5DX ALGORITHM SETTINGS FOR BGA VOIDING - Keysight · 2005. 4. 6. · IPC-7095 Design and Assembly Process Implementation for BGAs-August 2000 BGA ball size: 24 mils Current Agilent

5DX ALGORITHM SETTINGS FOR BGA VOIDING

March 1, 2005

4/6/2005

Slide 2

Objective

• Assess the current setting for the Agilent 5DX BGA2 Void Algorithm to detect voids.

• Compare and correlate the voiding measurement data between Laminography 5DX and Transmission 2D X-ray systems.

• Discuss how 5DX technology can be used for SMT process improvement, specifically to reduce BGA defects.

4/6/2005

Slide 3

Process Floor Flow

Start

Place ReflowPaste Insert WavePaste Place Reflow

FT5DX Test ICT5DX PLR FIFinish

All boards and all components are tested for the customer at 5DX system.

4/6/2005

Slide 4

IPC-7095

Accept/Reject criteria for BGA voids in Ball at interface

Need rework for class IIat ball level: > 10.8 mils (void diameter) or > 20.25% (void area)

at board level: > 8.4 mils (void diameter) or > 12.25% (void area)

IPC-7095 Design and Assembly Process Implementation for BGAs - August 2000

BGA ball size: 24 mils

Current Agilent software 8.2.3 doesn’t have capabilities for setting different BGA void Algorithm for each of the three slices. However, new version 8.3 will have the capability to set different thresholds for each slice.

Question from our customer:

• How do you know that you are catching all BGAs voiding above IPC-7095 class II level?

4/6/2005

Slide 5

Measurement for BGA voids

Result:

BGA voiding measurement data strongly depends on the Algorithm Sensitivity Setting.

Questions:Per Agilent default setting: 0.15 Per Customer: 0.45

• Package with Plastic BGA with 289pins

• Package size: 19.2mm x 19.2mm

• Pitch size: 1.0mm (40 mils)

• Ball size: 0.60mm (24 mils)

- Texas instruments

Example:

4/6/2005

Slide 6

Contrast Images - camera frame grabber number

• Setting the correct camera frame grabber number will give us images with good contrast between ball and background, which will help preventing escapes of void diameters of 10.8 mils or greater.

• Having images with good contrast will help operators visually confirm defect calls at the Repair Station.

Image of BGA voiding with the camera frame grabber setting of 3 (S/W 7.3)

9.37 mils 10.15 mils


4/6/2005

Slide 7

Agilent AXI 5DX BGA2 Void Algorithm


10.54 mils

Algorithm Settings:• Sensitivity of 0.25• MIN_VOID_DIA of 9 mils• PERCENT_VOID of 20• Camera frame grabber setting of 5 • S/W 8.1

This plastic BGA has 289 pins with pitch size of 40 mils on a PCB fab with 28 layers .

Action:

All 5DX program are converted to new S/W 8.2.3.

4/6/2005

Slide 8

Agilent 5DX limitations

Experiment:•Test a BGA nine times for each of the different sensitivity settings:The data from same 175 pins were collected for void “diameter” study, and data from same 164 pins for void “area percentage” study.

Issue:Repeatability of BGA voiding diameter and area percentage are not very good.

Questions:-What is the exact diameter of a void for a given void within a BGA? -What is the mean and standard deviation of the BGA’s voiding measurement?

4/6/2005

Slide 9

BGA voiding diameter data

BGA voiding diameter data of 175 pins with Sensitivity Settings of 0.25, 0.35, and 0.45

D25-1 represents diameter measurement 1, with sensitivity setting of 0.25.

Measure the BGA with 289 pins. Any pin with “0” reading is not including for analysis.

4/6/2005

Slide 10

Data for 9 measurements (Sensitivity = 0.25)

Nine retests on the same board (FCN600-00238-03, S/N 0018345) location U42, same pins. S/W 8.1

Building 3, line 3 with S/W 8.1

Void Diameter (mil)D1 D2 D3 D4 D5 D6 D7 D8 D9 Mean STDEV

9.76 10.15 8.98 9.76 10.93 10.15 10.15 9.37 10.93 10.02 0.659.76 10.15 10.93 10.15 9.76 10.15 9.76 9.76 10.15 10.06 0.389.76 9.37 10.15 9.76 9.76 9.37 9.76 9.76 8.98 9.63 0.3410.93 10.93 9.76 10.15 10.15 10.54 10.54 10.15 10.54 10.41 0.3910.54 11.71 10.15 10.93 11.71 10.54 10.93 10.15 11.32 10.89 0.609.76 9.37 9.37 9.76 10.15 8.98 9.37 9.76 10.54 9.67 0.4710.54 10.54 10.54 10.54 11.32 11.71 10.93 10.15 10.93 10.80 0.488.59 8.98 7.03 8.59 7.42 6.64 7.81 5.85 7.03 7.55 1.039.76 8.59 8.59 7.42 8.2 5.85 7.42 7.81 7.81 7.94 1.079.76 8.98 9.76 8.98 8.98 8.2 8.2 10.15 10.15 9.24 0.769.37 9.76 8.98 9.37 9.37 9.37 9.76 8.59 9.76 9.37 0.397.03 6.64 5.07 7.42 8.59 8.2 7.81 7.03 7.42 7.25 1.027.81 8.98 7.03 7.03 7.03 8.2 7.42 7.03 7.42 7.55 0.687.03 7.42 7.81 7.42 7.42 7.03 7.03 8.2 7.42 7.42 0.3910.54 10.54 10.54 10.15 11.32 10.54 10.15 10.15 10.93 10.54 0.3910.15 10.54 10.54 9.76 10.54 9.76 10.93 10.54 10.15 10.32 0.407.03 7.42 6.25 7.42 5.85 6.64 7.03 6.25 7.81 6.86 0.658.2 8.98 8.59 7.81 8.59 8.2 8.2 7.81 8.2 8.29 0.389.76 8.98 7.81 10.15 8.98 7.42 8.98 8.2 9.37 8.85 0.899.37 9.76 8.98 8.98 9.37 8.98 9.37 9.37 9.76 9.33 0.309.76 10.15 10.15 9.76 9.37 9.76 9.76 9.76 9.37 9.76 0.288.98 8.98 8.98 8.98 8.98 9.76 8.59 8.59 8.59 8.94 0.36

