SEMICONDUCTOR · Semiconductor Assembly & Test

Customer U YMS
Yield Management System

A full-chain yield management platform for the semiconductor assembly & test industry. It unifies heterogeneous data scattered across SAP, MES, CIM, S3, and equipment tools, achieving a paradigm upgrade in yield management — from Overall Yield overview to Single Defect root cause localization, and from manual Excel to AI-powered automatic analysis.

Yield Overview Defect Pareto Commonality Analysis FARCCA Auto Report AI Root Cause Recommendation
3 Days
→ 30 Minutes
95%
Reduction in Analysis Time
8 Steps
One-Click Defect Closure
PB-Level
Heterogeneous Data Fusion

The Yield Analysis Revolution: From "Three Days" to "Thirty Minutes"

Traditional yield analysis relies on Excel, JMP, and manual data extraction across systems — a complete analysis often takes nearly 3 days. YMS compresses this process to 30 minutes, transforming yield engineers from "data porters" into "problem insighters."

⚠️ Before · Traditional Mode
Annual/weekly/monthly data analysisExcel + JMP
Defect Mapping statistics8 hours
Cross-system document/data search2 hours
Abnormal batch data tracing4 hours
FARCCA report filling8 hours
IT script data extraction + Excel export4 hours
≈ 3 Days
✅ After · YMS Mode
Multi-dimensional yield & defect analysis3 minutes
Single defect PPM calculation and drill-down2 minutes
Sankey commonality analysis to locate root causes10 minutes
FARCCA auto-filling and approval15 minutes
Yield & Defect automatic alertsReal-time
Self-service data/JMP exportOne click
≈ 30 Minutes

Six Core Capabilities Covering the Full Yield Management Chain

From macro overview to micro root cause, from passive response to proactive alerting — building a complete closed loop for semiconductor assembly & test yield management.

📊

Overall Yield Overview

Automatically aggregates yield trends by product Phase, package type, and time dimension. Top 5 defect Pareto presented in one click, quickly identifying the most impactful defect types.

🔍

Single Defect Analysis

Automatically calculates single defect PPM, supports multi-date comparison and process station detail drill-down — from defect code straight to equipment, lot, and station-level data.

🕸️

Commonality Analysis

Visually traces defect distributions across material, equipment, site, and personnel dimensions via Sankey diagrams, quickly focusing on abnormal factors and narrowing the root cause scope.

📋

FARCCA Auto Report

Automatically fills in data and basic information based on system analysis results, generates the FARCCA report and routes it for approval — replacing 8 hours of manual filing.

🔔

Metrics Alarm

Automatically triggers alerts when yield/defect metrics exceed thresholds, combined with WeChat Work/email push notifications — shifting from "post-event review" to "in-process intervention."

🤖

AI Root Cause Recommendation

Builds a RAG knowledge base from historical issues, root causes, and countermeasures, combined with LLM for root cause intent recognition and intelligent solution recommendation.

Three-Phase Evolution: YMS → Traceability → AI

Guided by business value, building yield management, full-chain traceability, and intelligent decision capabilities in phases — reducing implementation risk and delivering value quickly.

PHASE 01

YMS Yield Management System

  • Enterprise data warehouse, connecting SAP/MES/CIM/S3/tool data
  • MP product complete yield analysis: Overall Yield, Top 5 defects, single defect, commonality analysis
  • FARCCA auto report and approval workflow
  • Low-code applications + self-service data dashboard building
PHASE 02

Traceability Full-Chain Tracing

  • Establish Unit/Lot level full lifecycle traceability records
  • Connect SPC and CIM key parameters for process parameter anomaly linkage
  • Support analysis of key parameters such as dot weight, viscosity, SPI, reflow profile
  • Trace back from defects to full records of process, equipment, material, and lot
PHASE 03

AI Intelligent Decision-Making

  • Automatic anomaly work order triggering when metrics exceed thresholds
  • Vectorize historical issue-root cause-countermeasure data to build a RAG knowledge base
  • Integrate LLM for root cause intent recognition and solution recommendation
  • NLP Agent natural language interaction, lowering the analysis barrier

Four-Layer Architecture, Robustly Supporting Massive Yield Data

From data source ingestion to application presentation, from offline data warehouse to real-time alerting — building a highly available, scalable technical foundation for yield management.

🔗
Data Ingestion Layer
SAP / MES / CIM / S3 / Tools / Local Files
🗄️
LeanFusion Data Fusion
ODS · DWD · DWS · Subject Data · Master Data
🔧
LeanCodee Low-Code
Model Design · Page Design · Process Engine · SDK
📊
LeanBI + Application Layer
Yield Analysis · Visualization · Auto Reports · AI Agent
1

Data Cleansing & Fusion

Multi-source heterogeneous data ingestion, completing data cleansing, association, matching, and fusion, unifying the yield/defect metric system.

2

Metric Calculation & Alerting

High-performance metric calculation based on the Doris distributed data warehouse, supporting real-time/offline alert push.

3

Low-Code Application Building

Rapidly build yield analysis pages, FARCCA workflows, and work order closed loops with LeanCodee; the business can iterate on its own.

4

AI Knowledge Accumulation

Accumulate historical analysis experience, root causes, and countermeasures into searchable knowledge assets, enabling organization-level capability reuse.

Delivery Method: Co-Creation & Knowledge Transfer

Not a one-time delivery of a closed-source system, but through joint teams, prototype-first approaches, and low-code platform empowerment — enabling customers to sustain self-growth capabilities.

🤝
Joint Team
Integrated project team of IT + business + Leansight
🧩
Prototype First
Prototype before development for complex requirements, reducing communication costs
📚
Knowledge Transfer
Key user training, giving customers the ability to replicate on their own
🚀
Continuous Iteration
Low-code supports self-service business adjustments; the system grows with the business

Start the Intelligent Upgrade of Semiconductor Yield Management

Whether you are an assembly & test house, a wafer fab, or an IDM, Leansight YMS can help you break down data silos and achieve rapid localization, root cause analysis, and closed-loop improvement of yield issues.