Copyright is owned by the Author of the thesis.  Permission is given for 
a copy to be downloaded by an individual for the purpose of research and 
private study only.  The thesis may not be reproduced elsewhere without 
the permission of the Author. 
 



The Development of A Low Cost 
Non-destructive Inspection System for 

Plating Quality Assessment of 
Plated-through Holes in Printed Circuit Boards 

A Thesis Presented in Partial Fulfilment of the Requirements for the 
Degree of Master of Technology 

in the 
Department of Production Technology 

at 
Massey University 

KAMCHIUMAK 

1995 



ll 

ABSTRACT 

The status of printed circuit board inspection is reviewed with special focus placed on 

the existing techniques of assessing plated-through hole quality. The need for developing 

a non-destructive method for plated-through hole inspection has been identified and is 

the major objective of this research. 

The results of the investigation into various methods that could lead to this objective are 

presented. This investigation has been concerned with the application of image 

processing techniques and the leakage light detection method. The hardware and 

software requirements for automatic visual inspection of stuffed board components are 

established initially using available equipment from the Department of Production 

Technology. 

Image processing techniques are found to be capable of discriminating copper-plated and 

unplated surfaces using the difference in reflectance between the surf aces. This suggests 

the possibility of applying such techniques to assess the quality of through-hole plating. 

The leakage light detection method can be implemented to assess the plating coverage of 

plated-through holes. A low cost inspection system demonstrating the principle of 

leakage light detection has been constructed. This system is particularly relevant to the 

small batch manufacturers in the printed circuit board industry. 

The performance of the demonstration system has illustrated the simplicity and reliability 

of the design. It is concluded that the leakage light detection technology offers a practical 

low cost solution for non-destructive plated-through hole inspection. 



w 

ACKNOWLEDGEMENTS 

I wish to thank my supervisors. Professor Bob Hodgson and Dr Roger Browne, for aU 

the guidance, the support, and the time they generously provided during the course of 

the work. I am also indebted to Dr Ross Nilson for his valuable comments and 

suggestions at various stages of this research. 

Messrs Dexter Muir, Ken Mercer, Farshad Nourozi. Gary Allen and Peter Haw are 

thanked for their helpful technical support. 

I also wish to thank my sister-in-law, Marlene, and her husband, Shaoquan (Dr Liu), for 

proof-reading this text. Finally, I shall always be grateful to my wife, Yvonne, for her 

understanding and tolerance to my frequent staying up late during the preparation of this 

thesis. 



CONTENTS 

ABSTRACT 
ACKNOWLEGEMENTS 
LIST OF FIGURES AND TABLES 
LIST OF ABBREVIATIONS 

CHAPTER 1 - INTRODUCTION 
1. 1 Background 
1.2 Printed Circuit Board Manufacturers 

1.2. l Example of .a small batch production facility 
1.2.2 Competitive strategy of small batch manufacturers 
1.2.3 Meeting the needs of the small batch manufacturers 

1.3 Objective, Scope and Relevance of this Research 
1.3.1 Main objective 
1.3.2 Scope 
1.3.3 Relevance 

1.4 List of Conference Papers Published 
1.5 Organisation of this Thesis 

CHAPTER 2 - PRINTED CIRCUIT BOARD MANUFACTURING 
2. I Introduction 
2.2 Standards and Specifications 

2.2. l Documentation 
2.2.2 Performance classification of PCBs 

2.3 Double-sided Plated-through Hole Printed Circuit Boards 
2.3.1 Material 
2.3.2 Manufacturing process 

2.4 Production of Plated-through Holes 
2.4. l Design requirements 

2.4.1.1 Hole size 
2.4.1.2 Plating thickness 
2.4.1.3 Acceptability criteria 

2.4.2 Key processes of plated-through hole production 
2.4.2.1 Drilling 
2.4.2.2 Deburring/desmearing 
2.4.2.3 Electroless copper plating 
2.4.2.4 Electroplating 

2.4.3 Summary 

CHAPTER 3 - PRINTED CIRCUIT BOARD INSPECTION 
3.1 Introduction 

3.1 . l In-process inspection 
3.1.2 Bare board testing and inspection 
3.1.3 The move towards automatic visual inspection 
3.1.4 Standard test patterns and test coupons 

11 

lll 

Vlll 

ix 

1 
2 
3 
3 
4 
4 
5 
5 
5 
6 
6 
6 

8 
9 

11 
11 
12 
13 
14 
15 
19 
20 
20 
21 
21 
22 
22 
23 
23 
24 
25 

26 
27 
27 
28 
29 
29 



3.2 

3.3 

3.4 
3.5 

3.6 

Surface Conductor Defects and Inspection 
3 .2.1 Common surf ace conductor defects 
3.2.2 Inspecting surface conductors 
Plated-through Hole Defects and Inspection 
3.3.1 Common plated-through hole defects 
3.3.2 Outgassing 
3.3.3 Inspecting plated-through holes 

3.3.3.1 Destructive testing 
3.3.3.2 Non-destructive testing 

Inspecting Stuffed Board Components 
Improving Plated-through Hole Inspection 
3.5.1 Disadvantages of using test coupons 
3.5.2 Leakage light detection 
Discussion 

CHAPTER 4 - PLATED-THROUGH HOLE INSPECTION USING 
IMAGE PROCESSING TECHNIQUES 

4.1 
4.2 

4.3 

4.4 

Introduction 
Spectral Reflectance Measurements 
4.2.1 Spectral reflectance of materials 
4.2.2 Equipment 
4.2.3 Sample preparation 
4.2.4 Results 
Methods to Discriminate between Plated and Unplated Regions 
4.3.1 Sample preparation 
4.3.2 Equipment 

4.3.2. l Light source 
4.3.2.2 CCD camera 
4.3.2.3 Frame grabber 
4.3.2.4 NIH Image software 

4.3.3 Image capture 
4.3.4 Image enhancement analysis 
4.3.5 Local histogram analysis 
Discussion 
4.4.1 Prospect for a prototype inspection system 

CHAPTER 5 - ALTERNATIVE SCHEMES FOR NON-DESTRUCTIVE 
INSPECTION OF PLATED-THROUGH HOLES 

5. I The PCB Panel After Electro less Plating 
5.2 Penetrant Flaw Detection 
5.3 Electromagnetic Radiation as Penetrant 
5.4 Visible Light 

5.4.1 Experiments 
5.4.2 Results 

5.5 Spectral Transmittance 
5.5.1 Transmittance of light in epoxy-glass substrate 

5.6 Microwaves 
5.6.1 Preliminary conjecture 
5.6.2 Expert advice 

31 
31 
33 
33 
34 
35 
36 
36 
37 
39 
40 
40 
41 
42 

43 
44 
44 
44 
45 
46 
48 
49 
50 
50 
50 
51 
51 
52 
53 
55 
56 
57 
59 

61 
62 
62 
63 
63 
63 
64 
65 
66 
67 
67 
68 



5.7 
5.8 

Radio Waves 
Discussion 

CHAPTER 6 - PLATED-THROUGH HOLE INSPECTION USING 
LEAKAGE LIGHT DETECTION 

6. I Leakage Light Detection 
6.1.1 The principle 
6.1.2 Assumptions 
6.1.3 Technical requirements 

6.2 Selection of a Suitable Sensor 
6.2.1 The criteria 
6.2.2 Experiments 

6.2.2.1 OPCON linescan system 
6.2.2.2 Using a phototransistor 
6.2.2.3 Using a photodiode with built-in amplification 

6.2.3 Discussion 
6.3 A Demonstration System for PTH Leakage Light Detection 

6.3 .1 Overview of the design 
6.3.2 The sensing head 
6.3.3 The illumination system 
6.3.4 Signal processing electronics 

6.4 Operating the System 
6.4.1 Inspection software 
6.4.2 Calibration 
6.4.3 Operating procedure 
6.4.4 Performance 
6.4.5 Limitations 

6.5 Discussion 

CHAPTER 7 - INSPECTION OF STUFFED BOARD COMPONENTS 
7. 1 Justification for Inspection 
7.2 Checking Component Orientations 

7 .2.1 Experimental setup 
7.2.2 Image processing techniques involved 
7 .2.3 Limitations 

7 .3 Selection from Available Equipment 
7 .3 .1 Available equipment 
7.3.2 Summary of results 

7.4 Discussion 

CHAPTER 8 - SUMMARY AND DISCUSSION 
8. I Overview 

8.1 .1 Electrical testing 
8.1.2 Visual inspection 
8.1.3 Automatic visual inspection systems 
8.1.4 Plated-through hole inspection 

8.2 Inspection Strategy of Printed Circuit Board Manufacturers 
8.2.1 Possible impact of low cost automatic inspection systems 

8.3 Research on Non-destructive Inspection of Plated-through Holes 

69 
70 

72 
73 
73 
74 
74 
75 
75 
76 
76 
78 
79 
81 
82 
82 
83 
84 
87 
88 
88 
89 
89 
90 
90 
91 

93 
94 
94 
95 
95 
98 
99 

100 
100 
100 

102 
103 
103 
103 
104 
104 
105 
106 
107 



8.3.1 Image processing techniques 
8.3.2 Leakage light detection 
8.3.3 Justification for focusing on a leakage light detection system 

CHAPTER 9 - CONCLUSIONS AND FUTURE WORK 
9 .1 Conclusions 
9 .2 Recommendations for Future Work 

APPENDICES 

107 
108 
108 

109 
109 
110 

Appendix 1-1 Some of the early printed circuit patents granted to Eisler 112 
Appendix 2-1 Panel size to manufacturing operation relationships 112 
Appendix 2-2 The manufacturing process for double-sided PCBs with PTHs 112 
Appendix 2-3 Steps of electroless copper plating 114 
Appendix 2-4 Chemistry of electroless copper plating 115 
Appendix 3-1 Sample of a PTH test pattern 116 
Appendix 3-2 Image processing techniques for surface conductor inspection 117 
Appendix 3-3 Standard micro-sectioning procedure 118 
Appendix 4-1 Listing of macros for the image enhancement method 120 
Appendix 4-2 Establishing an optimum number of pixels for each ROI 120 
Appendix 5-1 Expert comments on the application of microwave to PTH inspection 129 
Appendix 6-1 Response of the MFOD73 130 
Appendix 6-2 Characteristics of the IPL photodetector 131 
Appendix 7-1 NIB Image commands for component polarity inspection 132 
Appendix 7-2 Selection of equipment for stuffed board component inspection 132 

