Product lifecycle analytics - OptimalPlus · data analytics to prevent bad products from reaching...

Product lifecycle analytics

Eran Rousseau, VP Product

Solutions for Semiconductors

Month 2019

© OptimalPlus 2019. All rights reserved.

It’s a changed worldTechnological innovation has transformed our lives.

Products and devices are more intelligent and connected.

These products rely on thousands of electronic components that must be more reliable than ever before.

2

© OptimalPlus 2019. All rights reserved. 3

1) Automotive change drivers for the next Decade, EY, 20162) BMW - AEC Automotive electronics reliability workshop, 20173) NHTSA Recall Data4) Audi, DVCon Munich, 2017

90%

22%

3x

car innovations & new features are driven by electronics1

warranty costs related to electronics & semiconductors2

car recall increase from 2014-2016 due to electronics3

Autonomous car will drive 20x more than traditional car4

Automotive

innovation

exemplifies

the challenge

Reliable electronics is a must

1 car failure

every hour4

Audi says

$10Bn50Bn cars

Toyota’s gas pedals recall

Ford’s failure-to-park recall

Takata’s air bag recall

20x


Our roleWe deliver unprecedented manufacturing efficiency and reliability to the Automotive, Semiconductor and Electronics supply chains.

We provide actionable insights through unique data analytics to prevent bad products from reaching the market.

We call it lifecycle analytics


Analyzing huge volumes of data

5

100bn+devices per year

Semiconductor Automotive Electronics


Trusted by leading brands

6

Customers

Supply chain


Providing innovative solutions

7

Collect Detect Act• Data harmonization

• Product, machine and process data

• Prescriptive analytics

• AI / Machine learning

• 24x7 analytics engine

• Real-time

• Automatic

• Distributed

• Controlled

O+ solution

Customer methods

• Collect lots of data

• Use it primarily when there is a problem: Bad Yield, RMA (Returns), Etc.

• Find the problem but frequently not the root cause

• Process is often manual and reactive, not proactive

• Use of many tools, but not an integrated solution

A unique, automated and proactive integrated solution


Through the entire product lifecycle

8

Chip Board Product In use ReturnsECU modules

Machine • Process • Test • Rework • Genealogy • Performance • Reliability • Usage • Warranty


Semiconductor solution overview

9

Semiconductor solutions

Yield Productivity & Efficiency

Quality & Reliability

Ramp, NPI

Open platform

Big data highway


Cloud or on-prem.

O+ central analytics24x7 rule execution & orchestration

O+ data platformCentral repository

Central

System architecture

10

Factory C

Edge (factory floors)

Factory B

Factory A (internal or outsourced)

Actionable insights across all manufacturing and test processes

Tester C

Fab data

Other equipment & data sources

Tester A

Tester B

Assy. data

Portal+

Control Room+

PLM, ERP, CRM

Action

Data

MES

Action

Data

Action

Data

Actio

n

Data

O+ proxy

O+ proxy

O+ proxy

Action

Data

Action/R

ule

s

Data

Data

Data

O+ edge analytics24x7 rule execution & orchestration

O+ data platformEdge repository

Action

Data

Action/Rules

Data


Cloud or on-prem.

O+ central analytics24x7 rule execution & orchestration

O+ data platformCentral repository

Central

System architecture

11

Factory C

Edge (factory floors)

Factory B

Factory A (internal or outsourced)

Actionable insights across all manufacturing and test processes

Tester C

Fab data

Other equipment & data sources

Tester A

Tester B

Assy. data

Portal+

Control Room+

PLM, ERP, CRM

Action

Data

MES

Action

Data

Action

Data

Actio

n

Data

O+ proxy

O+ proxy

O+ proxy

Action

Data

Action/R

ule

s

Data

Data

Data

O+ edge analytics24x7 rule execution & orchestration

O+ data platformEdge repository

Action

Data

Action/Rules

Data


Common challenges we solve

12

• Minimize excursions

• Eliminate RMAs

• Protect your brand

• Product safety net



• Overall yield

• Site-to-site yield

• Re-test policy

• Equipment & H/W performance issues

• Excessive index & pause times

• Test time variations per tester

• Inefficient retest policies & execution

• Inconsistent tester availability & utilization

• Quality & consistency of data

• Minimize time to market

• No traceability

• Expensive mask costs

Ramp, NPI


Rules – Targeting challenges 24x7

13

• Library of standard rules accommodate most of the challenges faced by our industry