4/6/2005

Slide 11

One-way ANOVA: D25-1, D25-2, D25-3, D25-4, D25-5, D25-6, D25-7, D25-8, D25-9

Analysis of Variance

Source DF SS MS F P

Factor 8 3.79 0.47 0.20 0.991

Error 1566 3696.17 2.36

Total 1574 3699.96

Individual 95% CIs For Mean

Based on Pooled StDev

Level N Mean StDev ---------+---------+---------+-------

D25-1 175 8.380 1.511 (----------*----------)

D25-2 175 8.404 1.560 (----------*-----------)

D25-3 175 8.277 1.509 (-----------*----------)

D25-4 175 8.315 1.558 (-----------*----------)

D25-5 175 8.371 1.578 (-----------*----------)

D25-6 175 8.340 1.506 (----------*----------)

D25-7 175 8.308 1.592 (----------*-----------)

D25-8 175 8.257 1.511 (-----------*----------)

D25-9 175 8.391 1.499 (-----------*----------)

---------+---------+---------+-------

Pooled StDev = 1.536 8.20 8.40 8.60

ANOVA _ Shows No DIFF between measurement with a board

for diameter U42 (Sensitivity = 0.25) • Ho:Ho: Are the average voiding diameters the same for all 9 BGA voiding measurements.

• Ha: Ha: Are the average voiding diameters different for all 9 BGA voiding measurements.

The Null Hypothesis is represented by "Ho" and the Alternative Hypothesis is by "Ha".

The P value is probability of the means that are equal.

Software: MINITAB

ANOVA: Analysis of variance

4/6/2005

Slide 12


Nine retests on 1 same board (FCN600-00238-03, S/N 0018345) location U42, same pins.



10.93 12.1 11.32 12.1 11.71 11.71 11.71 12.1 11.71 11.71 0.3910.93 11.32 10.93 10.93 10.15 10.15 10.93 10.93 10.54 10.76 0.4010.54 10.15 10.54 10.15 10.93 9.76 10.15 10.93 10.93 10.45 0.4312.5 10.93 10.93 10.54 11.32 10.93 10.54 12.5 10.54 11.19 0.7812.1 11.71 12.1 12.1 12.5 11.32 11.71 12.5 11.32 11.93 0.4411.32 11.71 11.71 11.71 11.32 11.71 11.32 11.71 11.71 11.58 0.199.76 8.2 4.68 10.93 10.93 11.32 11.32 10.15 7.42 9.41 2.2510.54 7.81 10.15 9.76 10.93 8.98 7.42 11.32 10.15 9.67 1.3511.71 11.71 11.32 11.71 12.1 11.32 12.5 11.32 12.1 11.75 0.419.76 10.93 10.54 10.93 9.37 10.54 10.15 9.76 9.37 10.15 0.628.98 9.37 10.15 9.76 8.98 9.37 9.37 8.98 10.15 9.46 0.478.59 7.03 7.03 8.98 9.76 8.59 9.76 9.76 9.76 8.81 1.1211.32 11.32 10.54 11.71 11.71 11.32 10.93 11.32 11.71 11.32 0.3911.71 10.15 10.54 11.71 11.32 10.93 10.93 10.93 10.54 10.97 0.538.59 10.15 8.98 9.76 8.2 9.76 11.32 8.98 8.98 9.41 0.943.12 5.85 6.64 10.15 10.93 6.64 6.64 10.15 5.46 7.29 2.5910.54 6.25 10.15 8.59 11.32 10.54 10.15 10.54 10.93 9.89 1.568.98 8.98 8.59 9.37 9.76 8.2 9.37 9.37 10.15 9.20 0.5910.54 9.37 8.98 9.76 9.76 9.76 10.54 8.98 9.76 9.72 0.5711.32 7.42 6.25 7.03 7.03 7.42 6.25 3.9 7.81 7.16 1.9410.54 10.54 10.54 9.76 10.54 9.76 10.15 9.76 8.2 9.98 0.768.59 8.2 10.15 8.98 7.81 9.76 8.98 9.76 10.93 9.24 0.99

4/6/2005

Slide 13

ANOVA _ Shows No DIFF between measurement with a board



Source DF SS MS F P

Factor 8 3.91 0.49 0.25 0.981

Error 1566 3057.15 1.95

Total 1574 3061.07



Level N Mean StDev --------+---------+---------+--------

D35-1 175 10.070 1.215 (------------*------------)

D35-2 175 10.117 1.379 (------------*------------)

D35-3 175 10.130 1.315 (------------*------------)

D35-4 175 10.052 1.447 (------------*------------)

D35-5 175 10.043 1.571 (------------*------------)

D35-6 175 10.034 1.304 (------------*------------)

D35-7 175 10.010 1.424 (------------*------------)

D35-8 175 10.175 1.415 (------------*------------)

D35-9 175 10.095 1.472 (------------*------------)

--------+---------+---------+--------

Pooled StDev = 1.397 9.92 10.08 10.24

for diameter U42 (Sensitivity = 0.35)• Ho:Ho: Are the average voiding diameters the same for all 9 BGA voiding measurements.



4/6/2005

Slide 14


Nine retests on 1 same board (FCN600-00238-03, S/N 0018345) location U42, same pins.



