methods for active manipulation of fluid...

DEGREE PROJECT, IN , SECOND LEVELMECHATRONICS

STOCKHOLM, SWEDEN 2015

Methods for active manipulation offluid droplets

LUDWIG LINDEBRATT ERIK SEGERSTÉEN

KTH ROYAL INSTITUTE OF TECHNOLOGY

INDUSTRIAL ENGINEERING AND MANAGEMENT

Master of Science Thesis MMK 2015:72 MMA 515

Methods for active manipulation of fluid droplets

LUDWIG LINDEBRATT

ERIK SEGERSTÉEN

Approved

Examiner

LEI FENG Supervisor

BENGT ERIKSSON Commissioner

MYCRONIC AB Contact person

GUSTAF MÅRTENSSON

Abstract The deposition of functional materials utilising a jetting head has a large number of applications as can be seen by the growing interest in rapid prototyping. The ability to jet with precision and accuracy for high throughput application is at the core of the jetting technology. If this can be achieved at increasingly large altitudes over the substrate an increased portion of the business application space can be addressed, such as the ability to jet with smaller droplets, the ability to jet on complex PCB configurations, et cetera. A state of the art study is included in this project, which describes different potential methods that could be used for active manipulation of fluid droplets. In this project, two methods have been investigated for active manipulation of fluid droplets that could be implemented on the MY600 Jet Printer. The first method is based on using a converging air flow to focus the droplets towards the centre of a tube and thus reduce error positioning. From the simulations and analyses performed, it was concluded that this method was not suitable for implementation on the MY600 Jet Printer due to the excessively high air flow speed required to correct any misalignments. The second method investigated uses an electrostatic field to affect the positioning of charged droplets. From the performed experiments, it was concluded that the droplets can be affected by the electrostatic field, but that active charging of the droplets is necessary to achieve an adequate force required to correct any misalignments within a reasonable height.

Examensarbete MMK 2015:72 MMA 515

Metoder för aktiv manipulering av vätskedroppar

LUDWIG LINDEBRATT

ERIK SEGERSTÉEN

Godkänt

Examinator

LEI FENG

Handledare

BENGT ERIKSSON Uppdragsgivare

MYCRONIC AB Kontaktperson

GUSTAF MÅRTENSSON

Sammanfattning Den ständigt växande marknaden för kretskortstillverkning ställer allt högre krav på leverantörer när det kommer till tillverkningshastighet och noggrannhet. Detta är en stor utmaning för Mycronic som ständigt måste förbättra precisionen på sina produktionsrobotar för att behålla och eventuellt stärka sin position inom branschen. Genom att öka höjden från vilken skotten skjuts ifrån, och samtidigt bibehålla precisionen, skulle man kunna öppna nya dörrar för att till exempel minska volymen på skotten för att träffa allt mindre paddar eller skjuta lodpasta på komplicerade kretskort med färdigmonterade komponenter. Under projektet utfördes en undersökning om vilka metoder som idag används för att styra och manipulera droppar. Undersökningen resulterade i idéer på hur problemet kan lösas, vilka presenteras i rapporten. Två metoder har studerats och för att ta reda på om de kan bidra till en ökad precision för MY600 Jet Printer. Den första metoden baseras på ett fokuserande luftflöde där man påverkar kroppens rörelse med ett riktat luftflöde. Uträkningar och simuleringar visade att kraften, som luften verkar med, inte räcker till för de lufthastigheter som var realistiska att implementera. Den andra metoden baserades på fokuseringen av uppladdade droppar med hjälp av elektriska fält. Denna metod uppvisade positiva resultat. Efter ett flertal tester, med olika vätskor, drogs slutsatsen att det skulle vara möjligt att implementera en prototyp där dropparna laddas upp för att sedan fokuseras av ett elektriskt fält inuti en cylinder och på så sätt förbättra noggrannheten.

Terminology

Acoustophoresis - The induced motion of particles subjected to an acoustic field.

COMSOL Multiphysics - Software for simulating physics-based problems.

Faraday pail - A metal container used to register a charge by electrostatic induction.

Lyophobic - A liquid repellent substance.

Printed Circuit Board - A board that mechanically supports and electrically connects elec-tronic components.

Proof of Concept - A proof of concept is a realization of a certain method or idea to demonstrateits feasibility whose purpose is to verify that some concept or theory has the potential of beingused. A proof of concept is usually small and may or may not be complete.

Purgel - A fatty gel substance that is used to lubricate the MY600 ejectors.

Quad Flat No-lead - A surface mount IC that physically and electrically connects to the PCBwithout extended pins.

Quad Flat Package - A surface mount IC that physically and electrically connects to the PCBwith pins in the shape of gull-wings.

Satellites - Small droplets that are separated from the main deposit of jetting fluid.

Slew Rate - A measure of how fast the output voltage can change over time, [V/s].

iv

Abbreviations

ESD Electrostatic discharge

PCB Printed Circuit Board

Px Pixels

QFN Quad Flat No-leads

QFP Quad Flat Package

Re Reynolds number

SMC Surface Mount Components

SMT Surface Mount Technology

THC Through-Hole Components

THT Through-Hole Technology

TRL Technology Readiness Level

v

Acknowledgements

We would like to give special thanks to our supervisor at Mycronic, Gustaf Martensson, for hissupport and mentoring throughout this project. Our appreciations and gratitude also goes to therest of the team at R&D, for their technical assistance.

vi

Contents

1 Introduction 11.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Problem Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.3 Thesis Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.4 Thesis Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.5 Thesis Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.6 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 State of the Art 72.1 Swirling Air Flows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72.2 Soldering Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82.3 Electrostatic Forces to Deflect or Focus Droplets . . . . . . . . . . . . . . . . . . . 112.4 Droplet Charge Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132.5 Acoustic Focusing with Acoustophoresis . . . . . . . . . . . . . . . . . . . . . . . . 15

3 Analyses, Models and Simulations 173.1 Force Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173.2 Air Flow Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203.3 Electrostatic Field Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

4 Experimental Setup 294.1 Test Equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294.2 Analysis Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

5 Prototype - Electrostatic Focusing System 355.1 System Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

6 Results 376.1 Models and Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 376.2 Proof of Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 436.3 Prototype . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

7 Discussion 487.1 Air Flow Models and Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 487.2 Force Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 487.3 Proof of Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497.4 Prototype . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 507.5 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 517.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

8 Appendices 54

A - Graphs of the Results from Section 6.2 55

B - Risk Analysis 70

C - TRL 71

D - Image Analysis Matlab Code 73

E - Graphs Matlab Code 75

vii

List of Figures

1.1 A photograph of a) a shot from the jetting head is shown and b) the jetting head. 1

1.2 The MY600 Jet Printer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.3 Schematics of a PCB with SMT (top) and THT (bottom) [1]. . . . . . . . . . . . . 2

1.4 Graph showing the weight and volume for different SMC and THC components [1]. 3

2.1 A model of the air swirling nozzle with all its characteristics, e.g. the injection angle,θ, for the compressed air [2]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.2 The flow inside the nozzle with different injection angles, where (a) θ = 35◦; (b)θ = 45◦; and (c) θ = 55◦ [2]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.3 Common issues with screen printing, where dry joints occur when using thick sten-cils, lean joints when using thin stencils and a component distance that is too largefor stepped stencils [3]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.4 A pie-chart showing the percentage of related defects due to the printing process,where solder placement in form of opens, insufficients and bridges are the majordefects. [4]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.5 The issues with screen printing is solved with jet printing, where the right amountof solder paste is given to each component through the ability of controlling thedroplet volume [3]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2.6 Volume histograms showing the deposit repeatability for jet printing versus screenprinting [5]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2.7 A comparison of the volume standard deviation, σV, for jet printing versus screenprinting [5]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.8 Two examples of solder paste application on PCBs for a) jet printing and b) screenprinting evaluated in a solder paste inspection device, where the upper figures arecomponent 1 (QFP 144) and the lower are component 2 (0.5 mm QFN) [5]. . . . . 11

2.9 By introducing a high voltage probe near the droplet stream it is possible to deflectthe droplet stream by introducing a voltage [6]. . . . . . . . . . . . . . . . . . . . . 12

2.10 The continuous droplet stream is collected in the trap and recycled when the highvoltage probe is off. The droplets only end up on the substrate when the high voltageprobe is turned on [6]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.11 Satellite droplets are deflected and pulled off of the main stream using the highvoltage probe [6]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.12 Schematic of the set-up used by Orme et al. [7]. . . . . . . . . . . . . . . . . . . . . 13

2.13 The experimental setup of the electrostatic charge evolution arrangement, containinga syringe as a droplet generator, a high voltage charging ring, four cylinder sensorsand a Faraday pail to detect the droplet charge [8]. . . . . . . . . . . . . . . . . . . 14

2.14 Theoretical prediction of the droplet charge evolution during the time it falls throughthe sensors [8], where the decreasing time for each cylinder depends on the increasingvelocity of the droplet. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.15 The conclusions in (b), that shows the theoretical prediction in comparison withthe experiments, proves that the charge remains the same and (a), which shows thedroplet charge evolution with respect to the liquid resistivity, proves that the chargestayed in the same order independent of liquid characteristics. . . . . . . . . . . . . 15

2.16 A sketch of particle focusing in a microchannel with acoustic resonant elements fromCai et al. [9]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

3.1 Acting forces on the droplet where mg is the gravitational force, FDx and FDy arethe drag force in x- and y-direction, and Freq is the applied force from the electricfield. h is the height between the ejector’s nozzle and the PCB, σ is the positioningerror and α is the departure angle of the droplet. . . . . . . . . . . . . . . . . . . . 17

ix

List of Figures

3.2 Comparison of the high and low Re approximations of the drag forces from Eq.3.5and Eq.3.3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

3.3 Comparison between the air flow velocity needed to realign a cylinder shaped dropletand a sphere shaped droplet. Both droplets had a volume of 2.1 nL, a speed of 20 m/sand a misalignment angle of 1.76◦. . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

3.4 The drag force acting on a) a sphere and b) a cylinder with the same volume. . . . 20

3.5 Sketch of a swirling air flow inside a straight cylinder and the proof of concept model,with 4 inlets with an incidence angle of 45◦, created in COMSOL Multiphysics. . . 21

3.6 Sketch of a swirling air flow inside a conical cylinder and the proof of concept model,with 4 inlets with an incidence angle of 45◦, created in COMSOL Multiphysics. . . 22

3.7 Sketch of the perforated cylinder model with the air inlet at the top of the outercylinder, side view to the left and top view to the right. . . . . . . . . . . . . . . . 23

3.8 A model of the air flow in the perforated cylinder model with 4 sets of 8 inlets in acircular formation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

3.9 A sketch of a droplet with charge q travelling through an electric field, E, with avelocity, v, over a distance h. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

3.10 A cylindrical tube with height h, radius b, symmetry axis z, and an applied voltageV , where a is the distance from the centre. . . . . . . . . . . . . . . . . . . . . . . 26

3.11 The electric field, E, inside a tube that acts on a charged droplet. . . . . . . . . . 27

3.12 CAD images of the mount used to attach the solenoid to the ejector. The solenoidwas fitted in the small hole in the centre of the mount. . . . . . . . . . . . . . . . . 27

4.1 Sketch of how the test reg is set up, seen from ejector’s perspective. . . . . . . . . 29

4.2 Sketch of how the test rig is set up with the internal distance, l, and height fromthe nozzle, h, seen from the perspective of the camera. . . . . . . . . . . . . . . . . 30

4.3 Electric field test module with 3D-printed plastic mounts and thin copper platesmounted on an adjustable track. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

4.4 The test rig with a recording camera, a flash and an ejector system, all connectedto a computer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

4.5 Representative images from the analysis. a) is the grey-scale image taken by thecamera and loaded into MATLAB, b) is the image adapted from grey-scale to a clearblack and white binary image, c) is the image inverted and d) is the final image withthe boundaries and geometric centre ( * ) displayed for three different object. . . . 32

5.1 Design idea of the electrostatic focusing prototype, not to scale. . . . . . . . . . . . 35

6.1 Plot showing velocity contours and streamlines for the swirling air flow model witha straight cylinder. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

6.2 Plot showing velocity contours and streamlines for the swirling air flow model witha conical cylinder. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

6.3 Three-dimensional trajectory plots of the swirling air flow model with a straightcylinder. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

6.4 Three-dimensional trajectory plots of the swirling air flow model with a conicalcylinder. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

6.5 Velocity contour plots for the perforated cylinder model, test 1, 4x8 holes. . . . . . 39

6.6 Three-dimensional trajectory plot of the perforated cylinder model, test 1, 4x8 holes,xz-plane to the left and xy-plane to the right. The droplet, the red dot, movestowards the centre of the cylinder. . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

6.7 Velocity contour plots for the perforated cylinder model, test 2, 4x4 holes. . . . . . 40

6.8 Three-dimensional trajectory plot of the perforated cylinder model, test 2, xz-planeto the left and xy-plane to the right. . . . . . . . . . . . . . . . . . . . . . . . . . . 41