REFERENCES 136 



LIST OF FIGURES AND TABLES 

FIGURES 
Figure 2.1 The key manufacturing processes of PCBs 18 
Figure 2.2 Definitions of terms 20 
Figure 2.3 Example of through-hole plating thickness 21 
Figure 3.1 Common PCB surface conductor defects 32 
Figure 3.2 Common plated-through hole defects 34 
Figure 3.3 Generation and effect of outgassing 35 
Figure 3.4 The backlight test 37 
Figure 3.5 Detecting leakage light in defective PTHs 41 
Figure 4.1 Principle of measuring absorbance by the MPS-5000 46 
Figure4.2 The four samples for absorbance measurements 47 
Figure 4.3 Difference in spectral absorbance between rough and smooth 

epoxy-glass surfaces 48 
Figure 4.4 Spectral absorbance of copper and epoxy-glass 49 
Figure 4.5 Setup for image capture under a microscope 53 
Figure 4.6 Area of a sample to be captured 54 
Figure 4.7 A pair of captured images 54 
Figure 4.8 Image enhancement result compared to the actual plating voids 55 
Figure 4.9 Scatter diagram for classifying plated and unplated surfaces 57 
Figure 5.1 Spectral transmittance of epoxy-glass 66 
Figure 6.1 Alternative arrangement for detecting leakage light in defective PTHs 73 
Figure 6.2 Setup for PTH leakage light detection using the OPCON system 77 
Figure 6.3 Results of PTH leakage light detection using the OPCON system 78 
Figure 6.4 Response of the IPL photodetector 80 
Figure 6.5 Construction of the sensing head 83 
Figure 6.6 Location of the three photodetectors 84 
Figure 6.7 A cross-sectional view of the sensing head 85 
Figure 6.8 The inspection action 86 
Figure 6.9 Basic circuit diagram for the IPL photodetector 87 
Figure 6.10 The signal processing electronics 87 
Figure 6.11 Curve for calibrating the inspection system 89 
Figure 7.1 Features indicating the polarities of DIPs 95 
Figure 7.2 Images of DIPs before and after processing 97 
Figure 7.3 Images of diodes before and after processing 98 

TABLES 
Table 2.1 Selected properties of two substrate materials 10 
Table 2.2 Characteristics and dimensions of PCBs 10 
Table 2.3 Standard laminate thickness for PCB manufacture 14 
Table 2.4 Common printed circuit base materials 14 
Table 2.5 Acceptability of through-hole plating 21 
Table 3.1 Some test coupon testable parameters 30 
Table 3.2 Some stuffed board component inspection systems 39 



LIST OF ABBREVIATIONS 

ANSI 

CAD 

CCD 

DIP 

DPI 

IPC 

IC 

IR 

LLD 

PCB 

PWB 

PTH 

ROI 

SNR 

UV 

American National Standard Institute 

Computer aided design 

Charge coupled device 

Dual in line packaging 

Dots per inch 

Institute for Interconnecting and Packaging Electronic Circuits 

Integrated circuit 

Infrared 

Leakage light detection 

Printed circuit board 

Printed wiring board 

Plated-through hole 

Region of interest 

Signal-to-noise ratio 

Ultra violet 

ix 



CHAPTER ONE 

INTRODUCTION 

Printed circuit boards (PCBs) are the building blocks of almost all modern electronic 

systems. They provide a convenient way of mounting and interconnecting electronic 

components to form circuits and systems. The primary functions of the PCB are 

component support and circuit interconnection. 



Chapter 1 - Introduction 2 

1.1 BACKGROUND 

According to [THE62] the concept of the printed circuits has earlier origins than it is 

commonly supposed. A US patent dated 1903 describes ribbon cables fonned in situ by 

electro-deposition. Between 1923-1929, a number of US patents were granted covering 

some of the techniques now still in use for printed circuit manufacture [THE62]. In 

1936, P. Eisler was granted a British patent under the title "Printed Circuits" [EIS85]. 

Although Eisler' s idea of etched foil circuits eventually became the most widely used 

method of producing circuits, the idea was not immediately successful [POL84]. Some 

of the early printed circuit patents granted to Eisler can be found in appendix 1-1. 

During the Second World War the heavy demand for electronic devices stimulated the 

investigation into methods of producing circuits quickly and cheaply [POL84]. The first 

major application of the printed circuit concept occurred in 1945 when mass production 

of a United States military proximity fuse began at 5000 units a day. The circuit pattern 

was screen printed on a ceramic wafer using a conducting ink. The ink consisted of a 

binding agent, a solvent and fine silver powder. The printed ceramic plate was then 

heated to 500-800 °C, driving off the solvent and leaving the silver fused to the ceramic 

surface. The silver thus formed an excellent solid conductor and was well adhered to 

the ceramic base plate [THE62]. 

The success of the proximity fuse stimulated the launch of the printed circuit industry 

after the war. By the middle of the 1950s, printed circuits were being used in consumer 

products as well as military and industrial equipment [POL84]. Since then, the basic 

elements making up a PCB have remained unchanged. These are the base and the 

conductors. The base is a thin insulating material supporting all the conductors and 

electronic components. The conductors are customised patterns of thin metal strips, 

usually copper, finnly bonded to the base to form a laminate. 

A PCB provides the necessary interconnections for the mounted components, it is also 

referred to as a printed wiring board (PWB) [LE081]. By shaping the conductor 

patterns appropriately, it is possible to incorporate passive components such as small 



Chapter l - Introduction 3 

inductors, small capacitors and resistors directly on the board. In this case it is always 

called a printed circuit board because the functional circuitry is also printed [HAI91]. 

To avoid confusion, this thesis uses only the term Printed Circuit Board (PCB) and 

PWB is treated as a synonym of PCB. 

1.2 PRINTED CIRCUIT BOARD MANUFACTURERS 

According to [TYL94], up to 60 percent of the PCB manufacturers world wide are in 

the prototype and small batch sector. They provide services "where price is less of an 

issue and customers are prepared to pay for things in a hurry" and that these small 

companies generally "enjoy much higher profit margins than their large competitors". It 

has been estimated that the small batch sector represents I O percent of the entire PCB 

market. On the contrary, 50 percent of the total PCB output is produced by only 7 

percent of the manufacturers [TYL94]. 

In a case study report prepared for the Ministry of Research, Science and Technology, 

McNaughton [MNA92] suggests that the main problem faced by the New Zealand 

electronics industry is a matter of scale when adapting overseas technology to New 

Zealand needs. In coping with New Zealand's small production runs, specialised 

technology has to be developed to counter the scale problems. 

1.2.1 Example of a Small Batch Production Facility 

The Department of Production Technology at Massey University has established a 

manufacturing pilot plant (MPP) for electronic products. The plant is capable of PCB 

design, bare board production and component assembly. As stated in [NIL91], one of 

the many aims of the MPP is "to focus on the technology and management of a 

strategic New Zealand industry". To be of New Zealand relevance, the MPP "is a low 

cost process capable of short runs and rapid response". 

The level of technology involved is a double-sided PTH capability. Output volume has 

been designed ·at up to fifty panels ( each 305 x 406 mm or 12 x 16 inches) a day. 

Various features of the MPP are described in [NIL9 l ]. 



Chapter 1 - Introduction 4 

1.2.2 Competition Strategy of Small Batch Manufacturers 

Tyler [TYL94] identifies four key investment targets that are becoming essential for 

contemporary PCB manufacturers to survive in competition. These are: 

• Yield improvement 

• Reduction of labour cost 

• Fine line and small hole technology 

• Guaranteeing product quality in the field. 

It has been estimated that typical investment in response to such demands would require 

US $0.3-0.5 million. Drilling and visual inspection systems are the two top areas of 

spending [TYL94]. 

Unlike major PCB manufacturers that can afford to stay ahead in technology and 

equipment, the small batch manufacturers' competitive strategy is to remain profitable 

on a low capital base. For example, commercial machine vision systems are designed for 

high volume inspections and require heavy capital investments (minimum US $ 0.3 

million) [TYL94]. The small batch manufacturers will find this level of spending 

prohibitive. Most importantly, such systems fail to match the needs of small scale 

production. This is a "matter of scale" problem no different to that as described in the 

McNaughton report (section 1.2). This helps to explain why in small scale production 

human visual inspection dominates. 

1.2.3 Meeting the Needs of the Small Batch Manufacturers 

Installation of suitable low cost PCB inspection systems could help the small batch 

competitors to achieve three goals: 

• improved yield 

• reduced labour cost 

• better quality of the product. 



Chapter 1 - Introduction 5 

A higher customer confidence and higher yield will combine to create a potential of 

increased profits. Here, low cost inspection systems constitute a significant investment 

advantage for the small manufacturers. This is due to: 

• A low level of injected capital generates relatively high return 

• Minimal increase in sales is required to adequately cover the injected capital 

• Overall capital level remains low to retain profitability. 

Such an investment strategy therefore falls in line with the low capital base strategy of 

the small batch manufacturers. It can be anticipated that low cost inspection systems, 

though less sophisticated than their expensive counterparts, would meet the needs of 

most small batch PCB competitors. 

1.3 OBJECTIVE, SCOPE AND RELEVANCE OF THIS 
RESEARCH 

1.3.1 Main Objective 

The main objective of this research was to explore and develop non-destructive 

methods for PTH inspection. These methods are relevant to small batch PCB 

production such as the facility in the Department of Production Technology at Massey 

University. 

1.3.2 Scope 

This research covers only double-sided PCBs with PTHs. Work has been focused on 

low cost solutions only. The PCB manufacturing process described in this thesis is 

based on the pilot plant facilities in the Department of Production Technology at 

Massey University. Although some of the work involved may be extended to cover 

certain categories of multilayer boards, such tests have not been conducted. 



Chapter 1 - Introduction 6 

1.3.3 Relevance 

The low cost approach towards non-destructive PTH inspection is targeted at providing 

technology support to the prototype and small batch sector of PCB manufacturers. 

Such a strong orientation towards serving the small business sector is relevant to the 

New Zealand industry. 

1.4 LIST OF CONFERENCE PAPERS PUBLISHED 

The following conference papers have been published in connection with the work 

described in this thesis: 

1. K.C.Mak, RM.Hodgson, RF.Browne, RR.Nilson, 'Image processing 

techniques for non-destructive testing of printed circuit boards'. Proc. 2nd 

New Zealand Conference on Image Vision & Computing, Palmerston North, 

August 1994, pp.2.5.1-2.5.5. 

2. K.C.Mak, RM.Hodgson, RF.Browne, RR.Nilson, 'Image analysis of 

copper plated surface - towards non-destructive inspection of plated through 

holes in printed circuit boards'. Proc. Inaugural New Zealand Postgraduate 

Conference for Engineering & Technology Students, Palmerston North, 

August 1994, pp.277·-281. 

3. K.C.Mak, 'Identifying defective through-hole plating in printed circuit boards 

using the leakage light detection method'. Proc. 2nd New Zealand 

Postgraduate Conference for Engineering & Technology Students, 

Auckland, August & September 1995, pp.195-200. 

1.5 ORGANISATION OF TIIlS THESIS 

A literature review on PCB manufacture and inspection is presented in chapters 2 and 

3. Chapter 2 starts with a description of double-sided bare board manufacturing. The 



Chapter 1 - Introduction 7 

key processes of PTH production are then examined. Chapter 3 describes the status of 

PCB inspection, focussing on the conventional method of testing PTHs. The need to 

develop non-destructive inspection methods for PTHs is identified. 

The use of image processing techniques to discriminate between copper-plated and 

unplated surfaces is reported in chapter 4. The potential requirements and implications 

of applying these techniques to assess the through-hole plating quality are then 

discussed. 

Chapter 5 examines alternative schemes for non-destructive inspection of PTHs and 

highlights the potential use of visible light as a penetrant for surface flaw detection. 

Following the development in the previous chapter, chapter 6 focuses on the leakage 

light detection method and describes the development and performance of a low cost 

leakage light detection system for PTH inspection. 

Chapter 7 presents the preliminary results on establishing the hardware and software 

requirements for a stuffed board component inspection system. 

