a study of 5dx inspection effectiveness of bga opens ... · pdf filea study of 5dx inspection...
TRANSCRIPT
Page 1
A Study of 5DX Inspection Effectiveness of BGA opens –
Circular vs Oval Pads
Page 2
Introduction
•Selcom Elettronica has been using Agilent 5DX since 1996 for process control and PCBA defect detection
•In particular, BGA inspection has initially been a driver for the adoption of an AXI technology
•Although the 5DX proved to be an excellent tool for control and repair of most of the SMT components, the results on BGA detection of marginal open joints have been considered un-satisfactory until the adoption of Rev 8
Page 3
The “BGA Inspection Problem”Prior to Rev 8 5DX could easily detect macroscopic BGA defects
Page 4
The “BGA Inspection Problem”Prior to Rev 8 5DX opens were detected by changes in shape or heavily insufficient joints
Page 5
The “BGA Inspection Problem”BGA Ball diameter is the main indicator for Opens Detection, however, if we compare each ball’s diameter to all other pins diameters we might miss the real open
Page 6
How can we improve our BGA Opens detection?
•We tried to solve the problem in parallel using two different approaches:
•Agilent R&D has enhanced the BGA2 Algorithm with the “Opens Outlier” indicator (available from Rev 8.06)
•Selcom has studied the adoption of elongated pads using the Flattening value as open indicator
Page 7
BGA Open Outlier
Good Joint
Open Joint
Page 8
BGA Open Outlier
Notice the differences in height and diameter of these good joints.
Page 9
Examples of “Near Neighbors” for 3 types of joints (corner, edge, and center.)
EVERY joint on the BGA is compared to its Near Neighbors.(Just like the 3 examples shown)
BGA Open Outlier
Page 10
BGA Open Outlier
Page 11
Oval Pads – Flattening
Pin 97
Pad Major Axis: M=40mils
Pad Minor Axis: m=24 mils
Page 12
Oval Pads - Flattening•The differentiator between good and bad joints is the Pad Slice Flattening Open Signal (F)
•F = 1- m/M (where m=minor axis M=major axis)
•In the example of the previous slide, m=24; M=40
•Theoretical F value for good joints•F = 1- 24/40 = 0.4
•Theoretical F value for bad joints (which remain circular)
•F = 1- d/d = 0 (where d is the diameter of the ball at pad slice level)
Page 13
Selcom Study on BGA Opens Defects
•The Set-up included 3 slices:
•Chip Slice – Open Outlier Active (Threshold 3.0)
•Ball Slice – Open Outlier Active (Threshold 3.0)
•Pad Slice – Open Oulier Disabled - Minimum Pad Slice Flattening value ~ 35% of the Flattening Theoretical value (Review Measurement)
•Flattening Direction forced at 45 degrees
Chip Slice (Open Outlier)
Ball Slice (Open Outlier)
Pad Slice (Flattening)
Page 14
Oval Pads - Flattening
Pin 97
Threshold: 0.15
Average ~ 0.3
Page 15
Selcom Study on BGA Opens Defects•Selcom has created Open defects on a number of pins by adding a thin layer of solder resist on selected pads
•The design of the board was such that the dummy BGAs had a test point for each pad and were electrically tested to verify the presence of each defect
•The PCB was a single sided 12 layers board with gold pad plating
•Selcom build 3 boards – 1 with no defects and 2 with a total of 128 defects
•The BGA types were:
•Bga 169 pin, pitch 1.5mm (59 mils), ball 0.75 mm (29,5 mils)
•Bga 225 pin, pitch 1.5mm (59 mils), ball 0.75 mm (29.5 mils)
•Bga 484 pin, pitch 1.0mm (39 mils), ball 0.60 mm (23.6 mils)
•Bga 208 pin, pitch 0.8mm (31 mils), ball 0.45 mm (17.7mils)
Page 16
Selcom Study on BGA Opens Defects•The Pads on each BGA type were:
•For the 1.5mm pitch (59 mils) components with 0.75mm (29.5mils) ball:
