Automatic Optical Inspection
I have been involved in Automatic Optical Inspection (AOI) for close on 30 years during which time the technology has moved on dramatically, but I am still amazed, as I was then, why a system as sophisticated as this is abused by its owners. That may sound harsh judgement but I still see people using this fantastic piece of technology as a goal keeper rather than a process control tool which is its real value. Anyone who follows sport knows that a good goalkeeper is a benefit but if he is constantly having to save shots, the manager would change the defence so he has less to do. Even goalkeepers make errors. Well, this applies to AOI. How so? You may ask. Well, by use of simple statistical analysis, the faults found can be categorised, and traced back to their process source and rectified there before they are moved on to the next process. I know this sounds obvious but I assure you that there are still a lot of companies out there who do not do fault analysis on rejects. This is a waste of a fantastic resource. In my view, this is like having 10 attacking players and wondering how the team keeps losing.
We all are familiar with zero defects which is a noble aim. Some people may think it is not a reality but I always took it with a view that if you shoot for the stars you might get to the moon. The more defects that can be cured at source, the less defects the AOI is presented with equals less rework but as importantly, less chance of anything slipping through to the end user.
I know a lot of people will read this and think ”that’s not me” but just spend a moment and think, could you do a bit more fault analysis and take the pressure off the goal keeper?
Your views would be welcomed as I am still after all this time, crusading for Zero.
I will also welcome technical reports to publish on this website if you think you can add constructively to our industry.
Next month…Unless someone suggests a better topic I will be discussing surface finish on bare PCBs.
John Barry…. Johnbarry@Clarydon.com