Martin,
We run cell manufacturing for electronic assemblies for PCB level products. (Circuit boards with components on).
Each cell makes a different product and within each cell is an ATE system. (Automatic Test Equipment).
Each unit is placed on the ATE and receives a verification check, (are all the bits in the right place and right value), once it has passed this, it is functionally tested, (powered up and inputs applied and outputs measured).
Now, when a unit fails, the screen shows a big "FAIL" on it and a printed ticket is issued which is attached to the unit.
This is just a bit of background to start.
Before I joined the company, all of these fails were collected daily and then delivered to the "Repair Area".
No one looked at the tickets in any great detail and it became a way of life.
What I have done over the past six months is to download the previous days failures each morning, from each ATE.
Using excel, I then plotted failure mode by date & quantity and the paretod this to find what the top (say) 10 failure modes were.
Sure enough patterns were emerging. With a few design changes the top 4 were eliminated quite quickly.
Once the data had been acquired, there were obvious popular causes for certain failure modes.
These have been anotated onto a full colour laminated layout and secured in each cell.
The operators can now quickly fix 95% of the faults themselves.
Due to the nature of the overall "process", we had set a target of 96% pass rate....we are now there in 3 out of 5 cells with the operators reapairing there own units using the data that is collected daily by myself and updated.
Problems can be seen "coming in on the radar", and eliminated before disaster strikes.
This involved a lot of data gathering, setup and configuring initially, but now the monitoring takes me about 20 minutes a day for all 3 cells.
As for the Repair Area, well....the dept has been scrapped and the guys are doing more interesting cost effective work.
Due to standard costs, any unit not working after 10 mins rework is binned.
Hope I havent rambled too much Martin and did cover some part of your question anyway.
Regards,
Chris