it is to supply an electrical motor. For the sole purpose of designing a turntable and make it as EMC as possible, these types of motors should however be considered.

2.2.4 Motor driver types

One of the most common ways of controlling the velocity or position of motors is with a variable voltage through a Pulse Width Modulation signal (PWM). If the motors are to be controlled by a variable voltage, a motor driver is necessary to be placed between the motors, controller and the power supply. The need for a motor driver is both because of the limited small current output of a microcontroller, as well as the more advanced logic needed to drive a BLDC or AC motor.

Special care and caution must be placed in the choice of motor driver since this can often be a substantial source of EMI in the system, according to the manual by STMicroelectronics [2]. Fast switching frequencies in an H-bridge naturally cause quick currents and voltages, and thus creates EMI.

A special circuit called snubber can be embedded into the design of a motor driver [2] [35]. A snubber is used for reducing the rate of change in currents and voltages. Two types of snubbers are the most common: the RC damping network (resistor and capacitor) and the RCD turn-off snubber (resistor, capacitor and a diode). These snubber circuits are placed across the switches to get the desired effect.

If the driver has either an insulated-gate bipolar transistor (IGBT) or a metal oxide semiconductor field effect transistor (MOSFET), there is more flexibility in tuning the trade-off between power losses and on-/off-transient behaviour of the switches through an appropriate choice of the gate resistor [2]. This entails the opportunity to slow down the switching.

Brushed and brushless DC motor drivers

A brushed DC motor and a BLDC motor cannot use the same driver. However, they can both allow a PWM signal, and hence the input to the driver is a duty cycle, which reflects on how long (0-100% of the switching frequency) the signal shall be ON. The duty cycle corresponds to the amplitude of the average voltage.

An H-bridge in the driver, usually consisting of four switches, controls the duty cycle and in which direction the current is running through the motor, to change the rotational direction.

(39)

The difference between brushed and brushless drivers is that a DC motor only requires two cords: supply and ground. A BLDC motor has a more advanced electrical commutation method and this entails a more complex driver routine. See [36] for such an example. Besides the inputs of supply and ground, a BLDC motor driver usually also has three inputs for the Hall sensors feeding back from the motor, and three outputs to each winding of the BLDC motor [37]. The Hall sensor inputs [38] are evaluated internally in the drivers and are the necessary signals to determine which windings are to be energised and not.

AC motor driver

An AC motor driver, or VFD (Variable frequency driver), can change the frequency of the input signal to the motor, thus changing the velocity accordingly [39]. Sim- ilar to the BLDC driver, an AC driver includes the logic necessary for energising the correct phases. The input is then simply the desired velocity of the motor or frequency of the power supply, depending on the driver.

2.2.5 Encoders

Encoders are widely used as positioning sensors for motors, as the SOTA analysis shows. There are three main technologies of encoders, namely: optical, mechanical and magnetic. When choosing between these types, the optical encoder technology is the only one which will not add either EMI or mechanical wear and tear.

The optical encoder usually has two squarewave pulses, often called A and B, which are typically 90^◦ (electrical degrees) apart. Knowing that, both the position and direction of the rotation can be known, with proper reading of the pulses [40].

The velocity can naturally be derived from the change in position. An absolute rotary encoder, unlike an incremental rotary encoder, has knowledge of exactly what position it is in, since it has unique readings for each position [41].

A quite common technique when using encoders is to use two encoders on the same motor [42]. One of the encoders can then be allowed to have a rather low resolution, and used only for sensing the direction of rotation. The second encoder can then read the position from only one of the usually two pulses that encoders have. The software implementation of this relieves the processor and thus it can afford having a much higher resolution encoder.

(40)

2.2.6 Choice of electronic components

Both of the motors in the turntable are decided to be DC motors. The AC motors should not be used when building an EMC turntable due to the harmonic signals which generates EMI. The BLDC type is then favoured, since it would otherwise entail unnecessary wear and tear, as well as possible EMI sources from the sparks of the brushes. The pneumatic or hydraulic motors are discarded because of implementation reasons. These types of motor would yield the best EMC results, since there are no electrical currents needed to rotate them. However, Ericsson agreed that an electrical motor is easier to implement and maintain, as well as supply with power and shield where necessary. In summary, the BLDC motors are the most optimal choice for this project.

For accuracy, and thus repeatability, the motors have to be in a servo structure, i.e. inside a closed loop control with the angular position being fed back from optical encoders. Because of what this thesis is set out to find, a pre-built servo motor could not be bought, since these have pre-built controllers inside them.