The important issues identified during this research are summarised in chapter 8. This 

chapter also discusses the inspection strategy and then examines the possible impact of 

a low cost inspection system on PCB manufacturers in the prototype and small batch 

sector. Chapter 9 concludes this thesis and recommends some directions for future 

work. 



CHAPTER TWO 

PRINTED CIRCUIT BOARD 

MANUFACTURING 

This chapter describes the major steps in bare board manufacturing by focusing on the 

production process of double-sided PCB with PTHs. This is the level of technology 

used in the Manufacturing Pilot Plant in the Department of Production Technology at 

Massey University. The production of PTHs will be described in more detail since it is 

the key step in bare board manufacture. 



Chapter 2 - Printed Circuit Board Manufacturing 9 

2.1 INTRODUCTION 

Earlier PCB constructions were single-sided, so named due to the presence of 

conductors on only one side of the board. These PCBs are still a cost-effective method 

for constructing simple low density circuit designs. Single-sided boards are the simplest 

variety of PCBs and are the least costly to manufacture [FLA92]. In the late sixties, 

processes were developed for plating copper on the walls of drilled holes in PCBs. This 

technique allowed PCBs to have conductors on both surfaces, interconnected by PTHs. 

These double-sided PTH PCBs allowed for the interconnection of medium density 

circuit designs and quickly became the industry standard [FLA92]. 

Subsequent technological development has brought many improvements to the design 

and manufacture of PCBs and alternative technologies and materials have been 

introduced. This can be illustrated by the move to multilayer technology, the reduction 

of conductor width and the use of epoxy-glass and more advanced base materials. 

Multilayer technology allows higher interconnection density than double-sided boards 

by bonding more than two conductive layers together. The multilayer boards are used 

when very dense and complex circuits are required in a limited space [MA T90]. Boards 

of up to 42 layers were manufactured in 1990 and the layer count is reported to have 

increased to 48 in the 1994 literature [HAI91,0ST91,ANG94]. 

Conductor line width has been reduced from 0.3 mm in the mid 1960's to less than 0.1 

mm in the mid 1980's [WEI89]. This trend combines with the move to multilayer 

boards to realise a significant size reduction in PCBs having complicated 

interconnections. For example, Fujitsu's M-780 computer uses a 42-layer board with 3 

Km of signal wiring at 60 µm width. The CRA Y-3 ''supercomputer'' has 40 µm wide 

conductors on its main board [NAK.92]. 

The replacement of phenolic-paper by epoxy-glass base materials gives better 

mechanical, electrical and chemical properties at lower costs [KEA87]. Table 2.1 gives 

a comparison of some properties of a typical phenolic-paper substrate, FR-2, with an 



Chapter 2 - Printed Circuit Board Manufacturing 10 

epoxy-glass substrate, FR-4. Phenolic-paper materials usually suffer from brittleness 

and high moisture absorption - a serious drawback in a humid operating environment. 

Table 2.1 Selected properties of two substrate materials 

Water Flexural Dielectric Electric Alkali 
Laminate absorbency strength breakdown arc resistance 

% si kV resistance 
FR-2 0.75 11250 60 poor poor 

(Phenolic-paper) 
FR-4 0.25 55000 45 good excellent 

Source: [ BOS83, KEA87 J 

Depending on the end-product application, the characteristics and dimensions of 

modem PCBs vary considerably [ANG94]. Table 2.2 provides an overview of such 

variations: 

Table 2.2 

Feature 

Board size 

Board thickness 

La.yercount 

Characteristics and dimensions of PCBs 

Typical value or 
material 
10 X 10 to 610 X 910 
mm 
0.3 to 6 mm 

I to 48 

Remarks 

includes multilayer boards and 
flexible circuits 
single-sided to multilayer boards 

Conductor width/spacing 0.3 to 0.04 mm 

Laminates 

Pm diameter 

Via hole diameter 

Electroless plating 

Coating materials 

various resins 

0. 7 to 1.2 mm 

0.15 to 0.7 mm 

copper, nickel 

depends on the thermal/dielectric 
requirements 
for mounting components 

for interconnection only 

thickness from 3 to 30 µm 

Sn, SnPb, NiSn, NiAu, optional 
solder masks, organics 

Source: [ ANG94 J 

Driven mainly by growing demands for very dense and complicated interconnections 

within limited space, production of multilayer boards and surface mount PCBs are the 

expanding sector of the PCB industry [l'Yl..94 ]. Nevertheless, the high accuracy 

double-sided PCB with PTHs is still one of the major categories of modem PCBs 



Chapter 2 - Printed Circuit Board Manufacturing 11 

manufactured world wide each year [FLA92,TYL94]. Compared with multilayer boards 

and surface mount boards, the double-sided PTH PCBs can be manufactured in a lower 

cost process using less sophisticated equipment. A double-sided board costs 

approximately one-fifth as much as a multilayer board [MAT90]. This advantage is 

highly significant for many inexpensive electronic products where the cost of 

production is the major consideration for manufacturers and customers. 

2.2 STANDARDS AND SPECIFICATIONS 

Printed circuit board specifications set to establish uniform quality for the finished 

products. Government organisations and professional institutions such as the US 

Department of Defense and the Institute for Interconnecting and Packaging Electronic 

Circuits (IPC) are recognised as the primary sources of initiation, implementation and 

control of PCB specifications [FLA92]. These specifications cover every sector of PCB 

manufacture: materials, design, workmanship standards, acceptability and methods of 

testing. 

2.2.1 Documentation 

The following are some documents often referred to by manufacturers and customers 

and so are regarded as industry-wide standards for bare board production: 

• Material 

This set of specifications defines the electrical, mechanical, and chemical properties 

of the material involved in PCB manufacturing such as laminate and solder. 

Examples are: 

IPC-AM-361 Specification for rigid substrates for additive process 

in printed boards 

MIL-P-13949 

IPC-SP-819 

Military specification, plastic sheet, laminated, 

copper clad, for printed wiring 

General requirements for electronic grade solder paste 



Chapter 2 - Primed Circuit Board Manufacturing 

• Design Criteria 

These specifications set the guidelines for the PCB designer: 

ANSI / IPC-D-275 Design standard for rigid printed boards and rigid 

printed board assemblies 

MIL-STD-275 Military specification, printed wiring for electronic 

equipment 

• Workmanship/ Quality assessment 

12 

These specifications quantify a11 the finished board attributes and set the acceptance 

guidelines: 

ANSI/ IPC-RB-276 Qualification and perfonnance specification for rigid 

printed boards 

MIL-P-55110 

IPC-A-600 

• Methods of Testing 

Military specification, printed wiring boards 

Acceptability of printed boards 

A collection of specific test methods and procedures including environmental 

conditioning on all fonns of printed circuits. Individual test methods are updated 

separately. The collection is intended to achieve uniformity and reproducibility for 

the testing of PCBs: 

IPC-TM-650 Test methods manual 

The customers often set additional specifications that reflect their dedicated service 

requirements in a particular product and their policy on quality and reliability. 

2.2.2 Performance Classification of Printed Circuit Boards 

Different performance requirements are set depending on the end-use of the products. 

According to the IPC specifications three general end-product classes have been 

established to reflect progressive increases in sophistication, functional performance 

requirements and inspection frequency. The definitions of this performance 



Chapter 2 - Printed Circuit Board Manufacturing 13 

classification can be found in all IPC specifications document under the section "scope". 

The following is a summary of the classification. 

Class 1 - General electronic products 

The major requirement is function of the product and cosmetic defects are not 

important. These include: 

• Consumer products 

• Some computer and computer peripherals 

• General military hardware. 

Class 2 - Dedicated Service Electronic Products 

High performance and extended life of the product is required. Unintenupted service is 

desirable but not critical. Certain cosmetic defects are allowed. Examples are: 

• Communications equipment 

• Sophisticated business machines and instruments 

• Certain military equipment. 

Class 3 - High Reliability Electronic Products 

Continued performance or performance on demand 1s vital. These include critical 

equipment such as: 

• Life support systems 

• Flight control systems 

• Special military equipment. 

Performance requirements as specified in the IPC documentation are separated so that 

PCBs may be tested to any one of the three classes. 

2.3 DOUBLE-SIDED PLATED-THROUGH HOLE 
PRINTED CIRCUIT BOARDS 

A double-sided PTH PCB has electrical conductors bonded on each side of the board 

interconnected by plated-through holes. These holes provide electrical .interconnections 



Chapter 2 - Primed Circuit Board Manufacturing 14 

between each surface of the PCB or are designated for mounting and soldering the 

component leads. The term "via hole" is reserved for a PTH intended for 

interconnection only. 

2.3.1 Material 

Material used for this type of PCB manufacture is usually copper cladded laminates. 

These laminate sheets have copper foil on both surfaces and are therefore referred to as 

double-clad. Laminates are available in various standard thicknesses. Non-standard 

thickness materials can also be obtained but are more costly [LE08 l]. Table 2.3 is a list 

of standard laminate thickness based on the military specification MIL-P-13949. 

Table 2.3 Standard laminate thickness for PCB manufacture 

LAminate Thiclazess ( based on MIL-P-13949 specification) 

mm 0.8 1.2 1.6 2.0 2.4 3.2 

inch 1/32 3/64 1/16 5/64 3/32 1/8 

Source: [MAT 90] 

Base materials for the laminate are mainly epoxy glass and polyimide glass. Polyimide 

glass is the usual material for flexible printed circuits. Table 2.4 shows the laminate 

designators and description for some of the most common base materials for rigid 

PCBs. 

Table 2.4 Common printed circuit base materials 

Laminate Description 

FR-2 Phenolic/paper 

FR-4 Epoxy/glass fabric, flame-retardant 

G-10 Epoxy /glass fabric, general purpose 

G-11 Epoxy/glass fabric, heat-resistant version of G-10 

FR-5 Epoxy/glass fabric, flame-retardant version of G-11 

Source: [ KEA87 J 



Chapter 2 - Printed Circuit Board Manufacturing 15 

The selection of a laminate type is based on costs and the specifications of the product 

[ARA89]. Hostile operating conditions such as extreme temperatures, humidity, 

corrosion, shock and vibrations must be taken into consideration in the selection 

process. 

2.3.2 Manufacturing Process 

The process starts from large sheets of laminate material being cut into panels of the 

required size. Panels are used to standardise the working dimensions of all equipment, 

clamps and fixtures in the manufacturing process [ARA89] but the ideal panel size 

mainly depends upon the plating equipment and for many manufacturers it is about 305 

x 406 mm or 12 x I 6 inches [LE08 1]. Appendix 2-1 shows the maximum panel size 

al1owed with different manufacturing operations. A panel may contain one or more 

PCBs. The number of boards in a panel is determined by the following factors 

[ARA89]: 

• Physical dimensions of the PCBs 

• Optimal geometrical placement of boards on a panel 

• Physical process equipment constraints 

• Economics of the process within panel yield. 

A panel goes through all the manufacturing steps until finally being cut up into 

individual circuit boards and the remaining part of the panel is scrapped. A typical 

manufacturing process for double-sided bare boards with PTHs involves up to fifty 

steps. Appendix 2-2 gives the sequence of operations involved. The following is a 

simplified description of the principal steps: 

• Panel drilling 

This is a mechanical process in which tooling holes and functional holes are drilled. 