6.9 Velocity contour plot of the perforated cylinder model, test 3. . . . . . . . . . . . . 41


6.11 Velocity contour plot of the perforated cylinder model, test 4. . . . . . . . . . . . . 42

x

List of Figures



6.14 A histogram of the distribution of volume for the droplets in test case 1, with u =0 V. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

6.15 A histogram of the distribution of volume for the droplets in test case 2, with u =+3600 V. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

xi

List of Tables

3.1 The boundary conditions for the force analysis of the droplet. Where v0 is a typicalspeed of the droplet in the MY600 Jet Printer. . . . . . . . . . . . . . . . . . . . . 19

3.2 Test specifications that applies for all models. . . . . . . . . . . . . . . . . . . . . . 213.3 Test specifications developed for the straight cylinder model, on the right hand side

of Figure 3.5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223.4 Test specifications developed for the conical cylinder model, on the right hand side

of Figure 3.6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223.5 The created models and simulations for the perforated cylinder. . . . . . . . . . . . 243.6 Test specifications for the perforated cylinder model. . . . . . . . . . . . . . . . . . 24

4.1 Tests cases performed in the test rig, where the distance l and height h can be seenin Figure 4.2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

6.1 Test case description for all 13 cases with respect to fluid, distance, l, voltage, u,and height, h. See Figure 4.2 for reference to distance l and height h. . . . . . . . . 44

6.2 The results of the mean volume, the median volume, the mean x-position and thedeviation from zero electric field for the tests cases performed in the test rig. . . . 44

6.3 Boundary cases when shooting with MY600 Jet Printer. . . . . . . . . . . . . . . . 47

7.1 Risk analysis for the electrostatic focusing system. . . . . . . . . . . . . . . . . . . 50

xiii

1 Introduction

This thesis discusses the possibility of implementing an alignment mechanism in order to increasethe positioning accuracy of the droplets ejected using the MY600 Jet Printer. Chapters 1 and 2present the current state of the electronics production industry and the importance of increasingthe positioning accuracy of the MY600 Jet Printer.

1.1 Background

This section will describe the background of the project and how it is defined. The deposition offunctional materials utilising a jetting head has a large number of applications as can be seen bythe growing interest in rapid-prototyping. The ejection of fluid from a jetting head, see Figure 1.1,is dependent on a number of physical properties, such as the viscosity of the fluid, the geometricdescription of the jetting head and the specifics of the jetting actuation. The ability to jet withprecision and accuracy is at the core of the jetting technology. If this can be achieved at increasinglylarge altitudes over the substrate an increased portion of the application space can be addressed,such as the ability to jet with smaller dots.

Figure 1.1: A photograph of a) a shot from the jetting head is shown and b) the jetting head.

1.1.1 Mycronic AB

Mycronic AB is a Swedish high-tech company engaged in the development, manufacture and mar-keting of innovative production equipment to the electronics industry. The company has twobusiness areas, pattern generators and surface mount technology (SMT) equipment. The pat-tern generators include mask writers and direct writers, whereas the SMT equipment includespick-and-place and jet printing solutions.

1.1.2 MY600 Jet Printer

The MY600 Jet Printer, seen in Figure 1.2, allows SMT producers to achieve high-precision solderjoints. The machine has an acceleration of 3g and a deposition speed of more than one milliondots per hour, making it extremely fast. Since the foundation is a 2000 kg granite module and thebeam is made out of carbon fiber, accuracy at the micrometer level can be achieved at full speed,1 m/s. The ability to achieve both volume and positioning accuracy, as well as repeatability, is ofprimary importance for the application, and something that the customers highly value [3].

1

Chapter 1. Introduction

Figure 1.2: The MY600 Jet Printer.

1.1.3 THT - Through-Hole Technology

Through-hole technology is the technique of producing printed circuit boards (PCB) with com-ponents that are entered through drilled holes in the PCB and soldered onto the copper pads onthe opposite side which completes the circuit. This differs from SMT, where the components aresoldered right on to the copper tracks. In Figure 1.3, the two ways of mounting components on toPCBs can be seen.

Figure 1.3: Schematics of a PCB with SMT (top) and THT (bottom) [1].

Since THT requires that all components go through the PCB, this method is limited to a maximumof two copper layers, one top and one bottom layer. However, when it comes to SMT, multiplelayers may be utilised, which not only offers you more space on the PCB, but also improves therobustness of the PCB since a layer can be used for only one thing, e.g. ground or power plane.

1.1.4 SMT History

In the mid-1960s, the first platforms for SMT emerged. This was due to the advantage of beingable to mount components on both sides of PCBs. It was not until the end of the 1970s that SMTmade its revolutionary change in the electronics industry. The main reason for this revolutionwas the fact that THT encountered increasing difficulty in meeting the constant need for higherdensities, mostly due to increasing cost for drilling additional holes. By mounting the components

2


directly on the PCBs surface, instead of drilling holes, SMT allows automation in a much largerscale, cost-effectiveness, lowers weight, and raises circuit density with improved performance. Agraph showing the weight compared to the volume of through-hole components (THC) and surfacemount components (SMC) can be seen in Figure 1.4. It is quite clear from this that THC is notused in modern electronic designs since they become too big and too heavy when the pin-numberof the components increases [1].

Figure 1.4: Graph showing the weight and volume for different SMC and THC components [1].

Another benefit, with SMT, due to its small size and weight, is that shipping costs are lower, butthe storage space also becomes smaller compared to THT.

1.2 Problem Definition

The ability to ensure accurate positioning for shots ejected from larger distances from the PCBwould enable a larger array of jetting applications. The objective of the project is to test thefeasibility of focusing droplets of fluid ejected from the nozzle of the jet printer’s ejector usingdifferent external forces. The purpose of utilizing this strategy is to ensure accurate positioningfor shots of solder paste from a larger height above the PCB. A larger jetting height would enableshots of solder paste to be placed on PCBs with pre-mounted components, as well as decreasing oursensitivity to height variations due to irregular flexible boards. At present, it has been establishedthat positioning accuracy decreases with increasing jetting height. The goal of this project is toutilise a converging air flow in order to focus solder paste deposits accurately on to a substrate.Currently, the distance between the PCB and the jetting nozzle is 650µm and the goal of thisproject is to investigate a method that can be used to shoot solder paste from at least a height of2 mm above the PCB and improving the accuracy for small deposits shot from a height of 650µm.At the end of the project, a report, simulation models of different air flows, a proof of concept ina larger scale and a prototype in correct scale for the MY600 Jet Printer will be produced.

1.3 Thesis Scope

The overall aim of this thesis is to provide a method that can be used to align the travelling dropletejected from the MY600 Jet Printer so that the droplet is focused toward a target position thusincrease the positioning accuracy of the jet printer. The following limitations in regard to thecontents of this thesis have been developed:

• The emphasis is placed on methods that will work for the MY600 Jet Printer with appropriatesolder paste. There are many methods that can be used to focus particles. However, sincethis thesis is focused on the MY600 Jet Printer, not all methods are of interest.

3


• Mycronic wanted to investigate if it is possible to utilise air flows in order to focus the dropletsfrom the MY600 Jet Printer. Small experiments with air focusing had been performed, butthey were inconclusive, but according to calculations made prior to this project it could bepossible. This is the main reason why the first method investigated was the use of air flows.

• Due to the complexity of the MY600 Jet Printer, an existing test rig will be used for thisproject. This test rig is connected to a computer with control software. Small modificationscan be made to the test rig, but due to the complexity of the machine major modificationswill not be made.

1.4 Thesis Objective

The goal of this project is to investigate the possibility of utilising an external force in order tofocus solder paste deposits accurately on a substrate. Based on the problem definition and thescope of this thesis, a main research question and hypothesis was defined, as well as some questionsand hypotheses derived from the main research question.

Main Research Question

Can an alignment mechanism (air guidance, electrostatic forces et cetera) be implemented inthe jetting system of the MY600 Jet Printer to cause an advantageous positioning correctionof the travelling droplet, such that a 2◦ misdirection angle is corrected within a height of h≥ 0.65 mm and a volume of 1 ≤ V ≤ 5 nL, and a height of h ≥ 2 mm and a volume of 5 ≤V ≤ 15 nL?

Main Research Question Hypothesis

An alignment mechanism can be implemented on the MY600 Jet Printer. With larger vol-umes it will be more difficult to affect the droplet, but an increased height will provide alonger distance and more time to affect the droplet.

Derivative Research Question

Is it possible to achieve a positive effect on the positioning accuracy of a droplet of fluidthrough the utilisation of a guiding air flow from a controlled mechanical system?

Derivative Research Question Hypothesis

A positive effect on the positioning accuracy of a droplet of fluid can be achieved through theutilisation of a guiding air flow from a controlled mechanical system. It will depend on thespeed, shape and mass of the droplet, as well as the flow of the air. Limitations regardingthese factors may make this type of system difficult to implement on the MY600 Jet Printer.


What factor (eg. shape, mass, speed et cetera) has the strongest effect on the behaviour ofthe droplet with respect to alignment?


When using an air flow, it is believed that the speed of the droplet is the most importantfactor affecting alignment, since this will determine the speed of the air needed to affectthe droplet. When using an electric field, it is believed that the mass of the droplet is thedeciding factor, since the charge to mass ratio will dominate the control of the droplet withan electric field.

4



How will an electric field alignment mechanism affect Purgel, adhesive and solder paste?


Purgel will become slightly charged, adhesive slightly more charged and solder paste minutelycharged when leaving the nozzle. Furthermore, the solder paste has a higher density thanthe Purgel and adhesive. This leads us to believe that the droplets made of Purgel will bemoderately affected by an electric field, droplets of adhesive will be slightly more affectedand the droplets of solder paste will be only slightly affected by the electric field without anactive charging mechanism for the droplets.

1.5 Thesis Approach

The results from this thesis are based on two main alignment mechanisms have been investigated.

1. The first method investigated was to focus the droplet with air guidance. This method wasfirst investigated based on the advice and previous experience of Gustaf Martensson andDaniel Grafstrom at Mycronic AB. Mycronic wanted to investigate the possibility of usingguided air flows to focus a droplet. Previous research had been done by Guo et al. [2], wherethey used swirling air flows to focus yarn into the centre of a tube.

2. The second method investigated was to focus the droplet using electrostatic forces. Initialattempts to measure the charge of the solder paste were made with 3D-printed mounts anda solenoid, however the signal was too weak to enable the detection of the charge. A proof ofconcept was designed to show the effect of an electric field on a droplet shot from the MY600ejector. The proof of concept consisted of two copper plates connected to a high voltagepower supply which created an electric field between the plates. Purgel, an adhesive, andsolder paste were all ejected through the electric field. Comparisons were made between thedifferent fluids with a positive electric field, a negative electric field, as well as zero electricfield. Previous research had been performed by Orme et al. [7], where they investigated thismethod to deflect satellites from ejected droplets of molten metal.

1.5.1 Project Methods

Throughout the project, multiple of methods have been used to keep track of the progress andto have a good overview of what is left to be done. The methods chosen for this project wereTechnology Readiness Level (TRL), visual planning, and weekly meetings with the company su-pervisors. These three methods contributed in their own way to keep the project on track and inmotion. The TRL system was useful as a reminder internally and externally of the present stateof technology readiness, as well as the goal that is to be reached. The visual planning systemcontributed with the needed overview of the project that kept the work rate balanced. The weeklymeetings with the company supervisors was a great opportunity for ambiguities and questions tobe unravelled and answered. This was also a good time to discuss if the project was heading inthe right direction and to assemble action plans if this was not the case.

1.6 Thesis Outline

This report describes the problem at hand, the theory behind it, the models created and how allparts of the project were implemented. It also contains the results from the different parts of theproject, as well as a discussion of both the project itself and the results of this project. At the endof the report a bibliography and appendices are attached.

5

2 State of the Art

This chapter presents the state of the art in focusing a medium in air with an alignment mechanism.It is separated into three parts. The first part presents the use of swirling air flows used in thetextile industry in order to increase the yarn quality by spinning it. The second presents theability to deflect or focus droplets with the aid of electrostatic forces, which is commonly usedin ink jet printing application. Last, but not least, ways of focusing particles with the help ofacoustic vibrations are presented. All parts discuss the opportunities and limitations for each ofthe focusing methods that could be relevant for the MY600 Jet Printer implementation.

2.1 Swirling Air Flows

The method of using a swirling air flow is studied for applications in the textile industry foryarn spinning. Guo et al. [2], studied a turbulent swirling air flow and its characteristics for thepurpose of spinning textile fibres in order to increase the quality of the yarn. They investigated howdifferent nozzle properties, such as twisting chamber diameter, the number of injectors and theirdiameter, and the incidence angel of the injectors, were related to the quality of the yarn. Theseanalyses were performed using the k − ε turbulence model, which is one of the most frequentlyused computational fluid dynamics (CFD) models used to simulate turbulent flows. A 3D-modelof the nozzle can be seen in Figure 2.1.