The two rotational loads are decided to be driven by the same type of motor, for the simplicity of implementation. If a higher torque is needed for one of the motors, the velocity can easily be traded away with a suitable gear box.

A brake system, rated for holding the worst case scenario of the antenna load (at peak angle), is added to the motor which will rotate the antenna. A dedicated brake system is needed because of the unwanted emissions from an otherwise running motor during the radio measurements when the antenna has to be held still at an angle. Without a brake system, the motor would have to continuously work against gravity affecting the unbalanced load, and thus have current running through which can interfere with the measurement data.

2.3 Control strategies

What the manufacturers of the turntables do not publish is their control strategy, i.e. what type of controller they implement in their products. This creates a need for this thesis to search for effective control strategies for turntables similar to ours.

Moreover, an exploratory research is needed to find what control strategies that could affect the EMI in the controller.

There are undoubtedly many different control strategies, e.g. the classical PID design, Loop shaping, H-inf design, Quantitative feedback theory, amongst many

(41)

more [43]. They all have their positive features and advantages, but are usually accompanied by some drawbacks as well. Strengths of a controller can be anything from robustness, simplicity or adaptability, while drawbacks can be cost inefficiency, complexity, narrow applicability, etc.

Nowhere can there be found research or academic topics about the EMI of a controller. The control theory problems usually put their focus on getting a faster response, reducing the overshoot, or similar. These aspects even come at the cost of stronger and faster control signals, which then only worsen the EMI.

To answer the research question about what control strategy to use for this EMC turntable, this section of the research’s focus is put on exploring the many different control strategies available today, while filtering away those who would not present a possibility of affecting the reduction of EMI.

2.3.1 Traditional PID

A PID controller contains the terms: Proportional, Integrative and Derivative gains, and is the most frequently used controller, mostly because it is possibly the most intuitive one. The parameters all represent more or less a physical characteristic behaviour [44] and therefore tuning them to achieve the desired result usually works quite well. Knowing the desired response characteristics presents many theories as well as rules of thumb to help take an initial guess on the PID parameters [44].

How a PID controller affects the system’s time and accuracy aspects is quite clear from its theory. In rough terms, the proportional gain simply increases the control signal, the derivative gain can be said to be sensitive to measurement noise, whereas the absence of an integral term may prevent the system from reaching its target value [44]. These explanations are however quite simplified. Sometimes the integration part of a PID controller leads to undesired/unstable behaviour if the control signal reaches a saturation point, thus leading to a wind-up in the integrator [45].

A PID controller is unfortunately very limited in the sense of the strict structure of the controller, see Equation 2.1 for the so called parallel form. To call it a PID controller, the parameters would have to be derivable from this transfer function, denoted here as G_{P ID}(s), where s indicates the Laplace transform variable.

GP ID(s) = u

e = P + I1

s + D N s

s + N (2.1)

The low pass filter coefficient N does not have to be implemented with every PID controller, or can sometimes be seen as T_f = _N¹ instead.

(42)

A PID controller cannot reduce the EMI much in itself. Since EMI stems from rapidly changing currents, it can be directly correlated with the control signal. The control signal is the output voltage to the motor, which depends on the measured error for conventional controllers. Since the error is instantaneously changed from 0 to the desired set point, the control signal will induce a rapid change in current.

All controllers which act directly on the error will show this EMI prone behaviour.

A PID controller is however fully capable of controlling a turntable. A paper on a control system for a small scale turntable [46] was found during the research. In this paper, a DC motor together with a traditional PID controller, containing only the proportional gain, was implemented. It was found to gain an accuracy in position of around 70-80% in an open loop implementation. There were no tests depicting what the accuracy obtained with a closed loop would be, it is only noted that it would greatly benefit from it. Conclusions can also be drawn that since this was a much smaller system (in terms of forces, weights, etc.), the usage of a proportional controller was sufficient for their needs, but might not work as well for larger scale and complex implementations. They did not consider any EMC properties in their design, so as to how the control signal looked like was not presented. It was however an example of how a simple P-controller can be sufficient to solve simple turntable positioning tasks.

If the structure of a PID controller is used, it can be easy to design the controller to achieve the desired properties. However, the issue of rapidly changing control signals must be solved somehow. Some promising control strategies are presented below.