Tooling holes are larger holes placed on the panel for subsequent alignment 

purposes. A common drilling practice is to stack several panels to minimise handling 

costs [LE081,KEA87]. Entry and backup materials are utilised to sandwich the 

stacked panels. These are usually aluminium or phenolic sheets and the purpose is to 



Chapter 2 - Printed Circuit Board Manufacturing 16 

reduce hole burrs. Block [BL089] provides a discussion on how these sheets are 

essential for successfully drilling PCBs. 

• Electroless copper plating 

Electroless plating is also known as chemical plating. It is a method that deposits a 

thin layer of copper, typically 5 to 7 µm over the entire panel surface as well as on 

the hole walls. This process can produce very even plating thickness [NAK.92]. Prior 

to plating, it is essential that the panels are deburred (see section 2.4.2.2) to remove 

traces of contaminants and burrs generated during the drilling operation. Otherwise, 

the success of the electroless plating will be severely affected [LE08 l,BL089]. The 

copper coating enables the through-holes to conduct electricity, hence connecting 

the conductors on both surfaces of the board. Since electroless plating is a slow 

process, usually a very thin layer of copper is deposited. Then the panels are 

transferred to an electrolytic bath to build up a further 3 to 7 µm of copper (flash 

plating). This is to condition the panel surface for the next manufacturing step 

[LE08 l ]. Details of electroless copper plating will be further described in section 

2.4.2.3. 

• Application of photoresist and expose 

This is an imaging process in which the circuit layout is transferred to surface of the 

panel. Photoresist is an organic compound that changes its solubility to certain 

solvents when it is exposed to UV light [HAI91]. Image transfer is initiated by 

applying photoresist to the panel surface and then it is exposed to UV with a mask 

carrying the circuit pattern. Unexposed photoresist is washed away using a suitable 

solvent, leaving behind the required circuit pattern on the copper surfaces. The 

function of the photoresist is to protect specific areas of copper from subsequent 

processing. All copper areas including the PTH walls that will remain on the finished 

PCB are unprotected by the resist. 

• Electroplating 

The panel is electroplated with the specified thickness of copper, usually 20 to 25 

µm is required (see section 2.4.1.2) on the uncoated circuit pattern and hole walls. 



Chapter 2 - Printed Circuit Board Manufacturing 17 

This is followed by overplating with tin-lead (or solder). The solder will protect the 

circuit pattern and the hole walls from the subsequent etching actions. The deposited 

solder must be at least 8 µm thick to be an effective etch resist [LE08 l ]. 

• Strip and Etch 

The photoresist is then removed from the panel using solvents, sometimes 

augmented by mechanical scrubbing [ARA89]. This is often referred to as resist 

stripping. After this, the redundant copper is no longer protected and can be 

removed by a chemical etchant. The etchant does not attack solder hence leaving the 

circuit pattern intact. Leonida [LE081] provides a detailed account of the process. 

• Application of solder mask and legends 

The solder mask ( or solder resist) is generally an epoxy-based resin that seals off the 

entire PCB surfaces except for those areas that require to be soldered or to make 

electrical connection externally. The solder mask serves as a barrier to solder, thus 

safeguarding a PCB against unwanted solder bridgings or short circuits during 

component soldering [ARA89]. This is essential for high specification contemporary 

PCBs as conductor widths and spacings have been decreased due to miniaturisation. 

Another function of the solder mask is to protect the condition of the PCB from 

degrading over its life span. For example, oxidation and dendritic growth of copper 

crystals are greatly reduced under the protection of solder masks. 

Legends are markings on a PCB to aid component assembly or to provide other 

information as required. Both the solder mask and legends can be applied by screen 

printing.- Wall [W AL86] gives an account of screen printing covering the types of 

product used, the areas of application and suggests methods for obtaining best 

results. It has been reported that positioning tolerances of 250 µm can be achieved 

on modem screen printing equipment [WEI89]. 

Figure 2.1 (a) to (g) illustrates the process of bare board manufacturing from drilling to 

finish. 



Chapter 2 - Printed Circuit Board Manufacturing 

copper 

epoxy-glass 

deposited 
copper 

desired circuit 
pattern 

photo 
resist 

(a) copper cladded laminate 

FR-4 

(b) panel drilling 

(c) electroless copper plating 
A vel}' thin layer of copper 
is deposited on the panel 
surface including the hole 
walls. 

( d) apply photo resist 

The desired circuit pattern 
is exposed. 

Figure 2.1 (a)-(d) Steps of a typical bare board manufacturing process 

18 



Chapter 2 - Printed Circuit Board Manufacturing 

copper plating & 
solder overpfate 

substrate 

solder mask 

(e) electroplating 

D~posite copper to specified 
thickness and overplate 
with solder 

(f) strip and etch 

Photo resist & unwanted 
copper are removed 

(g) apply solder mask 

All surfaces covered except 
areas to be soldered 

Figure 2.1 ( e )-(g) Steps of a typical bare board manufacturing process 

2.4 PRODUCTION OF PLATED-THROUGH HOLES 

19 

Plated-through holes are found in many types of PCBs. These include single-sided, 

double-sided and multilayer boards. Single-sided PCBs with PTHs are uncommon, this 



Chapter 2 - Printed Circuit Board Manufacturing 20 

particular type of board provides better soldering of component leads than ordinary 

single-sided boards [LE08 l]. 

2.4.1 Design Requirements 

The design requirements for PTHs can be found in PCB design specifications such as 

the IPC-D-275, "Design standard for rigid printed boards and rigid printed board 

assemblies". Figure 2.2 is a cross-sectional view of PCB that depicts the definitions of 

the more frequently used terms in PCB manufacturing and inspection. 

_J_ 

F 

T 

A: P11I diameter 
B: Conductor width 
C: Conductor spacing 

-11-
0 

D: P11I plating thickness 

Figure 2.2 

2.4.1.1 Hole Size 

A I- -I B 

-+ C T 

-I 1-
E 

E: Annular ring 
F: Board thickness 
G: Conductor thickness 

Definition of terms 

PTHs intended for insertion of component leads will have diameters dependent on the 

size of component leads. Automatic insertion of components requires larger PTH sizes 

than manual insertion [MAT90]. All the dimensions are referred to the finished product 

and hence the design must make allowance for the plating thickness of hole walls. 



Chapter 2 - Printed Circuit Board Manufacturing 21 

2.4.1.2 Plating Thickness 

According to the IPC-RB-276 specification, the thickness of through-hole copper plating 

should be an average minimum of 20 µm for class 1 products and 25 µm for class 2 and 3. 

Figure 2.3 depicts the plating criteria for the commonly used 1.6 mm thick substrate. 

Dimension of fearures: 

1.6 mm substrate 

35 µm base copper 

20-25µm through-hole plating 

( drawing not to scale) 

plated copper 

base copper 

substrate 

> 125 µm 

Figure 2.3 Example of through-hole plating thickness 

2.4.1.3 Acceptability Criteria 

The following table shows the requirements for acceptable through-hole plating 

according to the IPC-RB-276 specification: 

Table 2.5 Acceptability of through-hole plating 

Property Class 1 Class 2 Class 3 

Copper plating 3 voids allowed per None allowed None allowed 
voids hole. No 

circumferential voids 
over 90° allowed 

Final coa.ting 3 voids not exceeding 3 voids not exceeding 1 void not exceeding 
5% of hole wall area 5% of hole wall area 1 % of hole wall area 
oer void per void 

In addition, there shall be no separation of plating layers and no plating cracks. 

Definitions for class 1 to 3 products have been described in section 2.2.2. 



Chapter 2 - Printed Circuit Board Manufacturing 22 

2.4.2 Key Processes of Plated-through Hole Production 

The production of PTHs has been briefly outlined in section 2.3.2 on the manufacturing 

process of bare boards. The following is an account focusing on the key processes of 

PTH production. 

2.4.2.1 Drilling 

The majority of substrate materials used for this type of PCBs is epoxy-glass. Holes can 

be produced by drilling, punching or more recently plasma etching and laser ablation. 

Punching epoxy-glass laminate does not leave the clean and smooth hole required for 

plating [LE081,HED87]. As a result, drilling is the preferred method in PTH technology 

although it is a more expensive process than punching [ANG94]. Given the tight 

tolerances allowed for PTH locations and the presence of large amount of PTHs in the 

modem day PCBs, highly specialised NC drilling machines have been developed 

[KEA87]. Kea [K.EA87], Block [BL089] and Angstenberger [ANG94] discuss how vital 

the drilling process is towards the production of reliable PTHs. 

A detailed description of panel drilling can be found in [KEA87] and [LE081]. 

Vandervelde [V AN88] identifies the key variables that need to be controlled in the hole 

drilling process as: 

• Choice of drilling machine 

• Stacking and pinning of panels 

• Drilling parameters, eg. feeds and speeds 

• Drill bits 

• Entry and backup materials 

When properly controlled and maintained, bole drilling can be accomplished with 

consistent and acceptable quality. Methods to optimise key parameters have been 

discussed in [BER84,SEN86,TSU88,V AN88]. A list of criteria for assessing the drilled 

hole quality is given in [TSU88]. These are: 

• Epoxy smear (see next section) 



Chapter 2 - Printed Circuit Board Manufacturing 23 

• Burring at entry and exit points 

• Hole wall roughness 

• Hole positional accuracy 

Inspection is usually carried out immediately after drilling to extract relevant information 

about the drilled hole quality. If done properly, this would remove most of the sub­

standard holes from being plated in the electroless copper stage. 

2.4.2.2 Deburring / Desmearjng 

Burrs often occur near the drilled hole edges. A burr will eventually lead to cracked 

plating and will cause a large concentration of current locally during the electroplating 

process. Burrs can be removed using mechanical methods or an abrasive liquid or both. 

Smear refers to thin layers of substrate material generated by overheated drill bits due to 

excessive friction during the downstroke. The resin first softens, then distorts and 

smears. If left unremoved, resin smear will contaminate the plating baths and once 

metallised, lead to large reduction in the finished PTH diameter [LE081 ]. Angstenberger 

[ANG94] describes methods of desmearing and examines the selection of the optimal 

desmear method with respect to the substrate material. Smearing is usually considered as 

a problem with multilayer boards but some manufacturers also run double-sided boards 

through a desmear process to improve the adhesion of plating to the hole walls 

[LEA86c]. 

2.4.2.3 Electroless Copper Plating 

This is typically a 15-step chemical plating process. The sequence of steps is given in 

appendix 2-3. The primary purpose of the process is to make the drilled hole walls 

conductive. Electroless copper plating makes use of special activators to sensitise the 

hole walls. A panel is dipped into a solution containing stannous and palladium ions. 

Stannous ions are first deposited on the surfaces to be catalysed. Immediately following 



Chapter 2 - Printed Circuit Board Manufacturing 24 

that, palladium ions are deposited, creating metallic sites for copper deposition to take 

place. Such a panel is referred to as being activated [HAI9 l ]. 

After activation, the panel is transferred to a copper bath where copper ions are reduced 

to the metallic form with the palladium acting as a catalyst. Common plating rates are 

several micro-inches per minute or equivalent to a few µm per hour [HAI9 l] and the rate 

of deposition tends to decrease as the thickness of copper increases [LE081]. Due to 

this slow rate, . copper deposition is stopped when the thickness of the coating is 

sufficient for electroplating, or 5 to 7 µm [LE081,NAK92]. The chemistry involved in 

electroless plating can be found in appendix 2-4. 

