Master Thesis

Electrical Engineering
Thesis no: MEE 06-18
June 2018

Performance and
Improvement Investigation of
Accelerated Temperature
Change Test

Vandana Narri

This thesis is presented as part of Degree of

Master of Science in Electrical Engineering

Blekinge Institute of Technology

June 2018

Blekinge Institute of Technology Scania CV AB

School of Engineering Automotive Industry Company
Department of Applied Signal Processing RECT Department
Supervisor: Prof. Wlodek J. Kulesza Supervisor: Per Nordling
Examiner: Dr. Sven Johansson Group Manager: Jon Lindholm

i
This thesis is submitted to the School of Engineering at Blekinge Institute of Technology in
partial fulfillment of the requirements for the degree of Master of Science in Electrical
Engineering with Emphasis on Signal Processing. The thesis is equivalent to 20 weeks of full
time studies.

ii
Contact Information:
Author:
Vandana Narri
Blekinge Institute of Technology
Address: SE – 15132 Södertälje, Sweden
E-mail: [email protected]

Supervisors:
Prof. Wlodek J. Kulesza
School of Engineering, Blekinge Institute of Technology
Address: SE – 371 79 Karlskrona, Sweden
Phone: +46 455 385898
Email: [email protected]

Per Nordling, Development Engineer.

Scania CV AB, RECT
Address: Stockholm, Sweden.
Phone: +46 855 385575
Email: [email protected]

iii
Abstract
This thesis is focused on Accelerated Temperature Change Test (ATCT). This test is
performed in order to deteriorate and age a product in a faster than natural way. The ATCTs
are primarily controlled with four parameters: the temperature range (∆T), ramp rate (RR),
dwell time (DT) and number of cycles in test (𝑁𝑡 ). A comprehensive investigation was
carried out to analyze the performance and functionality of the cabinet, used for performing
ATCT. This was achieved by performing repeated measurements in the empty cabinet with
different combinations of temperature range (∆T) and dwell time (DT). The test parameters
are normally adjusted according to values given in the standards common in automotive
environmental testing and also according to the thermal properties and size of the tested
components. In this investigation, four different standards for accelerated testing were taken
into consideration. They are TB1900, ISO 16750-4:2010 (International Organization for
Standardization), IEC 60068-2-14 (International Electrotechnical Commission) and The
GMW3172: User Guide. From each standard, each parameter definitions are illustrated.

Practical tests were executed on three different types of components and with different test
conditions. A suitable experimental-setup was prepared to record the temperature
measurements on/in the DUT (Device Under Test). This experimental-setup was designed
using thermocouples (Type - K) and DEWESoft. The results from the test were used to
analyze the deviation between the standards and practical testing. This comparison helps in
understanding the required improvements in the test parameters values i.e. the cabinet
parameter settings or the test conditions. The values assigned for each parameter before the
test are called Cabinet Parameter Settings (CPS). The temperature readings from the DUT
are recorded and plotted after ATCT. Based on these ATCT measurements, optimal values
of the test parameter are estimated. These are called Estimated Parameter Values (EPV). A
significant deviation is observed between CPS and EPV.

From these EPV, an acceleration factor (AF) for each test is calculated using two different
life prediction models i.e. (i) Coffin-Manson and (ii) Norris Landzberg. And using this AF,
an evaluation of the “number of cycles in the field (𝑁𝑓 )” with certain “number of cycles in
the test (𝑁𝑡 )” is made. This evaluation helps in understanding the effect of parameter values
during the test on the acceleration conditions. A simple aluminum box (one of the test
component) is replicated into a simple structure, to implement in COMSOL Multiphysics
Simulations. When the simulation results show good agreement with practical results, then
simulations are recommended to be used to find the proper test conditions and test parameter
values. Further, the simulations are used to find the sensitive point in/on the component.
These simulations take some reasonable efforts.

Index Terms— Acceleration factor, Dwell time, Fatigue failure, Accelerated temperature
change test, Stress level, Temperature range, Coefficient of Thermal Expansion.

iv
Acknowledgment
This thesis was carried out at Scania CV AB, Södertälje, Sweden, under the supervision of
Per Nordling. Many thanks for his support and patience.

I would like to thank Prof. Wlodek Kulesza for his support and guidance throughout this
work.

Also, special thanks to Dan Magnusson for giving me the opportunity to develop my thesis
in a great environment surrounded by the best team of scientists in the field.

Thanks to RECT department for their help during measurements day. I would also like to
thank Mattias Forslund and Nulifer Ipek for their happy and knowledgeable support.

v
vi
Table of Contents

Abstract ................................................................................................................................ iv
Acknowledgment .................................................................................................................. v
Table of Contents ................................................................................................................ vii
List of Figures ....................................................................................................................... x
List of Tables ...................................................................................................................... xiv
List of Abbreviations .......................................................................................................... xv
1 INTRODUCTION ......................................................................................................... 1
2 BACKGROUND AND RELATED WORKS ............................................................. 4
2.1 SN-Curve............................................................................................................................. 4
2.2 Accelerated Testing – Temperature Change Test ............................................................... 5
2.3 Temperature Sensors ........................................................................................................... 6
2.4 Modes of Heat Transfer....................................................................................................... 7
2.5 Dependency of Accelerated Temperature Change Test Parameters ................................... 8
2.6 Failures Modes due to Mismatch in Coefficient of Thermal Expansion............................. 9
2.7 Standards of Accelerated Temperature Change Test ........................................................ 12
2.8 The Coffin-Manson and Norris Landzberg Models .......................................................... 13
2.9 Review of Related Works ................................................................................................. 18
3 PROBLEM STATEMENT AND MAIN CONTRIBUTIONS ................................ 22
4 WORKING PRINCIPLES OF THE TEMPERATURE SHOCK CHAMBER
AND TEST SET-UP ........................................................................................................... 26
4.1 Features and Working Principles of the Chamber............................................................. 26
4.2 Test Set-Up to Record Temperature Change in the Cabinet. ............................................ 28
4.3 Test Cycles to Record Temperature Change in an Empty Cabinet. .................................. 31
4.4 Experiment Set-up for Component Testing....................................................................... 35
4.4.1 Data Acquisition System - DEWESoft ........................................................................ 35
4.4.2 Experiment Setup ......................................................................................................... 36
5 CURRENT SCENARIOS OF COMPONENT TESTING IN TEMPERATURE
SHOCK CHAMBER .......................................................................................................... 39
5.1 Motor: Three DUTs at the Same Time. ............................................................................. 40
5.2 Aluminum Box .................................................................................................................. 46

vii
 5.3 Test on Plastic Box under Similar Test Conditions as Aluminum Box ............................ 52
5.4 Comparison Between Standards and Current Scenario of Testing ................................... 55
5.5 Conclusions from Current Scenario Testing ..................................................................... 55
6 ESTIMATION OF THE AF FROM CURRENT TESTING SCENARIO ............ 58
6.1 Effect on AF When Compared to Position of the Component .......................................... 58
6.2 Effect on AF When Compared to Size and Material of the Component ........................... 61
6.3 Effect on AF when Test Parameters Varying – Norris Landzberg Model ........................ 63
6.4 Summary ........................................................................................................................... 69
7 SOFTWARE VALIDATION ..................................................................................... 72
7.1 Simulation Models ............................................................................................................ 72
7.2 Physics Applied ................................................................................................................. 74
7.3 Simulation Scenarios ......................................................................................................... 75
7.4 Simulations Results and Comparison with Results from Current Test Scenario .............. 77
7.4.1 Simulation Scenario 1 - 2D Model ............................................................................... 79
7.4.2 Simulation Scenario 2 - 3D Model ............................................................................... 81
7.5 Observations ...................................................................................................................... 83
8 CONCLUSIONS AND FUTURE WORK ................................................................ 86
8.1 Conclusions ....................................................................................................................... 86
8.2 Future Work ...................................................................................................................... 88
References............................................................................................................................ 90
Appendix A: Illustration of Fixing of Thermocouples on the component surface ....... 92
Appendix B: Validation of Thermocouple Set-up ........................................................... 93
Appendix C: Calculation of Ramp Rate (RR) from the ATCT Results ........................ 95
Appendix D: Comparison of Temperature Response from the PCB and the Air Inside
the Plastic Box. .................................................................................................................... 97

viii
ix
List of Figures

Figure 2.1 S-N curve. .......................................................................................................................... 4

Figure 2.2 Temperature cycle test profile. .......................................................................................... 5
Figure 2.3 Thermocouple Type K (TCK). [8] ..................................................................................... 7
Figure 2.4 Illustration of conduction, convection and radiation[11] . ................................................. 8
Figure 2.5 Dependence relationship of thermal cycling parameters [13]. .......................................... 9
Figure 2.6 Effect of S-N slope on Accelerated Testing [14]. Note: The use of value ‘m’ is explained
in section 2.8. .................................................................................................................................... 10
Figure 2.7 Component made up of two different materials. 𝛼1 ≠ 𝛼2. ............................................ 11
Figure 2.8 Stress generated at the junction of both the materials due to different CTEs and increase
in temperature. 𝛼1 ≠ 𝛼2. ................................................................................................................. 11
Figure 2.9 𝛼1 ≅ 𝛼2. Even thought CTEs of both the materials are almost similar, stress can be
generated due to thermal gradient. It is not necessarily the way the figure depicts. ......................... 12
Figure 2.10 Definition of DT in different standards[16]. Here 𝑇𝐴 = room temperature, 𝑇𝐵 = 𝑇𝑚𝑎𝑥
and ∆𝑇 = stability state. The red line defines DT for TB1900 and IEC. The green line defines DT
for ISO and GMW3172. .................................................................................................................... 13
Figure 2.11 Plot of DT AF versus 𝐷𝑇𝑡𝑒𝑠𝑡. DT AFs are calculated using (2.5) and (2.6).
Assumptions provided in Table 2.3 are used in calculations. ........................................................... 16
Figure 2.12 Plot of High temperature AF versus 𝑇𝑡𝑒𝑠𝑡 𝑚𝑎𝑥. High temperature AFs are calculated
using (2.7) and (2.8). Assumptions provided in Table 2.4 are used in calculations. ......................... 17
Figure 2.13 Plot of RR AF versus 𝑅𝑅𝑡𝑒𝑠𝑡. RR AFs are calculated using (2.9) and (2.10).
Assumptions provided in Table 2.5 are used in calculations. ........................................................... 18
Figure 4.1 Front view of the temperature shock cabinet [29]. .......................................................... 26
Figure 4.2 Interior of the two-chamber cabinet with basket in hot chamber. ................................... 27
Figure 4.3 Illustration of air flow in hot cabinet and cold cabinet respectively (side view). The broad
arrows indicate the temperature at respective chambers with different basket location in the cabinet.
........................................................................................................................................................... 28
Figure 4.4 Representation of sections in the basket. Each section is explained in the ...................... 28
Figure 4.5 Location of nine thermocouples in the basket. TCK 1, TCK 2, TCK 3 on section 1,
closed to the fan of the cabinet. TCK 4, TCK 5, TCK 6 on section 2, placed in middle of the basket
and TCK 7, TCK 8, TCK 9 on section 3, placed closer to the door of the cabinet. .......................... 30
Figure 4.6 Location of two thermocouples in constant positions. TCK 10 in the cold chambers and
TCK 11 in the hot chamber. .............................................................................................................. 31
Figure 4.7 Test 2: temperature response from thermocouples TCK 1, TCK 2 and TCK 3............... 32
Figure 4.8 Test 2: temperature response from thermocouples TCK 4, TCK 5 and TCK 6............... 33
Figure 4.9 Test 2: temperature response from thermocouples TCK 7, TCK 8 and TCK 9............... 33
Figure 4.10 Test 2: temperature response from TCK 11 and TCK 10 in hot and cold chamber
respectively. Green line is representing the temperature recording from TCK 11 and pink line is
representing the temperature recording from TCK 10. ..................................................................... 34
Figure 4.11 A top view of the hardware unit of data acquisition system (DEWE-43). .................... 35
Figure 4.12 Schematic diagram of the experimental set-up. ............................................................. 36

x
Figure 4.13 Picture of connections between thermocouples, adapters and DEWE-43 channels during
the ATCT. ......................................................................................................................................... 37
Figure 4.14 Picture of recording temperature response in ATCT. .................................................... 37
Figure 5.1 Motor used in ATCT........................................................................................................ 41
Figure 5.2 Outline of top view of the basket with three components parallel to the door. ............... 42
Figure 5.3 Outline of top view of the basket with three components perpendicular to door. ........... 42
Figure 5.4 Temperature recording during the test on component placed in the middle. Units of time
in minutes and temperature in ℃....................................................................................................... 43
Figure 5.5 Temperature recording of the air between two components in the basket. Units of time in
minutes and temperature in ℃........................................................................................................... 44
Figure 5.6 Schema of dismantled motor parts used for testing. ........................................................ 45
Figure 5.7 Temperature recordings from six different locations on parts of the motor including the
hot chamber log for reference. .......................................................................................................... 46
Figure 5.8 Drawing of aluminum box used for ATCT...................................................................... 47
Figure 5.9 Drawing of aluminum box placed horizontally in the basket. ......................................... 47
Figure 5.10 Drawing of aluminum box placed vertically in the basket. ........................................... 48
Figure 5.11 Photo of designed stand which is used to help the component stand vertically in the
basket................................................................................................................................................. 48
Figure 5.12 Drawing of aluminum Box with open lid. ..................................................................... 49
Figure 5.13 Drawing of aluminum Box with open lid placed vertically in the basket. ..................... 49
Figure 5.14 Temperature response measured on the outer surface of the aluminum box from ATCT.
The graph includes three different test cases on aluminum box. ...................................................... 50
Figure 5.15 Temperature response measured at the air inside the aluminum box from ATCT. The
graph includes three different test cases on aluminum box. .............................................................. 50
Figure 5.16 Temperature response measured on the plastic inside the aluminum box from ATCT.
The graph includes three different test cases on aluminum box. ...................................................... 51
Figure 5.17 Drawing of plastic Box used for ATCT. ........................................................................ 52
Figure 5.18 Outline of plastic box placed vertically in the basket. ................................................... 52
Figure 5.19 Temperature response measured on the outer surface of aluminum box and plastic box
from ATCT........................................................................................................................................ 53
Figure 5.20 Temperature response measured in the air inside the aluminum box and plastic box
from ATCT........................................................................................................................................ 53
Figure 5.21 Temperature response measured from the plastic inside the aluminum box and from the
PCB inside the plastic box from ATCT. ........................................................................................... 54
Figure 6.1 Temperature response of air side the aluminum box for AF calculations. ...................... 59
Figure 6.2 Estimated values of 𝑁𝑓 for aluminum box in three test cases using Coffin-Manson and
Norris Landzberg for lead free solder (m=2.65) and leaded solder (m=2.5). ................................... 60
Figure 6.3 Temperature response inside Aluminum box and Plastic box. ........................................ 62
Figure 6.4 Estimated values of 𝑁𝑓 for aluminum box and plastic box using Coffin-Manson and
Norris Landzberg for lead free solder (m=2.65) and leaded solder (m=2.5). ................................... 63
Figure 6.5 AF verses DT in ATCT. Where DT is varied from 0 minute to 100 minutes and AF is
calculated according to the DT value. ............................................................................................... 65
Figure 6.6 Number of cycles in field (𝑁𝑓) is calculated from the AF calculates in Figure 6.5 and
with x-axis representing the DT varying from 0 minute to 100 minutes. ......................................... 65

xi
Figure 6.7 AF verses RR in ATCT. Where RR is varied from 0 ℃/minute to 20 ℃/minute and AF
is calculated according to the RR value. ........................................................................................... 67
Figure 6.8 Number of cycles in field (𝑁𝑓) is calculated from the AF calculates in Figure 6.7 and
with x-axis representing the RR varying from 0 ℃/minute to 20 ℃/minute. .................................. 67
Figure 6.9 Number of cycles in field (𝑁𝑓) is calculated with varying ∆𝑇 in ATCT. ....................... 69
Figure 6.10 Flowchart describing the dependency of factor for improving testing .......................... 69
Figure 7.1 Structure and material of a component for simulations. .................................................. 73
Figure 7.2 Physics applied in 2D and 3D model. .............................................................................. 74
Figure 7.3 Input temperature signal for simulations – 3 cycles – Used in 3D model. ...................... 75
Figure 7.4 Input temperature signal for heating the component to the maximum temperature – Used
in 2D model. ...................................................................................................................................... 75
Figure 7.5 2D simulation model........................................................................................................ 76
Figure 7.6 3D simulation model without screws. .............................................................................. 77
Figure 7.7 3D simulation model with screws. ................................................................................... 77
Figure 7.8 Temperature response from current test scenario, when the component is placed
horizontally. ...................................................................................................................................... 78
Figure 7.9 Temperature response from current test scenario, when the component is placed
vertically. ........................................................................................................................................... 78
Figure 7.10 Temperature distribution in the box (2D model). .......................................................... 79
Figure 7.11 Simulated temperature response for first 50 minutes of heating (2D model). ............... 79
Figure 7.12 Temperature distribution in the box (2D model). .......................................................... 80
Figure 7.13 Simulated temperature response for first 50 minutes of heating (2D model). ............... 80
Figure 7.14 Temperature distribution (3D model without screws) after 50 minutes of heating. ...... 81
Figure 7.15 Temperature distribution (3D model without screws) after 50 minutes of cooling ....... 81
Figure 7.16 Temperature graph for 312 minutes for 3D model without screws. .............................. 82
Figure 7.17 Temperature distribution (3D model with screws) after 50 minutes of heating. ........... 82
Figure 7.18 Temperature distribution (3D model with screws) after 50 minutes of cooling. ........... 83
Figure 7.19 Temperature graph for 312 minutes for 3D model with screws. ................................... 83
Figure 8.1 Example of usage of additional stand, which helps in improving the test. ...................... 87
Figure 8.2 Stand used in the testing. ................................................................................................. 87