10.93 11.32 11.71 11.32 11.32 11.71 12.89 12.89 13.28 11.93 0.8610.93 11.32 11.32 11.71 11.32 10.93 12.1 11.71 11.71 11.45 0.3911.71 11.32 12.1 11.32 11.71 10.15 12.89 12.5 12.89 11.84 0.8812.1 12.5 11.32 10.54 12.5 11.32 13.28 12.5 12.89 12.11 0.8812.1 12.1 12.1 11.71 12.1 12.1 13.67 13.28 13.28 12.49 0.7112.1 11.71 11.71 11.32 11.71 12.1 12.1 12.5 12.89 12.02 0.475.46 8.59 11.71 10.54 8.98 8.59 12.1 11.71 12.1 9.98 2.2511.71 12.1 8.98 9.76 9.37 9.76 11.71 12.1 11.71 10.80 1.2911.71 11.71 11.71 12.1 12.1 12.5 12.89 13.28 12.89 12.32 0.5910.54 9.76 11.32 11.71 12.5 10.15 10.93 10.93 10.54 10.93 0.8310.15 8.98 10.15 9.37 9.37 9.37 11.71 12.89 11.71 10.41 1.374.29 3.12 8.98 6.64 3.12 7.81 12.1 11.71 11.32 7.68 3.629.37 9.76 4.68 9.37 9.37 9.76 10.93 9.76 10.15 9.24 1.7811.32 10.93 10.93 10.93 11.71 12.1 12.1 12.1 12.1 11.58 0.5510.93 11.71 11.32 11.32 10.93 11.32 13.28 12.89 12.5 11.80 0.878.98 8.98 11.71 9.76 9.37 8.98 10.15 9.76 11.32 9.89 1.0110.15 9.37 7.42 7.81 6.25 9.76 10.93 10.93 10.54 9.24 1.6910.93 10.15 8.2 9.76 10.54 10.15 10.54 10.54 11.32 10.24 0.8910.54 10.93 8.98 11.32 8.98 7.42 10.54 10.15 10.54 9.93 1.249.37 8.59 9.76 10.15 9.37 10.15 11.71 11.71 10.54 10.15 1.056.64 3.9 3.9 8.2 8.98 8.59 10.93 10.15 10.93 8.02 2.7110.15 11.32 10.93 10.15 8.98 10.15 11.32 11.32 11.32 10.63 0.82

4/6/2005

Slide 15



Source DF SS MS F P

Factor 8 900.95 112.62 60.47 0.000

Error 1566 2916.43 1.86

Total 1574 3817.37



Level N Mean StDev ------+---------+---------+---------+

D45-1 175 10.184 1.353 (---*--)

D45-2 175 10.146 1.303 (--*--)

D45-3 175 10.293 1.318 (---*--)

D45-4 175 10.179 1.439 (---*--)

D45-5 175 10.262 1.335 (--*--)

D45-6 175 10.088 1.479 (--*---)

D45-7 175 11.856 1.243 (---*--)

D45-8 175 11.791 1.430 (---*--)

D45-9 175 11.724 1.366 (--*---)

------+---------+---------+---------+

Pooled StDev = 1.365 10.20 10.80 11.40 12.00

for diameter U42 (Sensitivity = 0.45)

ANOVA _ Shows DIFF between measurement with a board

• Ho:Ho: Are the average voiding diameters the same for all 9 BGA voiding measurements.


With sensitivity set at 0.45 the measurements are different between the first 6 and last 3. As we include the sensitivity the average increases and there’s more variation between the same samples.


4/6/2005

Slide 16

Average Void Diameter U42 with different Sensitivity setting

1. Same 175 pins measurement data are taken at U42 for 9 times with each sensitivity setting.

2. The MEAN and STDEV are from 175x9=1575 data

4/6/2005

Slide 17

D45

Mea

n

D35

Mea

n

D25

Mea

n

13.5

12.5

11.5

10.5

9.5

8.5

7.5

6.5

5.5

4.5

Sensitiv ity

Mea

n

Boxplots of Mean by Sensitiv(means are indicated by solid circles)

ANOVA for Mean of Void Diameter

One-way ANOVA: Mean versus SensitivityMean versus Sensitivity

Analysis of Variance for Mean

Source DF SS MS F P

Sensitiv 2 534.95 267.48 176.74 0.000

Error 522 790.00 1.51

Total 524 1324.96



Level N Mean StDev ---------+---------+---------+-------

D25Mean 175 8.339 1.378 (-*--)

D35Mean 175 10.093 1.269 (-*-)

D45Mean 175 10.725 1.015 (-*-)

---------+---------+---------+-------

Pooled StDev = 1.230 8.80 9.60 10.40

Ho: Are all the average voiding diameters the same Ha: Are all the average voiding diameters different

The average voiding measurements with 0.25, 0.35 and 0.45 are statistically different. Sensitivity 0.25 shows the smallest average measurement.


4/6/2005

Slide 18

ANOVA for StDev of Void Diameter

D45

ST

DE

V

D35

ST

DE

V

D25

ST

DE

V

4

3

2

1

0

Sensitiv ity _1

ST

DV

Boxplots of STDV by Sensitiv(means are indicated by solid circles)

One-way ANOVA: STDV versus Sensitivity_1STDV versus Sensitivity_1

Analysis of Variance for STDV

Source DF SS MS F P

Sensitiv 2 27.409 13.704 55.21 0.000

Error 522 129.573 0.248

Total 524 156.982



Level N Mean StDev ---+---------+---------+---------+---

D25STDEV 175 0.6345 0.3418 (---*--)

D35STDEV 175 0.6273 0.5435 (--*---)

D45STDEV 175 1.1155 0.5766 (---*--)

---+---------+---------+---------+---

Pooled StDev = 0.4982 0.60 0.80 1.00 1.20

Ho: Are all the average standard deviations the same Ha: Are all the average standard deviations different

The average standard deviations for voiding measurements with 0.25, 0.35 and 0.45 are statistically different. Sensitivity 0.25 shows the smallest variation overall.