Figure 2.1: A model of the air swirling nozzle with all its characteristics, e.g. the injection angle,θ, for the compressed air [2].

The conclusions were that all characteristics of the nozzle affected the quality of the yarn. Byincreasing the injection angle, the velocity in the upward direction of the injectors grew consid-erably, and the breakdown of the vortex moved swiftly downward in the nozzle. This caused alarge reversed jet in the nozzle, which is undesirable since fibres could get stuck. However, if theinjection angle were to small, the spiral torque could be too low, resulting in a vortex with a steeppitch. It was also noted that a larger diameter of the twisting chamber results in a flow that ismore turbulent. The flow path for three different injection angles can be seen in Figure 2.2.

7

Chapter 2. State of the Art

Figure 2.2: The flow inside the nozzle with different injection angles, where (a) θ = 35◦; (b)θ = 45◦; and (c) θ = 55◦ [2].

What can be seen in Figure 2.2 is that the flow inside the tube varies considerably, when theinjection angle changes. To obtain the desired swirling flow, and thus achieve the best quality ofthe yarn, one must tweak all the parameters for the nozzle characteristics in the right way in orderto succeed. Understanding the effect of these features would be of importance for the applicationof focusing droplets using a similar air flow.

2.2 Soldering Processes

2.2.1 Screen printing

Screen printing has been the leading and most commonly used method by companies around theworld for many years for the positioning and placement of solder paste for SMC assembly appli-cations [4]. The method works such that a stencil, that is designed for the specific PCB beingworked with, is placed on to the PCB and a squeegee pushes the solder paste over the stencilfilling all its apertures. The manufacturing of the stencils are done with three different methods,laser cutting, electroforming and chemical etching, depending on technical and cost requirements.Since every stencil is unique for each and every PCB revision, every new version of the PCB willtherefore require a new stencil. This could be very costly with multiple revisions, not only becausethe company needs to order new stencils, but also but also due to lost production time since thedelivery time of a new stencil is normally 1-3 days [4]. In order to maintain a flow in the produc-tion, keeping the stencils clean and intact is crucial. Today’s printers can do this automatically,but they require manual refilling of cleaning agents or paper. The need for storage area of allstencils is something that is unwanted, where storing and handling them correct can be costly.However, if the stencils are kept clean and intact, then they could have a lifespan of approximately10,000 strokes, a number that increases with the stencil thickness [10]. Taking these factors intoconsideration screen printing is not the best solution for mounting SMC on to PCB’s.

A common problem is that the PCBs often have components with large differences in size. In Figure2.3, this problem is illustrated. When a thick stencil, which is optimized for large components, isused a large amount of solder paste can get stuck inside the hole for the small components of thestencil and lead to dry joints. However, if a thin stencil, which is optimized for small components,is used too little solder paste will be placed on the larger components leading to lean joints. Thisproblem could be solved by using a stepped stencil, where both small and large components receivesthe right amount of solder paste. Even this type of stencil has a problem, since the distance betweenthe varying size of components has to be larger resulting in larger board areas.

8


Figure 2.3: Common issues with screen printing, where dry joints occur when using thick stencils,lean joints when using thin stencils and a component distance that is too large for stepped stencils[3].

Even though screen printing is a well proven method, the assembly process still has the largesteffect on the soldering results. The placement of solder paste is considered to be the biggest causeof soldering failure [4]. Up to 70% of all assembly flaws are directly connected to the placementof solder paste using screen printing, see Figure 2.4 [4]. Defects of this kind can lead to seriousreliability issues if undetected.

Figure 2.4: A pie-chart showing the percentage of related defects due to the printing process,where solder placement in form of opens, insufficients and bridges are the major defects. [4].

2.2.2 Jet printing

A solution or complement to the weakness of screen printing is a stencil-free version of printingcalled jet printing. This method could be compared with ink jet printers and how they shoot inkon a paper in order to create a document. The user can create the CAD data for the wanted PCBon any computer and later send it to the machine for printing. With the power to regulate thedroplet volume, it is possible to give each pad the optimal amount of solder paste, see Figure 2.5.

9


Figure 2.5: The issues with screen printing is solved with jet printing, where the right amount ofsolder paste is given to each component through the ability of controlling the droplet volume [3].

The deposit repeatability, regarding the droplet volume, is a crucial part of ensuring that thecomponents get perfect joints. In a study made by Martensson [5], a comparison between jetprinting and screen printing was carried out with respect to the volume repeatability of solderpaste for different components. This study showed, as can be seen in Figures 2.6 and 2.7, that therepeatability for jet printing outperforms screen printing for a number of component configurations.In Figure 2.6 below, the deposit repeatability regarding the droplet volume is shown for twodifferent components, one quad flat package (QFP) and one quad flat no-leads (QFN), where they-axis represents the percentage of the total amount of pads and the x-axis represents the dropletvolume.

(a) Histogram of the volume for componentQFP 144.

(b) Histogram of the volume for component0.5mm QFN.

Figure 2.6: Volume histograms showing the deposit repeatability for jet printing versus screenprinting [5].

In Figure 2.7 below, the standard deviation of the droplet shots is shown, where the y-axis repre-sents the deviation per mean volume and the x-axis represents the test number.

10


(a) σV from Figure 2.6a. (b) σV from Figure 2.6b.

Figure 2.7: A comparison of the volume standard deviation, σV, for jet printing versus screenprinting [5].

With a higher degree of repeatability per deposit, the correct amount of solder paste for each andevery pad will be obtained, which can be seen in Figure 2.8.

Figure 2.8: Two examples of solder paste application on PCBs for a) jet printing and b) screenprinting evaluated in a solder paste inspection device, where the upper figures are component 1(QFP 144) and the lower are component 2 (0.5 mm QFN) [5].

As can be seen in Figure 2.8 above, different coverage strategies are used for jet printing and screenprinting, where jet printing uses less solder paste on the pads relative screen printing.

2.3 Electrostatic Forces to Deflect or Focus Droplets

The use of electrostatic forces to focus or deflect droplets of fluid is commonly used in the areaink jet printing. Doak et al. [6] discuss the usability of a high-voltage dielectrophoretic probe todeflect a high-velocity droplet stream. A high voltage probe is used to deflect the droplet stream,as seen in Figure 2.9. The stream is deflected only when the voltage is turned on.

11


Figure 2.9: By introducing a high voltage probe near the droplet stream it is possible to deflectthe droplet stream by introducing a voltage [6].

A trap can be used in order to collect and recycle droplets as the default mode, see Figure 2.10.The high voltage probe can be switched on whenever you wish to print on the substrate. This willenable rapid on and off switching in order to increase the control of the ink jet.

Figure 2.10: The continuous droplet stream is collected in the trap and recycled when the highvoltage probe is off. The droplets only end up on the substrate when the high voltage probe isturned on [6].

According to Doak et al. [6], another use for this method is to separate and trap satellite droplets,as seen in Figure 2.11. This will ensure the accuracy and consistency of the volume of dropletssent to the substrate below.

Figure 2.11: Satellite droplets are deflected and pulled off of the main stream using the highvoltage probe [6].

12


Orme et al. [7] investigate the method of electrostatically charging a stream of droplets and thendeflecting it. A charging electrode, as seen in Figure 2.12, is used to both charge the droplets, aswell as help separate each droplet from the stream. The charged droplet is then deflected via thedeflection electrodes. The molten solder is connected to ground and a positive periodic waveformpotential is applied to the charging electrode, as seen in Figure 2.12. This causes a negative chargeto be induced on the droplet and a horizontal movement can be achieved using the deflectionelectrodes.

Figure 2.12: Schematic of the set-up used by Orme et al. [7].

The experiments were deemed accurate when compared with the theoretical charge predictionprovided by Schneider [7].

2.4 Droplet Charge Evolution

For the utilisation of electrostatic forcing to be feasible, the existing charge on droplets, as well astheir constancy, must be known.

A study of this was carried out by Thulin et al. [8]. They built a system that could eject droplets,charge them and then let them fall free through four metallic cylinders acting as sensors, beforeending up at the bottom in a Faraday pail. The goal with this experiment was to establish ifa charged droplet would lose some of its charge during the time it flew through the air or if itremained the same. The droplets charged the sensors by induction when they travelled throughthem, which made it possible to measure the droplets’ charge and thus detect if there were anydifferences in the charge during the fall. A sketch of the general setup for the system can be seenin Figure 2.13.

13


Figure 2.13: The experimental setup of the electrostatic charge evolution arrangement, containinga syringe as a droplet generator, a high voltage charging ring, four cylinder sensors and a Faradaypail to detect the droplet charge [8].

If the charge of the droplets remained the same during the time they fell in the air, the sensorswould theoretically measure values from the charge as shown in Figure 2.14.

Figure 2.14: Theoretical prediction of the droplet charge evolution during the time it falls throughthe sensors [8], where the decreasing time for each cylinder depends on the increasing velocity ofthe droplet.

The experiments were performed with several liquids with varying material characteristics e.g.liquid resistivity. The results showed that all the liquid charges have the same order of magnitude.It was concluded from the experiment that a charged droplet maintains its charge while movingthrough the air. In Figure 2.15, the charge distribution of the various liquids are presented togetherwith theoretical predictions.

14


(a) (b)

Figure 2.15: The conclusions in (b), that shows the theoretical prediction in comparison withthe experiments, proves that the charge remains the same and (a), which shows the droplet chargeevolution with respect to the liquid resistivity, proves that the charge stayed in the same orderindependent of liquid characteristics.

2.5 Acoustic Focusing with Acoustophoresis

The use of acoustic waves to focus or measure the size of particles is used in the life science field,especially medicine.

Cai et al. [9] propose a method of particle focusing by using acoustic waves in a microfluidicchannel. It is suggested that piezo-actuators and acoustic resonant elements could be used to focusthe particles in a microchannel, see Figure 2.16.

Figure 2.16: A sketch of particle focusing in a microchannel with acoustic resonant elementsfrom Cai et al. [9].

It is proposed that an evanescent wave field is to be used instead of a standing wave field, since the

15


gradient of the evanescent field would be higher than that of a standing wave field. The particleswould therefore experience a greater acoustic radiation force. The method used is independenton the type of incident wave and size of microchannel. Even though the results showed that it ispossible to focus particles, the time span in which this was accomplished is considerably larger thanthe time span in which the MY600 Jet Printer operates within. However, this does not excludethe possibility of focusing droplets with the aid of acoustic.

16

3 Analyses, Models and Simulations

In this chapter all models for the different investigated concepts and the methods used for theircreation will be presented. In addition, the calculations regarding the acting forces on the fallingdroplet, and the development of a solenoid, in order to detect the charge from the solder pasteshots, will be handled.

3.1 Force Analysis

The order of magnitude of the horizontal force needed to adjust the misdirected shots from theMY600 Jet Printer is of critical importance in order to gauge the feasibility of the method. Inorder to calculate this force, Newton’s second law was used. In Figure 3.1, the acting forces on thedroplet are shown as the droplet moves through the air.

Figure 3.1: Acting forces on the droplet where mg is the gravitational force, FDx and FDy arethe drag force in x- and y-direction, and Freq is the applied force from the electric field. h is theheight between the ejector’s nozzle and the PCB, σ is the positioning error and α is the departureangle of the droplet.

To calculate how large the applied force Freq needs to be, the following second order differentialequations have to be solved.

my =∑i

Fyi= mg − FDy (3.1)

mx =∑i

Fxi = −FDx − Freq (3.2)

In this approximation, we assume the droplet to be a sphere with radius r. The drag force, FD isthe force that acts on this spherical object when it falls through a viscous fluid. Assuming small

17

Chapter 3. Analyses, Models and Simulations

Reynold’s numbers, Re, and using the Stokes’ assumption, the drag force of a droplet traveling inair then becomes,

FD = 6π · µair · r · v , (3.3)

where µair is the dynamic viscosity of the air, r is the radius of the droplet, and v the speed of theair. The mass of the droplet, is expressed as

msphere =4

3π · r3 · ρ , (3.4)

where ρ is the density of the spherical droplet. For droplets with higher speeds, a large Reapproximation, outlined in Orme et.al. [7], can be used. In this case,

FD = Cd · ρair · v2 ·A , (3.5)

where ρair is the density for air, v is the relative velocity of the droplet to the fluid, and A is thecross sectional area of the droplet. The factor Cd is the drag coefficient and is expressed as

Cd =24

Re+

6

1 +√Re

+ 0.4 , (3.6)

where Re is given by

Re =2r · v · ρair

µair. (3.7)

A comparison of drag forces in the low and high Re cases is shown in Figure 3.2.