xii
xiii
List of Tables
Table 2.1 : Comparison between standards. ...................................................................................... 12
Table 2.2 : Representation of parameters when used in field and test respectively. ......................... 14
Table 2.3 : DT AF versus 𝐷𝑇𝑡𝑒𝑠𝑡 plot.............................................................................................. 15
Table 2.4 : High Extreme Temperature AF versus 𝑇𝑡𝑒𝑠𝑡 𝑚𝑎𝑥 plot. ................................................ 16
Table 2.5 : RR AF versus 𝑅𝑅𝑡𝑒𝑠𝑡 plot.............................................................................................. 17
Table 4.1 : Description for each basket section................................................................................. 29
Table 4.2 : Placement description of the thermocouples in the cabinet along with motivation for
logging of temperature in an empty cabinet, to obtain the temperature distribution in the cabinet. . 30
Table 4.3 : CPS for test cycles in an empty cabinet. ......................................................................... 31
Table 5.1 : Types of components tested. ........................................................................................... 39
Table 5.2 : Types of test cases performed to understand different test scenarios. ............................ 40
Table 5.3 : CPS for testing of the motor in test case(i) and test case(ii). .......................................... 41
Table 5.4 : CPS for case(iii) i.e. dismantled components testing. ..................................................... 44
Table 5.5 : The position of all the thermocouples used during the test case (iii). ............................. 45
Table 5.6 : CPS for testing of the aluminum box. ............................................................................. 47
Table 5.7 : Comparison of CPS and EPVs for the aluminum box for three test cases. For the all the
test cases the DT is 50 minutes. ........................................................................................................ 51
Table 5.8 : CPS for testing of the plastic box.................................................................................... 52
Table 5.9 : Comparison of CPS and EPV for aluminum box and plastic box. They are positioned
vertically in the basket during the ATCT. ......................................................................................... 54
Table 5.10 : Comparison between standards and current scenario of testing.................................... 55
Table 6.1 : CPS programmed for the test. These values are applicable to all the three test cases for
the aluminum box. ............................................................................................................................. 59
Table 6.2 : Comparison CPS and EPV for temperature response obtained from ATCT of aluminum
box. EPVs estimated from the temperature graphs (in Figure 6.1). .................................................. 60
Table 6.3 : CPS programmed for the test. These values are applicable for vertical positioned testing
of the aluminum box and plastic box ................................................................................................ 61
Table 6.4 : Estimated parameter values from the temperature graphs (in Figure 6.3), obtained from
the tests. ............................................................................................................................................. 62
Table 6.5 : Assigned parameter values used in Norris-Landzberg model with varying 𝐷𝑇𝑡𝑒𝑠𝑡 ...... 64
Table 6.6 : Assigned parameter values used in Norris-Landzberg model with varying 𝑅𝑅. ............ 66
Table 6.7 : Assigned parameter values used in Norris-Landzberg model with varying 𝑇𝑡𝑒𝑠𝑡 𝑚𝑎𝑥
and 𝑇𝑡𝑒𝑠𝑡 𝑚𝑖𝑛. .................................................................................................................................. 68
Table 7.1 : Simulations were performed using following scenarios and test cases. ......................... 72

xiv
List of Abbreviations

∆𝑇 Temperature Range.
∆𝑇𝑓𝑖𝑒𝑙𝑑 Temperature Range for Field Usage.
∆𝑇𝑡𝑒𝑠𝑡 Temperature Range for Test Usage.
AT Accelerated Testing.
ATCT Accelerated Temperature Change Test.
AF Acceleration Factor.
𝐴𝐹𝐿𝑒𝑎𝑑−𝑓𝑟𝑒𝑒 𝑠𝑜𝑙𝑑𝑒𝑟 Acceleration Factor for Lead-Free Solder.
𝐴𝐹𝐿𝑒𝑎𝑑𝑒𝑑 𝑠𝑜𝑙𝑑𝑒𝑟 Acceleration Factor for Leaded Solder.
ALT Accelerated Life Testing.
CPS Cabinet Parameter Setting.
CTE Coefficient of Thermal Expansion.
DT Dwell Time.
𝐷𝑇𝑓𝑖𝑒𝑙𝑑 Dwell Time for Field Usage.
𝐷𝑇𝑡𝑒𝑠𝑡 Dwell Time for Test Usage.
DUT Device Under Test.
EPV Estimated Parameter Value.
FEA Finite Element Analysis.
ISO International Standards Organization.
MLE Maximum Likelihood Estimation.
𝑁 Number of Cycles to Fail.
𝑁𝑓 Number of Cycles in the Field (Real Time Cycles).
𝑁𝑡 Number of Cycles in the Test.
PCB Printed Circuit Board.
RR Ramp Rate.
𝑅𝑅𝑡𝑒𝑠𝑡 Ramp Rate for Test Usage
SN-Curve Stress verses Number of Cycles to Fail.
𝑇𝑚𝑎𝑥 High Extreme Temperature.
𝑇𝑓𝑖𝑒𝑙𝑑 𝑚𝑎𝑥 High Extreme Temperature on Field.
𝑇𝑡𝑒𝑠𝑡 𝑚𝑎𝑥 High Extreme Temperature in Test.
𝑇𝑚𝑖𝑛 Low Extreme Temperature.
𝑇𝑓𝑖𝑒𝑙𝑑 𝑚𝑖𝑛 Low Extreme Temperature on Field.
𝑇𝑡𝑒𝑠𝑡 𝑚𝑖𝑛 Low Extreme Temperature in Test
TCK Thermocouple Type-K.
TE Thermoelectric.
UR Usage Rate.
UTF Usage to Failure.

xv
xvi
1 INTRODUCTION

Accelerated Temperature Change Tests (ATCTs) are performed in order to deteriorate and
age a product in a faster than natural way [1]. The tests are primarily controlled by four
parameters: the Temperature Range (∆T), Ramp Rate (RR), Dwell Time (DT) and number of
cycles in test (𝑁𝑡 ) [2]. Failures, which occur under conditions of fluctuating stress are called
fatigue failures [3]. The increasing complexity of electrical and electromechanical systems
in vehicles has prompted the significant development in automotive industry. The usage of
such complex electrical systems in commercial vehicles has led to increase in the number of
EEE-components (Electronics Electric and Electro-mechanics).

Natural ambient thermal fluctuations or induced thermal fluctuations (such as in or around

an engine) frequently or constantly occur on vehicles and causes stress in components mainly
by variation in thermal expansion, which in turn affects their reliability and life length.
Accelerated stress test simulates the reality but much more accelerated by increasing the rate
or impact of one or more of the environmental stresses, which a product is exposed to during
its lifetime. By performing these accelerated stress tests, one can determine how these
environmental stresses affects the product and sometimes makes it possible to predict its life
length.

However, despite being accelerated, the tests are relatively time consuming and, especially
for large components, the relevance of the tests can be questioned due to the complexity or
uncertainty of the heat transfer. The main motive of this thesis is to understand temperature
changes in and on the Device Under Test (DUT) during ATCTs. And to find a way to
implement these accelerated tests in an efficient procedure.

In an ATCT, it is expected that the temperature cycle measurement from a DUT resembles
the temperature cycle measurement from the cabinet. But from the experiments preformed
on three different types of components, they exhibited a huge difference between temperature
cycle measurements from the cabinet, on the DUT surface and inside the DUT. This
difference in the measurements resulted in obtaining a bad Acceleration Factor (AF) for each
DUT, which is not agreeable. It is very important to tailor the test parameter values for each
product. The tailoring can be done by using simulation, in a fast and efficiently manner.
Further, from the simulations the physical characteristics of each DUT can be analyzed. This
will be advantageous in assessing failures prone points in the DUT.

Accelerated testing is especially necessary for the products which are expected to have a long
life length, since the producer cannot wait for the actual life length of the product, for example
several decades, to see if the product has a sufficient endurance[4]. The accelerated testing
can be performed with different environmental stresses. The environmental stresses can
consist of thermal cycling, vibration, electrical stress, thermal shock, humidity, and others.
Each component on the vehicle will be exposed to a tailored set of tests depending on the
product functionality, materials used in the product and position of the product on the vehicle.

1
The presented thesis consists of 9 chapters. The first one is an introduction to the problem
considered in the thesis. Chapter 2 consists of a survey of related work and background data,
and then is followed by the problem statement in chapter 3. Chapter 4 describes the working
principles of the thermal shock chamber and experiment set up for recording the temperature
change during the accelerated test. Chapter 5 gives an overview of the results obtained during
the temperature change tests performed on different types of electro-mechanical components.
Chapter 6 is devoted to calculation of AF. Chapter 7 describes the simulations in COMSOL
Multiphysics and comparison of simulation results and practical results. The last chapter
summarizes the results obtained in the thesis and lists future work.

2
3
2 BACKGROUND AND RELATED WORKS

This thesis focuses on investigating the possibility to improve the Accelerated Temperature
Change Tests, ATCT. These tests are performed on electronic components present in heavy
automobile vehicles. Engineers in manufacturing industries have used Accelerated Test (AT)
experiments for many decades. The purpose of AT experiments is to acquire reliability
information quickly. Tested units of a material, component, subsystem or entire systems are
subjected to higher-than-usual levels of one or more accelerating variables such as
temperature or stress. Then the AT results are used to predict weaknesses of the Device Under
Test, DUT at field conditions.

2.1 SN-Curve
A SN-Curve (Stress verses Number of cycles to fail) is a plot of the magnitude of an
alternating stress versus the number of cycles to fail for a given material as shown in Figure
2.1. Typically both the stress and number of cycles are displayed on logarithmic scales [5].
Different materials have their own S-N curve.

Figure 2.1 S-N curve.

The usual procedure is to test the specimen at a high peak stress where failure is expected in
a fairly short number of cycles [6]. At normal stress levels, the component can have fairly
large number of cycles, as the stress is increased the results can be obtained at less number
of cycles. In the ATCT, high stress levels are desirable. For a given component, the high
levels of stress depends on the mismatch between Coefficient of Thermal Expansion (CTE)
of each material used for building the specific component and the environment for the
component on the vehicle. But one should not apply unnatural loads on a component during
the test.

4
2.2 Accelerated Testing – Temperature Change Test
The term acceleration has many different meanings within the field of reliability, but the
term generally implies making time (on whatever scale is used to measure device or
component life) go more quickly, so that reliability information can be obtained more rapidly
[1]. Today’s electronic packaging continues to shrink in size and reaches higher packing
density and higher reliability. Accelerated tests are needed to save test time and cost, and to
reduce time to market. Various temperature cycling profiles are employed to evaluate the
effect of stress on product life and to detect the unknown failure modes [2]. The temperature
cycle profile can be characterized by:

• High extreme temperature (Tmax),

• Low extreme temperature (Tmin),

• Temperature Range (∆T), ∆T = Tmax – Tmin,

• Ramp Rates (RR),

• Dwell Times (DT) at the extreme temperatures.

The product life could be significantly affected by the temperature levels i.e. minimum and
maximum temperature, the temperature range (∆T), DT and RR. The larger Tmax, ∆T, and
RR indicate the higher stress level. Temperature cycling tests are used to characterize the
product capability and to detect unknown failure modes (e.g., die crack, via crack) during
technology development, product design verification, and product qualification. Figure 2.2
describes a temperature cycling profile.

Figure 2.2 Temperature cycle test profile.

5
The stress level generated in the component, due to the test temperature profile usually
exceeds the product field application stress level. The test parameters for each type of
component are specified by the International Standards Organization, ISO. Components can
vary according to size, shape, material, and weight. But the stress levels set by the standards
or manufactures might not be correctly assessed in the lab environment for all types of
components. In today’s testing, the choice of stress levels is set based on standards and
experience.

There are different methods of accelerating a reliability test [7]:

1. Increasing the use rate of the product: This method is appropriate for products that are
ordinarily not in continuous use. For example, the median life of a bearing for a certain
washing machine agitator is 12 years, based on an assumed use rate of 8 loads per week
[7].

2. Increasing the intensity of the exposure to radiation: Various types of radiation can lead
to material degradation and product failure. For example, organic materials (ranging from
human skin to materials like epoxies and polyvinyl chloride or PVC) will degrade when
exposed to ultraviolet (UV) radiation [7].

3. Increasing the aging rate of the product: Increasing the level of experimental variables
like temperature or humidity can accelerate the chemical processes of certain failure
mechanisms such as chemical degradation (resulting in eventual weakening and failure)
of an adhesive mechanical bond or the growth of a conducting filament across an
insulator (eventually causing a short circuit) [7].

4. Increasing the level of stress (e.g., amplitude in temperature cycling, voltage, or pressure)
under which test units operate: A unit at a high stress will generally fail more rapidly than
it would have failed at low stress [7].

2.3 Temperature Sensors

Thermocouples are the most commonly used temperature sensors; accurate temperature
measurements can be made with a typical low-level voltmeter. A thermocouple consists of
two different metal wires (for example, copper and constantan) which are welded together at
one end, and then separated from each other with insulated leads. With the influence of heat
at the welded junction, a DC voltage (thermocouple voltage) is produced between the two
metals which can be measured and used to provide information about the prevailing
temperature. The voltage generated by the thermocouple is largely proportional to the
difference between the temperature of the device under test and the reference temperature
[8]. In this thesis, thermocouples are used to measure temperature in the cabinet and on the
component during the ATCT. Thermocouples Type-K (TCK) are used, which are
manufactured by “Pentronic” [9]. Figure 2.3 illustrates the structure of the welded junction.

6
 Figure 2.3 Thermocouple Type K (TCK). [8]

2.4 Modes of Heat Transfer

The three main modes of heat exchange are: conduction, convection and radiation.

Conduction is a heat transfer through solids or stationery fluids. There are two mechanisms
of heat transfer by conduction: lattice vibration and particle collision. Conduction through
solids occurs by a combination of these two mechanisms and heat is conducted through
stationery fluids by molecular collisions. An example of conduction in solid is shown in
Figure 2.4. There are two stages of heating through conduction: (1) The steady state, where
there is no temporal change in system temperatures and (2) The transient state, where the
system temperature is still changing with time [10].

Convection uses the motion of fluids to transfer heat. In a typical convective heat transfer, a
hot surface heats the surrounding fluid, which is then carried away by fluid movement such
as wind. The warm fluid is replaced by cooler fluid, which can draw more heat away from
the surface. Since the heated fluid is constantly replaced by cooler fluid, the rate of heat
transfer is enhanced. There are two types of heating processes in convection: (1) Natural
convection, the flow is induced by the differences between fluid densities, which result due
to temperature changes and (2) Forced convection uses externally induced flow, such as wind
[10].

Radiation heat transfer does not require a medium to pass through; thus, it is the only form
of heat transfer in vacuum. It uses electromagnetic radiation (photons), which travels at the
speed of light and is emitted by any matter with temperature above 0 K (-273 °C). Radiative
heat transfer occurs when the emitted radiation strikes another body and is absorbed [10].

7
 Figure 2.4 Illustration of conduction, convection and radiation[11] .

2.5 Dependency of Accelerated Temperature Change

Test Parameters
Electronic systems utilize many different metals and plastics in the manufacturing and
assembly processes. The physical properties of these materials must be known, as well as the
operating environments, in order to produce a cost effective and reliable product [12]. Natural
ambient thermal fluctuations or induced thermal fluctuations (such as in or around an engine)
are frequently or constantly occurring on vehicles and are causing stress in components,
mainly by variation in thermal expansion. This in turn affects their reliability and life length,
since the highest failure rates in electronic systems are generally associated with high
component temperature [12].

Electronic devices can experience significant changes in temperature during normal

operation due to changes in both their internal temperature (heat dissipation) and the ambient
temperature. For instance, when component boards are exposed to temperature changes,
strains and stresses are concentrated in the solder interconnections due to the mismatch of
Coefficients of Thermal Expansion (CTE) of the various component board materials [13].
ATCTs are most commonly used to study the effect of thermal changes to assess the
reliability of electronic products. Figure 2.5 shows the dependency of different setup
parameters [13].

