Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for...

36
Test Strategies x6000 Version 1.1 Page 1 Thoughts on Test Strategies for Medium to High Complexity boards April 2007

Transcript of Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for...

Page 1: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 1

Thoughts on

Test Strategies

for Medium to High

Complexity boards

April 2007

Page 2: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 2

Do you select the same test strategy?

11 components26 joints

1,689 components10,239 joints

Board A

Board B

Board B magnified

Page 3: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 3

Complexity Index

Complexity Index = [( #C + #J ) / 100] * D * M * S

#C = Number components

#J = Number jointsS = Double sided = 1, Single sided = 1/2M = High Mix = 1, Low mix = 1/2

D = Joint density = Joints per square inch / 100or Joints per square cm / 15.5

Low complexity Medium complexity High complexity

< 50 >= 50 and < 125 >= 125

6 inch

2 inchSide A, 800 joints

Side B, 600jointsD = [(800+600)/(2*6)] / 100 = 1.17

Example

Page 4: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 4

Many test strategy options....

AOI solder paste

AXI pre-wave

inspection

Manual Visual

Inspection

AOI post-reflow

AOI pre-reflow

Functional test

In-circuit test

AXI post-wave

inspection

.. which combination is

optimal?

Page 5: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 5

Other considerations

bDouble sided boards

bHigh reliability concerns – few warranty defects

bHigh quality requirements – few field failures

bWants high yield at ICT and Functional Test

Page 6: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 6

• Many test / inspection solutions available

• Defect prevention – defect containment

• Medium to high complexity boards –

many defect opportunities

• This presentation shares result from latest research and studies

Key points....

Page 7: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 7

Topics

• Introduction

• Defect levels, defects found by 3D AXI

• Defect prevention and random defects

• Test effectiveness

• Summary

Page 8: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 8

3D AXI RepairUser A

3D AXIUser B

User N 726,091 Boards Inspected

3,675,298,930 Joints Inspected

14 users

2 in Americas

Repair operator

3.7 Billion Solder Joints

3D AXI

Repair

Repair

::

4 in Europe8 in Asia

Most of the data is from early 2007

1,138 Board Types

Page 9: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 9

Data “clean up”

• Removed debug boards

• Serial Number = “Family”, “Component”

• Boards inspected > 1 time

• Serial Number obvious debug

• Data over around 500 million joints inspected

Page 10: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 10

Solder Joints inspected by company

• 14 companies/sites, total = 3.7 billion

0

100,000,000

200,000,000

300,000,000

400,000,000

500,000,000

600,000,000

A B C D E F G H I J K L M N

100M

200M

300M

400M

500M

600M

A B C D E F G H I J K L M N

Page 11: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 11

DPMO Repaired and process indicators

327 DPMO repaired

245 DPMO process indicator

572 DPMO combined

Average numbers

0

500

1000

1500

2000

2500

3000

A B C D E F G H I J K L M NA B C D E F G H I J K L M N

500

1000

1500

2000

2500

3000

PI

Rep.

Page 12: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 12

Calls per 10 topped called algorithms

This is showing calls after ART, only for the top 10 algorithms based on # of calls.

(Number of opportunities per algorithm is not recorded, therefore DPMO values can not be calculated)

0

50,000

100,000

150,000

200,000

250,000

FPGullwing BGA2 Connector Gullwing Chip Res Paste PCAP PTH Socket

PI

Rep.

Page 13: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 13

Estimated yields

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

2500 5000 7500 10000 12500 15000 17500 20000 22500 25000

50

100

200

327

500

1,000

Solder joints

DPMO

Yields

Yield = (1 – DPMO/1,000,000)^Defect opportunities

Page 14: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 14

Key points....

• 2007 data – 14 companies

• 3.7 billion solder joints in study

• Defect levels are still high

• 327 DPMO Defects• 245 DPMO Process Indicators

• 572 DPMO Combined

• A board with 3,500 solder joints has an average of > 1 defects per board

• Yields after reflow for medium and high complexity boards typically very low

• Defects on all type of solder joints

Page 15: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 15

Topics

• Introduction

• Defect levels, defects found by 3D AXI

• Defect prevention and random defects

• Test effectiveness

• Summary

Page 16: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 16

Question

Are defects systematic?

And if so what percent are systematic?

1 billion solder joint study used to analyze this.

Page 17: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 17

Boards available for analysis

566 board types in data base – 1999 DPMO study

Selection criteria:>= 100 boards manufactured

>= 1000 joints inspected>= 20 components inspected

(# components inspected not in data, therefore >=20 referencedesignator in calls)

237 board types meet criteriaProduction volume from 103 to 188,360

10 boards were selected randomly

Page 18: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 18

What is a systematic defect?

Same reference designator called* on >= 3% of the boards

*Called with repair, repair later, or process indicator actions.

Factor that could increase # of systematic defectsSeveral RefDes same basic problem (DFM issue)

Factor that could decrease # of systematic defectsAll defect types for one RefDes considered systematic

In this analysis the following was used:

Objective with study is only to indicate ball park numbers.For precise analysis availability to boards with defects are needed.3% was selected on engineering judgment basis.