Droplet velocity [m/s]

0 5 10 15 20 25 30 35 40 45 50

Dra

g f

orc

e [

uN

]

0

5

10

15

20

25

30

35

High Re

Low Re

Figure 3.2: Comparison of the high and low Re approximations of the drag forces from Eq.3.5and Eq.3.3.

By using Eq. 3.8, 3.9, and 3.10 together with the specifications presented in Figure 3.1, thehypotenuse, s, and the departure angle of the droplet, α, can be calculated.

s =√h2 + σ2 (3.8)

sin(α)

σ=

sin(90◦)

s(3.9)

18


α = arcsin(σs

)(3.10)

With a jetting height, h = 650µm, and a typical positioning error for the MY600 Jet Printer,σ = 20µm, the departure angle is calculated to be α = 1.76◦.

The second order differential equation is solved using the boundary and initial conditions expressedin Table 3.1 below.

Table 3.1: The boundary conditions for the force analysis of the droplet. Where v0 is a typicalspeed of the droplet in the MY600 Jet Printer.

x(t=0) = 0 y(t=0) = 0x(t=0) = v0 · sin(α) y(t=0) = v0 · cos(α)x(t=0) = 0 y(t=0) = 0v(t=0) = v0 = 20 m/s

The solution for Eq. 3.1 and 3.2 then becomes

y(t) =1

m(mg − FDy) · t+ C1 (3.11)

x(t) =1

m− (Freq + FDx) · t+ C2 (3.12)

y(t) =1

2m(mg − FDy) · t2 + C1 · t+ C3 (3.13)

x(t) =1

2m− (Freq + FDx) · t2 + C2 · t+ C4 (3.14)

With the boundary conditions, from Table 3.1, the following coefficients are determined

C1 = v0 · cos(α)C2 = v0 · sin(α)C3 = C4 = 0

The solution for the differential equation then becomes

y(t) =1

2m(mg − FDy) · t2 + v0 · t · cos(α) (3.15)

x(t) = − 1

2m· (Freq + FDx) · t2 + v0 · t · sin(α) (3.16)

The time required to travel the distance h is obtained using Eq. 3.15,

1

2m(mg − FDy) · t2 + v0 · t · cos(α) = h . (3.17)

By simplifying Eq. 3.17, an expression for the time can be created with the use of the reducedquadratic equation,

t = −v0 ·m · cos(α)

mg − FDy

+(−)

√(v0 ·m · cos(α)

mg − FDy

)2

+h · 2m

mg − FDy. (3.18)

Inserting the values of each constant will result in a time of t1 = 32.516µs for the droplet to hitthe PCB. This time, t1, is now used to calculate how large the force, Freq, has to be in order toadjust the misalignment angle fo the shot. By putting Eq. 3.16 equal to zero and including t1, theapplying force can be determined,

Freq =40m · sin(α)

t1− FDx = 349.11 µN . (3.19)

19


The required force Freq is used together with Eq. 3.5 to obtain the required air speed, resulting inv = 227.2 m/s for a spherical droplet with a volume of 2.1 nL. In Figure 3.3 a comparison of thewind speed between a spherical and a cylindrical droplet, with the same volume, can be seen.

Air flow velocity [m/s]

100 150 200 250

Fo

rce

on

dro

ple

t [µ

N]

0

200

400

600

800

1000

1200

1400

1600

Cylinder shaped droplet 108.9 m/s

Sphere shaped droplet 227.2 m/s

Freq

349.11 µN

Figure 3.3: Comparison between the air flow velocity needed to realign a cylinder shaped dropletand a sphere shaped droplet. Both droplets had a volume of 2.1 nL, a speed of 20 m/s and amisalignment angle of 1.76◦.

Since the shape of the droplets differed, see Figure 3.4, they did not have the same drag coefficient,CD. The reason for this is because they have different surface areas, which results in the cylinderhaving a larger CD than the sphere.

Figure 3.4: The drag force acting on a) a sphere and b) a cylinder with the same volume.

3.2 Air Flow Models

The simulations performed were done to investigate the affect of manipulating droplets with the useof controlled air flows. All air flow models and simulations were created in COMSOL Multiphysics5.0 and were created in two different scales. The first was created to scale with the MY600Jet Printer and a second version as a scaled up version, roughly 10 times larger. In COMSOLMultiphysics, the flow of air was simulated, meaning that the models are not the actual structure

20


of the cylinders. In Table 3.2, recurring constants for the following models are presented. Thesevalues are taken from a typical test run of the MY600 Jet Printer, using solder paste as the fluid.

Table 3.2: Test specifications that applies for all models.

Spec. Value Unit Desc.ρdroplet 1400 kg/m3 Densityµair 18.6·10−6 Pa·s Viscosityrdroplet 0.5 mm Radiusxdroplet 2 mm Drop position in the x-directionydroplet 2 mm Drop position in the y-direction

3.2.1 Swirling flow models

The idea of having a swirling flow inside a cylinder was that the flow would influence any misalignedshots and steer them towards the centre. This idea had been tested using a simple prototype byDaniel Grafstrom without any equitable results. However, since the concept had not been fullytested and evaluated, the concept of focusing objects with a swirling air flow needed further study.A literature search yielded showing an article this type of air flow was used in order to spin yarn[2]. Parameters, such as injection angle of the compressed air, cylinder diameter and number ofinjectors, were all important to achieve a good vortex inside the cylinder. Based on these articles,sketches was developed and later constructed in COMSOL Multiphysics, both in the up-scaledversion and the actual scale version, see Figures 3.5 and 3.6.

Figure 3.5: Sketch of a swirling air flow inside a straight cylinder and the proof of concept model,with 4 inlets with an incidence angle of 45◦, created in COMSOL Multiphysics.

21


Figure 3.6: Sketch of a swirling air flow inside a conical cylinder and the proof of concept model,with 4 inlets with an incidence angle of 45◦, created in COMSOL Multiphysics.

The proof of concept models were supposed to be ten times larger than the actual scale version.The reason why a scaling factor of ten were chosen was because the proof of concept model had tobe large enough in order to make a 3D printed version of it, and in the same time small enough toensure that the behaviour of the air flows and droplets would stay the same as in the actual scaleversion. In Tables 3.3 and 3.4, the values used to create the models and simulations are presented.The results from these tests can be found in Chapter 6.

Table 3.3: Test specifications developed for the straight cylinder model, on the right hand sideof Figure 3.5.

Spec. Value Unit Desc.rcylinder 5 mm Radiushcylinder 50 mm Heightlinlet 16.67 mm Lengthrinlet 0.83 mm Radiuszinlet 45 mm HeightvzDroplet 0.1 m/s Initial speedvair 10 m/s Inlet air speed

Table 3.4: Test specifications developed for the conical cylinder model, on the right hand side ofFigure 3.6.

Specs Value Unit DescriptionrcylinderTop 5 mm RadiusrcylinderBottom 3 mm Radiushcylinder 50 mm Heightlinlet 16.67 mm Lengthrinlet 0.83 mm Radiuszinlet 45.25 mm HeightvzDroplet 0.1 m/s Initial speedvair 10 m/s Inlet air speed

3.2.2 Perforated Cylinder Model

The goal of the perforated cylinder model was to create a uniform air flow lengthwise along thewhole cylinder focused towards the centre. A sketch to describe the concept as an initial step inthe design phase before creating the models in COMSOL Multiphysics 5.0, is seen in Figure 3.7.

22


Figure 3.7: Sketch of the perforated cylinder model with the air inlet at the top of the outercylinder, side view to the left and top view to the right.

The flow was meant to travel downwards into a space between concentric cylinders where the innercylinder is perforated allowing the air to flow radially inward thus creating a centre focused flow.Models with varying amounts of holes and different heights were created, an example of which canbe seen in Figure 3.8. The holes were placed perpendicular to the surface of the inner cylinder.

Figure 3.8: A model of the air flow in the perforated cylinder model with 4 sets of 8 inlets in acircular formation.

Models and simulations for the perforated cylinder model are described in Table 3.5. In Table 3.6,the dimensions and specifications for the perforated cylinder model tests are shown.

23


Table 3.5: The created models and simulations for the perforated cylinder.

Sets Num inlets Configuration4 8 Circular4 4 Circular6 8 Circular10 8 Circular

The results from these tests can be found in Chapter 6.

Table 3.6: Test specifications for the perforated cylinder model.

Specs Value Unit DescriptionrinnerTube 5 mm RadiushinnerTube 50 mm Heightlhole 6 mm Lengthrhole 0.8 mm RadiusvzDroplet 0.5 m/s Initial speedvair 5 m/s Inlet air speed

24


3.3 Electrostatic Field Evaluation

To achieve the force determined in Eq. 3.19 the charge of the droplets need to be calculated.To determine the charge on a particle that is deflected a distance d, with mass m and velocity vtravelling through an electric field, E, with a length h Eq. 3.20 is used. Eq. 3.20 is simplified byignoring the drag force acting upon the droplet.

q =2 ·m · d · v2

E · h2(3.20)

In Figure 3.9, a charged droplet is travelling through an electric field. The electric field, E, isdefined as the applied voltage, V , divided by the distance between the two copper plates, l. Thisresults in Eq. 3.21.

E =V

l(3.21)

Figure 3.9: A sketch of a droplet with charge q travelling through an electric field, E, with avelocity, v, over a distance h.

A model that was proposed toward this end includes a charge, Q, that is distributed uniformlyover the wall from a circular tube of radius b, and height h. In Figure 3.10, the problem can beseen with a distributed voltage over the wall instead of a charge.

25


Figure 3.10: A cylindrical tube with height h, radius b, symmetry axis z, and an applied voltageV , where a is the distance from the centre.

An expression for Q and E could be composed as

Qcylinder =V · 4πε0h

ln

(z+√b2+z2

(z−h)+√

b2+(z−h)2

) (3.22)

Ecylinder =a ·Q

4πε0h

(1√

b2 + (z − h)2− 1√

b2 + z2

). (3.23)

From Eq. 3.23, it is shown that the strength of the electric field depends on the height of the tube,h, the charge, Q, the distance from the centre, a, and z. The field strength will increase linearlywhen the droplet moves closer to the interior walls of the tube, due to the constant a. This willprovide a valuable focusing effect towards the centre for the misdirected shots, which can be seenin Figure 3.11.

26


Figure 3.11: The electric field, E, inside a tube that acts on a charged droplet.

3.3.1 Measurement of Charge

The realigning process using electrostatic forces is dependent on the charge fo the droplet travellingthrough the field. A solenoid was created in order to attempt to measure the charge on the droplet.The idea was to shoot the droplet through the solenoid and measure the voltage. The inductanceof the manufactured solenoid is calculated using

L =µcore · µ0 ·N2 ·A

lsolenoid, (3.24)

where µcore = 1.00000037 is the permittivity of the core, µ0 = 1.2566 · 10−6 N/A2 is the permit-tivity for vacuum, N = 32 is the number of spool turns around the core, A = 1.3273 · 10−6 m2 isthe cross-sectional area of the solenoid and lsolenoid = 3 mm is the length of the spool.

A mount for the solenoid was created to fit the solenoid to the ejector, see Figure 3.12. The mountwas modeled in Solid Works and 3D-printed. Cables were fitted to the solenoid and connected toan oscilloscope (Tektronix - TDS 3032B, Beaverton, USA) where the voltage was measured.The charge of a droplet is determined from

Qdroplet =2π ·mdroplet · ε0 · Vc

ρair · r2droplet · ln(

rringrdroplet

) , (3.25)

where Vc is the potential used to charge the droplet and rring is the radius of the charging ring.

Figure 3.12: CAD images of the mount used to attach the solenoid to the ejector. The solenoidwas fitted in the small hole in the centre of the mount.

Noise from the ejector itself proved to be greater than the signal obtained during the measurementof the droplet. Rather than spending time on creating a filter to extract a signal for the charge ofthe droplet it was decided that the charge of the droplet would be analytically determined.

27

4 Experimental Setup

This chapter presents a proof of concept and the implementations made during this project, aswell as the methods used to gather the results for Purgel, adhesive and solder paste, which arepresented in Section 6.2.

4.1 Test Equipment

A test rig was constructed in order to test the feasibility of using electrostatic forces to affect adroplet. For the initial tests, Purgel was used as the fluid. Purgel was used due to the spontaneouscharging of the droplets when ejected from the nozzle of the MY600, most likely due to ionised airsucked past the fluid and thus charging the fluid as it exits the nozzle [11]. For this reason, it waspossible to test if even a spontaneously charged droplet could be affected by an electric field in ashort distance. Tests were also performed with an adhesive that becomes even more spontaneouslycharged, according to previous experience by Kenth Nilsson at Mycronic [11], and with solder pastewhich were thought to hardly be affected at all.

An overview of the complete test rig can be found in Figures 4.1 and 4.2. The test equipmentincluded:

• The electric field test module, Figure 4.3.