Dwell Time (DT) is the duration for the DUT to reach Tmax / Tmin in the test. The material
and package of the DUT vary from component to component depending on the functionality
and position of the component in the vehicle. Due to this, there is no fixed value for DT for
different components in ATCTs.

During the test, Stability state of the component has to be considered. When the component
reaches ± 3 ℃ to ± 5 ℃ of the Tmin and Tmax respectively, then it is called Stability state
of the component. The Transfer time is defined as the time taken to shift the components
from one chamber to another.

8
Ramp Rate (RR) is the rate at which the DUT ramps up to Tmax or ramps down to Tmin.
The value of RR varies according to the size, material and package of the DUT. Both the
values of RR and DT sum up to Cycle time.

The values for Tmax and Tmin in the test are decided in correlation to the temperature on
the vehicle in the real time. These values in the test, helps in calculating the temperature
range (∆𝑇) and RR.

Studying the effects of different ATCT parameters can be problematic due to the fact that
almost any change in one parameter (DT, RR and ∆𝑇), can lead to change in test profile
characteristics that influence the failure mechanism or accumulation of failures [13]. Several
different combinations of temperature differences (∆T), extreme temperatures, DTs, and RRs
are used to control the acceleration factor of the thermal cycling tests.

Figure 2.5 Dependence relationship of thermal cycling parameters [13].

2.6 Failures Modes due to Mismatch in Coefficient of

Thermal Expansion
The expansions and contraction of dissimilar materials resulting from changes in temperature
will produce cyclic fatigue in the material that is used to bond the dissimilar materials
together. This problem would not occur if the dissimilar materials had the same CTE. The
effect of thermal cycling will have differing levels of severity depending upon the slope of

9
S-N curve of the material being stressed. In the following example, a device contains a
stretched copper wire that is attached by a solder joint [14].

At the normal range of thermal cycling (43 ℃ for an interior device) both the solder and the
copper have the same life time. However, during testing we increase the stress by increasing
the range of thermal cycling (125 ℃ is typical for an interior device). The increased stress is
used to accelerate the test, but the steeper slope of the S-N line for copper results in the failure
of the copper before the solder. This illustrated in Figure 2.6. Thermal cycling has a more
profound effect upon the copper than the solder because of the difference in the slopes of the
two S-N lines. In this example, if the copper fails first, then it should be calculated to
determine if the copper would have failed before the requirement at normal stress [14].

Figure 2.6 Effect of S-N slope on Accelerated Testing [14]. Note: The use of value ‘m’ is
explained in section 2.8.

Thermal expansion is the tendency of matter to change in shape, area, and volume in response
to a change in temperature. If the temperature of a body is increased, that leads to thermal
expansion giving a change in dimension either in length, volume or area. Hence, there are
three types of thermal expansion - Linear expansion, Area expansion and Volume expansion
[15]. The CTE describes how the size of an object changes with a change in temperature.
Specifically, it measures the fractional change in size per degree change in temperature at a
constant pressure. Equation (2.1) gives the formula for linear thermal expansion.

𝐿1 − 𝐿0
= 𝛼𝑙 ×(𝑇1 − 𝑇0 ) (2.1)
𝐿0

10
where:

𝐿0 Initial or original object length (at 𝑇0 );

𝐿1 Final object length due to increase in temperature (at 𝑇1 );

𝑇0 Initial temperature;

𝑇1 Increased temperature;

𝛼𝑙 Linear – coefficient of thermal expansion.

There are two scenarios, where stress is generated in components due to different CTE of
materials. This stress can lead to failure of a component. Consider a component made up of
two different materials as shown in Figure 2.7. Case (1): when the component is made up of
two different materials with different CTEs i.e. 𝛼1 ≠ 𝛼2 and the surrounding temperature is
increased from 𝑇0 to 𝑇1 as shown in Figure 2.8, then stress is generated at the meeting point
of both the materials. This stress can lead to failures or cracks in the component. Case (2):
when the component is made up of two similar materials with almost equal CTEs i.e. 𝛼1 ≅
𝛼2 and with unequal increase in the temperature at both the surfaces as shown in Figure 2.9.
Then this leads to stress at the junction due to a thermal gradient, which in turn results in
component failure or crack.

Figure 2.7 Component made up of two different materials. 𝛼1 ≠ 𝛼2 .

Figure 2.8 Stress generated at the junction of both the materials due to different CTEs and
increase in temperature. 𝛼1 ≠ 𝛼2 .

11
Figure 2.9 𝛼1 ≅ 𝛼2 . Even thought CTEs of both the materials are almost similar, stress can
be generated due to thermal gradient. It is not necessarily the way the figure depicts.

2.7 Standards of Accelerated Temperature Change Test

The standards specify the procedure how to carry out the temperature change test to assess
the failures that a component could possibly obtain in the long run. These standards
recommend values for the test parameters i.e. cabinet parameter settings. Four different
standards were studied and compared as shown in Table 2.1. The standards are: (1) TB1900,
(2) IEC 60068-2-14: International Standard [16], (3) ISO 16750-4:2010 [17], (4) The
GWM3172: User Guide [14]. The procedures for the ATCTs in standards are similar, but the
main difference lies in the definition of each test parameter.

Table 2.1 : Comparison between standards.

PARAMETER TB1900 IEC 60068-2-14: ISO 16750-4: The GMW3172:

International 2010 User Guide
Standards

Dwell Time (DT) Complete test time Complete test time “After inside of the “Shall begin after
in one chamber. in one chamber. component reaches inside of the
stability.” component reaches
stability.”

Stability state for Any point on/in Any reference Inside of the Inside of the
the component the component point on the component should component should
should reach component. reach stability. reach stability.
stability.

Transfer Time ≤ 10 seconds. Not more than 3 Within 30 seconds Thermal lag time –
minutes. or less. time for the
component inside
reaches stability.

Number of Cycles Depends on 5 (unless Depends on type of Depends on type of

in test (𝑵𝒕) position of specified). component. component.
component on the
truck.

12
Before setting the values for the test parameters, one should be aware that different standards
suggest different definitions for each test parameter. For example, refer to the definitions of
the DT from Table 2.1. This difference can cause different test results and possibly also
confusion for test engineers. In Table 2.1, the four standards are compared based on the
factors which contribute to the total test time. The stability state corresponds to the state at
which the DUT reaches ± 3℃/± 5℃ of the 𝑇𝑚𝑎𝑥 or the 𝑇𝑚𝑖𝑛 during the test. Irrespective of
the standards, the stability state is necessary for the DUT to achieve the correct stress. The
test duration in one chamber consists of two parts. They are: (1) time for the DUT to reach
the stability state (𝑡𝑠 ) and (2) total time in the test (𝑡1 ) as shown in Figure 2.10. Dwell time
(DT) is one of the crucial factor affecting the duration and efficiency of the test. When the
two-chamber method is used for the ATCT, the transfer time is always less than 10 seconds.
This is observed from the study of the temperature cabinet.

Figure 2.10 Definition of DT in different standards[16]. Here 𝑇𝐴 = room temperature, 𝑇𝐵 =

𝑇𝑚𝑎𝑥 and ∆𝑇 = stability state. The red line defines DT for TB1900 and IEC. The green line
defines DT for ISO and GMW3172.

The time duration after the component reached stability state at 𝑇𝑚𝑎𝑥 or 𝑇𝑚𝑖𝑛 is called soak
time. There are two contrasting definitions of DT, i.e. (1) only the soak time after stability
state is achieved and (2) the DT is the complete duration at maximum/minimum temperature.
It could include the ramp time to reach stability and soak time after stability. For instance,
the GMW user guide recommends the soak time could be minimum of 10 minutes and can
go up to 15 minutes.

2.8 The Coffin-Manson and Norris Landzberg Models

The AF is defined as the ratio of cycles to failure in the field to cycles to failure in test. Cycles
to failure used in AF calculations are estimated upfront - prior to testing and product
deployment - using a life prediction model. In this project we consider two types of life
prediction models to calculate AF, i.e. Coffin-Manson Model and Norris Landzberg Model.
For calculating AFs, parameters used in test and parameters used in field are considered (refer
to Table 2.2)

13
Table 2.2 : Representation of parameters when used in field and test respectively.

Parameters Parameters for field usage Parameters for test usage

𝑻𝒎𝒂𝒙 𝑇𝑓𝑖𝑒𝑙𝑑 𝑚𝑎𝑥 𝑇𝑡𝑒𝑠𝑡 𝑚𝑎𝑥

𝑻𝒎𝒊𝒏 𝑇𝑓𝑖𝑒𝑙𝑑 𝑚𝑖𝑛 𝑇𝑡𝑒𝑠𝑡 𝑚𝑖𝑛

∆𝑻 ∆𝑇𝑓𝑖𝑒𝑙𝑑 ∆𝑇𝑡𝑒𝑠𝑡

DT 𝐷𝑇𝑓𝑖𝑒𝑙𝑑 𝐷𝑇𝑡𝑒𝑠𝑡

RR - 𝑅𝑅𝑡𝑒𝑠𝑡

N 𝑁𝑓 𝑁𝑡

The Coffin-Manson model is deliberately kept simple because the extra complications such
as rate and creep effects and may not actually increase the accuracy. Thus, the Coffin-Manson
model fails to distinguish the difference between different test temperature profiles with the
same ∆𝑇. The provided AF is a function of both ∆𝑇 and temperature test profile, but the
relation and effect of all test parameters on the AF must also be carefully addressed. The
Coffin Manson equation is given in (2.2). Some modified Coffin Manson models account for
cyclic frequency (inverse of cycle time), but the models were only verified for bulk solder
test structure. However, for the same temperature cycle, the different RR and DT may also
have an impact on the mechanical behavior of solder joints and therefore alter the fatigue life
[18].

Coffin Manson Model for AF:

𝑚
∆𝑇𝑡𝑒𝑠𝑡
𝐶𝑜𝑓𝑓𝑖𝑛 𝑀𝑎𝑛𝑠𝑜𝑛 𝐴𝐹 = ( ) (2.2)
∆𝑇𝑓𝑖𝑒𝑙𝑑

where,

𝑚= The slope value of the Stress-Life (S-N curve) relationship line for each
material. 𝑚 = 2.65 for lead-free solder and 𝑚 = 2.5 for leaded solder.

∆𝑇𝑡𝑒𝑠𝑡 = The temperature range, the component is exposed to during the test.

∆𝑇𝑓𝑖𝑒𝑙𝑑 = The temperature range, the component is exposed to, when it is used on the
field.

Researchers have found the Coffin-Manson model to yield somewhat conservative estimates
for fatigue life. IBM researchers K.C Norris and A.H. Landzberg modified it to compensate
for frequency-dependent and time-dependent anomalies [19]. Thus, the Coffin Manson
model takes less parameters into account and it is easy to calculate but might be less accurate.
In this thesis, this model is used to relate a number of cycles in the test (𝑁𝑡 ) to a number of
cycles in the field (𝑁𝑓 ), explained in (6.1). This helps in analyzing the efficiency of the tests.

14
To make this analysis more reliable, another model is taken in to account i.e. Norris
Landzberg model.

The Modified Norris-Landzberg model provides the relationship between the test parameters
and the field parameters. The Norris-Landzberg model are shown in (2.3) and (2.4) for lead-
free solder and leaded solder respectively. These models consist of four parts i.e. AF for ∆𝑇,
AF for DT, AF for RR and AF for 𝑇𝑚𝑎𝑥 . The damage developed during thermal cycling is the
product of these four parts of AFs. The AFs from these four parts are multiplied together to
form the total acceleration factor used in testing[14]. It can be seen that the first AF part in
Norris-Landzberg is similar to the Coffin-Manson AF. Considering the last three AFs
separately, the following content describes how the models are related to each other.

Norris-Landzberg Model for AF:

Norris-Landzberg 𝐴𝐹𝑙𝑒𝑎𝑑−𝑓𝑟𝑒𝑒 𝑠𝑜𝑙𝑑𝑒𝑟

2.65 0.136
∆𝑇 𝐷𝑇
= (∆𝑇 𝑡𝑒𝑠𝑡 ) × (𝐷𝑇 𝑡𝑒𝑠𝑡 ) ×[1.22×(𝑅𝑅𝑡𝑒𝑠𝑡 )−0.0757 ]×
𝑓𝑖𝑒𝑙𝑑 𝑓𝑖𝑒𝑙𝑑 (2.3)
1 1
〈2185×( − )〉
𝑇𝑓𝑖𝑒𝑙𝑑 𝑚𝑎𝑥 +273 𝑇𝑡𝑒𝑠𝑡 𝑚𝑎𝑥 +273
[𝑒 ]

Norris-Landzberg 𝐴𝐹𝐿𝑒𝑎𝑑𝑒𝑑 𝑠𝑜𝑙𝑑𝑒𝑟

2.5
∆𝑇𝑡𝑒𝑠𝑡
= ( ) × [0.768×(𝐷𝑇𝑡𝑒𝑠𝑡 )0.0667 ]×[0.80094×(𝑅𝑅𝑡𝑒𝑠𝑡 )0.0964 ]×
∆𝑇𝑓𝑖𝑒𝑙𝑑
(2.4)
1 1
〈1414×( − )〉
𝑇𝑓𝑖𝑒𝑙𝑑 𝑚𝑎𝑥 +273 𝑇𝑡𝑒𝑠𝑡 𝑚𝑎𝑥 +273
[𝑒 ]

Dwell Time AF:

0.136
𝐷𝑇𝑡𝑒𝑠𝑡
𝐷𝑇 𝐴𝐹𝐿𝑒𝑎𝑑 𝑓𝑟𝑒𝑒 𝑠𝑜𝑙𝑑𝑒𝑟 =( ) (2.5)
𝐷𝑇𝑓𝑖𝑒𝑙𝑑

𝐷𝑇 𝐴𝐹𝐿𝑒𝑎𝑑𝑒𝑑 𝑠𝑜𝑙𝑑𝑒𝑟 = [0.768×(𝐷𝑇𝑡𝑒𝑠𝑡 )0.0667 ] (2.6)

Table 2.3 : DT AF versus 𝐷𝑇𝑡𝑒𝑠𝑡 plot.

Test Parameter Assigned Value

∆𝑻𝒕𝒆𝒔𝒕 −40 ℃ to + 80 ℃

𝑫𝑻𝒇𝒊𝒆𝒍𝒅 270 minutes

𝑫𝑻𝒕𝒆𝒔𝒕 Varying from 0 minutes to 270 minutes

15
 Figure 2.11 Plot of DT AF versus 𝐷𝑇𝑡𝑒𝑠𝑡 . DT AFs are calculated using (2.5) and (2.6).
Assumptions provided in Table 2.3 are used in calculations.

High Extreme Temperature AF:

1 1
〈2185×( − )〉
𝑇𝑚𝑎𝑥 𝐴𝐹𝐿𝑒𝑎𝑑 𝑓𝑟𝑒𝑒 𝑠𝑜𝑙𝑑𝑒𝑟 = [𝑒 𝑇𝑓𝑖𝑒𝑙𝑑 𝑚𝑎𝑥 +273 𝑇𝑡𝑒𝑠𝑡 𝑚𝑎𝑥 +273
] (2.7)

〈1414×(
1
−
1
𝑇𝑓𝑖𝑒𝑙𝑑 𝑚𝑎𝑥 +273 𝑇𝑡𝑒𝑠𝑡 𝑚𝑎𝑥 +273
)〉 (2.8)
𝑇𝑚𝑎𝑥 𝐴𝐹𝐿𝑒𝑎𝑑𝑒𝑑 𝑠𝑜𝑙𝑑𝑒𝑟 = [𝑒 ]

Table 2.4 : High Extreme Temperature AF versus 𝑇𝑡𝑒𝑠𝑡 𝑚𝑎𝑥 plot.

Test Parameter Assigned Value

∆𝑻𝒕𝒆𝒔𝒕 −40 ℃ to + 80 ℃

𝑻𝒇𝒊𝒆𝒍𝒅 𝒎𝒂𝒙 30 ℃

𝑻𝒕𝒆𝒔𝒕 𝒎𝒂𝒙 Varying from 30 ℃ to 80 ℃

16
 Figure 2.12 Plot of High temperature AF versus 𝑇𝑡𝑒𝑠𝑡 𝑚𝑎𝑥 . High temperature AFs are
calculated using (2.7) and (2.8). Assumptions provided in Table 2.4 are used in
calculations.