Page 19: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 19

Result

For 10 boards types randomly selected. Number of random defects 74%

Systematic

Random

(“Same” defect on less than 3% of the boards)

Page 20: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 20

Result

For 10 board types randomly selected. Number of systematic defects

>=3%

0%

10%

20%

30%

40%

50%

60%

70%

80%

1 2 3 4 5 6 7 8 9 10 11AverageBoard types

%Systematic

defects

Page 21: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 21

• In this study around 75% of all defects are random.

• Should defect prevention strategies be implemented? ABSOLUTELY, efforts should definitely be done to minimize systematic defects.

• However an effective test strategy for random defects is needed, especially for medium and high complexity boards.

• What is an effective test strategy?

Key points....

Page 22: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 22

Topics

• Introduction

• Defect levels, defects found by 3D AXI

• Defect prevention and random defects

• Test effectiveness

• Summary

Page 23: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 23

Defect coverage

• Poor wetting

• Marginal Joints

• Voids

• Excess

• Shorts

• Opens

• Hidden parts

• Missing

• Gross Shorts

• Lifted Leads

• Bent Leads• Insufficient

• Bridging

• Tombstone

• Misalignment

• Bypass Caps, L’s

• Extra Parts

•Orientation

• Missing non-electrical parts

•Mark Inspection

•Paste Deposition Defects

• Polarity

• Inverted

• Missing Socketed Parts

• Dead part

• Wrong part

• Bad part

• PCB short/open

• Functionally bad

AOI

ICT AXI

ICT: In-circuit Test

AXI: Automated X-ray Inspection

AOI: Automated Optical Inspection –

including solder paste inspection

(SPI), pre-reflow and post-reflow.

This is a conceptual

diagram!

How do you get

accurate data? ….

Page 24: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 24

What is Test Effectiveness?

Test Effectiveness is a measure of the “goodness” of your test and your test /inspection equipment and programs.

TESTER(S)

10 DEFECTS 3 DEFECTS

Test Effectiveness = (10-3)/(10) = 70%

Page 25: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 25

Test or Inspection

Manually Categorizethe test inspection results

into defects, process indicators, and false calls

NO REPAIRS

2.

3.

Customer has final say on what is a defect, process indicator, or false call.

Building of boards,including wave.

1.

Test Effectiveness Evaluation Flowchart

No inspection/test or repairs for the boards in the

Study.

AllTest or Inspections

Steps done?4.

No

Only defects are included in Test Effectiveness data.

Yes

Page 26: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 26

Case Study

• Compilation of nine similar studies performed in nine separate

manufacturing facilities

• 421 assemblies tested

• No defects repaired between test/inspection steps (except

shorts at ICT)

• Only one defect per component

• 3D AXI and ICT always present in study

• One of AOI, Functional Test, and Manual Visual also present

Most of these studies have been done on medium to high complexity

boards with medium to low manufacturing volumes.

Page 27: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 27

Summary TE Studies – 421 boards

3D AXI

AOI, MVI, or FT

ICT67

536

73

30

3

53103

779

209

226

Total Defects: 865

Page 28: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 28

Summary TE Studies – Defect Coverage

3D AXI

AOI, MVI, or FT

ICT

90.1%

24.2%

26.1%

Total Defects: 865

Page 29: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 29

Test Effectiveness studies - Summary

Key learnings:

- 3D AXI Test Effectiveness 85% - 95%- AOI “ “ 25% - 75%- ICT “ “ 20% - 60%- FT “ “ 5% - 20%

These numbers includes both Agilent and non-Agilent systems

(3D AXI is Agilent 5DX only)

All systems are not equal!!

Page 30: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 30

A visual presentation

Of

Test/Inspection

Coverage

Page 31: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 31

Summary TE Studies – Defect Coverage

Defect universe – All defects – No coverage

Page 32: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 32

Defect universe – ICT and AOI, MVI, or FT coverage

Defects not covered

Red =

ICT

AOI, MVI or FT

Summary TE Studies – Defect Coverage

Page 33: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 33

Defect universe – AXI 3D, ICT and AOI, MVI, or FT coverage

3D AXI coverage

90.1%

Defects not covered

Red =

ICT

AOI, MVI or FT

Summary TE Studies – Defect Coverage

Page 34: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 34

• 3D AXI has by far the highest defect coverage, around 90%

• Without 3D AXI around 50% of defects are shipped to the end customer. Some of those defects will create warranty and field failures.

• For medium and high complexity boards an effective implementation of 3D AXI in the test strategy is strongly recommended.

Key points....

Page 35: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 35

Topics

• Introduction

• Defect levels, defects found by 3D AXI

• Defect prevention and random defects

• Test effectiveness

• Summary

Page 36: Thoughts on Test Strategies for Medium to High Complexity ... · Thoughts on Test Strategies for Medium to High Complexity boards April 2007. Test Strategies x6000 Page 2 Version

Test Strategies x6000

Version 1.1Page 36

Summary – Medium & High Complexity boards

• Defect levels are still high > 300 DPMO

• High defect coverage is important and 3D AXI is by far the most effective test / inspection method.

• An effective defect prevention strategy is recommended

• A majority of defects are random

• To achieve high yields at ICT and FT and to minimize warranty and field failures,

3D AXI is strongly recommended