• A camera - Pike F-032B (Allied Vision, Stadtroda, Germany) with a Zoom 7000 lens (Navitar,New York, USA), Figure 4.4.

• A MY600 ejector system, Figure 4.4.

• A flash, Figure 4.4.

• A high voltage power supply A3K6-50R (Oltronix, Leek, Netherlands).

• A computer with control software (inhouse) as the interface.

Figure 4.1: Sketch of how the test reg is set up, seen from ejector’s perspective.

29

Chapter 4. Experimental Setup

Figure 4.2: Sketch of how the test rig is set up with the internal distance, l, and height from thenozzle, h, seen from the perspective of the camera.

The electric field test module, see Figure 4.3, consists of a set of rails and two copper plates thathave been mounted on a pair of 3D-printed plastic mounts. The distance between these plasticmounts, and thus the copper plates, can be adjusted easily. The top part, which includes thecopper plates and plastic mounts, can be adjusted along the axis of the module, enabling the easyadjustment of the plates directly under the nozzle of the ejector.

Figure 4.3: Electric field test module with 3D-printed plastic mounts and thin copper platesmounted on an adjustable track.

The test rig was connected to a computer used to control the system. This computer controlledthe triggering time for the flash, the settings for the ejector, the amount of shots to be fired etcetera.

30


Figure 4.4: The test rig with a recording camera, a flash and an ejector system, all connected toa computer.

4.2 Analysis Method

A visual inspection of the images produced from the test rig was made in order to determineif the electrostatic field affects the droplets. As the droplets travelled through the electrostaticfield, images of the droplets were taken with the camera. The resulting images were analysed inMATLAB, where the volume was calculated with the help of Eq. 4.2. The images are processed inMATLAB and the colours are inverted, as seen in Figure 4.5, after which the number of objects ineach image is determined. Once the number of objects have been determined, the size and positionof each object is calculated. Each object is then split into a left half and a right half, each half’svolume is then approximated by using Guldin’s rule [12]

Vi = 2π ·Ri ·Ai , (4.1)

where i is the right or the left droplet, Vi is the volume for the i:th droplet, Ri is the x-distancefrom the centre of the droplet to the centre of the i:th droplet and Ai is the area for the i:th droplet.From MATLAB the x- and y-positions of the geometric centre, the areas and the volumes for eachdroplet were output to data files and saved for further analysis. The total volume of each dropletis then approximated by

Vtot =Vleft + Vright

2(4.2)

31


a)

x-position [px]

20 40 60 80

y-p

ositio

n [px]

50

100

150

200

250

300

350

400

450

b)

x-position [px]

20 40 60 80

y-p

ositio

n [px]

50

100

150

200

250

300

350

400

450

c)

x-position [px]

20 40 60 80

y-p

ositio

n [px]

50

100

150

200

250

300

350

400

450

d)

x-position [px]

20 40 60 80y-p

ositio

n [px]

50

100

150

200

250

300

350

400

450

Figure 4.5: Representative images from the analysis. a) is the grey-scale image taken by thecamera and loaded into MATLAB, b) is the image adapted from grey-scale to a clear black andwhite binary image, c) is the image inverted and d) is the final image with the boundaries andgeometric centre ( * ) displayed for three different object.

32


4.2.1 Experiments Conducted

Multiple tests were performed with the test equipment listed in Section 4.1. These tests wereperformed in order to see if the electric field created between the two copper plates affected thePurgel-droplets at all. Tests were performed where the parameters were altered to create a fieldas strong as possible without clogging up the area between the two copper plates. The parametersused results for these tests can be seen in Table 4.1.

Table 4.1: Tests cases performed in the test rig, where the distance l and height h can be seen inFigure 4.2.

Test case Fluid No. of tests Images per test Distance l Voltage Height hTest case 1 Purgel 10 ≈ 1750 3 mm 0 V 11 mmTest case 2 Purgel 10 ≈ 1750 3 mm +3600 V 11 mmTest case 3 Adhesive 3 ≈ 4850 5 mm 0 V 11 mmTest case 4 Adhesive 3 ≈ 4850 5 mm +3600 V 11 mmTest case 5 Adhesive 10 ≈ 1600 3 mm -3600 V 11 mmTest case 6 Adhesive 10 ≈ 1600 3 mm 0 V 11 mmTest case 7 Adhesive 10 ≈ 1600 3 mm +3600 V 11 mmTest case 8 Adhesive 10 ≈ 1600 3 mm -3600 V 7 mmTest case 9 Adhesive 10 ≈ 1600 3 mm 0 V 7 mmTest case 10 Adhesive 10 ≈ 1600 3 mm +3600 V 7 mmTest case 11 Solder 10 ≈ 1600 3 mm -3600 V 7 mmTest case 12 Solder 10 ≈ 1600 3 mm 0 V 7 mmTest case 13 Solder 10 ≈ 1600 3 mm +3600 V 7 mm

33

5 Prototype - Electrostatic Focusing System

In this chapter, a prototype of an alignment mechanism utilising electrostatic forces will be pre-sented and explained. The risks and limitations of this concept will be discussed in Section 7.4.

5.1 System Overview

The prototype is based on a focusing system that uses electrostatic forces to manipulate misdirectedshots of solder paste toward the central vertical axis. This is obtained by applying a charge to thedroplet when it exits the nozzle, and then acting with induced electrostatic forces on the dropletwhen it is traveling downwards toward the PCB. This is achieved with the help of a charging ringand an electrostatic focusing cylinder, see Figure 5.1.

Figure 5.1: Design idea of the electrostatic focusing prototype, not to scale.

A droplet of solder paste receives a charge via the charging ring that is placed at a position wherethe droplet has separated from the ejector nozzle. The droplet then travels through a focusingcylinder in which there is an electric field with the same charge as the droplet. This field thenfocuses the droplet towards the centre of the cylinder due to the symmetric electrostatic forcesacting on the droplet. The force acting upon the droplet increases linearly with the distance fromthe centre of the cylinder, meaning that the closer to the walls the droplet gets the larger the forcebecomes. It is crucial that the proportion between the droplet charge, Qdroplet, and the electricfield inside the cylinder, Ecylinder, results in an acting force Fdroplet ≥ Freq as calculated in Section

35

Chapter 5. Prototype - Electrostatic Focusing System

3.1.

A specific voltage that has been chosen specifically with respect to the goal volume of the droplet,which is known, can be applied to each droplet when they pass the charging ring so that the chargeto mass ratio always stay the same. This will result in the correct amount of force acting on thedroplet during the time it travels inside the focusing cylinder.

If the ring voltage would be turned on and off between every shot, the slew rate and the dischargetime of the power supply would have to be faster than the shot frequency. Since the MY600 JetPrinter has a shot frequency above 200 Hz, this has to be done in less than 50 ms, which could bedone with modern high voltage power supplies. However, switching the voltage on and off at thisspeed, and thus charging and discharging the ring, will create a large magnetic field, which willproduce high voltage peaks that could damage the surrounding equipment.

36

6 Results

This chapter presents the results from the project, including simulation results from COMSOLMultiphysics, proof of concept tests and calculations for the prototype. The results from the proofof concept tests are divided in to three parts, 1) Purgel, 2) adhesive and 3) solder paste. All thisis later evaluated in Chapter 7.

6.1 Models and Simulations

6.1.1 Air Flows

Swirling flows

When the swirling air flows were simulated, the velocity stream lines behaved as expected. InFigures 6.1 and 6.2, it can be seen that a distinct vortex is built up inside the cylinders where thevelocity close to the walls and in the centre is very low, but much larger in between.

Figure 6.1: Plot showing velocity contours and streamlines for the swirling air flow model witha straight cylinder.

37

Chapter 6. Results

Figure 6.2: Plot showing velocity contours and streamlines for the swirling air flow model witha conical cylinder.

The simulations showed that both the straight and conical model did not improve the accuracy ofthe droplet, but rather worsened the performance since the particle followed the swirling air flowinstead of being focused towards the centre, see Figures 6.3 and 6.4. It was therefore decided thatswirling flows were not going to be further investigated.

Figure 6.3: Three-dimensional trajectory plots of the swirling air flow model with a straightcylinder.

38

Chapter 6. Results

Figure 6.4: Three-dimensional trajectory plots of the swirling air flow model with a conicalcylinder.

Perforated Cylinder Model

The results from the perforated cylinder model looked promising when simulated in COMSOLMultiphysics. In Figure 6.5, the velocity contour plots can be seen for the first test of the perforatedcylinder model. This test was conducted on the model with 4 sets of 8 circular holes between theouter and inner cylinder. The air flow is focused towards the centre and down through the outlet.With an inlet speed of 5 m/s, the speed through the centre of the inner cylinder is around 25 m/s.

Figure 6.5: Velocity contour plots for the perforated cylinder model, test 1, 4x8 holes.

In Figure 6.6, the trajectory of a droplet can be seen for the first test of the perforated cylindermodel. The particle is introduced off centre, see Table 3.6 for test specifications, with an initialvertical speed of 0.5 m/s. The particle is focused towards the centre of the inner tube.

39

Chapter 6. Results

Figure 6.6: Three-dimensional trajectory plot of the perforated cylinder model, test 1, 4x8 holes,xz-plane to the left and xy-plane to the right. The droplet, the red dot, moves towards the centreof the cylinder.

In Figure 6.7, the velocity contour plots can be seen for the second test of the perforated cylindermodel. This test was conducted on the model with 4 sets of 4 circular holes between the outer andinner cylinder. The air flow is focused towards the centre and down through the outlet. With aninlet speed of 5 m/s, the speed through the centre of the inner cylinder is around 30 m/s, but notvery stable.

Figure 6.7: Velocity contour plots for the perforated cylinder model, test 2, 4x4 holes.

In Figure 6.8, the trajectory of a droplet can be seen for the first test of the perforated cylindermodel. The particle is introduced off centre with a vertical speed of 0.5 m/s, see Table 3.6 for testspecifications. The particle is not focused towards the centre of the inner tube.

40

Chapter 6. Results

Figure 6.8: Three-dimensional trajectory plot of the perforated cylinder model, test 2, xz-planeto the left and xy-plane to the right.

In Figure 6.9, the velocity contour plot can be seen for the third test of the perforated cylindermodel. This test was conducted on the model with 6 sets of 8 circular holes between the outer andinner cylinder. The air flow is focused towards the centre and down through the outlet. With aninlet speed of 5 m/s the speed through the centre of the inner cylinder is, slower than on the testwith 4 sets, around 20 m/s.

Figure 6.9: Velocity contour plot of the perforated cylinder model, test 3.

In Figure 6.10, the trajectory of a droplet can be seen for the first test of the perforated cylindermodel. The particle is introduced off centre with a vertical speed of 0.5 m/s, see Table 3.6 for testspecifications. The particle is focused towards the centre of the inner tube.

41

Chapter 6. Results


In Figure 6.11, the velocity contour plot can be seen for the fourth test of the perforated cylindermodel. This test was conducted on the model with 10 sets of 8 circular holes between the outerand inner cylinder. The air flow is focused towards the centre and down through the outlet. Withan inlet speed of 5 m/s the speed through the centre of the inner cylinder is, slower than on thetest with 4 sets, around 10 m/s.

Figure 6.11: Velocity contour plot of the perforated cylinder model, test 4.

42

Chapter 6. Results

In Figure 6.12, the trajectory of a droplet can be seen for the fourth test of the perforated cylindermodel. The particle is introduced off centre with a vertical speed of 0.5 m/s, see Table 3.6 for testspecifications. The particle is focused towards the centre of the inner tube.


In Figure 6.13, the trajectory of a droplet can be seen for the fourth test of the perforated cylindermodel. The particle is introduced off centre with a vertical speed of 20 m/s. The particle is notfocused towards the centre of the inner tube since the particle speed is too high.


6.2 Proof of Concept

This section will present the results from the experiments performed on the proof of concepttest rig. The methods used to achieve these results are presented in Chapter 4. For a graphicalrepresentation of the results, see Appendix A. Tables 6.1 and 6.2 present the parameters used foreach test case as well as a summary of the results from each test case.

43

Chapter 6. Results

Table 6.1: Test case description for all 13 cases with respect to fluid, distance, l, voltage, u, andheight, h. See Figure 4.2 for reference to distance l and height h.

Test case Fluid Distance l Voltage u Height hTest case 1 Purgel 3 mm 0 V 11 mmTest case 2 Purgel 3 mm +3600 V 11 mmTest case 3 Adhesive 5 mm 0 V 11 mmTest case 4 Adhesive 5 mm +3600 V 11 mmTest case 5 Adhesive 3 mm -3600 V 11 mmTest case 6 Adhesive 3 mm 0 V 11 mmTest case 7 Adhesive 3 mm +3600 V 11 mmTest case 8 Adhesive 3 mm -3600 V 7 mmTest case 9 Adhesive 3 mm 0 V 7 mmTest case 10 Adhesive 3 mm +3600 V 7 mmTest case 11 Solder 3 mm -3600 V 7 mmTest case 12 Solder 3 mm 0 V 7 mmTest case 13 Solder 3 mm +3600 V 7 mm

Table 6.2: The results of the mean volume, the median volume, the mean x-position and thedeviation from zero electric field for the tests cases performed in the test rig.