Ramp Rate AF:

𝑅𝑅 𝐴𝐹𝐿𝑒𝑎𝑑 𝑓𝑟𝑒𝑒 𝑠𝑜𝑙𝑑𝑒𝑟 = [1.22×(𝑅𝑅𝑡𝑒𝑠𝑡 )−0.0757 ] (2.9)

𝑅𝑅 𝐴𝐹𝐿𝑒𝑎𝑑𝑒𝑑 𝑠𝑜𝑙𝑑𝑒𝑟 = [0.80094×(𝑅𝑅𝑡𝑒𝑠𝑡 )0.0964 ] (2.10)

Table 2.5 : RR AF versus 𝑅𝑅𝑡𝑒𝑠𝑡 plot.

Test Parameter Assigned Value

∆𝑻𝒕𝒆𝒔𝒕 −40 ℃ to + 80 ℃

𝑹𝑹𝒕𝒆𝒔𝒕 Varying from 0 ℃/minute to 20 ℃/minute.

17
 Figure 2.13 Plot of RR AF versus 𝑅𝑅𝑡𝑒𝑠𝑡 . RR AFs are calculated using (2.9) and (2.10).
Assumptions provided in Table 2.5 are used in calculations.

From the above graphs in Figure 2.11, 2.12 and 2.13, it is observed that when the DT is
increased and 𝑇𝑚𝑎𝑥 is not reached then the respective AF comes close to 1. With this, there
is a chance of approximating Norris-Landzberg model to Coffin-Manson model.

2.9 Review of Related Works

Many researchers have focused on improving the Accelerated Tests by using statistical
analysis, defining a procedure and simulations. To assess the life length of a Thermoelectric
(TE) module, finite element analysis (FEA) software ANSYS was used to simulate the
thermoelectric modules under the mechanical bending load. During the operation of a
Thermoelectric (TE) module, one side of it has thermal expansion and the other side has
cooling shrinkage. The corresponding resulted tensile and compressive stresses at the
interfaces of its constituent materials will often cause the fatigue failure, especially during
the repeated operations of the module. For the life assessment of TE modules, it is necessary
to validate the corresponding deformation, stress, and strain with those in the real-world
operation of the TE modules [20].

Usually accelerated tests are performed at high stress levels, but a method was proposed to
test the products at both higher stress levels and Usage Rate (UR). An UR effect model is
proposed to describe the dependence of Usage to Failure (UTF) on UR. The relationship
between UTF, and stress and UR is established, and used to estimate the UTF distribution at
design stresses and usual UR. The model parameters are estimated by maximum likelihood
method [21].

18
There are two other methods using simulations to optimize the test plan for Accelerated
Testing. The types of simulations used are: (1) Monte Carlo Simulations [22] and (2) MLE
(Maximum Likelihood Estimation) simulation [23]. A method is defined to optimize the test
design for ADT – Accelerated Degradation Test. It is performed by using the criterion of
minimizing the estimate of the mean-squared error of the estimated 100pth percentile of the
product’s lifetime distribution. There are three key steps in this method: simulating; modeling;
optimizing. A complex optimization problem was transformed into a statistical problem by
Monte Carlo simulation [22].

On the other hand, a simulation technique is used to investigate the required sample size for
using the large sample Gaussian approximation s-confidence interval and the properties of
the ML estimators in the finite sample situation with different fitting models. The maximum
likelihood (ML) method is commonly used in estimating model parameters, and its
asymptotic variance is the key quantity used in searching for the optimum design of ALT
(Accelerated Life Tests) plans and in making statistical inferences [23].

There is one researcher focusing on Design for Reliability instead of Accelerated Testing.
Here the aim is not to simulate the operating conditions in the field but to identify failure
modes and estimate the margin to the operating conditions in the field. This concept uses the
physics of failure approach. The physics of failure analysis approach should be used focusing
on each failure mode separately. An optimized step by step design analysis and test process
is explained, to reduce the number of potential failure modes [24].

For accelerated reliability testing of electronic device packaging, temperature cycle profiles
at various stress levels were investigated. Coffin-Manson model was used to model crack
growth due to repeated temperature cycling. Using the correlation between Coffin-Manson
model and the test results, Activation Energy was determined [2].

Choong R. Yang & Jin T.Kim conducted Accelerated Life Test (ALT) for a DUT. They
evaluated the failure rate of a DUT from the expedient information by applying the stress
levels more severe than those specified by standard condition. Generation of failure rate data
was obtained by applying a temperature higher than the operational temperature of a DUT.
This helped to obtain the data within a short period of time to estimate the failure rate during
actual system operation [25].

A step-stress accelerated life test for two stress variables was developed by Chenhua Li and
Nasser Fard. Failure time follows the Weibull distribution, and the test was terminated at a
pre-determined time, which leads to a censored failure data. An optimum test plan was
developed for determining the test interval for each combination of stress levels. The scale
parameter of the Weibull distribution for each combination of stress levels is defined as a
linear logarithmic function of the stress levels. To minimize the asymptotic variance of the
maximum likelihood estimator of the life for a specified reliability, an optimal criterion was
defined [26].

Accelerated thermal cycling methodology was utilized to estimate and describe the actual
fatigue lifetime of advanced packaging. Along with ATCT, the test results and acceleration
factor (AF) was integrated to assess the fatigue span of the actual use of the vehicle. Original

19
AF model presented by Norris and Landzberg (N-L) is not adequate for predicting the life
reliability when the tests are subjected to different RRs. A modified AF model was proposed
by considering the creep behavior of solders and the RR effect. The 𝑇𝑚𝑎𝑥 is modified by the
mean temperature because of the accumulation of creep strain and the thermal loading
frequency separated into two portions, the RR and the DT, which is adopted for adjusting the
RR effect. Peak value effect of applied temperature effect on the wafer level chip packaging
under ATCT test was explored by two fatigue span estimated models, Coffin-Manson
empirical formula and energy-based Darveaux model [27].

AF empirical equation has been developed for many years to predict product lifetime based
on ATCT result and loading condition, such as well-known Norris-Landzberg equation with
operating temperature and frequency terms. This model indicated opposite results when
higher RR is used. To address such phenomenon, a modified AF empirical equation has been
characterized with SnAg based lead-free solder on the wafer level chip package under various
thermal cycling profiles. A Finite Element Method (FEM) with a temperature dependent
Young’s modulus and Garafalo-Arrhenius creep equation to describe the strain rate and creep
of solder properties under various thermal cycling profiles. Modified AF empirical formula
shows a good correlation between the experimental results, including tests with various DT
and RR profiles [28]

20
21
3 PROBLEM STATEMENT AND MAIN
CONTRIBUTIONS

The ATCTs are primarily controlled with three parameters; ∆𝑇, DT and 𝑁𝑡 . Many standards
and guidelines that define the test parameters, have been formulated several decades ago and
they may not be applicable for today’s EEE-components (Electronic, Electric and Electro-
mechanical) [13]. Despite being accelerated, the tests are still relatively time consuming and,
especially for large components, the relevance of the tests can be questioned due to difficulty
or uncertainty of the heat transfer.

Natural ambient thermal fluctuations or induced thermal fluctuations such as in or around an

engine are frequently or constantly occurring on vehicles and are causing stress in
components, mainly by variation in thermal expansion. This in turn affects their reliability
and life length. ATCTs can partly emulate the real vehicle environment and are commonly
performed in dedicated cabinet consisting of two-chamber i.e. hot chamber and cold chamber.
These tests are important in assessing the risks of failure in EEE-components. The
components can be very complex since they are designed for a multitude of applications and
vary in used materials, weight and size. To understand the accuracy of the ATCT, the
temperature changes on and in the component, can be recorded during the accelerated test on
selected component samples. To address the mentioned problems five research questions are
considered in this thesis.

The first research question is: What is the temperature response in and on a component
during the ATCT? And how this temperature response can be estimated for a specific
component in ATCT? The hypothetical answer to this question is that during the ATCT, the
temperature responses are uneven, both from inside and on the surface of the component.
The reason of this are the variations in the physical attributes of a component, which lead to
uneven temperature response. These uneven responses lead to a variation in the temperature
stress on and in the components. The difference in temperature responses can be evaluated
by comparing CPSs (Cabinet Parameter Value) and EPVs (Estimated Parameter Values).
CPSs are assigned values for each parameter before the test based on the type of component
and the environment faced by the specific component on a vehicle. And EPVs are estimated
from the temperature recordings during the ATCT at different locations in and on the DUT.
Then EPVs estimated for different locations of the DUT are compared with CPS.

The second issue can be addressed by the following research question: How the problem of
uneven temperature response in and on a component can be overcome when aiming to reduce
the ATCT test time? To reduce the accelerated test time for a complex component, the
component could be dismantled into parts which should be tested separately. However,
although testing a dismantled component can reduce the test time, there is a risk of overseeing
a few stress points or fragile points. On the other hand, the test of dismantling component
can provide accurate results for the electronics inside the component. Another way is to place
the component in a position, which will help in having maximum of the area of the

22
component exposed to the simulated environment in the cabinet. Exposing maximum area
will help in having fast conduction of heat or fast cooling effect on the component. This can
lead to having reduced test time.

The third research issue is: How can the DT be adjusted for different/complex components,
for an ATCT? For different types of components, depending on the environment on the
vehicle, CPS values are proposed in the standards. However, the temperature levels and the
DT values set by the standards for the test need to be adjusted to the specific lab environment.
Because of this, in today’s testing, the choice of stress levels is heuristic and set mainly based
on experiments. DT can be set with the use of pre-tests on the component and analyze what
the components’ will experience at maximum stress levels. From the pre-tests, the
temperature change log at different locations on/in the component can be obtained.

The fourth research issue is: How can the DT estimate be validated? What can the validation
of DT estimation be used for? The DT values obtained from a pre-test for a specific
component can be applied to calculate AFs based on different test models, where the AF is
defined as the ratio of cycles to failure in the field to cycles to failure in test. Coffin-Manson
model or Norris-Landzberg model are suitable for this purpose. Based on the estimated AF
values, we can analyze and validate DT estimate. If too long DT (longer than required for a
specific component) is estimated then there is high chance of obtaining low accelerating
conditions i.e. a wrong AF value. The proposed validation of DT choice via calculating the
AF can be applied to check the applicability of different methods to reduce the test time of
ATCT i.e. dismantled component for the test or assembled components for the test. Moreover,
the proposed validation of DT choice via calculating the AF can help to estimate 𝑁𝑡 and relate
𝑁𝑓 in real time.

The last research question is: In which way one can improve the ATCT that can reduce test
time, test efforts and contribute reliable test results? The target can be met by implementing
software simulations on the desired component, to estimate the choose of test parameters.
These simulations include: (1) creating the cabinet test environment, (2) designing the
structure and defining the materials of the component and (3) defining the physics applied
for the test. For a particular component, the iterative simulations can lead to optimal
parameters’ values. In addition, the simulations providing data regarding the temperature
response of the complete component and temperature graph at any particular point inside or
on the component, can be useful for monitoring and comprehensive understanding of the test.
The advantage of using the simulations for ATCT design is saving time compare to the
regular pre-tests.

The main contributions of this project are:

(1) Analysis of different standards. Four standards are compared based on the factors,
which contribute to the total test time.

(2) Investigation of the temperature change in the empty basket during an ATCT, which
can be used in future for any kind of reference.

23
(3) Design of an experimental set-up, which can be used to recorded temperature of a
DUT during an ATCT.

(4) Preforming test cases, same as the currently used scenario, with three different types
of components. The temperature readings from the component were recorded during
the test.

(5) Investigation the deviation between CPS and EPV.

(6) Analyzing how well the tests are carried out today and how the current testing agree
with the standards.

(7) Estimation of AFs from the EPV to investigate the reliability of the currently used
scenario of testing.

(8) Establishing a good agreement between component testing results and simulation
results. With help of simulations, recommendations are proposed with a target to aid
in the choice of test parameters and set-ups.

24
25
4 WORKING PRINCIPLES OF THE
TEMPERATURE SHOCK CHAMBER AND
TEST SET-UP

In this chapter, features and working of the temperature change cabinet is explained. Also,
the tests performed on the cabinet along with their results are mentioned. Observations from
these test results are described.

4.1 Features and Working Principles of the Chamber

A two-chamber cabinet was used. The typical temperature cycling equipment consists of a
cabinet with one hot chamber and one cold chamber. The test samples are automatically
transferred between the two chambers by an elevator-type mechanism. It consists of a basket,
where the components can be placed for undergoing the test. This basket is transferred from
one region to another within 3 seconds. The temperatures of the chambers are maintained
with the help of fans. Each chamber consists of one fan, located opposite to the door.

Figure 4.1 Front view of the temperature shock cabinet [29].

The hot chamber is set to +10 ℃ above the nominal test temperature, when the basket is in
the cold chamber. This setting helps to achieve a faster change in the chamber temperature
when the basket is transfer from the cold to the hot chamber. Similarly, the cold chamber is
maintained at −10 ℃ below the nominal test temperature, when the basket is in the hot
chamber, which is illustrated in Figure 4.10. It is also possible to perform temperature cycling
in a single compartment chamber where the temperature is ramped between hot and cold.

26
This generally produces a slower rate of temperature change compared to the two-chamber
method. The temperature is maintained at a constant temperature, as programmed.

Figure 4.2 Interior of the two-chamber cabinet with basket in hot chamber.

The temperature inside the two-chamber cabinet is maintained at two different temperatures,
as programmed with respect to the type of chamber. One test cycle comprises of two periods
of DT, first in the hot chamber and second in the cold chamber. The first test cycle comprises
of ramping time to the respective chamber temperatures from the room temperature and two
portions of DT. Each chamber in the cabinet has its own fan, which generates the air-flow to
maintain at the programmed temperatures. In the hot chamber, the air is heated to the
maximum temperature (according to the CPS) and is blown out into the chamber through the
gap between hot chamber ceiling and basket’s ceiling (but also from the side walls of the
basket). Then into the basket from the door side, as illustrated in Figure 4.3. Similarly, in the
cold chamber, the air is cooled to the minimum temperature (according to the CPS) and is
blown out into the chamber from the gap between cold chamber floor and the basket floor.
And then into the basket as shown in Figure 4.3.

27
 Figure 4.3 Illustration of air flow in hot cabinet and cold cabinet respectively (side view).
The broad arrows indicate the temperature at respective chambers with different basket
location in the cabinet.

4.2 Test Set-Up to Record Temperature Change in the

Cabinet.
To study the temperature change in the cabinet, test cycles were performed with an empty
basket. A test set-up is designed using thermocouples to perform reliable test cycles. To find
the appropriate locations for the thermocouples in the chamber, the basket was initially
divided into nine sections as shown in Figure 4.4. There are (i) three sections parallel to door
(1, 2, 3 from Figure 4.4), (ii) three perpendicular to the door (7, 8, 9 from Figure 4.4), and
(iii) three parallel to floor of the basket (4, 5, 6 from Figure 4.4).

Figure 4.4 Representation of sections in the basket. Each section is explained in the
Table 4.1

28
Table 4.1 : Description for each basket section.

Section Position Description

1 Parallel to door/fan Always close to the fan in hot chamber or cold chamber.

2 Parallel to door/fan Equidistance from door and fan.

3 Parallel to door/fan Always closer to the door in hot chamber and cold chamber.

4 Parallel to the floor Helping in measuring the temperature when the basket is in
the cold chamber and the top layer of the basket is closer to
the hot chamber. This helps in investigating if there is any
heat radiating from the hot chamber to the cold chamber
during the test cycle.

5 Parallel to the floor It is in middle, equidistance from top section and bottom
section

6 Parallel to the floor Helping in measuring the temperature when the basket is in
the hot chamber and the bottom layer of the basket is closer
to the cold chamber. This helps in investigating if there any
cold air flow from the cold chamber to the hot chamber
during the test cycle.

7 Perpendicular to door This section is placed to the extreme left.

8 Perpendicular to door Middle layer.

9 Perpendicular to door This section is placed to the extreme right.

With combination of these nine sections, there are twenty-seven locations in the basket. But
only nine locations were chosen by excluding some similar locations. These locations are
expected to have similar measurements. To record the temperature during the test cycles,
eleven thermocouples were placed inside the cabinet. (i) Nine of them are placed in the basket,
in a diagonal fashion and (ii) two of them are placed at a constant position, one in the hot
chamber and the other in the cold chamber. The locations are visually explained in Figure
4.5 and 4.6.

29
Table 4.2 : Placement description of the thermocouples in the cabinet along with motivation
for logging of temperature in an empty cabinet, to obtain the temperature distribution in the
cabinet.

Number of Position in the cabinet Motivation

thermocouples

9 In the basket These thermocouples move along

with the basket and records
temperature during the test cycle.

1 Constantly fixed in the hot chamber This helps in continuous logging of

the temperature in the hot chamber
even when the basket is in the cold
chamber.

1 Constantly fixed in the cold chamber This helps in continuous logging of

the temperature in the cold chamber
even when the basket is in the hot
chamber.