Electroless plating is sometimes immediately followed by a flash plating of electrolytic 

copper, usually 3 to 7 µm [LE08 I]. This is the thickness required to protect the 

electroless copper if the panels are to be cleaned by a mild etch before full electroplating 

(refer to the sequence in appendix 2-2 on bare board manufacture). Flash plating also 

allows storage of the panels for some time before further processing [HED87]. 

2.4.2.4 Electroplating 

After the circuit pattern has been transferred on to the panels, electroplating of copper 

follows. This process is necessary to add thickness to the plated hole walls to meet a 

required standard. A minimum of 20 to 25 µm of copper is required (section 2.4.1.2). 

The overall metal thickness will determine the thermal and current-carrying capabilities 

of the finished PTHs [HAI91]. 

The deposition rate of electroplated metal is a function of current density and current 

distribution. Current density is greatest at the comers, edges and isolated holes but is 

lowest in recessed areas [BID87]. Rough hole walls are therefore undesirable as non­

uniform current density affects the plating thickness. The presence of organic 

contaminants affects the current distribution. All these factors contribute to the variation 

of plating thickness across a panel [LE081]. 



Chapter 2 - Printed Circuit Board Manufacturing 25 

2.4.3 Summary 

Electroplating of copper is followed by the plating of tin/lead or etch resist. The 

subsequent steps are common to the production of PCBs either with or without PTHs. 

These steps have already been outlined in section 2.3.2. 

It can be seen that the production of PTHs involves a complicated multi-step process 

which is highly susceptible to drifts. Thus the inspection of PTHs plays an important role 

in assuring the quality of PCBs. This will be discussed in the next chapter on PCB 

inspection. 



CHAPTER THREE 

PRINTED CIRCUIT BOARD 

INSPECTION 

This chapter is a description of PCB inspection concerning bare board surface 

conductors, PTHs and stuffed board component placement. Focus will be on the 

destructive and non-destructive methods of PTH inspection. A wide range of other 

inspection issues such as board warpage, hole pattern accuracy, annular rings, 

solderbility and solder joints are not covered. 



Chapter 3 - Primed Circuit Board Inspection 27 

3.1 INTRODUCTION 

The main objective to inspect any product is to ensure that its quality meets the 

specifications. A secondary but equally significant objective is to extract information 

associated with the production process for the benefit of improving the process. Also 

trends are monitored to allow process control and so the key parameters can be kept 

within specification. 

Inspection is costly since considerable time and labour are consumed in the process. 

Therefore the choice of features to inspect, the way in which samples are taken and the 

way in which inspections are conducted all need to be considered carefully. For 

example, trying to inspect every board of an inexpensive product may not be cost­

effective. Kear [KEA87] lists four factors that influence the degree and the extent of 

the inspection to be used. These are: 

• Cost of the circuit board. 

• Function of the circuit board. 

• Design of the circuit board. 

• Statistical quality history. 

3.1.1 In-process Inspection 

As described in chapter 2, the production of bare PCBs is a multi-step process. Each 

step has its variability and so contributes to the overall product reject rate. Apart from 

the final inspection that qualifies a certain product as acceptable, in-process inspection 

must also be conducted. This is a sequence of inspection tasks at various stages of the 

production cycle. Any defects identified could help to determine whether a panel should 

be scrapped or reworked at that stage. Otherwise the final reject rate could be very 

high. As Tyler [TYL94] puts it: .. boards rejected at the end of the production process 

contain the maximum added value of a company's profit and loss account". The 

information extracted from in-process inspection also helps to monitor any drifts in the 

manufacturing process. Thus, in-process inspection is an important means to reduce 

scrap losses and to minimise the cost of rework. 



Chapter 3 - Printed Circuit Board Inspection 28 

3.1.2 Bare Board Testing and Inspection 

To verify for product acceptability, the finished bare boards are normally subjected to 

both electrical testing and visual inspection. Acceptability is established by meeting the 

various requirements defined by the customer specifications, or any such standards 

recommended by government and industrial organisations. One of these, for example, 

is IPC-B-276, "Qualification and Performance Specification for Rigid Printed Boards". 

This specification will be frequently referred to in this chapter. 

Another imponant role of end-product inspection is to feed back information to the 

personnel in charge of production. In this way, corrective measures can be implemented 

as soon as possible should products fail to meet the specifications. 

Electrical testing is particularly efficient in detecting open or short circuits. The testing 

facility can be automated to enhance the testing effectiveness for high density 

interconnections. This usually involves a fixed array of probing pins (so called "bed-of­

nails") each connected to the automatic test station. Resistances at different points of 

the finished product can be measured according to a testing sequence [HAI9 l]. 

The commonly used "bed-of-nails" systems are criticised as having major drawbacks in 

[ANG87]. These are: 

• Testing component patterns in small grids may become difficult. 

• With low voltage equipment, any line width violation may not be reponed; 

running the assembly in its final designation with higher voltages may cause 

interrupts of the affected interconnections. 

• There is no possibility of 100% testing of closely toleranced line widths. 

It follows that relying on electrical testing alone to verify the product quality is not 

sufficient. Currently, visual inspection is the major means to identify both existing and 

potential defects in a PCB. Pau [PAU90] points out that up to 80 percent of PCB 

inspections are visually based. Naturally human vision, as a well-developed and 

sophisticated system, can be employed to accomplish the task. 



Chapter 3 - Printed Circuit Board Inspection 29 

3.1.3 The Move towards Automatic Visual Inspection 

With growing interconnection density and miniaturisation of modern day electronic 

circuits, human visual inspection has become increasingly inefficient and error prone 

[D0Y84]. Double-sided and multilayer PCBs with line widths less than 150 µm and 

PTHs smaller than 0.6 mm in diameter are already manufactured in large quantities 

[WEI89]. Regardless of whatever magnifying devices are used, the human visual 

system is subject to fatigue. According to [ANG87], the limits of human visual 

inspections related to fatigue of human inspectors are as follows: 

• The capability of finding any defects decreases exponentially with time. 

• Depending on the complexity of a given PCB, after one hour of intensive 

inspection as many as 70% of the existing defects will be missed. 

Another factor that discourages employing manual inspection is merely the huge 

volume of PCBs produced each year. In 1994, the world wide output was valued 

around US $26 billion [CUS94]. In coping with large scale automatic production, the 

only workable solution would be to also automate PCB inspection [AND88]. Tyler 

[TYL94] points out that the US PCB industry is currently spending 1.4 percent of its 

annual turnover on automatic optical inspection equipment and the installation of this 

equipment is growing by 14 percent each year. Such demand to perform visual 

inspection efficiently and reliably has already brought the application of digital image 

processing techniques into the subject of printed circuit board inspections. As will be 

seen in section 3.4, a number of machine vision systems have been designed for such 

applications. 

3.1.4 Standard Test Patterns and Test Coupons 

Test patterns are special conductor patterns used to define test coupons or test boards. 

They are intended as a reference in evaluating a PCB manufacturer's material and 

processes. A test coupon is defined as "a sample or test pattern usually made as an 

integral part of the printed board, on which electrical, environmental and 

microsectioning tests may be made to evaluate board design or process control without 



Chapter 3 - Primed Circuit Board Inspection 30 

destroying the basic board" [C0088]. The use of test coupons and their significance 

towards PCB quality control are highlighted in [MC092]. The coupon results are taken 

as a «true" indication of the board quality although this may not always be the case. 

Appendix 3-1 shows a standardised test pattern for PTHs and illustrates how test 

coupons can be located on a production panel. Table 3.1 summarises the range of 

parameters that can be checked with the use of test coupons. 

Table 3.1 

Manufacturing Process 

Drilling Quality 

Electrokss Copper 

Imaging 

Etching 

Desmearing 

Plating 

Solder Fusion 

Solder Resist 

Legend 

Some test coupon testable parameters 

Example Parameters 

Hole Location 
Drill Smear 
Burring 
Drill Wear out 

Plating Thickness 
Adhesion 
Bath Decomposition 

Registration Accuracy 
Exposure/ Print Quality 

Bath Activity 
Degree of Etching 
Definition 

Hole Cleanliness 
Etchback Quality 

Bath Chemical Balance 
Cracking 
Adhesion 

Wetting Quality 
Solderability 

Registration Accuracy 
Adhesion 
Definition 

Registration 
Character Definition 

Source: [MC092] 



Chapter 3 - Printed Circuit Board Inspection 31 

Many manufacturers always include test coupons in the design and production of circuit 

boards. Test coupons are selectively removed from panels for inspection at convenient 

stages of the production process. As described in [MC092], "the test coupon is taken 

as a 'true' indication of the board quality and thus is inspected thoroughly while the 

main board is briefly checked". However, whether a coupon may be taken as a "true" 

indication of the board quality is a subject to investigate. A "major US corporation" has 

carried out an investigation on its PCB manufacturing facility and revealed that "the 

extent of agreement (of PTH quality) between the various test coupons and 

corresponding intelligent board areas was not satisfactory within or across boards" 

[HAY89]. The results of the investigation suggest that the coupon-to-board correlation 

has to be established using statistical process control. 

3.2 SURFACE CONDUCTOR DEFECTS AND INSPECTION 

For a double-sided PCB, the surface conductors appear on both surfaces of the board 

that form the circuit interconnection. Any surface conductor defects will likely affect 

the circuit reliability by introducing unexpected open or short circuits. Common 

occurring surface conductor defects are well documented [KEA87,PAU90]. These 

features have been made a requirement for bare board inspection in most specifications 

such as the IPC-RB-276. 

3.2.1 Common Surface Conductor Defects 

The defects found in bare PCBs can be grouped into the following three types: 

1. Dimensional 

These may occur anywhere on a board as unacceptable variations of board or 

conductive feature geometries. Examples are warped board, wrong board thickness, 

improper conductor width and wrong hole size. The dimensional defects violate the 

PCB design specifications. 



Chapter 3 - Printed Circuit Board Inspection 32 

2. Cosmetic 

Such defects affect the appearance but not the function of the circuit board. Examples 

are discoloured or scratched base material and smeared markings [KEA87]. Normally, 

cosmetic defects only need some touch-up work. 

3. Functional 

These are often more difficult to evaluate than the other types of defects [KEA87]. In 

defining functional defects, many PCB customers specify their own requirements along 

with the usual industry-wide specifications. Different forms of functional defects exist 

but they all affect the integrity of the circuit interconnection. Figure 3.1 illustrates some 

common defective features: (a) protrusion; (b) nick; (c) dent; (d) improper edge 

definition; (e) unetched copper; (f) short circuit; (g) open circuit; (h) intrusion; (i) 

pinhole; G) void; (k) conductor peeling. 

a 

conductor 

substrate 

C 

b 

vertical section 

f 

conductor 
tracks 

/ ---
· oio .: . .. ,·J ' 

.. : . . ~:; 

Figure 3.1 Common PCB surface conductor defects 



Chapter 3 - Printed Circuit Board Inspection 33 

3.2.2 Inspecting Surface Conductors 

According to the IPC-RB-276 regarding test methods, visual inspections are carried 

out at a minimum of 1.75x magnifications. Conventional PCB surface conductor 

inspection has been carried out by human inspectors. Human visual inspection is still 

widely in use today, especially in small production volume facilities. The majority of 

PCB manufacturers in this category derives profits on a low capital base. They cannot 

afford expensive investment in automatic inspection equipment (see section 1.2.2). 