•Circular Pad: 24 mils
•Oval Pad1: 40x24 mils
•Oval Pad2: 47x24 mils
•For the 1.0mm pitch (39 mils) components with 0.6 mm (23.6 mils) ball:
•Circular Pad: 16 mils
•Oval Pad1: 23x20 mils
•Oval Pad2: 45x22 mils (not implemented due to via-in-pad problem)
•For the 0.8mm pitch (31 mils) components with 0.45mm (17.7 mils)ball:
•Circular Pad1: 14 mils
•Circular Pad2: 16 mils
•Oval Pad: 33x18 mils (not implemented due to via-in-pad problem)
Page 17
Structural and Stress considerationsThe results on this slide are extracted from the paper “A Study of High Density and Reliability BGA Package with Solder Ball Lands of Oval Type” by S. J. Kim, C. H. Lee, and S. G. Lee
The table shows the results of the Shear Strength over hundreds of thermo cycles for a BGA with balls of 0.76 mm mounted on circular pads of 0.635 mm of diameter and elongated pads (0.764 x 0.47 mm)
The measures of Shear Strength were obtained using Keller Technology’s Bond Test-30 and the thermo cycles were executed according to the level C of the JEDEC A104-A Standard (-65°C/150°C). The X direction is the minor axis and the Y the major axis
Results show an increase of 7% in Shear Strength along the minor axis and a decrease of 3% along the major. In general the strength of this oval pad is at least equal if not greater than the circular pad
Page 18
Selcom Study on BGA Opens Defects: ResultsComponent
Type Pad Type #
Defects Created
# Open Outlier
Detected Defects
# Flattening Detected Defects
Total # of Detected Defects (open
outlier OR flattening)
Total Effectiveness
%
Circular 24 mils 10 9 N.A 9 90
Oval 40x24 mils 10 10 10 10 100
BGA 169 pin, pitch 1.5mm, ball 29,5 mils
Oval 47x24 mils 10 10 10 10 100
Circular 24 mils 10 8 N.A 8 80
Oval 40x24 mils 10 9 10 10 100
BGA 225 pin, pitch 1.5mm, ball 29,5 mils
Oval 47x24 mils 10 9 10 10 100
Circular 16mils 15 12 N.A 12 80 BGA 484 pin, pitch 1.0mm ball 23,6 mils Oval
23x20 mils 15 14 N.A 14 93
Circular 14 mils 19 15 N.A 15 79 BGA 208 pin, pitch 0,8 mm ball 17,7 mils Circular 16 mils 19 16 N.A 16 84
No False Calls occurred during the test !No False Calls occurred during the test !
Page 19
Conclusions•Open Outlier on circular pads has an efficiency greater than 80%
•The 20x23 mils elongation has not proved to be sufficient for the use of Flattening
•A ratio of 1.6 between major and minor axis is giving 100% effectiveness results, therefore it is not necessary to use higher elongations
•A Flattening call is very easy to confirm visually at the repair client
•False Calls were not found – The test board did not have neither voiding nor shading, which are the two main causes for FC on BGAs
•The open outlier efficiency is greater on oval pads than circular (this is probably due to the greater area of the pad with oval shape, the ball collapses on a greater area surface thus reducing its mid-ball diameter. In this way an open ball will have a larger difference in diameter, also on circular pads the 16 mils diameter has had a better result than the 14 mils)
Page 20
Conclusions•We must remember However that:
•The study has been carried out on prototype and not production boards
•The boards were single sided with no shading or voiding
•The way defects were created is different from real production defects
•The defects were in known positions (this makes it easier to program)
•Selcom has defined a library of oval pads to use as a standard for the design of its own future products.
• At the moment oval pads have been implemented only on few products.
• Real open defects have not been found jet.