4/6/2005

Slide 19

ANOVA for Mean of Void Area Percentage

P45

Mea

n

P35

Mea

n

P25

Mea

n

40

30

20

10

Boxplots of P25Mean - P45Mean(means are indicated by solid circles)

One-way ANOVA: P25Mean, P35Mean, P45Mean


Source DF SS MS F P

Factor 2 9025.45 4512.72 516.01 0.000

Error 489 4276.52 8.75

Total 491 13301.97



Level N Mean StDev -----+---------+---------+---------+-

P25Mean 164 16.543 2.888 (*-)

P35Mean 164 23.706 3.038 (-*)

P45Mean 164 26.763 2.944 (*-)

-----+---------+---------+---------+-

Pooled StDev = 2.957 17.5 21.0 24.5 28.0

Ho: Are all the average voiding percentage the same Ha: Are all the average voiding percentage different

The average voiding percentage with 0.25, 0.35 and 0.45 are statistically different. Sensitivity 0.25 shows the smallest average percentage.


4/6/2005

Slide 20

ANOVA for StDev of Void Area Percentage

P25

ST

DE

V

P35

ST

DE

V

P45

ST

DE

V

0

5

10

15

Boxplots of P25STDEV - P45STDEV(means are indicated by solid circles)

One-way ANOVA: P25STDEV, P35STDEV, P45STDEV


Source DF SS MS F P

Factor 2 770.78 385.39 116.10 0.000

Error 489 1623.25 3.32

Total 491 2394.03



Level N Mean StDev -+---------+---------+---------+-----

P25STDEV 164 2.200 0.953 (--*--)

P35STDEV 164 2.834 1.602 (-*--)

P45STDEV 164 5.115 2.546 (--*--)

-+---------+---------+---------+-----

Pooled StDev = 1.822 2.0 3.0 4.0 5.0

Ho: Are all the average standard deviations for percentage the sameHa: Are all the average standard deviations for percentage different

The average standard deviations for voiding percentage areas with 0.25, 0.35 and 0.45 are statistically different. Sensitivity 0.25 shows the smallest variation overall.


4/6/2005

Slide 21

5DX (auto mode) vs. 2D X-ray (manual mode)

Pin # MeanS25_SW7.3 MeanS3_SW7.3 MeanS25_SW8.1 MeanS35_SW8.1 MeanS45_SW8.1 2D X-ray

3 8.33 9.76 9.63 10.67 11.84 11.79 9.98 10.45 10.80 11.97 12.32 13.815 8.07 8.94 9.37 10.80 11.80 12.131 8.07 8.59 8.29 10.93 10.93 10.371 9.63 10.97 10.11 12.36 12.84 13.085 9.02 10.24 10.06 11.45 11.76 12.592 8.98 9.76 9.54 11.32 11.23 12.1106 8.72 9.54 9.24 11.32 11.36 11.0181 7.64 7.98 9.20 10.93 10.67 9.8197 8.89 9.28 9.89 11.19 11.36 10.8243 8.55 9.63 9.76 10.93 12.28 11.4247 8.72 9.80 9.46 10.54 11.02 10.5261 9.98 10.67 10.80 12.24 12.80 13.8

Voiding diameter (mils) measurement from 5DX and 2D X-ray machine

The 2D X-ray measurements are user dependent--- i.e. inconsistent.

Notes: measurement data for 2DX from a vendor site at San Jose with manual mode.

The 2DX readings are larger than 5DX reading with sensitivity 0.25 setting for this BGA.

The 2DX reading are smaller than 5DX reading with sensitivity 0.25 setting for another BGA.

4/6/2005

Slide 22

Gage R & R Definition

• Repeatability R1 : Variation in measurements obtained with one measurement instrument when used several times by an appraiser while measuring the identical characteristic on the same part.

• Reproducibility R2 : Variation in the average of the measurement made by different appraisers using the same measuring instrument when measuring the identical characteristic on the same part.

Gage R & R:

GR&R = R12 + R2

2

Square root of the sum of the squares of Repeatability and Reproducibility.

Reference: Measurement Systems Analysis – GM, Ford, and Chrysler

4/6/2005

Slide 23

Gauge R & R

Gauge R & R

0

10

20

30

40

50

60

70

BGADiameter mils

(slice1)

BGADiameter mils

(slice2)

FPGullwingHeel

Thickness

FPGullwingFuillet Lenth

mils

PTH milAverage

Thickness

Chip PadThickness

Resistor PadSolder

Thickness

Gau

ge R

& R

Year 2002Year 2004

Chip

• Under 10% error - Acceptable

• 10% to 30% error - Marginal

• Over 30% error - Measurement system needs improvement.

Make every effort to identify the problems and have them corrected

4/6/2005

Slide 24

BGA Ball Diameter Gage R&R -1

The units for diameter are: 2D (mm), and 5DX (mil)

Gage R&R was performed on the ball diameters of 5 BGAs. 3 operators performed three measurements on all pins. The first 30 pins of each BGA are used for analysis.

R1=Repeatability R2=Reproducibility

R1 (%) R2(%) R & R (%) R1 (%) R2(%) R & R (%)U5 3.93 0.63 3.97 13.82 10.29 17.19U13 3.02 0.64 3.08 12.35 9.31 15.46U23 3.4 1.77 3.83 11.31 10.91 15.71U46 7.69 3.11 8.29 12.90 11.16 17.06U57 6.19 3.44 7.08 10.98 10.23 15.01Average 4.85 1.92 5.25 12.27 10.38 16.09

FCN600-298 Components

2D 5DX

Tolerance is +/- 20% of the average ball diameter.