The major manufacturers are more inclined towards using machine vision systems to 

inspect PCBs [AND88,PAU90,TYL94). These machines use three alternative image 

processing approaches [TH088,P A U90]: 

1. Comparison Technique 

2. Feature Detection Technique 

3. Design Rule Verification Technique. 

A brief description on the application of these techniques in PCB inspection is given in 

appendix 3-2. 

3.3 PLATED-THROUGH HOLE DEFECTS AND INSPECTION 

As pointed out in [BL089], many PTH defects can be traced back to the original 

drilling process. In [MIL84], some typical through-hole plating defects are analysed and 

suggestions on taking certain preventive measures during panel drilling are given. The 

causes leading to the occurrence of plating voids are identified in [DOU85] as: 

• cavities on hole walls originated from drilling problems 

• inefficient removal of drilling debris from the hole walls 

• problems with desmearing procedures 

• drifts in the electroless process. 



Chapter 3 - Printed Circuit Board Inspection 34 

3.3.1 Common Plated-through Hole Defects 

Common PTH defects are: 

• Cracked hole walls 

• Voids and pinholes in plating 

• Blistering 

• Contaminated hole walls 

• No connection between hole plating and surface plating layers 

• Hole not plated. 

Some examples of defects taken under a microscope are shown in figure 3.2. The 

samples were taken out after the electroless plating stage of the production process. 

Cracks, voids and pinholes are revealed using back light. 

Cracks and voids Voids and pinholes 

Blistering Contamination 

Figure 3.2 Common plated-through hole defects 



Chapter 3 - Printed Circuit Board Inspection 35 

3.3.2 Outgassing 

Outgassing refers to the release of gas from the PCB substrate underneath the through­

hole plating when the components are soldered. This is due to areas of porosity or 

weakness in the PTH wall electroless plating [LE081,HOW86]. Leonida [LE081] 

explains how this problem can be traced back to the roughness of drilled hole walls: a 

nick in the wall will not be covered by the thin electroless plating. Subsequent electro­

overplating will be uneven due to concentration of current on sharp edges; the cavity 

will be closed by the overplated copper, but leaving a pocket in it. During soldering, the 

thermal expansion of the substrate opens the pocket, releasing any trapped compound. 

This is outgassing and is also known as blowholing. Figure 3.3 (a)-(d) depicts such a 

sequence of events. 

(a) 

(c) 

hole 
wall 

substrate 

pocket 
formed 

.----/ electroplated 
copper 

(b) 

(d) 

~ uneven growth 
of copper 

blowhole due to 
thermal expansion 
during soldering 

Figure 3.3 Generation and effect of outgassing 

Source: [LE081] 

It has been established in [FEL88] that the degree of outgassing faults is proportional 

to the hole wall roughness and inversely proportional to the copper plating thickness. 

Given that the drilling process bas been optimised, the electroless plating stage is the 

most important process to be controlled in order to eliminate the outgassing or 

blowholing problem. In a study when all the drilling parameters were kept constant, it 



Chapter 3 - Printed Circuit Board Inspection 36 

was found that the outgassing problem varied by a factor of 20, and the variation 

correlated well with the degree of electroless copper voids present [LEA87]. The study 

concluded that the outgassing problem could be improved by: 

• ensuring a minimum average thickness of 20 µm copper on the hole walls 

• the complete absence of electroless voids. 

It therefore follows that the electroless plating stage is the single key process in the 

production of reliable PTHs. 

3.3.3 Inspecting Plated-through Holes 

Detecting the presence of plating voids is a prime concern in PTH inspection. In a 

comprehensive survey of possible void formation mechanisms involving 20 panels each 

containing 3000 PTHs, it was found that virtually all the voids observed were 

associated with exposed glass fibre [LEA86c]. The large voids were associated with 

exposed lengths of glass and the small voids with exposed ends of glass. 

3.3.3.1 Destructive Testing 

The current standard practice of inspecting PTHs is destructive. The common 

procedure is micro-sectioning and then the backlight test is applied to the sectioned 

sample. After the through-holes are plated, a row of PTHs is first selected and removed 

for micro-sectioning. This is done through the centre of the row as shown in figure 3.4 

(a), resulting in a thin slice with several half-holes. Appendix 3-3 gives the standard 

procedure of micro-sectioning as referred to by the IPC-RB-276 standard. 

Micro-sectioning can provide a direct linear measurement of the plating thickness, but 

the process also yields additional information on the nature, uniformity and presence of 

voids or nodules of the plating [LE081 ]. In detecting plating voids, hole walls are 

exposed and the plated surfaces are inspected directly under a microscope using back 

illumination as shown in figure 3.4(b). Since copper deposition is opaque to light, 

plating voids can easily be detected as bright regions against a dark background. This 



Chapter 3 • Printed Circuit Board Inspection 37 

is known as the backlight test. Figure 3.4(c) shows a PTH with plating voids revealed 

by the backlight test. 

line of 

(a) Section through the centre of PTHs 

! Viewing 

f Illumination f 
(b) View under microscope ( c) Plating voids revealed 

Figure 3.4 The back.light test 

The micro-sectioning process is destructive and time-consuming. According to 

[DAW87], the repeatability, the time required to micro-section and the costs are the 

major problems of this manual operation. The task of producing acceptable micro­

sections has been identified by manufacturers as being a major bottleneck [DA W87]. 

Typically it takes 1 to 2 hours to prepare a micro-section and requires a highly skilled 

operator [LE08I]. lf applied incorrectly, this elaborate process could induce damage to 

the plating before inspection. The nature of the test also makes it difficult to automate 

the process [DA W87]. 

3.3.3.2 Non-destructive Testing 

In contrast to the destructive testing, four non-destructive methods for testing through­

hole plating have been found in the literature review. These are: 



Chapter 3 - Printed Circuit Board Inspection 38 

• Manual inspection using PTH viewer 

A PTH viewer is essentially a fish-eye lens that allows a 360-degree view of the hole 

wall from outside the hole. This instrument has been specially designed for the 

inspection of drilled and plated-through holes in PCBs. Obviously, trained inspectors 

equipped with this specialised tool and working under suitable illumination can 

readily report the presence of cracks or voids. 

The board to be inspected should be placed on a light-table or similar source of 

diffuse, below-surface, illumination. A typical viewer such as the Mexter™ is 

positioned on the board until a hole is at the centre of the field of view. By manually 

zooming in, the wall of a hole can be observed at various levels. The magnification 

of the viewer varies with the position of the zoom but the maximum is 

approximately x60. Holes up to a depth of 3 mm can be inspected. 

• Beta ray backscatter 

This method relies on the emission of electrons from a metallic layer subjected to a 

beam of beta ray. The emission of electrons decreases with increasing plating 

thickness [LE081,WIC87]. The actual arrangement is said to be rather complicated 

and careful calibration with known standards is essential. This method is reported as 

popular in the testing of precious metal coating such as gold [LE081]. 

• Micro-resistance 

This method uses pulses of direct-current to measure precisely the true resistance of 

the through-hole copper barrels [WIC87,LAT93]. The resistance values involved are 

typically a few hundred micro-ohms. This method is sensitive in finding 

circumferential cracks but has limited capability in detecting small but significant 

voids scattered around the through-hole copper barrel. 

• Leakage light detection 

This is a method that can be fully automated. Further description will be found in 

section 3.5.2. 



Chapter 3 - Printed Circuit Board Inspection 39 

3.4 INSPECTING STUFFED BOARD COMPONENTS 

Stuffed boards are PCBs assembled with electronic components. According to 

[PAU90], the break-down of all electronic assembly errors is as follows: 

• 

• 

• 

• 

insertion errors 

missing components 

wrong polarities 

wrong components 

55% 

20% 

15% 

10% 

It is reported that no vision systems for checking wrong component values such as 

reading resistor bands existed then [ROB89]. Table 3.2 lists several commercial stuffed 

board component inspection systems currently available: 

Table 3.2 Some stuffed board component inspection systems 

System Inspection/or Hardware Comments 

CA/V-1000 missing components; CCD camera 64 grey levels; 
component leads ( 128x 128 pixels) 80 components /sec 

ORS-1000 missing components x-y table and template matching; 
CCDcamera inspects 90 locations 

per minute 

Octek component insenion; TV camera; section-by-section 
component polarity; positioning table scanning 
reading IC part numbers 

CHECK- component leads; require specialised 
POINT missing, damaged or unknown (CHECKPOINT) 
PCB wrong components; processors 

reading IC part numbers 

IntellVue component leads; uses component 
DR2000 missing components; unknown design rules for 

component polarity inspection 

Fujitsu rectangular capacitors 3 TV cameras; highly specialised 
and insertion, positioning He-Ne laser expensive equipment 
and polarity 

Source: [ZUE 87,AND88,ROB89 J 



Chapter 3 - Printed Circuit Board Inspection 40 

These inspection systems are intended for high volume production facilities. Each 

system is designed for a very specific task. Completely automatic inspection of stuffed 

boards would mean major capital investment in a number of such systems to cover the 

wide range of possible assembly errors. 

3.5 IMPROVING PLATED-THROUGH HOLE INSPECTION 

In 1990, 92.1 percent of the US output of rigid PCBs were either double-sided or 

multilayer and worth 5 billion US dollars [FLA92]. In 1993, the European Electronic 

Component Manufacturers Association (EECA) reported that double-sided and 

multilayer technologies accounted for 80 percent of the total demand within the 

European Union PCB market valued at 3.5 billion Ecus [INT93]. 

Notwithstanding the advance of surface mount technology, a huge volume of PCBs 

having PTHs will be produced each year. Any improvements in the current methods of 

PTH inspection (excluding via holes in surface mount boards) will have prominent 

economic significance. 

3.5.1 Disadvantages of Using Test Coupons 

As mentioned earlier in section 3.1.4, there has been a study on the variation of 

through-hole plating quality across a panel. It is reported in [HA Y89] that the use of 

test coupons could not truly reflect such variations unless the coupon-to-board 

correlation had been established properly. The article describes the effort and 

methodology used to tackle the problem. Approximately US $250,000 was eventually 

spent by the corporation towards establishing coupon-to-board correlation for a 

particular family of PCBs. This can be justified by the annual cost saving of about US $ 

5 million for the high volume PCB manufacturer. As for the small volume 

manufacturers, this level of spending will be out of their reach. There are also economic 

considerations involved in setting up and running a coupon-based process control 



Chapter 3 - Printed Circuit Board Inspection 41 

system. Such facts are significant to the small PCB manufacturers and warrant the 

research in non-destructive methods for inspection of PTHs. 

3.5.2 Leakage Light Detection 

In reviewing the literature on PTH inspection, there have been very few publications on 

non-destructive methods. A "leakage light detection" system for inspecting PTHs is 

described in [AND88]. The principle of leakage light detection for PTHs is illustrated in 

figure 3.5. 

leakage 
light 

plating voids 

illuminatio" I I 

Figure 3.5 Detecting leakage light in defective PTHs 

A PTH inspection . system has been built and is reported to be in use by the major 

Japanese PCB manufacturer Fujitsu. The system detects voids, cracks and pinholes in 

the through-hole plating [AND88]. An area of 300 mm square can be inspected in three 

minutes. The major specifications for this PTH inspection system are: 

Type of board 

Size of board 

Detectable defects 

Speed of inspection 

Physical size of the system 

Multilayer 

300 mm square 

voids, cracks, pinholes 

20s per l 00 mm square of panel 

1.4 X 1.6 X 1.3 m 



Chapter 3 - Printed Circuit Board Inspection 42 

3.6 DISCUSSION 

From the literature review, a number of automatic visual systems exist for the 

inspection of PCB surf ace conductors and stuffed board component placement. In 

contrast, the conventional inspection of PTHs relies largely on the micro-sectioning 

method. This destructive method is the standard practice recommended by the 

industrial specifications such as the IPC-RB-276. 