2.3.2 Reference tracking

When a position controller is wanted, a Reference tracking control strategy might be useful. It involves a Trajectory planner and a Model follower. The Trajectory planner adds smoother characteristics of the overall system behaviour, as well as prohibiting the control signal to saturate. What it means to design a Trajectory planner is ultimately to change the reference signal fed to the controller, deforming it into a smoother control signal out to the plant, instead of an infinitely quick step response. The following segment explains this control strategy with the help of parts of the material from the course Dynamic and Motion Control [47], given by KTH.

Originally, this control strategy is used to tailor-design a controller to known limitations of the plant. Such limitations can be the maximum acceleration and velocity allowed by the actuator, or forces and torques. This is partly done through

(43)

changing the reference signal to something more complex than a step. What this thesis found interesting with this strategy is the fact that a smooth incline of the reference signal, see Figure 2.5, will in turn mean a smoother, less erratic electrical control signal, i.e. less EMI. Since this particular reason is not discussed when talk- ing about Reference tracking, this thesis researches if this can be used to control the EMI of the system, which the result of is shown in the implementation chapter through simulations.

Figure 2.5: The reference tracking idea. r_s is the step input reference, and r is the adjusted reference signal. Source: KTH course, Dynamics and Motion control.

This method might entail somewhat slower dynamics depending on the design of the reference trajectory. When designing the Reference tracking, it must be done with this in mind, since it can ultimately become a direct trade-off between response time and a smoother reference signal.

One part of this method is the Model following control block, seen in the Simulink example in Figure 2.6. The ideal goal is to have a perfect follower of the model, i.e.

having the Model following control block, called G_m(s), be the inverse of the plant, since if

G_m(s) = G_p(s)⁻¹ (2.2)

the model following would be perfect: G_p(s)⁻¹G_p(s) = 1. The plant would have to be fully known and predictable if this should become a truly perfect model

(44)

follower, i.e. the quality of the model follower is directly dependant on how accurate the plant can be put into mathematical terms. Moreover, the Model following control block will be improper if it would simply be the inverse of the transfer function of the plant [47].

Figure 2.6: Standard block diagram for model following. Source: KTH course, Dynamics and Motion control.

Rearranging the block diagram would solve the second issue. If the new control design would be arranged as Figure 2.7 depicts instead of Figure 2.6, no blocks are improper, as long as G_m(s) is of higher degree or equal to G_p(s).

Figure 2.7: Rearranged block diagram for model following. Source: KTH course, Dynamics and Motion control.

However, notice that Reference tracking is a control structure, and not an actual controller. The control block, which is cut off in Figure 2.7, can be any type of controller, e.g. a PID controller. Implementing a Reference tracker structure can entail the control of how the velocity and acceleration trajectories look like. The torque could therefore be designed to never go beyond the maximum torque of the motor. Torque is proportional to acceleration, which in turn is proportional to the current in the motor. So to control the acceleration would mean that control of

(45)

the control signal is possible, thus controlling the EMI. This is why the Reference tracking strategy is of high interest for this thesis.

2.3.3 Neural Networks

Many published papers on turntable control design were found implementing a PID controller together with an Artificial Neural Network (ANN). One of these [48]

used the conventional error feedback structure for their closed loop system, and implemented an improved BP (Back Propagation) Neural Network and let that tune the PID parameters online to suit their simulated models of a turntable. They state that it helped them achieve a more ideal controller than what traditional manual tuning would accomplish. The results that they achieve are promising, such as a good rise time, and thus they show the positive effects of using a Neural Network implementation to tune a PID controller online, i.e. while it is acting on a plant.

They further claim that it is because of difficulties developing a precise mathematical model for their turntable servo structure that the Neural Network made such an impact on the results.

This paper does not take into consideration the EMC of the system, but it does show that designing an ANN to try to reduce rapid changes in control signals might be something that can be done. This would in that case be a direct way to reach lower EMI in the system. A more straight forward approach to do just this was encountered in the research, called Fuzzy logic.

2.3.4 Fuzzy logic

The search results for PID structures with implementations of "Fuzzy logic" control can also be found in abundance during the research. It shares similarities with ANNs in the way that they can both behave differently to changing scenarios. They however differ in the sense that a Fuzzy controller has something called static, pre- defined rules it follows [49]. These rules receive the input(s) and, through the logic specified, it presents the output(s) according to these rules. This is implemented conveniently with a look-up table (LUT).

A conference paper [50] found during the research implemented what is by them called a "fuzzy PID controller" using Matlab’s application, the Fuzzy Logic Toolbox [49]. They made a design of a self-tunable PID controller with the output error and the error rate of change as input parameters, in order to tune the PID parameters (K_P, K_I and K_D) as output. An illustration of the structure from that paper is presented in Figure 2.8.

