1 / 23

Preliminary Application of Principal Components Analysis (PCA) to a Microchip Test Process for MS/RF Test Data.

Preliminary Application of Principal Components Analysis (PCA) to a Microchip Test Process for MS/RF Test Data. APACT Conference ’03 York. Personnel. Introduction. Dynamic test process Batchwise testing < cycle times > testers, handlers Many constituent vars. Continuous / discrete data

yale
Download Presentation

Preliminary Application of Principal Components Analysis (PCA) to a Microchip Test Process for MS/RF Test Data.

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Preliminary Application of Principal Components Analysis (PCA) to a Microchip Test Process for MS/RF Test Data. APACT Conference ’03 York

  2. Personnel

  3. Introduction • Dynamic test process • Batchwise testing • < cycle times • > testers, handlers • Many constituent vars. • Continuous / discrete data • Non-standard formats, dB extraction • Dependencies • Lot, batches, wafer • Platform & tester

  4. Philosophy ‘Real World’ Methodologies: • SPC • FDI • R2R (RbR) • PCA A combination of these can be used to analyse and model a process for the purpose of quality improvement.

  5. Quality Quality can be broken down into: • Product • ‘conformance to requirements’ • Fitness for use • Process • Monitor, control and minimise variation(s) for a given process One is controllable

  6. Variation • Chance • Natural variation inherent in a process. Cumulative effect of many small, unavoidable causes. • Assignable • Variations in raw material, machine tools, mechanical failure and human error. These are accountable circumstances and are normally larger. These both generate process variation.

  7. The Process Microelectronic Division • µelectronic product test facility • Pass or Fail • Yield based on test process & products • Multiple Device Under Test (DUT’s) • Multiple test cells • Large product base

  8. Test Cell • Handler • Device Under Test (DUT) • Device Interface Board (DIB) • DIB Socket • Picker • Tester • Test Program • Pass/Fail is function of imposed limits • Of greater interest is test variation • Common Infrastructure

  9. Data Reduction • Principal Components Analysis (PCA) • Linear data reduction technique • Explain as much variation as possible in as few components • Decorrelate vars. Transforms m correlated vars. Into m new vars. These are uncorrelated Using a matrix of Eigenvectors Transformed vars. are the PC’s of The ith PC is

  10. PCA • Reduction of high volume data sets • Generates combinations that describe the process • 1st Principal Component (PC) accounts for max variance • Succeeding PC’s account for remaining variance • Too few – poor model, incomplete representation of the process • Too many – over parameterised, includes noise

  11. PCA • z -m dimension projected vector • U -PCA projection vectors • x -original, d dimension data vector • m < d, usually m << d

  12. PCA Where R -Correlation matrix  -Eigenvalue I -Identity matrix v -Eigenvector

  13. PCA • Eigenvalues (λ) are variance of the original components • 1st PC has largestλ • 2nd PC has 2nd largest λ etc. • How many components? • Disregard λ < 1 • Scree plots • Subjective process

  14. PCA • Latent vars can sometimes be interpreted as measures of physical characteristics of a process i.e., temp, pressure. • Var reduction can increase the sensitivity of a control scheme to assignable causes • However, PCA as a process monitoring scheme can not always detect process mean shifts

  15. PCA • The application of PCA to SPC / R2R monitoring is increasing • Start with a reference set defining normal operation conditions, look for assignable causes • Generate a set of indicator variables that best describe the dynamics of the process • PCA is sensitive to data types

  16. Results • From a controlled experiment, data were extracted from a MS/RF testing cell • Matlab • Data distributions • Descriptive stats. (mean, std, skew) • Parameter plots • Pass / Fail • Data reduction (PCA)

  17. Results

  18. Results

  19. Results

  20. Results

  21. Results

  22. Results • 178 columns of variables • Approx. 1000 cycles • Dissimilar data • Approx. 50% of variance can be explained through the initial 10 PC’s • Scree-plot showing similar trend throughout

  23. Conclusion • Results suggest possible candidate indicators for process monitoring • Reduction in data volume helps analysis • A robust model is important • A PCA model will be applied to production data in the form of a statistical monitoring scheme, SPC & R2R

More Related