Gauge Repeatability & Reproducibility Results of BGA Diameter

4/6/2005

Slide 25

BGA Void Percentage Gage R&R - 1

Gage R&R was performed on the measurement of the percentage of the BGA ball area that is comprised of voids. Three operators performed three measurements on all pins of these 5 BGAs. The pins with zero reading are not included.

5DX doesn’t provide reliable readings if the BGA voiding percentage is set less than the Thresholds (20%) setting.

Tolerance is 20%Repeatability Reproducibility R & R (%)

U5 10.31 1.18 10.37U13 15.71 3.55 16.11U23 14.25 1.03 14.28U46 13.77 0.44 13.78U57 12.27 3.54 12.77

Component2D

4/6/2005

Slide 26

BGA Void Percentage Gage R&R - 2

5DX Gage R & R results of BGA voiding percentage from the pins with void percentages larger than 20%.

Ga ug e Typ e Ga ug e Na me Ga ug e S /N

Laminography X-ray Agilent 5DX Series 3 (S/ W 8.2.3) US40690588 at line 2 Da te P e rfo rme d S p e c ific a tio n P e rfo rme d By

2/ 9/ 2005 FCN600_00238, U42, BGA, void above 20 percenta Building 3

Opera torSample # 1st 2nd 3rd Range 1st 2nd 3rd Range 1st 2nd 3rd Range

Tria l Tria l Trial Tria l Tria l Tria l Trial Tria l Tria l1 22.14 20.21 20.36 1.93 20.57 21.67 20.09 1.58 20.09 20.36 20.57 0.482 24.18 24.18 26.79 2.61 25.82 27.51 24.18 3.33 25.82 24.18 25.82 1.643 26.63 26.63 28.32 1.69 25 26.63 26.63 1.63 28.32 28.32 26.63 1.694 27.51 25.82 24.18 3.33 25.82 24.18 25.82 1.64 24.18 24.18 25.82 1.645 22.6 21.06 21.06 1.54 22.62 21.06 21.31 1.56 22.6 21.8 21.09 1.516 23.38 24.15 25.85 2.47 25.85 24.15 25 1.7 23.36 25.85 24.15 2.497 22.6 22.6 21.06 1.54 22.6 24.21 22.6 1.61 24.18 22.6 21.06 3.128 25 25.1 25.23 0.23 25.82 25.41 25.1 0.72 25.1 26.25 25.82 1.159 26.69 25 26.69 1.69 28.44 25.82 24.18 4.26 24.18 25 22.6 2.4

10 23.36 23.36 25 1.64 25 25 25 0 25 25 25 011 23.41 25.79 26.63 3.22 23.93 23.41 26.63 3.22 28.32 26.63 26.63 1.6912 21.67 20.09 21.67 1.58 20.09 21.67 21.67 1.58 20.09 23.3 21.67 3.2113 25.85 27.6 23.36 4.24 25.85 25.85 27.6 1.75 27.6 27.6 25 2.614 25 26.75 23.3 3.45 25 25 23.3 1.7 23.3 25 25 1.715 26.63 28.32 26.63 1.69 25 28.32 28.32 3.32 25.79 25.79 25.79 0

Tota ls 366.65 366.66 366.13 32.85 367.41 369.89 367.43 29.6 367.93 371.86 362.65 25.32Sum 1099.44 R bar A 2.19 Sum 1104.73 R bar B 1.9733 Sum 1102.44 R bar C 1.688

X bar A 24.43 X bar B 24.55 X bar C 24.50

R bar A 2.19 # Trials D4 Max X bar 24.55R bar B 1.97 2 3.27 R dbl bar x D4 = UCLr Min X bar 24.43R bar C 1.69 3 2.58 UCLr 5.03 X bar Diff 0.12

Sum 5.85R dbl bar 1.95 S p e c ific a tio n To le ra nc e 20

Note : A dimension tolerance of +/ - 0.001 equals a Specification Tolerance of 0.002

Me a s ure me nt Unit Ana lys is % To le ra nc e Ana lys is

Re p e a ta b ility - Eq uip me nt Va ria tio n (E.V.)% E.V. = 100 x [(E.V.) / (Tolerance)]

E.V. = R dbl bar x K1 Trials 2 3

E.V. = 5 .95 K1 4.56 3.05 % E.V. = 29 .74%

R e p ro d uc ib ility - Ap p ra is e r Va ria tio n (A.V.)

A.V. = Square Root ([(X bar diff) x (K2)]̂ 2 - [(E.V.)̂ 2 / (n x r)])n = # of parts 15 % A.V. = 100 x [(A.V.) / (Tolerance)]r = # of tria ls 3 Operators 2 3

A.V. = 0 .83 K2 3.65 2.7 % A.V. = 4 .14%

R e p e a ta b ility a nd Re p ro d uc ib ility (R & R)

R & R = Square Root [(E.V.)̂ 2 + (A.V.)̂ 2] % R & R = 100 x [ R & R / Tolerance]

R & R = 6 .01 % R & R = 30 .03%

Gage Re pe atability and Re producibility Data She e t

A B C

Repeatability: 29.74%

Reproducibility : 4.14%

Gage R & R: 30.03%

Tolerance is 20%

4/6/2005

Slide 27

5DX BGA void measurement

Algorithm Joint Type : BGA2Characteristic Name : Voiding PercentM-1 : % Voiding Sampling Interval : 1