• Custom rules available for unique monitors and actions including support for R and Python scripts

• Deployed at any level of your supply chain (central vs edge)

• Rules engine running 24x7


Rules turning challenges into actions

14

Action categories

Equipment actions

• Pause

• Engineering tool alert

Process actions

• Put materials on hold

• Re-binning

Recipe adjustments

• Re-test skip/add

• Adaptive testing

Data augmentation

• Feed-forward

• Feed-backward

• Virtual operation

Alerts

• Quality outlier alerts

• Yield alerts

• Predictive/Anomaly alerts

SolutionsDetails


Quality & Reliability – Solutions

16

• Outlier Detection

• Adaptive testing using Machine Learning

• Escape Prevention

• Test data reliability

• Minimize excursions

• Eliminate RMAs

• Protect your brand

• Product safety net



Driving quality improvements

17

Approaches

• Yield impact budgetProactive outlier screening

• Cycles of learningRMAs

Best practices

• Escape preventionDeterministic

• Outlier detectionGeographical & Statistical

• Quality indexDie “goodness” score

• Data feed forwardDrift detection & ML scoring

• Auto-Hold


Yield impact budget – Balancing quality and yield

18

• Comprehensive and flexible library of algorithms, supporting R & Python

• Can be applied and adjusted to meet yield/quality targets

• Results are saved in the database which allows you to visualize and monitor the impact


RMA/Quality cycles of learning

19

Cycles of

learning

Data analysis

Monitoring recipe

execution

Publishing &

activating

Recipeconfiguration

Simulation &

validation


Customer Use Case: Quality monitoring – Test escapes

20

Device: Automotive

Problem: Possible quality returns

Issue: Penalties from customer

Problem discovered: No PRR monitoring & 12 possible test escapes

Fix: Rules & special binning

12 devices

Result: Insurance and improvement reports to customer


Yield – Solutions

21

• Baseline yield & SBL monitoring

• Test equipment performance

• Test and retest policies and execution

• Tests limits validation

• Cross-operation correlation

• Targets against any measure/KPI

Yield

• Overall yield

• Site-to-site yield

• Re-test policy

• Equipment & H/W performance issues


Overall yield contributors

22

DesignTest & Ops

Process

Currently and historicallyour main focus

Active PoC’s started forFab process data yield

Interest from design partners for our solution


Customer Use Case: Operational yield – Site issue

23

Device: Network

Problem: Yield loss

Issue: Yield by tester varies

Standard O+ rules foundWith no monitoring – Site-Site issue not detected – This case is 16 lots

Iteration 0

Iteration 1


Customer Use Case: Operational yield – Site issue (continued)

24

Operation issue: Same-site problem over 4 re-tests

Problem discovered: Potential Good devices tested as Bad

Fix: Rules monitoring catches issues on 1st pass

Iteration 2

Iteration 3

Result: Increased overall Yield by 2%


Productivity & Efficiency – Solutions

25

• Adaptive Testing using Machine Learning

• Test equipment performance

• Test and retest policies and execution

• Testers availability & utilization (OEE analysis)

• Classical Test Time Reduction (TTR analysis, ROA)

• Adaptive Test Time Reduction (ATTR)

• Cross-operation correlations

Productivity & Efficiency

• Excessive index & pause times

• Test time variations per tester

• Inefficient retest policies & execution

• Inconsistent tester availability & utilization


Productivity & Efficiency opportunities

26

Better resolution of time during test

Retest optimization

Test time consistency

Tester utilization


Customer Use Case: Productivity & Efficiency – Increasing test time

27

Device: Microcontroller with Flash

Problem: Capital avoidance

Issue: Needed 10 more test stations

Problem discovered: Issue with test program

Fix: Improved O+ rule for monitoring for all future testers/devicesResult: Saved 8 test stations = $12M in CapEx & OpEx Savings

Standard O+ rules foundTesters had different throughputsTest Time Increasing from 120 Sec to 300 Sec