Test case Vmean [nL] Vmed [nL] x [px] Deviation σ [µm]Test case 1 0.69 0.0016 50.12 0Test case 2 1.36 0.31 32.24 -113.40Test case 3 2.65 3.18 30.20 0Test case 4 2.28 2.65 30.32 +0.76Test case 5 3.37 3.63 33.52 +5.72Test case 6 3.21 3.53 32.70 0Test case 7 3.24 3.55 32.24 -3.21Test case 8 0.32 0.27 32.57 +4.39Test case 9 0.29 0.24 31.94 0Test case 10 0.33 0.27 31.67 -1.88Test case 11 0.27 0.82 28.50 -20.56Test case 12 2.65 0.87 31.45 0Test case 13 2.70 0.84 33.69 +15.61

6.2.1 Purgel

In both test case 1 and 2, data was gathered from 10 test sequences. For test case 1, the sequenceswere performed with zero electric field, u = 0 V, and for test case 2 they were performed throughan electric field, u = +3600 V. When comparing Figure 6.14 to Figure 6.15, it is shown that thereis a much larger percentage of low volume droplets, 0 - 0.16 nL, for u = 0 V. However, with anapplied electric field, the amount of low volume droplets decreases from 73% to 45%. What is alsoseen in Figures A.1 and A.2 is the mean value of the volumes for the droplets, Vmean = 0.69 nLfor test case 1, with u =Vmean, and Vmean = 1.34 nL for test case 2, with u = Vmean. This furthershows that the mean volume of droplets is higher when applying an electric field.

44

Chapter 6. Results

Volume [nL]0 2 4 6 8 10 12 14 16

Fre

quen

cy [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Distribution of droplet volumesMean volumeMedian volume

Figure 6.14: A histogram of the distribution of volume for the droplets in test case 1, with u =0 V.

Volume [nL]0 2 4 6 8 10 12 14 16

Fre

quency [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Figure 6.15: A histogram of the distribution of volume for the droplets in test case 2, with u =+3600 V.

From Figures A.3, A.4, A.5 and A.6 it is shown that when filtering the smallest droplets (V<0.02 nL), the distribution of the volumes is very similar for both test case 1 and 2. This impliesthat the electric field removes some droplets that are smaller than 0.02 nL in volume.

45

Chapter 6. Results

6.2.2 Adhesive

Test Case 3 and 4

Test cases 3 and 4 were performed with adhesive at a distance of l = 5 mm between the copperplates. A voltage of u = +3600 V and u = 0 V were compared. A total of about 14500 imageswere used for each test. The mean volume when the electric field was turned on (u = +3600 V)was Vmean = 2.282 nL and Vmean = 2.648 nL when the electric field was turned off (u = 0 V). Thedistribution of volumes is quite similar for both test cases, as seen in Figures A.7 and A.8.

The x-positioning of the recorded shots are rather similar when the electric field is on and when itis off, as seen in Figures A.9 and A.10. With the electric field applied, the average x-position is atx = 30.32 and with the electric field off the average x-position is at x = 30.20. With a pixel sizeof 6.97µm this distance is only 0.84µm.

Adhesive - Test Case 5, 6 and 7

For test case 5 (u = -3600 V), 6 (u = 0 V) and 7 (u = +3600 V), a distance of l = 3 mm betweenthe copper plates was used. Ten test sequences were performed for each test case resulting in atotal of around 16000 images per test case. Between each test sequence, the copper plates werecleaned to remove any residual adhesive. When the electric field was off, it was clear that bothcopper plates had residual adhesive stuck on them. When using u = -3600 V to power the electricfield only one of the copper plates had adhesive on it and when powering the electric field with u= +3600 V, the opposite copper plate had adhesive on it.


For test case 8 (u = -3600 V), 9 (u = 0 V) and 10 (u = +3600 V) a distance of l = 3 mm betweenthe copper plates was used. Ten test sequences were performed for each test case resulting in atotal of around 16000 images per test case. Between each test sequence, the copper plates werecleaned to remove any residual adhesive.

6.2.3 Solder Paste

Test Case 11, 12 and 13

For test case 11 (u = -3600 V), 12 (u = 0 V) and 13 u = (+3600 V) a distance of l = 3 mm betweenthe copper plates was used. Ten test sequences were performed for each test case resulting in atotal of around 16000 images per test case. Between each test sequence, the copper plates werecleaned to remove any residual solder paste.

6.2.4 Summary of the Proof of Concept Results

As seen in Table 6.2, all the performed tests showed a difference in x-position when the electricfield was turned on or off. Purgel showed the largest difference, however this was likely due to thefact that the small droplets affected the mean x-position. Tests performed with the solder pasteshowed a larger deviation than tests performed with the adhesive. It is clear that the directionof the field has an impact on the average position of droplets. With respect to positioning in thex-direction, solder paste showed the strongest effect. From the results above, see Table 6.2, it isalso interesting to note that the solder paste receive a charge opposite to that of the adhesive.

6.3 Prototype

To determine what will have the largest impact, regarding initial velocity for the droplet and thedroplet volume, on how well the misdirected shots gets realigned, the extreme cases for the MY600

46

Chapter 6. Results

were used to calculate how much applied force the different scenarios needs to get aligned. Table6.3, shows the divergence in the applied force.

Table 6.3: Boundary cases when shooting with MY600 Jet Printer.

Initial velocity Droplet volume Freq

10 m/s 4 nL 166.25µN10 m/s 20 nL 831.24µN100 m/s 4 nL 16 600µN100 m/s 20 nL 83 000µN

From these results it is clear that the velocity will affect the focusing ability much more in anegative way than the volume will, due to that the time for the electric field to act upon thecharge droplet will decrease at the same rate as the velocity increases.

47

7 Discussion

This chapter will present a discussion of the results presented in Chapter 6, but also the challengesmet and possible future work. All research questions and hypotheses stated in Section 1.4 will alsobe answered together with the conclusions that could be drawn from this report.

7.1 Air Flow Models and Simulations

One major issue with the air flow simulations was that COMSOL Multiphysics could not computesimulations with the actual values (droplet speed, air speed, droplet density et cetera) since themodel became too complex at higher speeds. Due to this issue, and since the focus of this projectwas not to create simulation models, the simulations were created with the speed scaled down.

7.1.1 Swirling Flow Models

Our hypothesis that the swirling air flows would centre any misaligned shots turned out to beerroneous. The vortex created inside both the straight and conical cylinders did not give anyforces towards the centre, and thus did not improve the accuracy of the falling droplets. Thedroplets simply got caught by the swirling air flow and collided with the inner walls, as can beseen in Figures 6.3 and 6.4, instead of being focused towards the middle.

7.1.2 Perforated Cylinder Model

When using lower speed on both the droplet and the air for the inlets the simulations seemedpromising. However, when increasing the speed of the droplet, the models showed that the dropletdid not become focused towards the centre. Using sets of 4 holes, instead of 8 holes, proved tobe less effective in focusing the droplet towards the centre. This is mainly due to fluctuating airflow velocities towards the centre of the tube. The velocity in the holes was much greater, thanin the centre, which caused them to counteract each other and create a flow upwards instead ofdownwards, as seen in Figure 6.7. Because of this, the following models were created with sets of8 holes in order to produce a more evenly distributed flow towards the centre.

7.1.3 Verification

From the force analysis, in Section 3.1, it became clear that the force exerted on the droplet fromthe air flow would not be enough to focus it towards the centre. This was further supported byperforming the same analysis on a droplet that was shaped like a cylindrical rod, as seen in Section3.1. From the tests performed for different drag forces, see Figure 3.2, it was concluded that witha more turbulent air flow Eq. 3.5 is more reliable whereas Eq. 3.3 is more simplified and onlyapplies for scenarios with a very low Re. Therefore, Eq. 3.3 was used to calculate the drag forcefor different objects in this project.

7.2 Force Analysis

The magnitude of the force needed to readjust misdirected shots was greatly influenced by theshape of the droplet. A larger surface area of the droplet implied a smaller of force required toalign them, given that the droplets had the same mass. This means that since the droplets inreality have a more cylindrical shape rather than a spherical, as in Figure 3.1, the drag force willhave a higher impact on the droplet and therefore easier align the misdirected shots. Calculationswere performed concerning the focusing of air flows for different shapes to see if this actually wasthe case. The calculations showed that the wind speed needed to correct a shot with a sphericalshape was approximately double that of a shot with a long cylindrical shape. In this report, it

48

Chapter 7. Discussion

has been assumed that the shape of the droplet remained constant during the fall, which does notcorrespond with how the droplet behaves in the real world. However, the shape will not changethat dramatically, making it justifiable to calculate with the droplet as a solid body.

7.3 Proof of Concept

This section will discuss the results from the proof of concept experiments conducted with Purgel,adhesive, and solder paste.

7.3.1 Purgel

Test Case 1 and 2

Both test case 1 and 2 were performed with ten test sequences each. Ten test sequences amountsto approximately 17,500 images for each of the two tests. With such a large amount of data theresults are significant. This initial experiment showed promising results and acted as a proof ofconcept. It showed a small, but significant effect on the droplets by moving the smallest dropletsoutside of the focus for the images. This successful proof of concept lead us to try the same ex-periments with adhesive instead of the Purgel.

7.3.2 Adhesive

Test Case 3 and 4

The results presented in Figures A.7, A.8, A.9 and A.10, imply that a distance of l = 5 mm betweenthe copper plates is most likely too big of a distance. The electric field becomes too weak at u =3600 V and l = 5 mm. Since we could not see any significant difference in the results when theelectric field was turned on or off, the next tests, test cases 5, 6 and 7, were performed at a distanceof l = 3 mm.


The results presented in Figures A.14, A.15 and A.16 indicate that the electric field is working.There is a significant difference in mean x-position when the field is turned on and when the fieldis turned off. Furthermore, the visual inspection of the copper plates, regarding the amount ofadhesive that got caught on them, also supports this.Even though the average mean x-position distances differ only a few µm, it is still a significantdifference since these droplets are only spontaneously charged. Also, since there was a clear differ-ence between the positive electric field and the negative electric field it is shown that electric fieldand the direction of the electric field has an affect on the position of the droplets. Furthermore,with the lack of small droplets, relative to the Purgel tests, it is most likely so that the adhesiveproduces less satellites overall than the Purgel.


Test cases 8, 9 and 10 were all performed with ten test sequences each, with each sequence con-sisting of around 1,600 images. This gave a total of around 16,000 images and a reliable amountof data to base the results on. These test cases show an even smaller difference in mean x-positioncompared to test cases 5, 6 and 7. This difference is likely due to the fact that the image frame iscloser to the ejector’s nozzle, thus giving the electric field less time to affect the droplet. However, asignificant difference in mean x-position can still be seen both for the negative and positive electricfield.

49


7.3.3 Solder Paste


Test cases 11, 12 and 13 were all performed with ten test sequences each, with each sequence con-sisting of around 1,600 images. This gave a total of around 16,000 images and a reliable amountof data to base the results on. At a plate distance of l = 3 mm and a height of h = 7 mm thetests with solder paste gave the largest difference in mean x-position, when compared to the otherfluids. Furthermore, the solder paste had the opposite charge to the adhesive, as seen in Table 6.2by comparing the mean x-positions for the negative and positive electric field for the two fluids.The reason for this difference in charging properties has not been studied in detail in this project.

7.4 Prototype

7.4.1 Risk Analysis

Having a high voltage system close to sensitive electric components can be dangerous, especiallywhen charged objects are being deposited on the PCB. Below, in Table 7.1, some of the risks thatcan occur when this system is implemented are presented.

Table 7.1: Risk analysis for the electrostatic focusing system.

Risk Likelihood Consequence Action planThe charged droplet dam-ages components on thePCB due to ESD.

1 9 Apply another method for focus-ing the misdirected shots.

Equally charged dropletsrepel each other on thePCB.

3 4 Switch the polarity of the dropletcharging between shots on thesame pad.

Memories on the ejectorresets due to the high volt-age.

4 9 Provide all sensitive equipmentwith proper shielding to with-stand the emissions from thehigh voltage.

The length of the focus-ing pipe becomes too long,due to the needed time ittakes to adjust the mis-aligned shots.

5 3 Increase the applied voltage onthe pipe to get a larger electricfield.

The operator gets a shockfrom the high voltage.

1 1 Ensure that the operator can notcome in contact with the highvoltage when it is turned on.

The charging on the sol-dering paste shots is nothigh enough.

6 5 Change from charge through airto direct charge on the droplet.

The inside of the focusingpipe gets soggy.