The absence of electroless voids is an important indicator of good quality PTHs apart 

from sufficient plating thickness (see section 3.3.2 on outgassing). Unconventional but 

non-destructive methods including the micro-resistance and beta ray backscatter are 

primarily directed at measuring the through-hole plating thickness rather than detecting 

the tiny and scattered voids. Manual inspection using a specialised viewer can only cope 

with a small output volume of PCBs. Even so the human visual inspection has certain 

limitations as already outlined in section 3.1.3. The Fujitsu leakage light detection 

system is automatic and carries out non-destructive inspection. However, this is a 

system designed for high output volumes and does not necessarily match the small tO 

medium sized production facilities. 

Consequently, researching new technology for PTH inspection has the potential of 

developing cost-effective in~pection systems for the small to medium PCB 

manufacturers. The concept of applying image processing techniques for the non­

destructive inspection of PTHs [MAK94 ], if proven practical in a real production 

environment, would be a most useful contribution to this subject. The research into this 

particular topic will be the focus of the next chapter. 



CHAPTER FOUR 

PLATED-THROUGH HOLE 

INSPECTION USING 

IMAGE PROCESSING TECHNIQUES 

This chapter reports on the experimental work targeted at the analysis of images 

obtained from PTH sections under a microscope. It will be shown that this approach 

has the potential to develop into a new and useful non-destructive inspection technique. 



Chapter 4 - Plated-through Hole Inspection Using Image Processing Techniques 44 

4.1 INTRODUCTION 

Digital image processing is a mature technology which has found applications in many 

areas. Hodgson [HOD92] gives a wide range of examples on applications that are 

relevant to New Zealand. The non-contact, non-destructive nature of image processing 

is particularly appealing for inspection related applications. It is one of the technologies 

that could bring considerable economic significance to the PCB industry. 

Structurally, the interior of a perfect PTH is a copper plated cylinder. This is often 

referred to as the copper barrel. Plating defects are violations of the integrity of the 

copper barrel. These defects are local unplated areas where the epoxy-glass substrate is 

still exposed even after the plating process (see section 3.3.1). 

When illuminated, a given surface reflects and absorbs various amounts of light at 

different wavelengths according to the physical properties of the material and the 

condition of the surface. Metallic copper and epoxy-glass each reflects light in a 

distinctive way. Hence a copper barrel with plating voids will exhibit irregular spectral 

reflectance when compared to a fully plated copper barrel. The degree of irregularity 

will vary according to the area of plating voids. This hypothesis led to an investigation 

into the possibility of detecting any such irregularities. The first step was to measure the 

spectral reflectance of copper and epoxy-glass. 

4.2 SPECTRAL REFLECTANCE MEASUREMENTS 

4.2.1 Spectral Reflectance of Materials 

The spectral reflectance of a surface can be defined as the distribution of returned light 

relative to the incident light at different wavelengths. Reflectance depends on the 

wavelength and the angle of incidence of the light, as well as the angle(s) at which 

reflected light is measured [MEY91]. 



Chapter4 - Plated-through Hole Inspection Using Image Processing Techniques 45 

Opaque materials reflect light in two ways. One is specular reflection and the other is 

diffuse reflection. Specular reflection depends mainly on the surface conditions of the 

material. Diffuse reflection is scattered and contains the absorption characteristics of the 

material. The spectral reflectance of materials other than highly polished surfaces are 

mainly diffuse reflectance and will vary selectively with wavelength [MEY91]. 

Measurements of spectral reflectance are made incrementally, wavelength by 

wavelength, relative to the reflectance of a calibrated standard. This eliminates the need 

to calibrate the spectral sensitivity of the detector. Spectral reflectance of materials is 

normally measured by a spectrophotometer. 

4.2.2 Equipment 

A spectrophotometer is the usual equipment for measuring spectral reflectance of 

materials. It records the spectral distribution of light reflected or absorbed relative to a 

standard reference sample at different wavelengths. 

A Shimadzu MPS-5000 spectrophotometer was used. This equipment measures the 

absorbance of incident light relative to a standard reference. The reference sample is 

usually a piece of white filter paper. Percentage absorbance can be related to percentage 

reflectance by the simple subtraction: 

% reflectance = 100 - % absorbance. 

According to the Shimadzu Instruction Manual, a tungsten-iodine lamp is used as the 

light source for visible and near infrared wavelengths (320 to 2500 nm) and a deuterium 

source for the ultraviolet wavelengths ( 185 to 320 nm). The detectors are a photo­

multiplier and a PbS cell. These are designated to operate in the UV-visible and near 

infrared regions of the spectrum respectively. 

When working in the reflectometry mode, the spectrophotometer is designed to 

eliminate almost all specular reflections using mirrors and a light trap. A large fraction 



Chapter 4 - Plated-through Hole Inspection Using Image Processing Techniques 46 

of the diffuse reflections is captured by the large detecting surface of a photomultiplier 

placed closely to the sample as depicted in figure 4.1. 

Photocathode 

Mirror 

Light 

Detector 

Source: Shimad.zu MPS-5000 Instruction Manual 

Figure 4 .1 Principle of measuring absorbance by the MPS-5000 

The manufacturer claims that this arrangement yields better results than those 

obtainable with an integrating sphere, particularly for glossy opaque samples. The 

range of incident wavelengths can be varied continuously within each of the selected 

working ranges: 

I. UV (185 to 320 nm) 

2. Visible and near IR (320 to 850 nm) 

3. Nea:- IR (850 to 2500 nm). 

Measurements are recorded automatically in real time through an analogue chart plotter 

built into the system. No digital output of measurements is available. 

4.2.3 Sample Preparation 

Four slabs of material were taken from a standard single-sided FR-4 material 1.6 mm 

thick. Those labelled 1 and 2 were samples of the copper-clad surface. Samples 3 and 4 

were two epoxy-glass samples. Sample 3 was the smooth substrate surface of the FR-4 

material. Sample 4 was made by sticking together thin slices of cut substrate using an 



Chapter 4 - Plated-through Hole Inspection Using Image Processing Techniques 47 

epoxy resin adhesive. The unpolished cut edges of epoxy-glass jointly formed the 

surface of sample 4. Figure 4.2 illustrates the three types of surfaces represented by the 

four samples. 

Copper foil (1 & 2) Smooth epoxy-glass (3) Rough epoxy-glass (4) 

Figure 4.2 The four samples for absorbance measurements 

The reasons for having two samples representing the epoxy-glass surf ace are: 

• There is a varying degree of roughness in the drilled hole walls. 

• The majority of plating voids are associated with exposed glass fibre: large and 

small voids are associated with the exposed lengths and cut ends of glass 

respectively (see section 3.3.3). 

Sample 3 was intended to represent the surface of unplated smooth hole walls 

associated with the lengths of glass. Sample 4, which consisted of unpolished cut edges, 

was intended to model unplated rough surfaces that are associated with the cut ends of 

glass fibre. It is believed that the range of surface conditions of any unplated substrate 

would not normally exceed the extremes set by samples 3 and 4. 

Samples 1 and 2 were taken from two different parts of the FR-4 laminate. By having 

two pairs of samples for each material, the extent to which spectral reflectance is 

affected by the surface roughness of either copper or epoxy-glass can be examined. All 

samples were cut into rectangular slabs of approximately 25 mm by 30 mm to fit into 

the sample slot of the spectrophotometer. 



Chapter 4 - Plated-through Hole Inspection Using Image Processing Techniques 48 

4.2.4 Results 

Measurements were taken when the incident light varied from violet to near infrared ( or 

350 to 800 nm in wavelength). Two independent measurements were made for each 

sample and were compared. It was found that: 

Result l The two measurements for each sample agreed to within 2 %. 

Result 2 The measured spectral absorbance of the two copper samples (samples I 

and 2) agreed to within 6 %. 

Result 3 The two epoxy-glass samples (samples 3 and 4) had similar variation of 

spectral absorbance within the sensitivity of the spectrophotometer. But 

the actual absorbance values between these samples at the same 

wavelength were different (see figure 4.3 below). The smoother surface, 

or sample 3, had lower values. This difference diminished as the 

wavelength was increased to the near IR region. 

0 4---------+----4---------l 
350 

Figure 4.3 

450 550 650 750 850 

Wavelength (nm) 

Difference in spectral absorbance between rough and 
smooth epoxy-glass surfaces 

Result 1 demonstrates that the spectrophotometer performance had been consistent. 

Result 2 shows that the two copper samples had very similar surface conditions. Result 

3 suggests that rough and smooth epoxy-glass surfaces have similar characteristics in 

spectral absorbance (or reflectance). Variations in surface roughness affect the 

magnitude but not the change in absorbance (or reflectance) across the incident light 

spectrum. The smoother the surface, the more light it reflects. The results for copper 

and rough epoxy-glass are superimposed and summarised in figure 4.4: 



Chapter 4 - Plated-through Hole Inspection Using Image Processing Techniques 

100 

CP 75 i 
(.) 
C 

50 ~ 0 
.0 
0 
"' .0 

25 I 
< 
~ 0 

J 
350 400 

Figure4.4 

450 

·----·---- ... ------------·------------ -. - .. 

500 

--Copper 
······Epoxy 

550 600 650 

Incident Ught Wavelength ( nm ) 

700 750 

Spectral absorbance of copper and epoxy-glass 

49 

800 

As seen from the above graph, a comparatively significant difference in spectral 

reflectance exists between metallic copper and epoxy-glass from 650 to 800 nm. This 

range is represented by the red and near IR region of the electromagnetic spectrum. It 

can be concluded that copper reflects red and IR radiation more strongly than epoxy­

glass does. This result is perhaps not unexpected given the reddish colour of the fresh 

copper plating. 

4.3 METHODS TO DISCRIMINATE BETWEEN PLATED AND 
UNPLATED REGIONS 

The ultimate objective of having any machine vision inspection systems is to recognise 

accurately any defects deemed to be unacceptable according to the specifications. This 

task has to be accomplished within a reasonable duration of time. A secondary but 

certainly important function of an inspection system is to provide additional 

information, such as indications of the nature and locations of the defects. In through­

hole plating quality assessment, the ultimate objective translates into discriminating 

between plated and unplated regions within a PTII in the shortest possible time. 

As already shown in section 4.2.4, there is a relatively large difference in spectral 

reflectance between copper and epoxy-glass at wavelengths 650 to 800 run. This 

difference offers a possibility to differentiate between the two types of surfaces. In the 



Chapter 4 - Plated-through Hole Inspection Using Image Processing Techniques 50 

case of PTHs, the smallest plating voids to be detected are tiny pockets of exposed ends 

of glass fibre engulfed in much larger areas of copper. The dimensions of such voids can 

be 20 µm across [HOW86], depending on the diameters of the glass fibre in the 

material. However, the intrinsic difference of plated and unplated surfaces in reflecting 

light creates cenain features that can possibly be extracted and classified. 