7 4 1

2
5
3
8
6
9

Figure 4.5 Location of nine thermocouples in the basket. TCK 1, TCK 2, TCK 3 on section
1, closed to the fan of the cabinet. TCK 4, TCK 5, TCK 6 on section 2, placed in middle of
the basket and TCK 7, TCK 8, TCK 9 on section 3, placed closer to the door of the cabinet.

30
 11

3
10
6
9

Figure 4.6 Location of two thermocouples in constant positions. TCK 10 in the cold
chambers and TCK 11 in the hot chamber.

4.3 Test Cycles to Record Temperature Change in an

Empty Cabinet.
After the test set-up of thermocouples inside the cabinet (as explained in section 4.2), ∆𝑇s
and DTs values were chosen to perform tests consisting of at least three cycles. This helps in
understanding the working principles of the cabinet with two chambers, test profile and
temperature change inside the basket. The ∆𝑇s and DTs are picked from standards. These
values are frequently used for testing.

Table 4.3 : CPS for test cycles in an empty cabinet.

Test number Temperature range Dwell Time (DT) Number of cycles in

(∆𝑻) minutes test (𝑵𝒕)

Test 1 -40 ℃ to 80 ℃ 20 3

Test 2 -40 ℃ to 80 ℃ 30 3

Test 3 -40 ℃ to 80 ℃ 40 3

Test 4 -40 ℃ to 100 ℃ 20 3

Test 5 -40 ℃ to 100 ℃ 30 3

Test 6 -40 ℃ to 100 ℃ 40 3

Test 7 -40 ℃ to 140 ℃ 20 3

Test 8 -40 ℃ to 140 ℃ 30 3

Test 9 -40 ℃ to 140 ℃ 40 3

31
Test Results

In total nine tests are performed on the cabinet. Each test has temperature graphs from eleven
locations in the cabinet. All tests had similar results, therefore only one test result is shown
below in Figure 4.7, 4.8, 4.9 and 4.10. Temperature readings from the thermocouples located
on the section 1 are displayed in Figure 4.7. Similarly, temperature readings from section 2
and section 3 are displayed in Figure 4.8 and Figure 4.9 respectively. Figure 4.10, displays
temperature readings from the constantly fixed thermocouples in hot and cold chamber.

Figure 4.7 Test 2: temperature response from thermocouples TCK 1, TCK 2 and TCK 3.

32
Figure 4.8 Test 2: temperature response from thermocouples TCK 4, TCK 5 and TCK 6.

Figure 4.9 Test 2: temperature response from thermocouples TCK 7, TCK 8 and TCK 9.

33
 Figure 4.10 Test 2: temperature response from TCK 11 and TCK 10 in hot and cold
chamber respectively. Green line is representing the temperature recording from TCK 11
and pink line is representing the temperature recording from TCK 10.

Observations

The following observations from comprehensive analyses of standards and performed

measurements are taken.

• Placing the DUTs closer to the centre of the basket is must for efficient testing.

• 5 minutes– 10 minutes time period in the cold chamber is not effective. This can be
seen in Figure 4.7 to Figure 4.10, that there are oscillations in the temperature reading
of the cold chamber indicating that it is taking 5 minutes – 10 minutes to reach
constant temperature. This deviation could be unique to the cabinet used in this thesis.
Therefore, it is appropriate to perform a pre-test with an empty basket in the cabinet
before testing any component.

• Stable state is when the component reaches ±3 ℃ to ±5 ℃ with respect to the

minimum or maximum temperature respectively.

Therefore, the complete test time period in one chamber should be estimated considering all
the above factors. In the cold chamber, the cooling of the chamber environment takes much
longer time than the heating in the hot chamber environment. One of the possible reason
could be due to less power capacity of the cold chamber. The DT of the chamber starts

34
without referring to the DUT temperature. But it is necessary that the DUT reaches stability.
The extra time needed for the component to reach stability has to be added to the chamber
DT. A pre-test for the DUT, to find out the stabilization time (extra time) for the DUT before
the rapid temperature change test is recommended. It is possible that the DUT needs different
extra time in hot and cold chambers.

4.4 Experiment Set-up for Component Testing

Previously, we learnt about the working principles of the temperature shock cabinet with an
empty basket. The next task is to design a proper test set-up for the component testing. The
thermocouples are fixed in different locations on the component and inside the component,
to observe the temperature change of the component during the test. This will help us to
inspect the temperature change inside and on the component. The thermocouples are very
sensitive, there is a possibility of recording the surrounding temperature. To avoid this
situation, they should be insulated properly on the surface. An insulation set up which has
been validated is explained in Appendix A.

4.4.1 Data Acquisition System - DEWESoft

A data acquisition system is used to record and store the temperature change. In this project
DEWESoft is used. DEWESoft features capabilities of being a companion for all kinds of
automotive testing applications [30].

DEWESoft – Hardware

DEWE-43 [30] has got eight dynamic analog inputs, each one backed up with a 24-bit AD
converter for simultaneous sampling at the highest resolution. The inputs are "universal" type,
which can handle strain gages and voltages up to +/-10V natively. DSI adapters are used to
collect the temperature recording from the thermocouples. A sync cable is provided to
increase the number of channels. This can be performed using another unit of DEWE-43 and
a sync cable.

Figure 4.11 A top view of the hardware unit of data acquisition system (DEWE-43).

35
DEWESoft – Software

The DEWESoft X software [30] can acquire signals simultaneously from different sources
(even with different sampling rates). It includes rich post-processing features with all the
powerful mathematics and analysis functions for various application needs.

4.4.2 Experiment Setup

Firstly, the thermocouples are connected to the DEWE-43 adapters. Then these are connected
to each channel respectively. As each DEWE-43 consists of 8 channels, this had to be
extended to 16 channels because 8 channels were not sufficient. This extension can be
performed by using a USB-hub. This set-up is visually illustrated in Figure 4.12.

Figure 4.12 Schematic diagram of the experimental set-up.

36
 Figure 4.13 Picture of connections between thermocouples, adapters and DEWE-43
channels during the ATCT.

Figure 4.14 Picture of recording temperature response in ATCT.

The basic set-up for every test has been described in this chapter. The procedure of insulation
of the thermocouple on the surface of the component has been verified through proper testing
in the lab through an experiment (results are mentioned in Appendix B).

37
38
5 CURRENT SCENARIOS OF COMPONENT
TESTING IN TEMPERATURE SHOCK
CHAMBER

The test profile for the component is adjusted, depending on the location of the component
on the truck. Many components are made up of thick layers, which raises an uncertainty if
the electronics inside the component change temperature according to the test profile or CPS.
In an ATCT, the positioning of the component in the basket may have an effect on the testing.
For example, consider a component with a shape of a box. If this component is placed
horizontally on the floor of the basket, then the surface facing the floor is less affected by the
temperature change. Another scenario of test is when many components are placed in the
basket at the same time, then all components may not have the same temperature change. To
study these scenarios during ATCTs, temperatures are recorded on three different
components (described in Table 5.1) with the experimental set-up described in chapter 4.4.
This section of the project describes the different scenarios of the component testing. These
test results are then compared to the standards. The test scenarios and cases are explained in
Table 5.2.

Table 5.1 : Types of components tested.

Physical Properties Motor Aluminum box Plastic box

Weight 6.108 kg 5.84 kg 0.551 kg

Shape Cylindrical Rectangular box Rectangular box

Material used for Consists of different Aluminum Plastic

outer layer types of materials.

Dimensions - Length = 28 𝑐𝑚 Length = 18 𝑐𝑚

Width = 20 𝑐𝑚 Width = 14 𝑐𝑚

Height = 8 𝑐𝑚 Height = 4 𝑐𝑚

Thickness of outer - 1.5 cm 0.8 cm

layer

Others Figure 5.1. Has plastic layer, glass Has lead-free soldered
fiber, resistors and Printed Circuit Board
connectors. Figure 5.8. (PCB) inside. Figure
5.17.

39
Table 5.2 : Types of test cases performed to understand different test scenarios.

Type of Number of Test scenario Test cases

component pieces used for
the test

Motor 3 When three motors are They are placed on the floor of the
tested at the same time. basket and adjacent to each other.
The maximum place of
the basket is occupied Case(i): Parallel to the door
by three pieces.
Case(ii): Perpendicular to the door.

Case(iii): Dismantled component.

Aluminum 1 When a single piece It is used to learn the difference in

box places in the basket. temperature change due to position
of the component in the basket.

Case(i): Horizontal

Case(ii): Vertical

Case(iii): Vertical with open lid

(partially dismantled).

Plastic box 1 A single piece is used To understand effect on temperature

during the test and a change, due to difference in outer
similar test profile as layer and size compared to the
the aluminum box. aluminum box. Even though the
structure and test profile are same as
the aluminum box.

Case(i): Outer layer – Vertically.

Case(ii): Air inside the component –

Vertically.

Case(iii): Material inside the

component – Vertically.

5.1 Motor: Three DUTs at the Same Time.

The test profile for this component is decided using the standards, shown in Table 5.3. Motor
is a bulky component (as shown in Figure 5.1), which can lead to having longer DT. The DT
in the test profile is chosen by trial and error method. A single motor was tested repeatedly
with different DTs. The trial and error method for DT, started at 20 minutes and went up to
50 minutes with an interval of 10 minutes. The component was dismantled and tested. It took
much less time than the complete component, but there will be a chance of overseeing few
types of failures. The results from dismantle component will be shown, later in this section.
Testing was performed on the component as a whole and there were three motors to be tested

40
at the same time. With several components in the chamber, there could be several ways to
orient the components with respect to each other and the chamber. Here, two cases are studied:
(i) Parallel to the door and (ii) perpendicular to the door. Visually explained in Figure 5.2
and Figure 5.3.

Table 5.3 : CPS for testing of the motor in test case(i) and test case(ii).

Test parameters Value

Temperature range (∆𝑻) +100 ℃ to -40 ℃

Dwell time (DT) 50 minutes

Number of cycles in test (𝑵𝒕) 3

Figure 5.1 Motor used in ATCT.

41
Case(i) – Parallel to door of the basket

There are three similar components. They are placed parallel to the door of the cabinet. All
three are aligned to each other. It is shown below in Figure 5.2.

FAN

Component 1

Component 2

Component 3

DOOR

Figure 5.2 Outline of top view of the basket with three components parallel
to the door.
Case(ii) – Perpendicular to door of the basket

There are three similar components. They are placed perpendicular to the door of the cabinet.
All three are aligned to each other. It is shown below in Figure 5.3.

FAN
Component 2

Component 1
Component 3

DOOR

Figure 5.3 Outline of top view of the basket with three components perpendicular
to door.
Comparison of results from parallel and perpendicular placement

The two cases specified above about the placement of the components in the basket have
different results in temperature recording. It is a huge component which consists of different
materials and accumulates a lot of heat in the hot cabinet. They radiate out the heat when
they are in the cold cabinet of the chamber. This radiation heat from the component is difficult

42
to cool down in the cold chamber. From Figure 5.4 it is seen that when the components are
placed perpendicular to the door, there is a bit difference in the RR of heating and cooling of
the component and better performance in the cold cabinet. From Figure 5.5, it is learnt that
there is a lot of difference in the air temperature during test cycle. When placed parallel to
the door, the air flow in the basket is obstructed. This leads to slow heating and cooling of
the air in between the components. This environment will be affecting the components test
results. Therefore, the components should be placed in such a manner that the maximum area
of the components get exposed to the simulated environment in the cabinet and there should
be a proper air flow between the components.

Figure 5.4 Temperature recording during the test on component placed in the middle. Units
of time in minutes and temperature in ℃.

43
 Figure 5.5 Temperature recording of the air between two components in the basket. Units
of time in minutes and temperature in ℃.

Case(iii) – Dismantled component

The motor is dismantled into three parts as shown in Figure 5.6. The part 1 is made up of
heavy metals and is considered to consumed long during to heat or cool down during the test.
Part 2 is the PCB present in the component and part 3 is the plastic part of the component
which has electrical unit mounted inside. These parts were tested repeated with changing the
DT. In this section the results of the test performed with CPS as shown in Table 5.4 are
displayed.

Table 5.4 : CPS for case(iii) i.e. dismantled components testing.

Test parameters Value

Temperature range (∆𝑻) +100 ℃ to -40 ℃

Dwell time (DT) 40 minutes

Number of cycles in test (𝑵𝒕) 0.5 cycle – only hot chamber duration of the test.

44
 Figure 5.6 Schema of dismantled motor parts used for testing.

Test results and observation

Table 5.5 : The position of all the thermocouples used during the test case (iii).

Thermocouple Location and reason

TCK1 It is placed inside the plastic part of the motor to log the temperature of the
electrical unit.

TCK2 It is placed on the PCB. The temperature log reflects the thermal conductivity of
the PCB

TCK3 Another electrical unit is placed on one side of the motor. This thermocouple is
placed on the electrical unit side.

TCK4 This is placed on motor surface.

TCK5 This is placed on the plastic surface.

TCK6 This thermocouple logs the temperature in the hot chamber during test

45
 Figure 5.7 Temperature recordings from six different locations on parts of the motor
including the hot chamber log for reference.

From the Figure 5.7, it is observed that the dismantling of the component helps in reducing
the test time. The parts in the test reach stability within 30 minutes and stays at a stable state
for 10 minutes, which is desirable of the ATCT. But when the component is tested as a
complete piece, it at least takes 40 - 50 minutes to reach stability. If the component is
dismantled for the test then the test time will be reduced by 30 % to 60 %. Moreover, all the
parts undergo stress to the complete extent and results in efficient testing. There could be
chances of overseeing fatigue failures at the junction of plastic and motor part. But the PCB
will be tested according to the requirements. After performing the test on the dismantled parts,
they can be mounted together and functional test can be performed.

5.2 Aluminum Box

A single aluminum box was tested. Major part of this component is hollow (filled with air).
DT for this component was decided by preforming trial and error method. There are three
types of test set-ups, depending on how much area of the component is exposed to the
temperatures during the tests. The possible test cases are (i) horizontally placed, (ii) vertically
placed and (iii) vertically placed with lid removed (partially dismantled). The test profile
used for these tests, is shown in Table 5.6. From this test, the temperature change inside the
component is observed and it is also analyzed how the component behaves when position is
changed. Further, the CPSs and EPVs for different test cases are compared in Table 5.7.

46
Table 5.6 : CPS for testing of the aluminum box.

Test parameters Value

Temperature range (∆𝑻) +80 ℃ to -40 ℃

Dwell time (DT) 50 minutes

Number of cycles in test (𝑵𝒕) 3

Figure 5.8 Drawing of aluminum box used for ATCT.

Case(i) – Horizontally placed

The component is placed in a horizontal manner as shown in Figure 5.9, the basket will
experience a temperature change. But the side of the component facing downwards against
the basket is shielded from airflow.

Figure 5.9 Drawing of aluminum box placed horizontally in the basket.

47
Case(ii) – Vertically placed

The component is placed vertically as shown in Figure 5.10, which exposes a bigger area of
the component to the cabinet’s climate change. A stand was designed and assembled
especially for this aluminum box (as shown in Figure 5.11).

Figure 5.10 Drawing of aluminum box placed vertically in the basket.

Figure 5.11 Photo of designed stand which is used to help the component stand vertically in
the basket.

Case(iii) – Vertically placed with lid removed (partially dismantled)

The walls of the component are thick, so the temperature of the air inside is heated and cooled
with slow RR. When we remove the lid of the component as shown in Figure 5.12, the air
temperature inside is changed with a higher RR.

48
 Figure 5.12 Drawing of aluminum Box with open lid.

Figure 5.13 Drawing of aluminum Box with open lid placed vertically in the basket.

Comparison of the results of the aluminum box

The results show the temperature recordings on the box, air inside and the plastic in the box.
From Figure 5.14 to Figure 5.16, it is seen that the RR is high in case (ii) and case (iii) when
compared to case(i). From Figure 5.14, it is observed that the outer layer of the component
reaches the maximum temperature values when placed vertical and vertical with open lid.

49
Figure 5.14 Temperature response measured on the outer surface of the aluminum box from
ATCT. The graph includes three different test cases on aluminum box.

Figure 5.15 Temperature response measured at the air inside the aluminum box from
ATCT. The graph includes three different test cases on aluminum box.

50
 Figure 5.16 Temperature response measured on the plastic inside the aluminum box from
ATCT. The graph includes three different test cases on aluminum box.

Table 5.7 : Comparison of CPS and EPVs for the aluminum box for three test cases. For the
all the test cases the DT is 50 minutes.