Ramp, NPI – Solutions

28

• Data loading rules

• Load and create conditions

• Sandbox to edit metadata

• Datasets

• Virtual “workbench”• Shared analyses and data augmentation

• Full chain of custody

• Limits, Correlation & GR&R Applications

• Report generation

• Quality & consistency of data


• No traceability

• Expensive mask costs

Ramp, NPI


NPI areas of focus and flow

29


• Analyze split lots

• Determine production limits

• Identify design sensitivities

Datacollection

Sandbox Datasetmanagement

Analysis Reporting

• Proxy• Drop Box• STDF• OTDF• SAF

• Data Cleansing• Mapping• Validation

• Association• Augmentation• Attributes• Templates

• Static/Scheduled• Flexible• Intuitive• Customized

PVT Analysis

Correlation App

Limits App

GRR App


Customer Use Case: Ramp, NPI – Limit Simulation App

30

Device: Cell phone

Problem: Limits not optimized

Issue: Would not fail questionable measurements

Fix: Run analysis using limits application

Result: Immediate feedback = Faster product launch

O+ standard tools found: Limits too wide


Voice of the market

32

“How can I combine, and do more with my siloed data systems?”

“Our data retention is at least 10 years for our

automotive products.”

“We already have a corporate license of Tableau, can we use

this to visualize O+?”

“I want to leverage fab/assembly data (i.e. defect & inspection) to improve my quality”

“My teams are proficient in Python or R and I want to leverage this”

“Can we have programmatic access

to O+ data?”

“I know we need to do ML, we just don’t know how to get started”

“How can we store old data so it doesn’t take so long to

reload & use?”


Consolidated challenges

33

CTO/CIOs & IT professionals

Concerned about enterprise TCO (Total Cost of Ownership

Product, Quality & Yield Engineering teams

Need a solution providing analytics that scale

Data Scientists & Engineering teams

Need a collaborativeecosystem


Platform goals

34

1Support bi-directional data integrationwith any customer data lake

2Enable easy consumption of OptimalPlus data by 3rd

parties and BI tools

3Integrating with machine learning data science frameworks,leveraging OptimalPlus deployed infrastructure

4Boost developers innovation by leveraging OptimalPlus rich API’s, algorithms and infrastructure


Column store databaseMetadata index +“Hot” cacheSQL over Hadoop

Optimal+ data pipeline

Portal+ desktop & web UI

Analytics engine & API

Data pipeline

Optimal+ platform

Customer data lake

IBM Streams

Customer platform

(public, private, hybrid)Cloud

Customer data

Industry focused open platform

35

Synergetic with any big data strategy

Connected to existing infrastructure

Open for all kinds of data

Accelerates innovation

Extensible through both data and algorithms

Optimized“Manufacturing”data


Summary

36

Lifecycle analytics solutions

turning data into actions for immediate ROI

Product-centric approach

for improved quality & reliability and operational efficiency

End-to-end supply chain visibility

across operations and industries

Open platform

indusry focused for seamless integration with any big data strategy

Domain expertise

applying data science to solve industry challenges


Significant business impact

37

50% case avoidance

up to 25%cost savings

increase up to

10% NPI2% HVM

from weeks to days NPI, TTM, RCA



Ramp, NPI


Ask our customers

38

“Escape Prevention enables us to identify specific manufacturing and test issues that drive advanced quality screening and comprehensive product management.”– Michael Campbell Senior VP of Engineering

“Global Ops for Electronics enables us to rapidly identify and respond to the source of any PCB and systems manufacturing issue, down to an operation, facility, line or station.”– Keith Katcher VP of Operations Engineering

“Escape Prevention helps enable Marvell to deliver best-in-class silicon solutions for the automotive market segment.”– Mark Jacobs VP Engineering Operations

“We went from ‘We can’t afford to do this [Optimal+] to ‘We can’t afford not to do this [Optimal+]”– Carl Bowen AMD Fellow

“Optimal+ gives us real-time visibility of our test operations, enabling us to monitor every critical parameter to ensure that every product is of the highest quality and performs as expected.”– Vincent Tong Senior VP, Global Operations & Quality

“

“

Thank you

Product lifecycle analytics - OptimalPlus · data analytics to prevent bad products from reaching...

Documents

Transcript of Product lifecycle analytics - OptimalPlus · data analytics to prevent bad products from reaching...