Posted by: Kim Wilkes
Posted on: Thursday, 31st January 2002, 7:15 AM.
1) Most video systems have to be "taught to see" defects. I.e. when installed, you must pass various levels of good and bad parts through - and designate to the system which are good, and which are bad. Keeping those "teaching" parts handy - along with some similar ones not used to calibrate the software, and using them as a test at production start may suffice. As to whether 5PPM is obtainable, it depends on the type of part and characteristic checked. Vision systems catch things like label presence very well, and things like a scratch on a shiny metal surface, not so well.
2) 5 PPM will be difficult to obtain through sampling. The ideal situation would be to eliminate the chance for the defect to occur by creating a hard countermeasure for the root cause of the defect. However, if you have already tried this - you can increase the chance of detection by having multipe inspections in the process.
3) You always hear there is a 20-25% error in human inspection. I don't know how this number was derived. I would say it depends on the nature of inspection. I.e. Is it easy to detect the defect? Is the lighting adequate? Are inspectors looking for many defects, or just one?
Two GREAT books that have helped me is "Make No Mistake" by C. Martin Hinckley, and "Zero Quality Control: Source Inspection and the Poke-yoke System" by Shigeo Shingo.
Good luck with your start-up!
Message Thread: 
| | 5 PPM? by Kim on Wednesday, 30th January 2002 |
| | Re: 5 PPM? by Kim Wilkes on Thursday, 31st January 2002 |
| | Re: 5 PPM? by Anand Jha on Friday, 1st February 2002 |
Return To Discussion ForumPost A New MessageRead the Forum Guide to Good Etiquette