Temperature response Test scenario 𝑻𝒎𝒂𝒙 (℃) 𝑻𝒎𝒊𝒏 (℃) ∆𝑻 (℃) 𝑹𝑹

location. (℃/𝒎𝒊𝒏𝒖𝒕𝒆𝒔)

CPS 80 ℃ −40 ℃ 120 ℃ −

Temperature response Horizontal 76 ℃ −33 ℃ 109 ℃ 3.92

on the outer surface of
aluminum box Vertical 79 ℃ −35 ℃ 114 ℃ 6.03

Vertical open lid 80 ℃ −40 ℃ 120 ℃ 10.8

Temperature response Horizontal 63 ℃ −15 ℃ 78 ℃ 1.86

from the air inside the
aluminum box Vertical 65 ℃ −24 ℃ 89 ℃ 1.88

Vertical open lid 74 ℃ −33.5 ℃ 107.5 ℃ 3.20

Temperature response Horizontal 52 ℃ −7 ℃ 59 ℃ 1.83

on the plastic layer
inside the aluminum Vertical 55 ℃ −8 ℃ 63 ℃ 1.89
box.
Vertical open lid 62 ℃ −21 ℃ 83 ℃ 2.13

51
The EPVs are calculated from temperature responses displayed Figure 5.14, 5.15, 5.16. These
values are compared with CPSs in Table 5.7. From the table, it can be observed that the test
cases with vertical position and vertical position with open lib results in higher value of RR.
It is favorable to have high RR and the component reaches maximum temperature values
during the test. Therefore, it is always preferable to place the component in such a manner
that maximum area is exposed during the test.

5.3 Test on Plastic Box under Similar Test Conditions as

Aluminum Box
The plastic box is tested with test conditions similar to test case(ii) of the aluminum box, but
with less DT. This component is similar to the aluminum box, as it has same structure and
almost 50 % is hollow. Test cases are same as the aluminum box and the results are compared
with each other. Table 5.8 displays the test profile for the plastic box. The plastic box is tested
with DT of 30 minutes and the aluminum box with DT of 50 minutes. This is because the
aluminum box is made up of thick outer layers and the plastic box has thin layers.

Table 5.8 : CPS for testing of the plastic box.

Test parameters Value

Temperature range (∆𝑻) +80 ℃ to -40 ℃

Dwell time (DT) 30 minutes

Number of cycles in test (𝑵𝒕) 3

Figure 5.17 Drawing of plastic Box used for ATCT.

Test set-up: Vertical Position.

Figure 5.18 Outline of plastic box placed vertically in the basket.

52
Comparison of results of plastic box and aluminum box

Case(i) : Comparison between box surface temperature of plastic box and aluminum box
– horizontal

Figure 5.19 Temperature response measured on the outer surface of aluminum box and
plastic box from ATCT.

Case(ii) : Comparison between air temperature inside the plastic box and aluminum box –
vertically.

Figure 5.20 Temperature response measured in the air inside the aluminum box and plastic
box from ATCT.

53
Case(iii) : Comparison between temperature on material placed inside the plastic box and
aluminum box – vertically with open lid.

Figure 5.21 Temperature response measured from the plastic inside the aluminum box and
from the PCB inside the plastic box from ATCT.

Table 5.9 : Comparison of CPS and EPV for aluminum box and plastic box. They are
positioned vertically in the basket during the ATCT.

Temperature Test scenario 𝑻𝒎𝒂𝒙 (℃) 𝑻𝒎𝒊𝒏 (℃) ∆𝑻 (℃) 𝑹𝑹 𝑫𝑻

response
location. (℃ (𝒎𝒊𝒏𝒖𝒕𝒆𝒔)
/𝒎𝒊𝒏𝒖𝒕𝒆𝒔)
Set manually*

CPS 80 ℃ −40 ℃ 120 ℃ - 30 𝑎𝑛𝑑 50

Temperature Plastic box 80 ℃ −39 ℃ 119 ℃ 10.72 30

response on
the outer
surface. Aluminum box 79 ℃ −35 ℃ 114 ℃ 6.03 50

Temperature Plastic box 80 ℃ −39 ℃ 119 ℃ 10.71 30

response from
the air inside
Aluminum box 65 ℃ −24 ℃ 89 ℃ 1.88 50
the box

Temperature PCB inside the plastic 80 ℃ −39 ℃ 119 ℃ 10.60 30

response from box
layer inside
the box Plastic layer inside 55 ℃ −8 ℃ 63 ℃ 1.89 50
aluminum box

54
The EPVs are calculated from temperature responses displayed Figure 5.19, 5.20, 5.21. These
values are compared with CPSs in Table 5.9. From the table, it is observed that the plastic
box has the higher RR when compared to the RR of the aluminum box, even though the
aluminum box has the longer DT than the plastic box. From this comparison it is clear that,
if two components have similar physical structure and same environment on the vehicle, the
value of the DT need not be same. The DT value will vary depending on the material used
and the thickness of the outer layer of the component. The DT value should be carefully
assessed by performing a pre-test. The pre-test will help in understanding the factors
influencing ATCT parameters.

5.4 Comparison Between Standards and Current

Scenario of Testing
Table 5.10 : Comparison between standards and current scenario of testing.

Factors Standard Current Scenario of testing

Dwell time (DT) It is the soak time for 10 minutes The time duration at 𝑇𝑚𝑎𝑥 / 𝑇𝑚𝑖𝑛 in
after reaching the thermal stability one cycle i.e. half cycle.
or complete test time.

Stability State When the DUT reaches May not reach stability state
±3 ℃/±5 ℃ of 𝑇𝑚𝑎𝑥 OR sometimes.
±3 ℃/±5 ℃ of 𝑇𝑚𝑖𝑛 is noted as
stability state.

Ramp rate (RR) RR for thermal shock > Depends on the bulkiness and
15 ℃/minute. RR for thermal materials of the component.
cycling < 15 ℃/minute.

(∆𝑻), 𝑻𝒎𝒂𝒙, 𝑻𝒎𝒊𝒏 Values are chosen depending on May not reach required temperature
the type of component and the range or temperature levels, due to
environment for the component on short soak time or low RR.
a vehicle.

5.5 Conclusions from Current Scenario Testing

The results show how important it is to have the correct settings for a certain component. The
measurements suggest the following recommendations:

• A need for adjustment of the placement of the components in the cabinet’s basket
depending on number of components, air flow in the chamber, thickness of the outer
layer of the component.

55
• Every test set-up should focus on obtaining stability state especially for the most
relevant part of the components and have at-least 10 minutes – 15 minutes of soak
time.

• A pre-test should be performed before starting the accelerated testing to assess the
optimal choice of test parameters and correct test setup for the component depending
on its shape and materials.

• The component should be placed in a position that will help exposing maximum area
of the component to the simulated environment in the cabinet. This will help in
achieving high RR and reaching the maximum or minimum temperature levels.

• The DT value will vary depending on the material used and the thickness of the outer
layer, even if the components have similar physical structure and same environment
on the vehicle.

• Tests can be carried out by dismantling the components, agreement has to be made
with the supplier of the component.

56
57
6 ESTIMATION OF THE AF FROM CURRENT
TESTING SCENARIO

In this project, AF for a component is estimated using results from a ATCT and life prediction
models i.e. Coffin Manson and Norris-Landzberg models (refer (2.2), (2.3) and (2.4)). From
the estimated AFs, the number of cycles in the field (𝑁𝑓 ) is obtained (refer to (6.1)). A
component with a life length of 15 years in a real application, is estimated to be exposed to
10000 cycles in the field. Therefore, to attain 10000 cycles in field, a sufficient number of
cycles in test (𝑁𝑡 ) should be performed depending on the acceleration factor.

Number of cycles in field (𝑁𝑓 ) = AF × Number of cycles in test (𝑁𝑡 ) (6.1)

In this chapter, AF is estimated from the results obtained during current scenario testing (from
chapter 5). This is done by calculating the EPV based on the obtained temperature graphs
and implementing it in (2.2), (2.3) and (2.4). Then implementing each AF value in (6.1), to
relate the 𝑁𝑡 to 𝑁𝑓 . The following three sections describe how the AF will be affected due to
(i) change in component position during the ATCT (from section 5.2), (ii) with same test
profile and change in size and material of the component (from section 5.3), (iii) change in
each test parameter value individually.

6.1 Effect on AF When Compared to Position of the

Component
Consider the temperature recording of “air inside” the aluminum box with different position
of the component. With same DT in each position, a difference in the RR and ∆𝑇 is found.
When a standalone PCB (without housing) is tested, it generally takes very little time to heat
up and cool down. The temperature measurements on the PCB during the test cycle follow
the cabinets temperature (For reference – Appendix D). In any electrical component, PCBs
are considered to be one of the crucial parts. Therefore, finding the life length of the PCBs
will be appropriate to estimate the effectiveness of the accelerated testing performed in the
lab. Air temperature inside the aluminum box is measured for the PCB temperature during
ATCTs. Two cases are studied: (i) PCB with lead-free soldering and (ii) PCB with leaded
soldering.

Using the results shown below in Figure 6.1. The AFs are calculated using both the models
i.e. Coffin-Manson model (2.2) and Norris Landzberg model ((2.3) and (2.4)). Then the 𝑁𝑓
are calculated using (6.1). Figure 6.2 represents the values for each model and different
soldering cases. It is observed that the vertical position with open lid is a better test set up for
lead-free soldering to obtain a high AF.

58
Estimation of AFs from the ATCT results of aluminum box.

Table 6.1 : CPS programmed for the test. These values are applicable to all the three test
cases for the aluminum box.

Parameters Programmed values for the test

Maximum temperature (𝑻𝒎𝒂𝒙 𝒕𝒆𝒔𝒕 ) +80 ℃

Minimum temperature (𝑻𝒎𝒊𝒏 𝒕𝒆𝒔𝒕 ) −40 ℃

Temperature range (∆𝑻 = 𝑻𝒎𝒂𝒙 𝒕𝒆𝒔𝒕 − 120 ℃

𝑻𝒎𝒊𝒏 𝒕𝒆𝒔𝒕 )

Dwell time (DT) 50 minutes

Figure 6.1 Temperature response of air side the aluminum box for AF calculations.

59
Table 6.2 : Comparison CPS and EPV for temperature response obtained from ATCT of
aluminum box. EPVs estimated from the temperature graphs (in Figure 6.1).

Parameters CPS Horizontal Vertical Vertical open-

position position lid position

Maximum temperature 80 ℃ 63 ℃ 65 ℃ 74 ℃
(𝑻𝒎𝒂𝒙 𝒕𝒆𝒔𝒕 )

Minimum temperature -40 ℃ −15 ℃. −24 ℃. −33.5 ℃.

(𝑻𝒎𝒊𝒏 𝒕𝒆𝒔𝒕 )

Temperature range 120 ℃ 78 ℃ 89 ℃ 107.5 ℃

(∆𝑻 = 𝑻𝒎𝒂𝒙 𝒕𝒆𝒔𝒕 −
𝑻𝒎𝒊𝒏 𝒕𝒆𝒔𝒕 )

Dwell time (DT) 50 minutes 50 minutes 50 minutes 50 minutes

Ramp rate (RR) – - 56 −(−13)

=
58 −(−21)
=
66 −(−30)
=
Appendix B 253−216 255−213 243−213

1.86 ℃/𝑚𝑖𝑛𝑢𝑡𝑒 1.88 ℃/𝑚𝑖𝑛𝑢𝑡𝑒 3.2 ℃/𝑚𝑖𝑛𝑢𝑡𝑒

Number of cycles in 300 300 300 300

test (𝑵𝒕)

The values shown in Table 6.2 are applied in (2.2), (2.3) and (2.4) to obtain respective AF
values for each model. Then these AF values are substituted in (6.1) to obtain number of
cycles in the field (𝑁𝑓 ).

Type of Component versus Number of cycles in field.

10000
9131
9000
8000
Number of Cycles in Field

7000 m = 2.65 m = 2.5

6000
4773 Horizontal
5000 4373 4119 Vertical
4000 3308 3552
Vertical_Open Lid
3000 2497 2449 2215
1760 1715 1592
2000
1000
0
Norris Landzberg Coffin-Manson Lead Norris Landzberg Coffin-Manson
Lead Free Free Leaded Leaded

Figure 6.2 Estimated values of 𝑁𝑓 for aluminum box in three test cases using Coffin-
Manson and Norris Landzberg for lead free solder (m=2.65) and leaded solder (m=2.5).

60
The AFs are estimates for the aluminum box in three different test cases, i.e. (1) when
aluminum box is placed horizontally (2) when the aluminum box is placed vertically and (3)
when the aluminum box is placed vertically with open lid. The two life predicting models,
Coffin Manson and Norris-Landzberg models are used to estimate the AFs. Using the
estimated AF values, 𝑁𝑓 for each case is calculated for 𝑁𝑡 = 300 as shown in Figure 6.2.
From the Figure 6.2 it is seen that Norris-Landzberg model yields high and accurate AF than
Coffin Manson model as the Norris-Landzberg model considers all the test parameters, not
just the temperature range as in Coffin Manson model. And moreover, the test conditions are
more favorable to results in high 𝑁𝑓 when the component is placed vertically with open lid.
This is because the position of the component in the basket helps in exposing maximum area
of the component to the simulated environment and result in high RR.

6.2 Effect on AF When Compared to Size and Material

of the Component
When the outside box material and size is changed, then we find drastic difference in the
temperature change of air inside the box. Similar to section 6.1, air temperature inside the
plastic box and aluminum box (Figure 6.3) are consider to estimate the number of cycles on
field (𝑁𝑓 ) for performed number of cycles in the test (𝑁𝑡 ) (Appendix D). Using the parameter
values from Figure 6.3, the AFs and 𝑁𝑓 are calculated and represented in Figure 6.4. A large
deviation in ‘𝑁𝑓 ’ can be observed when the outer layer is changed. This presents that, test
setups and parameter values differ drastically for similar components even when outer layer
is different.

Estimation of AFs for results of plastic box and comparing with the estimated AF values
for aluminum box.

Table 6.3 : CPS programmed for the test. These values are applicable for vertical positioned
testing of the aluminum box and plastic box

Parameters Aluminum box Plastic box

Maximum temperature +80 ℃ +80 ℃

(𝑻𝒎𝒂𝒙 𝒕𝒆𝒔𝒕 )

Minimum temperature −40 ℃ −40 ℃

(𝑻𝒎𝒊𝒏 𝒕𝒆𝒔𝒕 )

Temperature range (∆𝑻 = 120 ℃ 120 ℃

𝑻𝒎𝒂𝒙 𝒕𝒆𝒔𝒕 − 𝑻𝒎𝒊𝒏 𝒕𝒆𝒔𝒕 )

Dwell time (DT) 50 minutes 30 minutes

Outer layer thickness 1.5 cm 0.8 cm

61
 Figure 6.3 Temperature response inside Aluminum box and Plastic box.

Table 6.4 : Estimated parameter values from the temperature graphs (in Figure 6.3), obtained
from the tests.

Parameters CPS Aluminum box – Plastic box – vertical

vertical position position

Maximum temperature 80 ℃ 65 ℃ 80 ℃
(𝑻𝒎𝒂𝒙 𝒕𝒆𝒔𝒕 )

Minimum temperature -40 ℃ −24 ℃ −39 ℃

(𝑻𝒎𝒊𝒏 𝒕𝒆𝒔𝒕 )

Temperature range 120 ℃ 89 ℃ 119 ℃

(∆𝑻 = 𝑻𝒎𝒂𝒙 𝒕𝒆𝒔𝒕 −
𝑻𝒎𝒊𝒏 𝒕𝒆𝒔𝒕 )

Dwell time (DT) Aluminum box = 50 minutes 30 minutes

50 minutes

Plastic box = 30
minutes

Ramp rate(RR) – - 58 −(−21)

=
72 −(−35)
=
Appendix B 255−213 138−127
1.88 ℃/𝑚𝑖𝑛𝑢𝑡𝑒 9.72 ℃/𝑚𝑖𝑛𝑢𝑡𝑒

Number of cycles in 300 300 300

test (𝑵𝒕)

62
These values are shown in Table 6.4 are applied in (2.2), (2.3) and (2.4) to obtain respective
AF values for each model. Then this AF values are substituted in (6.1) to obtain number of
cycles in the field (𝑁𝑓 ).

Type of Component versus Number of cycles in field.

12000 11409

10000
Number of cycles in Field

7906
8000

6000 5393 Plastic Box

4553 4579
Aluminum Box
4000
2351 2352 2093
2000

0
Norris Landzberg Coffin-Manson Norris Landzberg Coffin-Manson
Lead Free Lead Free Leaded Leaded

Figure 6.4 Estimated values of 𝑁𝑓 for aluminum box and plastic box using Coffin-Manson
and Norris Landzberg for lead free solder (m=2.65) and leaded solder (m=2.5).

The AFs are estimates for the plastic box and the aluminum box when the component is
placed vertically in the basket of the cabinet. The two life predicting models, Coffin Manson
and Norris-Landzberg models are used to estimate the AFs. Using the estimated AF values,
𝑁𝑓 for each case is calculated for 𝑁𝑡 = 300 as shown in Figure 6.4. From the Figure 6.4, it
is observed that even if the components have the similar test conditions the accelerating
conditions or the AF vary depending on the materials used in the component. Even though
the structure of two components are similar, the DT values vary depending on the thickness
and the material of the outer layer. The DT should be assessed in such a manner that it
comprises of ramp time and soak time to achieve high accelerating condition.