4.3.1 Sample Preparation 

Double-sided PCBs of 1.6 mm thickness with PTHs of 1 mm diameters were used. The 

PCBs have been plated with electroless copper and then sectioned into thin stripes 

containing a row of half PTHs. These samples were identical to the those intended for 

the backlight test as already described in section 3.3.3.1. 

4.3.2 Equipment 

The essential equipment consisted of: 

• A microscope at X 10 magnification 

• A sample holder specially designed to hold the sample stripe under the 

microscope lens 

• A high power cold light source with a flexible circular light guide 

• A colour CCD camera fitted to the microscope 

• A Macintosh Quadra computer with SCION frame grabber and NIH Image 

software. 

4.3.2.1 Light Source 

The Schott KL 1500 cold light source was employed to illuminate the samples. This 

equipment uses a halogen bulb, operating at around 3000 K colour temperature. At this 

temperature a substantial proportion of the output will be in the near IR spectrum 

[W AK86]. The light intensity can be regulated without change in colour temperature. 

Standard optical filters or user-made special optical glasses may be fitted into the light 



Chapter 4 - Plated-through Hole Inspection Using Image Processing Techniques 51 

source, provided they are of the specified dimensions. Output light is delivered through 

optical fibre light guides. At the light guide outlet the maximum illumination is 10 Mix. 

4.3.2.2 CCD Camera 

A CCD is basically an integrated circuit consisting of an array of photosensor sites that 

converts incident photons into electrical charge. The charge is proportional to the 

amount of light absorbed [SCH94]. A row of photosensor sites can be seen as a shift 

register of capacitors. Each of these. holds a charge proportional to the incident light 

and can be shifted to one end of the row to be read. The CCD has evolved from its 

earliest form of a simple 8-bit shift register into a complicated integrated device suitable 

for various imaging applications. It has already out-performed photographic films in 

sensitivity, spectral range, stability, dynamic range and linearity [SCH94]. 

The CCD camera used was the Ikegami ICD-835 single chip colour CCD. According to 

the instruction manual, the camera uses the interline transfer architecture. This 

arrangement consists of a parallel array of line sensors separated by a transfer electrode 

from masked CCD read-out registers that lead in parallel into a single output register 

[SCH94]. The entire image from the line array is shifted simultaneously into read-out 

registers. The main advantage of this arrangement is fast image read-out. A new image 

is integrated while the previous-one is being read. The camera is equipped with built-in 

colour filters for capturing colour images. The single chip CCD contains 390,000 

pixels. 

4.3.2.3 Frame Grabber 

The primary function of a frame grabber is to digitise video frames from an imaging 

device such as the CCD camera. The output from a CCD camera is typically an 

analogue composite signal that conforms to a certain video standard. This analogue 

signal needs to be digitised by the frame grabber as a frame of pixels before it can be 

processed by a computer. 



Chapter 4 - Plated-through Hole Inspection Using Image Processing Techniques 52 

The SCION LG-3 was the frame grabber used for the work. It is designed for the use 

with RS-170 ( or CCIR) CCD cameras and RGB video sources. When connected to an 

ROB source, the LG-3 can capture frames from each of the colour signals. The LG-3 

performs 8 bit (256 levels) analogue to digital conversion. If the incoming signal is 

greater than or equal to the upper limit of input voltage, it will receive a digital value of 

255; conversely, an input less than or equal to the lower limit of input will be assigned a 

digital value of 0. 

The limiting voltage levels can be controlled by adjusting the analogue offset and gain 

of the digitisation process. For example, if the video signal is low, the limits may be 

lowered to "brighten up" the captured image. Similarly, if the input signal has poor 

contrast, the voltages can be squeezed together to increase the overall contrast in the 

captured image. After digitisation, some processing can be done on the digital image 

using a look-up table. A look-up table maps each of the possible pixel values to a new 

value. Up to eight input look-up tables are provided on the LG-3. Each may be read, 

written and used for processing the incoming video signal. 

Some specifications for the LG-3 

Video signal type: RS-170 or CCIR 

Digitising speed: 1/30 s (RS-170); 1/25 s (CCIR) 

Image resolution: 640x480 (RS-170); 768x512 (CCIR) 

Capture mode: field or frame 

Source: SCION LG-3 Technical Manual 

4.3.2.4 Nill Image Software 

NIH Image is a public domain image processing and analysis software for the 

Macintosh. It can acquire, display, edit, annotate, enhance, analyse, print and animate 

images. The software reads and writes a range of image file formats including TIFF, 

PICT, PICS and MacPaint. NIH Image supports many standard image processing 

functions such as histogram equalisation, contrast enhancement, edge detection, median 

filtering, density profiling and spatial convolution with user defined kernels up to 



Chapter 4 - Plated-through Hole Inspection Using Image Processing Techniques 53 

63x63. More complicated image processing functions including morphological 

operations can also be implemented. The software incorporates a PASCAL-like macro 

programming language to automate complicated and frequently repetitive tasks. 

The LG-3 will initially process the digitised image by inverting the pixel values to make 

the image compatible with the NIH Image software. In contrary to the usual practice, 

this software interprets grey scale values with black as 255 and white as 0. 

4.3.3 Image Capture 

The samples need to be illuminated by incident as welJ as back light. Incident light was 

provided by a flexible circular light guide placed directly above the sample. Back 

lighting was achieved by the built-in illuminating system of the microscope underneath 

the sample. The images captured under incident and back illumination are referred to as 

the frontlit images and the backlit images respectively. Figure 4.5 shows the setup for 

image capture under a microscope. 

Figure 4.5 

circular 
light guide 

sample 
holder 

Setup for image capture under a microscope 

A pair of frontlit and backlit images for each sample was captured. Although only the 

frontlit images are analysed, the results need to be verified using the corresponding 

backlit images. Each sample was placed under the microscope. At a magnification of 

x 10, the microscope will have less than 0.3 mm depth of focus. Since the samples were 

half-holes of approximately 0.5 mm depth, focusing was aimed at the bottom part of the 



Chapter4 - Plated-through Hole Inspection Using Image Processing Technit/ues 54 

groove while the upper parts remained out of focus. Figure 4.6 depicts the portion of 

each sample being in-focus. Effectively an area covering about 75 percent of a sample 

could be captured. 

area captured 

Figure 4.6 Area of a sample to be captured 

A backlit image of the sample was first captured using back illumination. Then with 

front illumination turned on and back illumination turned off, a corresponding frontlit 

image was captured. For perfect matching of the pair of images, no relative motion 

between the camera and the sample was allowed during the entire process. This 

procedure was repeated for a number of samples. A typical image pair is shown in 

figure 4.7. 

(a) frontlit 

Figure4.7 

(b) backlit 

A pair of captured images 

Since subsequent analysis could involve images at different wavelengths of the visible 

spectrum, a colour CCD camera was used. The NIH hnage software provides a 



Chapter 4 - Plated-through Hole Inspection Using Image Processing Techniques 55 

function to separate a colour image into its red, green and blue components if that is 

desirable. 

4.3.4 Image Enhancement Analysis 

It can be hypothesised that some correlation exists between a pair of frontlit and backlit 

images of the samples taken. When reflectance properties are concerned, at the visible 

red band of the spectrum, unplated regions (exposed epoxy-glass) should appear 

relatively dark against the plated copper. This reasoning led to the application of image 

enhancement techniques to the red component of frontlit images to highlight those dark 

areas. The actual processing was implemented by the NIH Image software in PASCAL 

like macro commands. Appendix 4-1 gives the listing of macros for the image 

enhancement method. 

Figure 4.8(a) shows the red component of a frontlit image covering a PTH section. The 

result of enhancement is given in figure 4.8(b). The dark regions indicate problem areas 

of the plated surf ace. This pattern can be compared to the actual plating voids as 

revealed by the backlit image, figure 4.8 (c). 

Figure 4.8 

r 

• 

(a) 

.. r; . .. 
r 

.,._ --.. ... . .. . . 
• I • 

(b) (c) 

Image enhancement result compared to the actual plating voids 

The pattern in figure 4.8(b) does not constitute an exact match to the voids shown as 

white in figure 4.8(c) but is fairly close for some of the voids to be detected. It indicates 

that the current image enhancement techniques still need some refinement. This method 



Chapter 4 - Plated-through Hole Inspection Using Image Processing Techniques 56 

is also sensitive to uneven illumination. To ensure repeatable and reliable results, 

illumination must be carefully controlled to minimise the presence of shadows when the 

frontlit images are captured. 

The same technique had been applied to the green and blue components of the frontlit 

samples taken under white light. Results were much less promising than the red 

component. It confirmed with the deductions from spectral reflectance measurement 

that images at the red band of the visible spectrum should yield the best discrimination. 

4.3.5 Local Histogram Analysis 

The variation of light intensity in the frontlit images can be analysed using histograms of 

pixel intensity values. Essentially each histogram is a frequency distribution of intensity 

levels within a clearly defined area on an image, or the region of interest (ROI). This 

distribution varies according to conditions such as illumination, material and surface 

roughness. Every histogram has certain features that can be analysed. Jain [JAI89] 

provides a list of common histogram features which includes moments, dispersion, 

mean, median, mode, variance, mean square value, skewness and kurtosis. 

In the case of PTHs, the presence of defects will affect the distribution of light 

intensities and hence some of the associated histogram features. Consequently the 

histogram for an unplated surface will exhibit certain features as distinct from the 

histogram for a surface that is completely plated with copper. These distinctive features 

can be regarded as signatures associated with either the plated or unplated surfaces. By 

plotting the appropriate histogram features in multi-dimensional feature space, a plating 

void may be readily recognised by its characteristic signature. Industrial applications of 

this technique have been found in the literature; one example is the inspection of IC and 

LSI (large scale integration) packages for potential defects such as holes, hollows and 

protrusions [EJI89]. 

In applying this technique to PTH inspection, the first step is to extract certain 

histogram features from ~e frontlit images and to construct the signatures associated 



Chapter 4 - Plated-through Hole Inspection Using Image Processing Techniques 57 

with the plated and unplated surfaces. Initially a total of 87 sample ROis were taken 

from different frontlit images. The corresponding backlit images were used as templates 

to guide the selection. In this way, each ROI can be classified as plated or unplated. The 

optimum ROI size suitable for discriminating between the two types of surfaces was 

found to be 55 pixels. The method used to establish this number is outlined in appendix 

4-2. The histogram features useful for constructing the signatures were found to be the 

sample size, the range (the number of grey levels or contrast) and the standard deviation 

(S.D.) of the pixel intensities. These features were then extracted from the histograms 

of those R0Is that consisted of more than 55 pixels. Data from 45 samples are plotted 

into a 2-dimensional scatter diagram as shown in figure 4.9. 

50.00 1 
++ 

+ * ++ + 
ci + + + 
en 40.00 , + + - 30.00 l + + + + ++ 
G) + + + 
N A + .;; i A + + G) 

20.00 l A A ++ a. ~iAP A 
++ 

E 

1000 t 0 ~ Cl) + Plated Surface 

A Unplated Surface 

0.00 

0 5 10 15 20 

Range/ S.O. 

Figure 4.9 Scatter diagram for cla