4 8 Paint the inside of the pipe witha lyophobic paint.

The pipe gets too close tothe PCB or the compo-nents causing a discharge.

4 9 Shield the outside of the pipe.

The most severe risk for the prototype is if the charged droplets damages the components on thePCB due to electrostatic discharge (ESD). All other risks, irrespective of the degree of likelihoodand consequence, are manageable and can be solved.

50


7.4.2 Limitations

One major limitation for this prototype, due to the high voltage applied, is the risk for sparkover, which is an unwanted electric discharge between two conductive objects. If this happens itcould harm components on the PCB or in worst case even destroy them. This does not become aproblem when using low voltages. However, a distance of approximately 1 mm per 3 kV (humidity= 96%) is needed to ensure that a spark over does not occur.

Another limitation for this prototype is that the droplets should not get a too high charge whenbeing ejected. The reason for this is because of the damage they can cause from the suddendischarge at impact close to vulnerable components.

7.5 Future Work

There are many things that can and need to be done in the future in order to achieve a workingprototype that increases the positioning accuracy of the MY600 Jet Printer. The first step is torecreate the tests on the MY600 while stationary, instead of on the test rig. These results willhopefully confirm the conclusions drawn from the test rig experiments. The next step will be toperform a design study of the focusing cylinder with simulations. To perform the design study,the requirements for the prototype, regarding material and size, need to be established in orderto obtain fair and correct results. The simulations will provide important information about howlarge the electric field inside the cylinder must be to achieve a relevant charge on the droplet. Fromthe design study, a decision can be made whether an active charging of the droplet is necessaryor if the passive charging of the droplet is sufficient. Today, it is still unclear how exactly thedroplets become spontaneously charged. However, a hypothesis might be that the air which isbeing sucked into the ejectors nozzle is ionized and thus charges the liquid. This question needs tobe investigated further in order to determine what actually causes the spontaneous charge, whichwill simplify the work to calculate how the charging of different liquids occurs. If it should be thecase that the prototype does not work as intended, there are still things that could be learned andused from this research. For example, the use of an electric field operating in only one direction,as in the experiments in Chapter 4, could be used to selectively affect satellites and thus preventthem from hitting the PCB. Even though this looked promising in the experiments, especially forPurgel, an additional investigation needs to be carried out to test its feasibility. Last, but notleast, an acoustic focusing system could be worth studying if the use of electrostatic fields does notprovide the desired outcome.

7.6 Conclusions

In Section 1.4, four research questions were presented. In this section, these questions will beanswered based on the research and results presented in this thesis.

Is it possible to achieve a positive effect on the positioning accuracy of a droplet of fluidthrough the utilisation of a guiding air flow from a controlled mechanical system?

The hypothesis for this derivative research question was that a positive effect on the positioningaccuracy of a droplet of fluid can be achieved through the utilisation of a guiding air flow from acontrolled mechanical system. In the hypothesis, it was also stated that the effect will depend onthe speed, shape and mass of the droplet, as well as the speed of the flow of air. Based on theresults found in Section 6.1, it was determined that a swirling air flow is not suitable to centre thedroplet of fluid. The centre-focused air flow showed promising results and showed a positive effecton the positioning accuracy of a droplet of fluid. However, the issue with this method was thatthe air speed needed on the air flow in order to focus a droplet of solder paste with the parameterspresented in Section 3.1 became too high. This led to the conclusion that using a guiding air flowfrom a controlled mechanical system to focus a droplet of fluid in the MY600 Jet Printer is not a

51


viable method.

What factor (e.g. shape, mass, speed et cetera) has the strongest effect on the behaviourof the droplet with respect to alignment?

The hypothesis for this derivative research question was that when using an air flow the speed ofthe droplet will be the most important factor since this will determine the speed of the air neededto affect the droplet, and when using an electrostatic field to focus the droplet the most importantfactor will be the mass of the droplet. The results presented in Section 6.1 and Section 6.3 showthat the velocity of the droplet is the most important factor for both the air flow method, as wellas the electrostatic field method.

How will an electric field alignment mechanism affect Purgel, adhesive and solder pastedifferently?

The hypothesis for this derivative research question was that the Purgel would be slightly affected,that the adhesive slightly more so, and finally that the solder paste would be only slightly affectedby an existing electric field. From the results in Section 6.2, we can see that all fluids are affected bythe electric field. Contrary to the hypothesis, the solder paste is affected more by the electric fieldthan the adhesive, by around 5 times. An initial thought was that perhaps the size and velocity ofthe solder paste droplets were different than those of the adhesive, but after examining the settingsand results it was seen that there was no considerable difference in size or velocity between thetests for the two pastes. An answer to the derivative research question is that, Purgel was affectedconsiderably by the electric field, with a clear focus on the small satellites, while the adhesive wasslightly affected. The solder paste was affected the most by the electric field. This points to thesolder paste droplets being the most charged of the three fluids examined. Another interestingdiscovery was that the charge of the solder paste droplets was the opposite of the charge of theadhesive droplets. This indicates that the charge of the solder paste droplets was considerablylarger than the charge of the adhesive droplets, since the solder paste droplets are around the samevolume but around four times denser.

Can an alignment mechanism (air guidance, electrostatic forces et cetera) be imple-mented in the jetting system of the MY600 Jet Printer to cause an advantageous repo-sitioning of the travelling droplet, such that a 2◦ misdirection angle is corrected withina length of h ≥ 0.65 mm and a volume of 1 ≤ V ≤ 5 nL, and a length of h ≥ 2 mmand a volume of 5 ≤ V ≤ 15 nL?

The hypothesis for this main research question was that an alignment mechanism can be imple-mented on the MY600 Jet Printer and that it will be more difficult to align larger volumes ofdroplets. From the results in this thesis, we have concluded that an alignment mechanism usingelectrostatic forces can be implemented in the jetting system of the MY600 Jet Printer. Furtherresearch needs to be done, as discussed in Section 7.5. With no active charging of the solder pastedroplets, it was possible to affect droplets with a static electric field up to 20 µm from a height of7 mm. With active charging and with a stronger electric field that focuses the droplets towardsthe centre, we believe this method can be used to focus droplets in a much shorter height.

52

Bibliography

[1] Lee, N.C., Reflow Soldering Processes and Troubleshooting SMT, BGA, CSP and Flip ChipTechnologies. Butterworth-Heinemann. Chapter 1 - Introduction to Surface Mount Technol-ogy, 2001.

[2] Guo, H.F, Chen, Z.Y, and Yu, C.W., “Simulation of the effect of geometric parameters ontangentially injected swirling pipe airflow,” Computer & Fluids, vol. 38, pp. 1917–1924, 2009.

[3] Mycronic AB, MY600 Jet Printing Brochure, 3 2014.http://mycronic.com/www2/elements.nsf/0 //740A1915011D20FAC1257E4A002E6780/$File/P-001-0265%20MY600%20Jet%20printing %20brochure%20Feb%202015.pdf.

[4] Mydata, Solder Paste Jet Printing - A New Approach to Solder Paste Application, 2008.http://mycronic.com/www2/elements.nsf/%28read%29/D8F055AE189F2A0AC12577AC0046605E/$file/Solder%20paste%20 jet%20printing,%20av%20Peter%20Grundy.pdf.

[5] Martensson, M., “Solder paste deposit volume repeatability - a comparison of jetting andscreen printing.” IPC APEX Expo, Sand Diego, USA, 2015.

[6] Doak, W.J., Donovan, J.P. and Chiarot, P.R., “Deflection of continuous droplet streams usinghigh-voltage dielectrophoresis,” Experiments in Fluids, vol. 54, no. 7, pp. 1577–1584, 2013.

[7] Orme, M., Courter, J., Liu, Q., Huang, C. and Smith, R., “Electrostatic charging and deflec-tion of nonconventional droplet streams formed from capillary stream breakup,” Physics ofFluids, vol. 12, no. 9, pp. 2224–2235, 2000.

[8] Luttgens, S., Luttgens, G., Thulin, A. and Touchard, G., “Electrostatic charge evolution of afalling droplet - application to splash filling,” Proceedings of the 2014 ESA Annual Meetingon Electrostatics, vol. 7, pp. G2–1 – G2–11, 2014.

[9] Cai, X., Guo, Q., Hu, G. and Yang, J., “Particle focusing in a microchannel with acousticmetafluid,” Applied Physics Letters, vol. 103, no. 3, pp. 031901–01 – 031901–04, 2013.

[10] Martensson, G., Mycronic AB, Private Communications, 2015.

[11] Nilsson, K., Mycronic AB, Private Communications, 2015.

[12] Rade, L. and Westergren, B., Mathematics Handbook for Science and Engineering. Studentlit-teratur. Chapter 7.3, p. 151, 2004.

53

8 Appendices

This chapter contains the appendices that may be useful for this report.- Graphs of the Results from Section 6.2: A- Risk Analysis: B- TRL: C- Image Analysis Matlab Code: D- Graphs Matlab Code: E

54

A - Graphs of the Results from Section 6.2

Table A.1: The parameters used for all 13 test cases, see Figure 3.1 for reference to distance land height h.

Test case Fluid Distance l Voltage Height hTest case 1 Purgel 3 mm 0 V 11 mmTest case 2 Purgel 3 mm +3600 V 11 mmTest case 3 Adhesive 5 mm 0 V 11 mmTest case 4 Adhesive 5 mm +3600 V 11 mmTest case 5 Adhesive 3 mm -3600 V 11 mmTest case 6 Adhesive 3 mm 0 V 11 mmTest case 7 Adhesive 3 mm +3600 V 11 mmTest case 8 Adhesive 3 mm -3600 V 7 mmTest case 9 Adhesive 3 mm 0 V 7 mmTest case 10 Adhesive 3 mm +3600 V 7 mmTest case 11 Solder 3 mm -3600 V 7 mmTest case 12 Solder 3 mm 0 V 7 mmTest case 13 Solder 3 mm +3600 V 7 mm

Purgel - Test Case 1 and 2

Volume [nL]0 2 4 6 8 10 12 14 16

Fre

quency [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Figure A.1: A histogram of the distribution of volume for the droplets in test case 1, with zeroelectric field (u = 0 V). Mean and median volume are marked as * and *.

55

Appendix A. - Graphs of the Results from Section 6.2

Volume [nL]0 2 4 6 8 10 12 14 16

Fre

quen

cy [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Figure A.2: A histogram of the distribution of volume for the droplets in test case 2, with anelectric field (u = +3600 V). Mean and median volume are marked as * and *.

x-position [px]0 10 20 30 40 50 60 70 80

Fre

quency [%

]

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

Distribution of droplet x-positionMean x-positionMedian x-position

Figure A.3: A histogram of the distribution of positions for the droplets in test case 1, with zeroelectric field (u = 0 V). The smallest droplets have not been removed.

56


x-position [px]0 10 20 30 40 50 60 70 80

Fre

quen

cy [%

]

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1


Figure A.4: A histogram of the distribution of positions for the droplets in test case 1, with zeroelectric field (u = 0 V). The smallest droplets have been removed.

x-position [px]0 10 20 30 40 50 60 70 80

Fre

quency [%

]

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1


Figure A.5: A histogram of the distribution of positions for the droplets in test case 2, with anelectric field (u = +3600 V). The smallest droplets have not been removed.

57


x-position [px]0 10 20 30 40 50 60 70 80

Fre

quen

cy [%

]

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1


Figure A.6: A histogram of the distribution of positions for the droplets in test case 2, with anelectric field (u = +3600 V). The smallest droplets have been removed.

Adhesive - Test Case 3 and 4

Volume [nL]0 2 4 6 8 10 12 14 16

Fre

quency [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1



58


Volume [nL]0 2 4 6 8 10 12 14 16

Fre

quen

cy [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1



x-position [px]0 10 20 30 40 50 60 70 80

Fre

quency [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Figure A.9: A histogram of the distribution of positions for the droplets in test case 3, with zeroelectric field (u = 0 V).

59


x-position [px]0 10 20 30 40 50 60 70 80

Fre

quen

cy [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Figure A.10: A histogram of the distribution of positions for the droplets in test case 4, with anelectric field (u = +3600 V).


Volume [nL]0 2 4 6 8 10 12 14 16

Fre

quency [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Figure A.11: A histogram of the distribution of volume for the droplets in test case 5, with anelectric field (u = -3600 V). Mean and median volume are marked as * and *.

60


Volume [nL]0 2 4 6 8 10 12 14 16

Fre

quen

cy [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1



Volume [nL]0 2 4 6 8 10 12 14 16

Fre

quency [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1



61


x-position [px]0 10 20 30 40 50 60 70 80

Fre

quen

cy [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Figure A.14: A histogram of the distribution of x-positions for the droplets in test case 5, withan electric field (u = -3600 V). An average x-position of 33.52.

x-position [px]0 10 20 30 40 50 60 70 80

Fre

quency [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Figure A.15: A histogram of the distribution of x-positions for the droplets in test case 6, withzero electric field (u = 0 V). An average x-position of 32.70.