6.3 Effect on AF when Test Parameters Varying – Norris

Landzberg Model
The AF can be described as the ratio of the product life at normal operating conditions to the
life at accelerated test conditions. So far, we have seen how the AF value varying depending
on the position of the components in cabinet and the thickness and the material of the outer
layer of the component. In this chapter, two issues are studied: (i) to analyze how the AF
value is effect when one parameter is varied with other two parameters are constant. The
main three parameters considered in this section are (1) DT, (2) RR and (3) ∆𝑇 and (ii) to
analyze the relation between parameters and how to achieve high accelerating conditions.
The Norris Landzberg model is used i.e. (2.3) and (2.4).

63
Main factors:

• Dwell time (DT).

The AF and the 𝑁𝑓 are estimated using Norris-Landzberg for leaded solder and lead-free
solder with constant values for 𝑅𝑅 , 𝑁𝑡 , ∆𝑇𝑓𝑖𝑒𝑙𝑑 , ∆𝑇𝑡𝑒𝑠𝑡 , 𝐷𝑇𝑓𝑖𝑒𝑙𝑑 , 𝑇𝑓𝑖𝑒𝑙𝑑 𝑚𝑎𝑥 , 𝑇𝑡𝑒𝑠𝑡 𝑚𝑎𝑥 ,
𝑇𝑓𝑖𝑒𝑙𝑑 𝑚𝑖𝑛 and 𝑇𝑡𝑒𝑠𝑡 𝑚𝑖𝑛 . The 𝐷𝑇𝑡𝑒𝑠𝑡 value is varied. The values for each parameter is mention
in Table 6.5.

Table 6.5 : Assigned parameter values used in Norris-Landzberg model with varying 𝐷𝑇𝑡𝑒𝑠𝑡

Parameters Assigned values

RR 12 ℃/𝑚𝑖𝑛𝑢𝑡𝑒

𝑫𝑻𝒕𝒆𝒔𝒕 0 𝑡𝑜 100 𝑚𝑖𝑛𝑢𝑡𝑒𝑠

𝑫𝑻𝒇𝒊𝒆𝒍𝒅 270 𝑚𝑖𝑛𝑢𝑡𝑒𝑠

∆𝑻𝒇𝒊𝒆𝒍𝒅 40 ℃

∆𝑻𝒕𝒆𝒔𝒕 120 ℃

𝑻𝒇𝒊𝒆𝒍𝒅 𝒎𝒂𝒙 +30 ℃

𝑻𝒕𝒆𝒔𝒕 𝒎𝒂𝒙 +80 ℃

𝑻𝒇𝒊𝒆𝒍𝒅 𝒎𝒊𝒏 −10 ℃

𝑻𝒕𝒆𝒔𝒕 𝒎𝒊𝒏 −40 ℃

𝑵𝒕 300

The effect on the AF reaches saturation as the DT is increased above 20 minutes. But these
20 minutes are considered to be the time after reaching stabilization, as considered in few
standards. High AF value indicates high accelerating conditions. From Figure 6.5 it is seen
that with small DT, the ATCT is highly accelerated but as the DT is increased the acceleration
conditions reduce. With longer DT, the test reaches saturation which is not favorable. In
Figure 6.6, the estimated AF is used to calculate 𝑁𝑓 and it is observed that the AF is directly
proportional to 𝑁𝑓 . Therefore, high the AF yield high 𝑁𝑓 .

64
Figure 6.5 AF verses DT in ATCT. Where DT is varied from 0 minute to 100 minutes and
AF is calculated according to the DT value.

Figure 6.6 Number of cycles in field (𝑁𝑓 ) is calculated from the AF calculates in Figure 6.5
and with x-axis representing the DT varying from 0 minute to 100 minutes.

65
 • Ramp rate (RR).

The AF and the 𝑁𝑓 are estimated using Norris-Landzberg for leaded solder and lead-free
solder with constant values for 𝐷𝑇𝑡𝑒𝑠𝑡 , 𝑁𝑡 , ∆𝑇𝑓𝑖𝑒𝑙𝑑 , ∆𝑇𝑡𝑒𝑠𝑡 , 𝐷𝑇𝑓𝑖𝑒𝑙𝑑 ,𝑇𝑓𝑖𝑒𝑙𝑑 𝑚𝑎𝑥 , 𝑇𝑡𝑒𝑠𝑡 𝑚𝑎𝑥 ,
𝑇𝑓𝑖𝑒𝑙𝑑 𝑚𝑖𝑛 and 𝑇𝑡𝑒𝑠𝑡 𝑚𝑖𝑛 . The 𝑅𝑅 value was varying. The values for each parameter is mention
in Table 6.6.

Table 6.6 : Assigned parameter values used in Norris-Landzberg model with varying 𝑅𝑅.

Parameters Assigned values

RR 0 𝑡𝑜 20 ℃/𝑚𝑖𝑛𝑢𝑡𝑒

𝑫𝑻𝒕𝒆𝒔𝒕 12 𝑚𝑖𝑛𝑢𝑡𝑒𝑠

𝑫𝑻𝒇𝒊𝒆𝒍𝒅 270 𝑚𝑖𝑛𝑢𝑡𝑒𝑠

∆𝑻𝒇𝒊𝒆𝒍𝒅 40 ℃

∆𝑻𝒕𝒆𝒔𝒕 120 ℃

𝑻𝒇𝒊𝒆𝒍𝒅 𝒎𝒂𝒙 +30 ℃

𝑻𝒕𝒆𝒔𝒕 𝒎𝒂𝒙 +80 ℃

𝑻𝒇𝒊𝒆𝒍𝒅 𝒎𝒊𝒏 −10 ℃

𝑻𝒕𝒆𝒔𝒕 𝒎𝒊𝒏 −40 ℃

𝑵𝒕 300

From Figure 6.7 it is observed that low RR is a high accelerating condition for Lead-free
soldering. But a high RR is a high accelerating condition for Leaded soldering. To obtain
high 𝑁𝑓 , the parameters should be assessed to obtain high RR for leaded solder PCB and low
RR for lead-free soldered PCB.

66
Figure 6.7 AF verses RR in ATCT. Where RR is varied from 0 ℃/minute to 20 ℃/minute
and AF is calculated according to the RR value.

Figure 6.8 Number of cycles in field (𝑁𝑓 ) is calculated from the AF calculates in Figure 6.7
and with x-axis representing the RR varying from 0 ℃/minute to 20 ℃/minute.

67
 • Temperature Range (∆𝑻).

The AF and the 𝑁𝑓 are estimated using Norris-Landzberg for leaded solder and lead-free
solder with constant values for 𝐷𝑇𝑡𝑒𝑠𝑡 , 𝑁𝑡 , ∆𝑇𝑓𝑖𝑒𝑙𝑑 , 𝐷𝑇𝑓𝑖𝑒𝑙𝑑 ,𝑇𝑓𝑖𝑒𝑙𝑑 𝑚𝑎𝑥 , 𝑇𝑓𝑖𝑒𝑙𝑑 𝑚𝑖𝑛 and RR. The
𝑇𝑡𝑒𝑠𝑡 𝑚𝑎𝑥 and 𝑇𝑡𝑒𝑠𝑡 𝑚𝑖𝑛 value was varying which results in varying ∆𝑇𝑡𝑒𝑠𝑡 . The values for
each parameter is mention in Table 6.7.

Table 6.7 : Assigned parameter values used in Norris-Landzberg model with varying
𝑇𝑡𝑒𝑠𝑡 𝑚𝑎𝑥 and 𝑇𝑡𝑒𝑠𝑡 𝑚𝑖𝑛 .

Parameters Assigned values

RR 12 ℃/𝑚𝑖𝑛𝑡𝑢𝑒

𝑫𝑻𝒕𝒆𝒔𝒕 12 𝑚𝑖𝑛𝑡𝑢𝑒𝑠

𝑫𝑻𝒇𝒊𝒆𝒍𝒅 270 𝑚𝑖𝑛𝑡𝑢𝑒𝑠

∆𝑻𝒇𝒊𝒆𝒍𝒅 40 ℃

𝑻𝒇𝒊𝒆𝒍𝒅 𝒎𝒂𝒙 +30 ℃

𝑻𝒇𝒊𝒆𝒍𝒅 𝒎𝒊𝒏 −10 ℃

𝑵𝒕 300

From Figure 6.9, it can be observed that decreasing the ∆𝑇 , leads to less accelerated
conditions. One main parameter effecting the ∆𝑇 is the DT .When DT is not assessed
according to the component under test, the DT could be too short or too long for the
component. When the component does not reach the temperature level (𝑇𝑡𝑒𝑠𝑡 𝑚𝑎𝑥 and 𝑇𝑓𝑖𝑒𝑙𝑑 𝑚𝑖𝑛 )
in the ATCT, leads to small ∆𝑇. Therefore, the assessed parameters values should be able to
simulate cabinet environment which can achieve high ∆𝑇 for the specific component.

68
 Figure 6.9 Number of cycles in field (𝑁𝑓 ) is calculated with varying ∆𝑇 in ATCT.

To summarize, the DT should as shorter as possible and should accommodate ramp time and
soak time for a specific component. The ATCT should be able to simulate test environment
in such a way that leaded soldered PCB have high RR and lead-free soldered PCB have low
RR. Reduce in the ∆𝑇 will lead to low AF which is bad accelerating conditions. The DT is
the key parameter which can easily change the simulating condition in the cabinet for ATCT.
The DT should be assessed according to the component under test with help of a pre-test.

6.4 Summary

High
10-15 Improves
More Accelerating
Increase in Increases the minutes in the chance
Favourable conditions
RR in test ∆𝑇. stabilization to find
Set-up and Good
state. defects
AF

Figure 6.10 Flowchart describing the dependency of factor for improving testing

• If the component has maximum area exposed to the environment in the ATCT, it will
be helpful to achieve high RR. With high RR, the temperature levels can be reached
during the test.

69
• As explained in Figure 6.10, 10-15 minutes in stability state will lead to high
accelerating conditions and improve the chances of detecting the failure or defects.

• Each model used for calculation of acceleration factor, gives its own estimated value.

• Parameter values are to be properly compared with the values obtained during the test
for calculating the AF.

• When similar components are analyzed, parameter values and accelerating conditions
differ drastically even if there are differences in material and thickness of the
components.

70
71
7 SOFTWARE VALIDATION

In this project, investigation is made to check how simulations can be used to assess the
optimal choice of test parameter values. A reduction of test time or test efforts without losing
information about the risk of failure is desirable. To perform these simulations, there are four
steps to be followed. (1) Structure and material properties have to be designed and
implemented, (2) The input to the component have to be declared i.e. the cabinet test
environment using CPS, (3) Type of stress used and physics involved have to be included,
and (4) The temperature distribution in the component and the temperature log from any
point can be obtained after running the simulation. COMSOL Multiphysics software was
used for simulations.

7.1 Simulation Models

To start off with simulation, aluminum box test component is considered. As the electrical
components are very complex in structure, a basic replication of aluminum box was designed
as shown in Figure 7.1. A virtual aluminum box with a plastic and a FR4 layer were included.
The structure and material used for simulations are described in Figure 7.1. These simulations
are carried out in two stages. At first stage, simulations are performed with 2D model and at
the second stage 3D model is used. At each stage two different cases are performed as
mentioned in Table 7.1.

Table 7.1 : Simulations were performed using following scenarios and test cases.

Scenarios Simulation Type of input Cases Motivation

model used used in
simulation

1 2D model – Ramping up the Case(i) – Horizontal The input is given only

similar to the temperature from position from three sides.
cross section of room
the component temperature to Cases(ii) – Vertical The input is given from
maximum position all four sides
temperature

2 3D model – Three test cycles, Case(i) – Without There is no medium of

complete similar from screws on the plastic conduction through the
structure of the maximum layer inside the aluminum box and the
component temperature to component plastic layer
minimum
temperature Case(ii) – With Helps to understand the
screws on the plastic important detailing in the
layer inside the simulation model
component

72
Figure 7.1 Structure and material of a component for simulations.

73
7.2 Physics Applied
The three modes of thermal energy transfer are applied to every model that has been
simulated in the COMSOL Multiphysics as shown in Figure 7.2.

Figure 7.2 Physics applied in 2D and 3D model.

The basket in the cabinet is constantly moved from hot chamber to cold chamber and vice
versa. This event consists of having a constant temperature for defined time period, ramping
down from high temperature to cold temperature and ramping up from cold temperature to
hot temperature. A signal similar to square function with three cycles is built as shown in
Figure 7.3. This signal is used as the input for the simulation model. As this helps in
simulating the real test scenario for the component. Another case with just ramping up the
temperature from room temperature is used. This input is defined in Figure 7.4.

74
 Temperature Cycles
100

80
Temperature (celcius)

60

40

20

0
0 50 100 150 200 250 300 350
-20

-40

-60
Time (min)

Figure 7.3 Input temperature signal for simulations – 3 cycles – Used in 3D model.

Ramping up temperature - Input

90
Temperature (celcius)

80
70
60
50
40
30
20
10
0
0 5 10 20 30 40 50
Time (min)

Figure 7.4 Input temperature signal for heating the component to the maximum temperature
– Used in 2D model.

7.3 Simulation Scenarios

Firstly, a 2D model is simulated. This 2D model is the cross-sectional view along the width.
Then this model is extended to a 3D model.

75
Scenario 1: 2D Model

In the 2D model there are two scenarios of simulations. The 2D model used for simulation is
shown in Figure 7.5.

Case(i): Horizontally placed – Giving input (i.e. temperature change) only to three side,
insulation one side.

Case(ii): Vertically placed – Giving input (i.e. temperature change) from all the four sides.
Height

Width
Figure 7.5 2D simulation model.

Scenario 2: 3D Model

The 3D model is an extension from the 2D model. There are two scenarios of simulations:

Case(i): Without Screws on the plastic plate – The input is given from all sides of the box.
But consists of aluminum box, plastic and FR4. The 3D model without screws used for
simulation is shown in Figure 7.7.

76
 Height
Length
Width

Figure 7.6 3D simulation model without screws.

Case(ii): With Screws on the plastic plate – The input is given from all sides of the box. But
even screws are included, which are connected to the bottom side of the box and the plastic
layer. The 3D model with screws used for simulations is shown in Figure 7.7.
Height

Length
Width

Figure 7.7 3D simulation model with screws.

7.4 Simulations Results and Comparison with Results

from Current Test Scenario
Considering three locations in the component i.e. on the box, on the plastic and air inside the
box, Figure 7.8 and 7.9 provides the result from the practical tests performed on the
component using the temperature change cabinet. From these results, it is observed that when
the component is placed horizontally, the box takes 25-30 minutes to reach stability in the
hot chamber. But when the component is placed vertically, then the box takes 10-15 minutes

77
to reach the stability in the hot chamber. This can be noted as a key situation to verify the
results from the simulation performed on different models. Further when 3D models are
performed, the ∆𝑇 i.e. the ∆T of the plastic and the air inside should be observed.

Figure 7.8 Temperature response from current test scenario, when the component is placed
horizontally.

Figure 7.9 Temperature response from current test scenario, when the component is placed
vertically.

78
7.4.1 Simulation Scenario 1 - 2D Model
Case(i) – Horizontally Placed

Input from Figure 7.4 is given on only three sides of the box. As the bottom is shield from
the air flow in the practical test. The bottom surface has reduced heat flux in the simulation.
It is heated for 50 minutes from room temperature (22 ℃) to 80 ℃.

Figure 7.10 Temperature distribution in the box (2D model).

Figure 7.11 Simulated temperature response for first 50 minutes of heating (2D model).

Observation: This resembles the result from the current test scenario. It is seen from Figure
7.8, the aluminum box takes 25-30 minutes to reach stability when placed horizontal position.
This is similar to the result from the simulated temperature response, as it can be seen that
the aluminum box takes 25-30 minutes to reach stability state when placed in horizontal
position.

79
Case(ii) – Vertically Placed

Input from Figure 7.4 is given on all four sides of the box. It is heated for 50 minutes from
room temperature (22 ℃) to 80 ℃.

Figure 7.12 Temperature distribution in the box (2D model).

Figure 7.13 Simulated temperature response for first 50 minutes of heating (2D model).

Observation: This resembles the result from the current test scenario. as the box take 10-15
minutes to reach stability. It is seen from Figure 7.9, the aluminum box takes 15-20 minutes
to reach stability when placed in vertical position. This is similar to the result from the
simulated temperature response, as it can be seen that the aluminum box takes 15-20 minutes
to reach stability state when placed in vertical position.