No. Layer Component Pin Su Ori B S Ru M-11 1 U42 1 2 180 N N 1 20.832 1 U42 4 2 180 N N 1 20.093 1 U42 5 2 180 N N 1 24.184 1 U42 6 2 180 N N 1 20.945 1 U42 9 2 180 N N 1 28.326 1 U42 25 2 180 N N 1 27.517 1 U42 29 2 180 N N 1 21.098 1 U42 38 2 180 N N 1 22.439 1 U42 46 2 180 N N 1 24.1510 1 U42 50 2 180 N N 1 20.9411 1 U42 67 2 180 N N 1 20.9412 1 U42 71 2 180 N N 1 22.6013 1 U42 88 2 180 N N 1 25.0014 1 U42 93 2 180 N N 1 25.8215 1 U42 101 2 180 N N 1 25.0016 1 U42 119 2 180 N N 1 25.0017 1 U42 133 2 180 N N 1 21.6718 1 U42 168 2 180 N N 1 21.6719 1 U42 205 2 180 N N 1 25.8520 1 U42 206 2 180 N N 1 20.8021 1 U42 224 2 180 N N 1 21.5522 1 U42 238 2 180 N N 1 20.8023 1 U42 247 2 180 N N 1 21.4924 1 U42 255 2 180 N N 1 25.0025 1 U42 261 2 180 N N 1 26.63

Algorithm Joint Type : BGA2

Characteristic Name : Voiding

M-1 : Void Diameter

Sampling Interval : 1

No. Layer Component Pin Su Ori B S Ru M-1

1 1 U42 1 2 180 N N 1 10.15

2 1 U42 2 2 180 N N 1 9.76

3 1 U42 3 2 180 N N 1 9.76

4 1 U42 4 2 180 N N 1 10.15

5 1 U42 5 2 180 N N 1 11.71

6 1 U42 6 2 180 N N 1 10.54

7 1 U42 9 2 180 N N 1 12.89

8 1 U42 25 2 180 N N 1 12.50

9 1 U42 29 2 180 N N 1 10.93

10 1 U42 37 2 180 N N 1 9.37

11 1 U42 38 2 180 N N 1 10.54

12 1 U42 46 2 180 N N 1 11.32

13 1 U42 50 2 180 N N 1 10.54

14 1 U42 67 2 180 N N 1 10.54

15 1 U42 71 2 180 N N 1 11.32

16 1 U42 85 2 180 N N 1 10.15

17 1 U42 88 2 180 N N 1 12.10

18 1 U42 92 2 180 N N 1 9.37

19 1 U42 93 2 180 N N 1 12.10

20 1 U42 101 2 180 N N 1 11.71

21 1 U42 106 2 180 N N 1 9.37

22 1 U42 119 2 180 N N 1 12.10

23 1 U42 128 2 180 N N 1 9.76

24 1 U42 133 2 180 N N 1 10.54

25 1 U42 166 2 180 N N 1 9.37

26 1 U42 168 2 180 N N 1 10.54

27 1 U42 169 2 180 N N 1 9.37

28 1 U42 185 2 180 N N 1 9.76

29 1 U42 205 2 180 N N 1 11.71

30 1 U42 206 2 180 N N 1 10.15

31 1 U42 207 2 180 N N 1 9.37

32 1 U42 224 2 180 N N 1 10.15

33 1 U42 238 2 180 N N 1 9.37

34 1 U42 247 2 180 N N 1 9.76

35 1 U42 255 2 180 N N 1 11.32

36 1 U42 261 2 180 N N 1 12.50

No. Layer Component Pin Su Ori B S Ru M-1Void percentage Void diameter

All 25 defects found by 5DX were verified by transmission X-ray.

All 25 defects found with void percent algorithm were found by the void diameter algorithm.

7 defects found with void diameter algorithm were not found by the void percent algorithm but were verified real defects by transmission x-ray.

4 defects found with void diameter Algorithm were false calls as was verified by transmission x-ray.

4/6/2005

Slide 28

BGA (U42) pinouts (2D & 5DX)

BGA pinouts (2D) BGA image (2D)

BGA pinouts (5DX)

5DX Pin 2D Pin1 A12 A23 A34 A45 A56 A67 A78 A89 A910 A1011 A1112 A1213 A1314 A1415 A1516 A1617 A1718 B119 B2

2DX limitation:

It is not 100% test coverage with auto mode for double sided board.

4/6/2005

Slide 29

Measurement data for catching BGA voiding

A1 Failed

Diameter 599.551 um

Overall void 24.432 %

Largest void 23.159 %

A2 Failed

Diameter 579.165 um



A3 Failed

Diameter 571.106 um



A4 Failed

5DX Pin 5DX Voiding% 2D voiding % 2D Pins1 20.83 24.432 A14 20.09 24.085 A45 24.18 28.593 A56 20.94 26.156 A69 28.32 26.2 Absent25 27.51 24.713 B829 21.09 28.604 B1238 22.43 23.793 C446 24.15 27.307 C1250 20.94 27.516 C1667 20.94 24.903 D1671 22.6 25.918 E388 25 23.491 E1793 25.82 22.766 F8101 25 22.766 F16119 25 30.576 G17133 21.67 25.085 H14168 21.67 27.013 K15205 25.85 23.327 N1206 20.8 22.85 N1224 21.55 21.561 P3238 20.8 20.905 P17247 21.49 24.098 Q9255 25 25.374 Q17261 26.63 30.161 R6

2D data from auto mode testing.

5DX and 2D data at the same pins that caught as defects because void area > 20% of ball area.

Pin is not tested by 2D due to chip under opposite side.

26.2 is from manual measurement.

5DX and 2D caught all real defects.

4/6/2005

Slide 30

2D Images from manual measurement

A9 – 2D can’t test in auto mode due to obstruction.

