MODULE 3:VISION SYSTEM-1

Enabling Objective #1
Decribe the Basic function,Theory of Opration and Specification of the Vision System

Vision System Overview

• LCD Monitor
• S-Scan Cameras 2x and 6x standard
  -NEXIS Cameras-Manufactured by Matrox
  -Transfer time at 100fps is 10 msec
  -Dual camera control unit(CCU) and frame grabber combined inside Matrox 4Sight-M unit
• Optics Assembly
• Matrox Operating System (PRS)
• IIIuminator Interface(LED Driver Circuit)
• Man Machine Interface (MMI)

Function of Vision System

Automatic Device Alignmnt

• PRS allows the bonder and MHS to automatically find specific reference points before each device is bonded

Automatic Bond Placement Alignment

• Video Lead Locator(VVL)finds the exact taught location of bond position on the outer leads of each device to ensure accurate placement

Automatic Pad Loation and Centering

• Design to increase speed and accuracy when traching the locations of bond pad

Automatic Ball Finder for post bond Inspection(PBI)

• Enable inspection of bond placement and ball diameters

Operator Interaction

• Provides a interface with the Man Machine Interface(MMI) keyboard and mouse and handles the monitor display

• Communicates with the bonder processing systemvia an Ethernet (LAN)interface using e-Xpert

  
  Vision System Layout
  Vision System Specifications

Optics Housing Assembly

Left view
right view
Dual Mag Optical Path

• The two Camera Heads are connected directly to the Vision System Unit.

• The dual Camera Control Unit (CCU) and frame grabber are integrated in a PC board housed inside the Vision System Unit.

  
  Matrox box

Matrox Vision System

• Windows XP-Embedded Operating System
• X-Server for Wire Bonder GUI Display
• Vision Application consists of:
  -Die and Lead Eyepoints for Alignment
  -Pad Finder and Lead Locator to teach bond programs
  -Ball Finder for Post-bond Inspection and Crosshair correction
• Vision Application utilizes
  -Normalized Gray Scale Correlation
  -Geometric Model Based Pattern Matching

Matrox vision Assembly

note:Do not remove the camera connector from Matrox when the power is NO.Failure to comply will cause the camers haed to short circuit.

Right side panel
Left side panel

Die Eyepoint-Acceptance Level

• Increase the acceptance level if there are a lot of "false finds"
• Lower the acceptance level if there are lot of "fail to find"

-The lower you set the level, the higher the risk of false matches
-The higher you set the level, the lower the risk of false matches but you could reject devices with eyepoints that vary from the taught

Teach Algorithm Comparison

Die Algorithm uses normalized grayscale correlation

• Measure of similarity between the model and an image pixel by pixel.
• Independent of any linear differences in model or image brightness.

Geometric Algorithm is feature-based pattern matching

• Edges are extracted and distinguishing features are detected from the edges to create a model during teach.
• Edges (lines) are extracted and matched with the model when finding eyepoints.
• Images with high contrast and strong edges will lead to a more robust geometric model.
  Geometric Algorithm vs. Normalized Grayscale Correlation
  Advantages:
• Effectively deals with the nonlinear factors:
• Lighting variations (including spectral reflection)
• Contour occlusion
• Non-uniform contrast changes
• Contrast reversal
• Handles 10 degree rotation angles
  Disadvantage:
• Slower if the teach image has a lot of edges

VLL Foreign Material Detection

• Foreign materials(FM)presented on the lead finger demaged the capillary and/or caused yield loss
• This feature enables customers to configure the VLL sensitivity to foreign materials presented on the lead fingers
• VLL will stop if the found FM coverage exceeds the set tolerance

• If turned OFF, no FM check will be performed