80
7.4.2 Simulation Scenario 2 - 3D Model
Case(i) – Vertically Placed and without Screws on the Plastic Plate

Input from Figure 7.3 is given on all six sides of the box. It is heated and cooled for 312
minutes. It is heated for 50 minutes and cooled for 50 minutes. The temperature cycle moved
from +80 ℃ to -40 ℃ and is comparable to three (complete) real test cycles.

Figure 7.14 Temperature distribution (3D model without screws) after 50 minutes of
heating.

Figure 7.15 Temperature distribution (3D model without screws) after 50 minutes of
cooling

81
 Figure 7.16 Temperature graph for 312 minutes for 3D model without screws.

Observation: This resembles the result from the practical test, as the box takes 10-15 minutes
to reach stability. But the plastic and air inside have smaller ∆𝑇 i.e. ∆T, than expected. This
leads to adding screws in the component model.

Case(ii) – Vertically Placed and with Screws on the Plastic Plate

Input from Figure 7.3 is given on all six sides of the box. It is heated and cooled for 312
minutes. It is heated for 50 minutes and cooled for 50 minutes. The temperature cycle moved
from +80 ℃ to -40 ℃ and is comparable to three (complete) real test cycles.

Figure 7.17 Temperature distribution (3D model with screws) after 50 minutes of heating.

82
Figure 7.18 Temperature distribution (3D model with screws) after 50 minutes of cooling.

Figure 7.19 Temperature graph for 312 minutes for 3D model with screws.

Observation: This resembles the result from the practical test, as the box take 10-15 minutes
to reach stability. The ∆𝑇s of plastic and air inside almost reached the values as in the
practical tests.

7.5 Observations
From these results of COMSOL simulations, it is seen that the practical tests can be replicated.
Almost similar results can be obtained by simulating detailed design of the component. The
DT period of these test can be increased or decreased to understand the behavior of the
component. This helps in assessing the DT period required for the component to reach the
stability during the test. These simulations can also be used to analyze the sensitive locations

83
in the component or the reason for the failures after the accelerated tests on the original
component. Other kinds of environmental stresses can be simulated in the same manner.
Using such kind of software simulation help in reducing human effort or test effort and obtain
the reliable results on the same hand.

84
85
8 CONCLUSIONS AND FUTURE WORK

8.1 Conclusions
• From the measurements that are recorded during this thesis, it is learnt that there is
deviation between the CPS and the EPV. This deviation indicates an uneven
temperature distribution in/on the component during the ATCT. The level of
deviation is different for different components but the deviation could be negligible
as well. To avoid the deviation between the CPS and EPV, the CPVs should be
adjusted according to the component size, structure, material and number of
components tested at the same time. The position of the component during the test
also affects the deviation of CPV and EPV.

• The DT is the main factor giving rise to the uneven temperature distribution in the
component during an ATCT. The value of DT can vary depending on the properties
of the DUT. But there are no strict guidelines to assess the value of the DT. One
possible way of assessing it is by preforming a pre-test on the DUT. It is advantageous
to perform at least one pre-test, to optimize the value of the DT and also to help in
understanding if there are any unknown factors affecting the test. For example, it was
detected that the cabinet used in this thesis requires an extra 5-10 minutes in the cold
chamber to reach a stable temperature in the basket.

• It is understood that adjusting the DT value according to the components physical and
material properties, will result in more robust testing. If the DT is considered to be
the complete duration in one chamber then the DT should be sufficient for ramp time
and soak time. The ramp time will help in reaching the stability state and the soak
time makes possible to stress the component at extreme temperatures. The ramp time
is different for different components. Pre-tests is useful for assessing the ramp time.
But if the DT is considered to be the duration after the component has reached stability
state, then some extra time has to be given for the component to reach stability. This
extra time could be long for bulky components.

• A bulky component would require a long duration for ramp time. In such a situation
the component can be broken down or dismantled to have a possibility of reducing
the ramp time.

• If there are different types of components in the basket at the same time. There are
high chances of radiating heat from each component in the hot chamber, this will lead
to difficulties in cooling down the components in the cold chamber. As a result, the
placement of the components in the basket is a crucial part when there are several
components for an ATCT at the same time. Therefore, placement of the components
should be adjusted in a way that there is proper airflow between the components.
Some results indicate that an additional 5 – 10 minutes extra time is needed in the
cold chamber to cool down the DUTs.

86
• Additional equipment such as stand, can be used to improve a test setup. The main
purpose is to expose maximum area of the component to the air flow in the cabinet.
In this thesis, a steel stand was used as shown in Figure 8.1 and Figure 8.2. But this
can be further improved by building a stand using low thermal conductivity material.

Figure 8.1 Example of usage of additional stand, which helps in improving the test.

Figure 8.2 Stand used in the testing.

• With longer DT, the Norris Landzberg model is approximated to the Coffin-Manson
model. When the DT is very long, then it leads to low RR. Therefore, the AF value
depends mainly on the extreme temperatures.

• In this thesis, it took around 6-7 rounds of pre-test for each component to assess the
CPS, which was time consuming. To speed up the acceleration test process and obtain
optimum values for CPS, software simulations can be performed on the desired
component. With a proper component model and exact material properties, suitable
results can be obtained in simulations.

• A decrease in ∆𝑇 due to large components, leads to less accelerated conditions.

Achieving a high RR for an accelerated testing is the main objective to improve a test
and attain high accelerating conditions.

87
8.2 Future Work
• There are few components in vehicles with heat sinks, such as lamps. The heat sink
helps in keeping the electrical part of the component cool. The ATCTs are also
performed on these components. One possible future work will be to learn how these
components with heat sink react in the ATCT.

• In this thesis, Coffin-Manson and Norris Landzberg models are used to calculate the
AF. The future work is to learn about other life-predicting models, which can be used
to evaluate the testing.

• Simulations were performed by manually designing the component structure. Instead

the design of the component should be imported to the simulation software using the
CAD design. It will help in obtaining more accurate dimensions of test component
and reduce efforts.

• ATCT’s are performed using different kinds of stresses, such as vibrations, humidity,
vibration with temperature etc. Simulation can be performed with other kinds of the
stress and the usefulness of the tests can be analyzed.

88
89
References
[1] L. A. Escobar and W. Q. Meeker, “A Review of Accelerated Test Models,” Stat. Sci.,
vol. 21, no. 4, pp. 552–577, Nov. 2006.
[2] H. Cui, “Accelerated temperature cycle test and Coffin-Manson model for electronic
packaging,” in Annual Reliability and Maintainability Symposium, 2005. Proceedings.,
2005, pp. 556–560.
[3] C. F. Dunn and J. W. McPherson, “Temperature-cycling acceleration factors for
aluminium metallization failure in VLSI applications,” in 28th Annual Proceedings on
Reliability Physics Symposium, 1990, pp. 252–258.
[4] N. Tirkey, S. Pilkhwal, A. Kuhad, and K. Chopra, “Fundamentals of Accelerated Stress
Testing.,” BMC Pharmacol., vol. 5, no. 1, p. 2, 2005.
[5] “What is a SN-Curve?,” Siemens PLM Community, 16-Jul-2016. [Online]. Available:
https://community.plm.automation.siemens.com/t5/Testing-Knowledge-Base/What-is-
a-SN-Curve/ta-p/355935.
[6] “S-N Fatigue.” [Online]. Available: https://www.nde-
ed.org/EducationResources/CommunityCollege/Materials/Mechanical/S-NFatigue.htm.
[Accessed: 20-Jul-2017].
[7] L. A. Escobar and W. Q. Meeker, “A Review of Accelerated Test Models,” Stat. Sci.,
vol. 21, no. 4, pp. 552–577, Nov. 2006.
[8] R. Huber and O. O. Semiconductors, “Temperature measurement with thermocouples,”
EE Publishers, 15-Aug-2016. .
[9] “Pentronic - Thermocouples.” [Online]. Available:
http://www.pentronic.se/home/temperature-sensors/thermocouples.aspx. [Accessed: 06-
Feb-2018].
[10] “EN43ME Lecture Notes.” [Online]. Available:
http://emerald.tufts.edu/as/tampl/en43/lecture_notes/ch1.html. [Accessed: 09-Jan-
2018].
[11] “Modes of Heat transfer,” Polytechnic Hub, 03-Mar-2015. [Online]. Available:
http://www.polytechnichub.com/modes-heat-transfer/. [Accessed: 24-Mar-2018].
[12] D. S. Steniberg, Preventing Thermal Cyclying and Vibration Failures in electronics
Equipment. John Wiley & Sons, Inc., 2001.
[13] H. Jussi, “Effects of accelerated lifetime test parameters and failure mechanisms on
the reliability of electronic assemblies.,” Aalto University, School of Electrical
Engineering., Helsinki, 2014.
[14] Larry Edson, The GMW3172 Users Guide - The Electrical Validation Engineers
Handbook Series - Electrical Component Testing., Revision-19. 2008.
[15] “Thermal expansion,” Wikipedia. 03-Feb-2018.
[16] “IEC 60068-2-14:2009 | IEC Webstore.” [Online]. Available:
https://webstore.iec.ch/publication/507. [Accessed: 05-Sep-2017].
[17] “ISO 16750-4:2010 - Road vehicles -- Environmental conditions and testing for
electrical and electronic equipment -- Part 4: Climatic loads.” [Online]. Available:
https://www.iso.org/standard/46043.html. [Accessed: 05-Sep-2017].
[18] C. J. Zhai, Sidharth, and R. Blish, “Board level solder reliability vs. ramp rate dwell
time during temperature cycling,” in 2003 IEEE International Reliability Physics
Symposium Proceedings, 2003. 41st Annual., 2003, pp. 447–451.

90
[19] Ron Weglinski, “Highly Accelerated Stress Screening For Air-Cooled Switching
Power Supplies. Part1: Understand Stress Test Methodology.,” TW0058, Feb. 2007.
[20] Y. S. Chen, Y. H. Hsu, and C. H. Ho, “Accelerateci life assessment methods for
thermoelectric modules with mechanical stress tests,” in 2014 9th International
Microsystems, Packaging, Assembly and Circuits Technology Conference (IMPACT),
2014, pp. 213–216.
[21] G. Yang, “Accelerated life tests at higher usage rates,” IEEE Trans. Reliab., vol. 54,
no. 1, pp. 53–57, Mar. 2005.
[22] Y. Wang, C. Zhang, X. Chen, and Y. Mo, “Simulation-based optimal design for
accelerated degradation tests,” in 2009 8th International Conference on Reliability,
Maintainability and Safety, 2009, pp. 1302–1306.
[23] E. O. McSorley, J.-C. Lu, and C.-S. Li, “Performance of parameter-estimates in
step-stress accelerated life-tests with various sample-sizes,” IEEE Trans. Reliab., vol.
51, no. 3, pp. 271–277, Sep. 2002.
[24] V. Loll, “Optimizing the number of failure modes for design analysis based on
physics of failure,” in 2008 Annual Reliability and Maintainability Symposium, 2008,
pp. 166–170.
[25] C. R. Yang and J. T. Kim, “Temperature accelerated life test (ALT) at the circuit
board level,” in Electronics Manufacturing Technology Symposium, 1995.
“Manufacturing Technologies - Present and Future”, Seventeenth IEEE/CPMT
International, 1995, pp. 158–165.
[26] C. Li and N. Fard, “Optimum Bivariate Step-Stress Accelerated Life Test for
Censored Data,” IEEE Trans. Reliab., vol. 56, no. 1, pp. 77–84, Mar. 2007.
[27] M.-H. Hsu, C.-C. Lee, and K.-N. Chiang, “A Modified Acceleration Factor
Empirical Equation for BGA Type Package,” 2017, pp. 1020–1026.
[28] K. C. Wu, C. H. Lee, and K.-N. Chiang, “Characterization of Thermal Cycling
Ramp Rate and Dwell Time Effects on AF (Acceleration Factor) Estimation,” 2016, pp.
251–256.
[29] “Weiss Technik: Environmental Simulation.” [Online]. Available:
https://www.weiss-technik.com/en/products/environmental-simulation/. [Accessed: 10-
Feb-2018].
[30] P. Rome, “DEWE-43: Award winning USB DAQ system. Small, powerful and
smart!” [Online]. Available: https://www.dewesoft.com/products/dewe-43. [Accessed:
11-Feb-2018].
[31] “Second Order Step Response.”
https://www.egr.msu.edu/classes/me451/jchoi/2007/handouts/ME451_S07_lecture17.p
df
[32] “Second Order Response.”
http://faculty.mercer.edu/jenkins_he/documents/2ndorderresponseMSD.pdf

91
Appendix A: Illustration of Fixing of Thermocouples on
the component surface
Table A.1: Insulated thermocouple on the component surface.

Step Step wise arrangement

1 The location on the

component has to be
selected and take a
thermocouple.

2 As we will be placing
the component in high
and low temperatures,
aluminum tape has to
be use to fix them at a
location

3 Use an insulating
sponge with at least 0.5
cm thickness.

4 Place it on the
aluminum tape, this
helps in reading the
exact temperature in
that location.

5 Place a bigger piece of

tape cover the complete
sponge.

92
Appendix B: Validation of Thermocouple Set-up
Test Set-up.

A component with inbuilt temperature sensor was used. Test was conducted in power-on
mode to obtain the temperature recording from the inbuilt sensor. Data from the inbuilt sensor
and data from the thermocouples were taken simultaneously.

Table B.1: CPS for testing of the plastic box.

Test parameters Values

Dwell time (DT) 30 minutes

Temperature range (∆𝑻) +80 ℃ to -40 ℃

Number of cycles in test (𝑵𝒕) 3

Thermocouple Settings.

Figure B.1: Plastic box had an inbuilt sensor on the PCB.

Figure B.2: Thermocouple type-K fixation.

Figure B.3: Thermocouples were taken inside through a hole.

93
 Figure B.4: Sealed hole (Insulation) as shown in Chapter 5.

Results and Validation.

Both the graphs shown in Figure B.5 and Figure B.6, have almost identical readings. Which
indicates that the temperature recording from the inbuilt sensor and temperature reading from
the externally fixed TCK are equal. This explains that the test set-up can be used for obtaining
reliable temperature readings from the components in the ATCT.

Figure B.5: Temperature recording from the inbuilt sensor on the PBC of the plastic box.

Figure B.6: Temperature recording from the thermocouples on the PCB of the plastic box.

94
Appendix C: Calculation of Ramp Rate (RR) from the
ATCT Results
The graph from the results resembles the output from a general second order system. The
input for the system is considered to be a step input, as the temperature is raised from room
temperature or 𝑇𝑚𝑖𝑛 to 𝑇𝑚𝑎𝑥 in less than 10 seconds and vice versa. For this input, the output
from the temperature recording on the component resembles the output from an overdamped
second order system [31].

For a step input, the transient response is characterized by rise time.

Rise time 𝑡𝑟 : time required for the response to rise from 10% to 90% for overdamped
systems, and from 0% to 100% for underdamped systems [32].

Therefore,

𝑇𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒 𝑟𝑎𝑛𝑔𝑒 𝑓𝑜𝑟 𝑟𝑖𝑠𝑒 𝑡𝑖𝑚𝑒 (C.1)

𝑅𝑎𝑚𝑝 𝑟𝑎𝑡𝑒 =
𝑅𝑖𝑠𝑒 𝑡𝑖𝑚𝑒 (𝑡𝑟 )

Represents the time constant calculations for overdamped second order system.

Example of Calculations.

Step 1: Finding the maximum and minimum peak values.

Figure C.1: Maximum and Minimum peaks values are considered.

Step 2: Crop the part from 2nd minimum peak to 3rd maximum peak.

95
 90% of the
maximum value

10% of the
minimum value

Figure C.2: Finding the rise time and ∆𝑇.

Step 3: Calculate from obtained values using (C.1).

81.34 − (−5.208)
𝑅𝑎𝑚𝑝 𝑟𝑎𝑡𝑒 (𝑅𝑅) =
17.98 − 2.4
86.548
𝑅𝑎𝑚𝑝 𝑟𝑎𝑡𝑒 (𝑅𝑅) =
15.58
𝑹𝒂𝒎𝒑 𝒓𝒂𝒕𝒆 (𝑹𝑹) = 𝟓. 𝟓𝟓 ℃/𝒎𝒊𝒏𝒖𝒕𝒆.

96
Appendix D: Comparison of Temperature Response
from the PCB and the Air Inside the Plastic Box.
Temperature in the basket (empty – without any components): input for the test as shown in
Figure D.1.

Figure D.1: Input temperature readings for an ATCT.

Temperature on PCB (on the surface and below the surface): during the test as shown in
Figure D.2.

Figure D.2: Output temperature readings from the PCB.

The PCB has almost the same temperature readings as the basket temperature readings.
Therefore, this PCB temperature profile can be used for estimating parameter values such as
∆𝑇, DT, RR and 𝑇𝑚𝑎𝑥 . These values can be used to calculate AFs with different life predicting
models.

97