2D Manual measurement: void =26.2%. (Real defect)D7– 2D original data (auto mode) is 20.56%2D Manual measurement: void = 16.7%. (False call)

C3 -- 5DX detected this as voiding diameter > 9 mils. 2D Manual measurement: void = 18.0%. (False call )

4/6/2005

Slide 31

Measurement data comparison

15

20

25

30

35

1 3 5 7 9 11 13 15 17 19 21 23 25

BGA Pins

BG

A v

oid

perc

enta

ge

Reag

ings

5DX Voiding% 2D voiding %

5DX and 2D measurements for same real defect pins (void area > 20% of ball area).

5DX and 2D BGA void percentage measurements

4/6/2005

Slide 32

Voids caught at 5DX and 2D systems

• 2D covered only 85% pins due to limitation X-ray for double sided board

• Both system caught voids which were larger than 20%. 2D didn’t catch 3 real defects because it didn’t test these pins with auto mode. However 2D is able to identify defect with manual mode.

• False Call Rate = False Calls / Real Calls

False call rates are very low for both systems.

For U42

X-Ray machine Tested pins on U42 Defect call # Real defect # Skip real defect # Defect false call # False call rateTransmission 247 53 29 3 24 0.83Laminography 289 36 32 0 4 0.13

4/6/2005

Slide 33

Process Improvement Tool

• 5DX variable measurement data can be used for SMT process improvement

• True example……

4/6/2005

Slide 34

5DX Data (FCN600-00216_02 for August 2004)

*** 5DX TEST SUMMARY ***

Panel Name U9 - BGA 44Date from 08/01/04 U2 - BGA 40Date to 08/31/04 U10 - BG 33Total boards tested 185 U4 - BGA 15Boards with defects 22 U73 - FP 11Board Yield (%) 88.11 U11 - GU 6Total Defect Pins 184 U3 - BGA 5Primary Components 629 C37 - CH 4Secondary Components 570 L9 - GUL 4Total Tested Pins 7043 U18 - FP 4DPMO 121Comment

Component DefectTotal Serial No. Total Open Short Void Excess Misalign Insuff Multiple Miss KnockOffSub Total 184 All 184 21 31 29 0 4 34 10 14 0U9 - BGA2 44 18519 22 0 1 20 0 0 0 1 0 0U9 - BGA2 44 25006 1 0 1 0 0 0 0 0 0 0U9 - BGA2 44 25039 13 0 5 0 0 0 0 3 0 0U9 - BGA2 44 25041 8 0 0 1 0 0 0 0 0 0U2 - BGA2 40 25039 22 0 2 4 0 0 1 1 0 0U2 - BGA2 40 25041 3 0 0 1 0 0 0 0 0 0U2 - BGA2 40 25047 1 0 0 0 0 0 1 0 0 0U2 - BGA2 40 25065 14 0 13 0 0 0 0 1 0 0U10 - BGA2 33 18589 31 0 0 0 0 0 31 0 0 0

FCN600_00216_02_SW81

Distribution of Major Defect Components

0

10

20

30

40

50

U9 - BGA2

U2 - BGA2

U10 - B

GA2

U4 - BGA2

U73 - F

PGULLWIN

GU11

- GULL

WING

U3 - BGA2

C37 - C

HIPL9

- GULL

WING

U18 - F

PGULLWIN

G

Tota

l Def

ect P

ins

The top 4 defect locations are BGA.

4/6/2005

Slide 35

The non-optimized bottom reflow profile

Peak temperature 210 °C. Time above 183 °C is 64 seconds. Soak time between 130 °C and 160 °C is 65 seconds.WS609 solder paste is from Alpha Metals

4/6/2005

Slide 36

Experiment Matrix

Gemini Experiment to eliminate voids Planned for Aug 7th, 2004Customer

BOARD S/N FIXTURE SOAK TIMETAL PEAK SOAK TAL peak delta T % Void Dia. % Void Dia.1 18404 Y 110 83 209 68 60s 196 20 Front <5% 3 MILS 21% 9.76 Verify measurement2 18400 N 126 75 206 65 78 206 10 Front Negligible Negligible 16.95% 7.81 Excellent Results3 18402 N 126 75 206 65 78 206 10 Back 11.58% 6.25 Repeat of #24 18398 N 126 75 206 65 78 206 10 Back 11.11% 6.25 Repeat of #25 18396 N 126 75 206 65 78 206 10 Back 13.49% 7.42 Repeat of #26 18389 N 126 75 206 65 78 206 10 Back 13.08% 6.64 Repeat of #27 18397 N 125 70 210 65 2208 18399 N 125 70 210 65 78 206 Back many 5% 3 mils 17.91% 8.29 N 125 70 210 90 20610 N 120 60 200 65 22011 18401 N 120 60 205 65 78 206 10 Back Negligible Negligible 14.06% 7.3 Best Combination11A 18392 N 120 61 205 65 78 206 10 Back Negligible Negligible 11.56% 6.64 Repeat of #1111B 18390 N 120 62 205 65 78 206 10 Back Negligible Negligible 16.95% 8.2 Repeat of #1112 N 120 60 205 90 20613 N 85 90 215 65 22014 18269 N 85 90 215 65 78 206 10 Back 12% 7.5 mils 24.52% 8.9815 N 85 90 215 90 206

L=90sec L=200C L=65sec L=206CH=120sec H=210C H=90sec H=220C

Note: 1 If board runs successfully w ithout voids or low voids run 2 additional boards to confirm2 If choosing betw een more than one profile for production release - choose board number 8 assuming it is one of the best combination3 Top side 5DX measurement is based on the settings of .25 sensitivity, 4 mils minmum dia. 8% minum voiding area4 TAL = Time above liquids (dw elling time )

Comments5DX-BOTTOM 5DX-TOPTrailing edgeBOTTOM TOP

Search optimized profile.

Measure BGA void diameter and percentage area after each new profile.