62


x-position [px]0 10 20 30 40 50 60 70 80

Fre

quen

cy [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Figure A.16: A histogram of the distribution of x-positions for the droplets in test case 7, withan electric field (u = +3600 V). An average x-position of 32.24.


Volume [nL]0 2 4 6 8 10 12 14 16

Fre

quency [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1



63


Volume [nL]0 2 4 6 8 10 12 14 16

Fre

quen

cy [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1



Volume [nL]0 2 4 6 8 10 12 14 16

Fre

quency [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1



64


x-position [px]0 10 20 30 40 50 60 70 80

Fre

quen

cy [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1



x-position [px]0 10 20 30 40 50 60 70 80

Fre

quency [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1



65


x-position [px]0 10 20 30 40 50 60 70 80

Fre

quen

cy [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1



Solder Paste - Test Case 11, 12 and 13

Volume [nL]0 2 4 6 8 10 12 14 16

Fre

quency [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1



66


Volume [nL]0 2 4 6 8 10 12 14 16

Fre

quen

cy [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1



Volume [nL]0 2 4 6 8 10 12 14 16

Fre

quency [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1



67


x-position [px]0 10 20 30 40 50 60 70 80

Fre

quen

cy [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1



x-position [px]0 10 20 30 40 50 60 70 80

Fre

quency [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1



68


x-position [px]0 10 20 30 40 50 60 70 80

Fre

quen

cy [%

]

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1



69

Risk Likelihood Consequence Action plan

Simulations show no

increase in accuracy. 1 9

Study alternative ways of

increasing the accuracy, i.e.

acoustic vibrations.

Proof of concept

does not increase the

accuracy.

3 9 Investigate weaknesses in the

manufacturing process.

Scaling up may lead

to incorrect

conclusions

regarding e.g.

viscosity.

3 3

Perform system scaling analysis

with the respect to fluid dynamical

dimensionless numbers together

with advisors.

3D-printing delays. 1 3

Perform another task of the project,

e.g. writing on the report, until the

printing is done. If it takes too long,

we will outsource the task.

Manufacturing

within tolerances not

possible.

1 3

Verification on the actual machine

will not be possible. Calculate the

needed tolerance and perform

simulations.

Control strategy does

not adapt to

differences in droplet

volume.

1 3 Investigate why and adapt. Too

slow communication?

Re-examination 9 1

Plan properly and study on

evenings and weekends so as not to

affect the project.

Accident/dropout 1 9

Rewrite the scope and set new

limitations in order to successfully

finish the project in time.

Sickness 3 1

Catch up on what has been missed

on evenings and weekends, and

rework the time plan.

Manufacturing time

longer than expected

of the end product.

3 9

Find out how long it will take to

manufacture the product and plan

for it in advance. Find another

supplier if it takes too long.

End product

manufacturing delay

(short)

3 3

Await the product and perform

another task of the project during

the time.

End product

manufacturing delay

(long)

1 9 Find another supplier.

B - Risk Analysis

TRL Definitions, Descriptions, and Supporting Information

TRL Definition Description Supporting Information

1

Basic principles observed and reported

Lowest level of technology readiness. Scientific research begins to be translated into applied research and development (R&D). Examples might include paper studies of a technology’s basic properties.

Published research that identifies the principles that underlie this technology. References to who, where, when.

2

Technology concept and/or application formulated

Invention begins. Once basic principles are observed, practical applications can be invented. Applications are speculative, and there may be no proof or detailed analysis to support the assumptions. Examples are limited to analytic studies.

Publications or other references that out-line the application being considered and that provide analysis to support the concept.

3

Analytical and experimental critical function and/or characteristic proof of concept

Active R&D is initiated. This includes analytical studies and laboratory studies to physically validate the analytical predictions of separate elements of the technology. Examples include components that are not yet integrated or representative.

Results of laboratory tests performed to measure parameters of interest and comparison to analytical predictions for critical subsystems. References to who, where, and when these tests and comparisons were performed.

4

Component and/or breadboard validation in laboratory environment

Basic technological components are integrated to establish that they will work together. This is relatively “low fidelity” compared with the eventual system. Examples include integration of “ad hoc” hardware in the laboratory.

System concepts that have been considered and results from testing laboratory-scale breadboard(s). References to who did this work and when. Provide an estimate of how breadboard hardware and test results differ from the expected system goals.

5

Component and/or breadboard validation in relevant environment

Fidelity of breadboard technology increases significantly. The basic technological components are integrated with reasonably realistic supporting elements so they can be tested in a simulated environment. Examples include “high-fidelity” laboratory integration of components.

Results from testing laboratory breadboard system are integrated with other supporting elements in a simulated operational environment. How does the “relevant environment” differ from the expected operational environment? How do the test results compare with expectations? What problems, if any, were encountered? Was the breadboard system refined to more nearly match the expected system goals?

C-

TR

L

TRL Definitions, Descriptions, and Supporting Information (Continued) TRL Definition Description Supporting Information

6

System/subsystem model or prototype demonstration in a relevant environment

Representative model or prototype system, which is well beyond that of TRL 5, is tested in a relevant environment. Represents a major step up in a technology’s demonstrated readiness. Examples include testing a prototype in a high-fidelity laboratory environment or in a simulated operational environment.

Results from laboratory testing of a prototype system that is near the desired con-figuration in terms of performance, weight, and volume. How did the test environment differ from the operational environment? Who performed the tests? How did the test compare with expectations? What problems, if any, were encountered? What are/were the plans, options, or actions to resolve problems before moving to the next level?

7

System prototype demonstration in an operational environment.

Prototype near or at planned operational system. Represents a major step up from TRL 6 by requiring demonstration of an actual system prototype in an operational environment (e.g., in an air-craft, in a vehicle, or in space).

Results from testing a prototype system in an operational environment. Who performed the tests? How did the test compare with expectations? What problems, if any, were encountered? What are/were the plans, options, or actions to resolve problems before moving to the next level?

8

Actual system completed and qualified through test and demonstration.

Technology has been proven to work in its final form and under expected conditions. In almost all cases, this TRL represents the end of true system development. Examples include developmental test and evaluation (DT&E) of the system in its intended weapon system to determine if it meets design specifications.

Results of testing the system in its final configuration under the expected range of environmental conditions in which it will be expected to operate. Assessment of whether it will meet its operational requirements. What problems, if any, were encountered? What are/were the plans, options, or actions to resolve problems before finalizing the design?

9

Actual system proven through successful mission operations.

Actual application of the technology in its final form and under mission conditions, such as those encountered in operational test and evaluation (OT&E). Examples include using the system under operational mission conditions.

OT&E reports.

% Volume calculator - 2015-04-24 % ERIK SEGERSTEEN %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Calculates the volume of objects % Place script in the folder with images % realV is the output volume in nL %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% clear all; close all; clc; color = ['r';'g';'b';'m';'c';'k';'r';'g';'b';'m';'c';'k']; pixelSizeInput = 143.4217; % Input from Test-rig. pixelSize = 1/(pixelSizeInput*10^3); % [m] imageDir = dir('*.BMP'); Vtotal = []; Area = []; x = []; y = [];

for imageNo = 1:length(imageDir) % load image imagefile = imageDir(imageNo).name; tmpimage = imread(imagefile, 'bmp');

% threshold data I_eq = adapthisteq(tmpimage); bw = im2bw(tmpimage, graythresh(I_eq)); bw = ~bw; %invert binary image.

% extract bodies [L, num] = bwlabel(bw);

b = bwboundaries(bw); % extract positions [B,L] = bwboundaries(bw,'noholes');

stats = regionprops(L,'all'); for i = 1:length(stats) Area = [Area, stats(i).Area]; end % find centroids centroids = cat(1, stats.Centroid);

for i = 1:length(B) x = [x, floor(centroids(i,1))]; y = [y, floor(centroids(i,2))]; end

index = length(stats); % number of objects in the image. for i = 1:index tmpcentroid = floor(stats(i).Centroid - stats(i).BoundingBox(1:2)); %

Center of the object tmppic = stats(i).Image;

% look at left side tmppicleft = tmppic(:,1:tmpcentroid(1));

D - Image Analysis Matlab Code

% calculate mass center, area and volume [rowwhite, colwhite] = find(tmppicleft == 1); % Coordinates where the

left img is white. leftcentroid = sum(colwhite)/length(colwhite); Aleft = length(colwhite); Vleft = 2 * pi * leftcentroid * Aleft;

% look at right side tmppicright = tmppic(:, tmpcentroid(1)+1:end);

% calculate mass center, area and volume [rowwhite, colwhite] = find(tmppicright == 1); % Coordinates where the

right img is white. rightcentroid = sum(colwhite)/length(colwhite); Aright = length(colwhite); Vright = 2 * pi * rightcentroid * Aright;

if isnan(Vleft) % To protect against Vleft/Vright NaN when object is

only 1 pixel in width Vtotal = [Vtotal, Vright]; elseif isnan(Vright) Vtotal = [Vtotal, Vleft]; else Vtotal = [Vtotal, ((Vleft + Vright)/2)]; end end end

% Write the Area to a .txt file avgArea = sum(Area)/length(Area); fidArea = fopen('AreaResult.txt','wt'); fprintf(fidArea,'%f\n\n',avgArea); for i = 1:length(Area) fprintf(fidArea,'%i\n',Area(i)); end fclose(fidArea);

realV = Vtotal * pixelSize^3/1e-12; % Scale volume to nL avgV = sum(realV)/length(realV); % avg volume per picture

% Write the total volume and average volume to a .txt file fidV = fopen('volumeResult.txt','wt'); fprintf(fidV,'%f\n\n',avgV); for i = 1:length(realV) fprintf(fidV,'%f\n',realV(i)); end fclose(fidV);

% Write the x, y positions to a .txt file fidXY = fopen('positionResult.txt','wt'); for i = 1:length(x) fprintf(fidXY,'%i\t%i\n',x(i), y(i)); end fclose(fidXY); disp('SCRIPT COMPLETE')

% Plots results from result files - 2015-04-28 % Created by: ERIK SEGERSTEEN %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% clear all; close all; clc; filterValue = 0.02; % Filter value for the smallest droplets

% Load the position results text file and x-positions. loadPos = load('positionResult.txt'); x = loadPos(:,1);

% Load the volume results text file. loadVolume = load('volumeResult.txt'); volume = loadVolume; volume(1) = []; % Removes the first element from the result file, the

first value is an average volume.

% Load the area results text file. loadArea= load('AreaResult.txt'); Area = loadArea; % Area is in pixels Area(1) = []; % Removes the first element from the result file, the

first value is an average area.

% Volume plots figure(1) h = histogram(volume, 100, 'Normalization','probability'); axis([0 16 0 1]); hold on plot(mean(volume), 0, 'r*') plot(median(volume), 0, 'g*') ylabel('Frequency [%]') xlabel('Volume [nL]'); legend('Distribution of droplet volumes', 'Mean volume', 'Median volume')

tooSmallValues = find(volume < filterValue ); % Filters out the smallest

droplets. volume(tooSmallValues) = []; figure(2) h = histogram(volume, 100, 'Normalization','probability'); axis([0 16 0 1]); hold on plot(mean(volume), 0, 'r*') plot(median(volume), 0, 'g*') ylabel('Frequency [%]') xlabel('Volume [nL]'); legend('Distribution of droplet volumes', 'Mean volume', 'Median volume')

% Position plots figure(3) h = histogram(x, 80, 'Normalization','probability'); axis([0 80 0 1]); hold on plot(mean(x), 0, 'r*') ylabel('Frequency [%]')

E - Graphs Matlab Code

xlabel('x-position [px]'); legend('Distribution of droplet x-position', 'Mean x-position')

figure(4) x(tooSmallValues) = []; % Filters out the smallest droplets. h = histogram(x, 80, 'Normalization','probability'); axis([0 80 0 1]); hold on plot(mean(x), 0, 'r*') ylabel('Frequency [%]') xlabel('x-position [px]'); legend('Filtered distribution of droplet x-position', 'Mean x-position')

% Area plots figure(5) h = histogram(Area, 100, 'Normalization','probability'); axis([0 1500 0 1]); hold on plot(mean(Area), 0, 'r*') ylabel('Frequency [%]') xlabel('Area [px]'); legend('Distribution of droplet Area', 'Mean Area')

figure(6) Area(tooSmallValues) = []; % Filters out the smallest droplets. h = histogram(Area, 100, 'Normalization','probability'); axis([0 1500 0 1]); hold on plot(mean(Area), 0, 'r*') ylabel('Frequency [%]') xlabel('Area [px]'); legend('Filtered distribution of droplet Area', 'Mean Area')

ISRN MMK 2015:72 MDA 515

www.kth.se

methods for active manipulation of fluid...

Documents