(46)

Figure 2.8: The structure of a "fuzzy PID controller".

Using a PID structure makes it easier to understand how the different parameters affect the resulting behaviour of the system. This way, rules and guidelines for how the toolbox shall decide the PID parameters upon different variants of the inputs can be established quite intuitively.

This method shows the potential of being able to have both accurate and fast responses, as well as keeping it EMI reduced. If implemented correctly, the behaviour can be tuned as however to satisfy the requirements. This thesis then wants to know if EMI can be significantly reduced using these rules for when what parameter in the PID structure increases or decreases in the controller.

2.3.5 Comparing control strategies

The factors that play a part in deeming a control strategy either promising or not for this thesis can be summarised as the following criteria:

• The possibility of affecting the EMI by reducing it.

• The possibility to tune the system behaviour, such as rise time, control signal limitation, etc.

• The applicability for controlling a heavy turntable structure in a smooth motion.

PID controllers have already proven themselves to be applicable in more than just turntable applications. The structure of a PID is good to adopt simply because of the easiness it presents in tuning the different control parameters. It makes it intuitive to get the desired behaviour of the system response, in terms of rise time, steady state error, etc. This made the PID a suitable structure to conduct simulations with.

(47)

A conventional controller in an error feedback loop would give aggressive signals when the error suddenly presents itself. This would entail heavy EMI to the system, and thus, not being desirable for this thesis. Control strategies were therefore only presented in these previous sections if they showed a possibility to affect the EMI.

Other strategies researched worth mentioning are e.g. the Pole Placement strategy [45] and the Loop shaping [51]. The Pole placement strategy can manipulate system characteristics, such as the rise time, overshoot, damping, etc., if the closed loop poles can be placed with enough freedom [52].

Loop shaping revolves around the Sensitivity function and the Complementary- Sensitivity function [51], and takes uncertainty into consideration, such as disturbances and noise entering the system. This method is often used for systems where robustness is critical [53]. If the system to be controlled is unknown to a point that it could ruin the stability, this method could ensure continuous stability, since this strategy involves shaping the system towards higher stability margins.

Neither of these two strategies present any EMI controllability however, and are therefore discarded from moving further with in this thesis, along with many other control strategies.

The Fuzzy logic and the Trajectory planner methods showed the most promise to control the EMI. However, in the Fuzzy logic strategy, it is the PID parameter rules that is of interest in this thesis. Both a Trajectory planner strategy and a rule-based controller will therefore be implemented into simulations, in order to explore their EMI controllability.

The rule-based controller is suitable since it can inherit the traditional PID control structure, while the controller’s parameters would be dynamically tuned online as the positional error changes. It is therefore interesting to know if this method can slow down the electrical signals feeding the motors, through adapting the controller to the thesis’ specific needs.

The Trajectory planner similarly shows the possibility to affect the overall EMI, through changing the reference signal from an instantaneous step input, into a smooth inclining curve. This shows promise since a smooth reference signal would mean that the error would also be smooth, and thus, the control signal could become less aggressive. Any control structure could theoretically be applied with the Trajectory planner, however the PID control structure entails much simplicity, as previously stated, and is therefore chosen to be implemented with this strategy as well.

(48)

2.4 EMC optimising and EMI reducing

Radio signals, especially for 5G, can be disturbed, distorted or even destroyed if too much interference is present [54]. Knowing that the origin of EMI is rapidly changing currents in electronic components and cords, a literature review collected the needed knowledge of how to reduce EMI and improve the EMC of the turntable during its implementation. This provided the thesis with multiple rules and guidelines [2] on how to prevent EMI sources early in different implementation stages.

2.4.1 EMI propagation

The most fundamental knowledge about EMI is that it propagates (couples) through multiple media [55]. In a turntable for handling 5G antenna measurements, the following propagations are possible:

• Conductive coupling

• Capacitive coupling

• Radiative coupling (hot loops) - can also come from harmonics of the original signal

• Magnetic (inductive) coupling

See also Figure 2.9 below for an illustration, where the victim in this thesis is the 5G radio unit.

Figure 2.9: The possible propagations of EMI, from source to victim. Source: Wikipedia, Electromagnetic compatibility [1].

(49)

Both (electric) motors and circuit boards generate Electro-Magnetic Forces (EMF) which can be coupled through all of these different paths. The following sections will explain how, and how to reduce these effects.