4/6/2005

Slide 37

The optimized bottom reflow profile

Peak temperature is 209 °C.Time above 183 °C is 75-85 seconds.Soak time between 130 °C and 160 °C is 110-120 seconds.WS609 solder paste is from Alpha Metals

4/6/2005

Slide 38

No Voids Present after optimized reflow profile

Algorithm Settings:- Sensitivity of 0.25

- MIN_VOID_DIA of 9 mils

- PERCENT_VOID of 20

- Camera frame grabber setting of 5

- S/W 8.1

BGA ball size: 24 mils IPC Class II

Need rework if voiding diameter > 10.8 mils or

> 20.25% area

5DX image after optimized reflow profile

4/6/2005

Slide 39

Realized Savings for 2004 For One Customer

• The BGA number above correspond to the BGA defect reduction from historical data for each assembly of the customer.

• Reduce BGA defect by 0.5% (from 0.8% to 0.3%) by analyzing 5DX measurement data and oven profiles on one customer’s assemblies. Which represents a 60% improvement on the assembly process.

• The labor cost to rework a BGA is $5.7, based on a $17 labor cost and 20 minutes per part.

• The average BGA cost is $108 per part. Historically only 70% of the BGA with failure were reworked with new BGAs.

Month BGA number1 Labor Material Total September 30 $169 $2,242 $2,412

October 31 $176 $2,329 $2,505November 349 $1,991 $26,410 $28,401December 108 $615 $8,155 $8,770

Total 518 $2,951 $39,137 $42,087

With 5DX as an SMT process improvement tool using variable measurement data, we saved above $40,000 from deduce BGA defects.

4/6/2005

Slide 40

Conclusions

1. There are many algorithm settings on 5DX that can effect the detection of solder defects. The Algorithm void settings are depending on the characteristics of both the PCB and the part, therefore they must be adjusted accordingly. Based on our data it appears that an Algorithm sensitivity setting of 0.25 is a good starting number for BGA voiding for most BGAs package. However, there is no universal number for catching one solder joint defect type.

2. Even though the 5DX test reports only provide attribute data, we are able to retrieve variable data (actual measurements) from the system as well. With variable measurement data from this automatic test method, the 5DX is a good SMT process improvement tool, as well as a tool for process defect detection. The variable data can help us to control SMT process effectively. After having optimized the oven profile there were significant BGA defect reductions.

3. Based on the result of our experiment, if we use Alpha Metals WS609 solder paste on Force10 Networks HASL (hot air solder leveling) finish board, we concluded that the profile with 110–120 seconds longer soak time from 130 to 160 degree Celsius, 200~205 degree Celsius low peak temperature, and 75~85 seconds TAL (Time Above Liquidus) produced the best result in terms of less solder voids.

4. Transmission X-ray machine used as a failure analysis tool, can help us verify 5DX settings, identify defects (BGA open, microcrack).

5. The results of this study will be present in the SMTA Technology Summit at NEPCON, Shanghai, China on April 11, 2005.

4/6/2005

Slide 41

Abstract - NEPCON China (April 11, 2005)• Inspection of Ball Grid Arrays (BGA) and the detection and measurement of BGA voids has become critical for SMT

Production Lines,and for the Engineering of Electronics Manufacturing Service (EMS) as the use of BGAs on circuit card assemblies has increased. How large can a BGA void be and still be considered acceptable? First, we must ensure that we have an accurate method to measure BGA voids. Second, we must ensure that the criteria we use in setting BGA void rejection is in agreement with IPC standards or Customer Specifications.

• The objective of our work and this paper is to:a. Assess the current setting for the Agilent AXI 5DX BGA2 Void Algorithm in detecting voids.b. Compare and correlate voiding measurement data between AXI Laminography and Transmission X-ray systems.c. Discuss how AXI technology can be used for SMT process improvement in reducing BGA defects.

• The detection of BGA voids with the AXI system and the accuracy of the measurement made is a function of one of the machine’s Void Algorithm thresholds called the “Sensitivity Setting”. This threshold sets the level of grey-scale at which pixels are to be considered part of a void. The larger is this setting, the larger is this number of pixels that are considered to be part of a void. The AXI measurement of BGA void size and percentage is dependent on the AXI Algorithm’s Sensitivity Setting number. We studied this measurement data (BGA voiding diameter and area percentage) with the SPC tool MINITAB for Analysis of Variance (ANOVA). The ANOVA showed the mean of the Standard Deviation (STDEV) for a selected BGA void diameter to be 0.6345±0.3418 mils, 0.6273±0.5435mils, and 1.1155±0.5766 mils with sensitivity setting 0.25, 0.35, and 0.45 respectively. The ANOVA showed the mean of the STDEV of the BGA void percentage to be 2.200±0.953%, 2.834±1.602%, and 5.115±2.546% with sensitivity setting 0.25, 0.35, and 0.45 respectively. Based on this data it appears that an Algorithm sensitivity setting of 0.25 is good starting number for BGA voiding. The Transmission X-ray measurement data was used for comparison.

• The AXI test reports attribute data, but the system can provide variable data as well. With variable measurement data from this automatic test method, the AXI is a good SMT process improvement tool, as well as a tool for process defect detection. The variable data can help us effectively control SMT processes. We will also discuss how to reduce BGA defects using AXI measurement data. Decreases in BGA defects can lead to manufacturing cost savings. As a result of our use of this AXI defect detection data, we were able to significantly reduce the occurrence of BGA defects, which resulted in a cost savings of over $40,000 in a period of four months.

Documents

5DX ALGORITHM SETTINGS FOR BGA VOIDING - Keysight · 2005. 4. 6. · IPC-7095 Design and Assembly Process Implementation for BGAs-August 2000 BGA ball size: 24 mils Current Agilent