2.4.2 Ground design on a PCB

Products with Printed Circuit Boards (PCB) can require extra care to their ground design. To practice EMC ground design, the paths that the current takes to reach the ground level should be as short as possible [2]. Having so called ground grids across the PCB, or even dedicated ground planes enables some sought after effects:

• it reduces common impedance coupling and lets the return current flow nearer to the source current.

• it reduces the effect of common impedance by having much lower self-inductance and resistance, with respect to ground traces.

By dedicating an entire layer of a PCB to ground signals ensures that all signals will take their “shortest path”, or more correctly put: the lowest impedance path, thus decreasing the overall power consumption which will in turn lead to less EMI [55].

Ground paths between different layers on a PCB are called vias, see Figure 2.10.

These shall form periodic connections to the actual ground plane as often as possible. Incoming disturbances can hit any point of a PCB where it will want to travel through to ground. If there are components between the point of impact and the ground level, they might be defected or even destroyed. This risk reduces with more frequent paths to the ground plane.

Figure 2.10: Vias on a PCB connecting to ground. Source: STMicroelectronics [2].

Ground loops, or “hot loops” is an effect from a PCB producing its own unwanted radiated emissions. These hot loops can occur when the signal path lengths are ex- cessive, or too narrow. They can be found physically in the design of the circuits by following the current’s paths and noticing that they move in something similar to a

(50)

loop. If currents are flowing through such loops, they can radiate energy to other components, much in the same way as how a coil works, and possibly destroy the most sensitive components in the process. External magnetic fields inducted into the loop can cause fluctuations of the true reference potential (i.e. the ground level) and common-mode current can then flow into the circuit [55].

Placing sensitive and high frequency tracks away from high-noise power tracks (current sensing, fault signals and protections), avoiding the use of wire jumpers and minimising layer transitions, will improve the overall functionality of the PCB.

Where necessary and possible, keeping the same number of vias on each signal track can also improve the EMC [2].

2.4.3 Shielding

Apart from the usage of fibre cables, shielding the electrical cables are the next best option. What shielded cables refer to is that signal wires are surrounded by an outer conductive layer which is grounded at either one or both ends. If cables are not shielded properly, this can be where interference signals are accessing the rest of the system.

A shielded housing is something that is surrounding either a sensitive component or a component that is prone to radiate signals itself. The housings are of a conductive metal material that acts as a shield from the interference signals.

The motors and their additional electronics will have to be encased in these shielded housings in the end product.

Decoupling or filtering [55] at critical points such as cable entries and high-speed switches is sometimes crucial. Filters such as low pass filters should always be used where the high frequency disturbance signals are produced. Decoupling is a method used to separate different areas on a PCB.

2.4.4 Software design to improve EMC

Since the sudden changes of current is the source of EMI, some actions can even be performed when designing the software.

Turning off external devices (peripherals) when not in use will lower the power consumption, which decreases the electric field strength, resulting in lower emissions.

Common software actions to reduce EMI can be to use interrupts instead of polling when applicable, to save power. Terminating pins not in use with the pull-

(51)

up/down resistors, usually found in most microcontroller units (MCU), will also save power. If the rise and falling times of signals are deliberately extended, the rate of change in current will lower, hence the emissions as well.

(52)

5G Turntable Test Rig

5G Turntable Test Rig

A Two Degree of Freedom System for Antenna Characterisation

MARTIN SJÖGREN

Sammanfattning

Abstract

Acknowledgments

Contents

List of Figures

List of Tables

Nomenclature

Chapter 1 Introduction

1.1 Background

1.2 Problem Description

1.2.1 Requirements

1.3 Research Questions

1.3.1 Research and Implementation Methodology

1.3.2 Ethical aspects

1.4 Scope and Limitations

Chapter 2

SOTA analysis

2.1 SOTA of Turntables

2.1.1 MVG

2.1.2 ETS-Lindgren

2.1.3 Maturo

2.1.4 Summary of the turntable SOTA

2.2 Electronic components

2.2.1 DC motors

2.2.2 AC motors

2.2.3 Non-electrical motors

2.2.4 Motor driver types

2.2.5 Encoders

2.2.6 Choice of electronic components

2.3 Control strategies

2.3.1 Traditional PID

2.3.2 Reference tracking

2.3.3 Neural Networks

2.3.4 Fuzzy logic

2.3.5 Comparing control strategies

2.4 EMC optimising and EMI reducing

2.4.1 EMI propagation

2.4.2 Ground design on a PCB

2.4.3 Shielding

2.4.4 Software design to improve EMC