A Process Trouble Shooting Example

A Process Trouble Shooting Example

When I was at Westvaco, one of the important quality constraints on our product was the cut to print registration. By this, I mean the way the print lines up with the die cut of the package. Naturally some customers had tighter tolerances than others and one customer was particularly picky about theirs.

On these boxes, we found that some jobs ran just fine but on others we had a severe oscillation in the cut to print registration. We even had a term for it: “register rocking.” As the production run progressed, the cut would move from one limit to the other. And as the press speed increased, the registration would even jump to one side and then gradually creep to the other limit before it would jump back.

Our presses were fitted with the very best electronic registration systems available in the world. At the time, these were made by the Swiss firm Bobst. We had numerous conversations with Bobst engineers and had the electronics tested over and over.

Bobst always blamed the mechanical interface which was made by our press manufacturer. Of course Bobst also made presses and they wanted us to use theirs. However the Zerand press we had was much better suited to our style of printing and cutting.

We looked into every detail of the registration system. We considered the way the photo cell detected the registration marks. We looked for slop in the mechanical linkages. We looked for errors in the control program.

The intermittent nature of the problem made it even harder to trouble shoot. Several press runs would go fine and then one would pop up were the registration would not hold under any conditions.

Eventually we found that by carefully controlling the web tension we could stabilize the rock enough to make good product. However, the setting that worked went against common good printing practice and it was difficult to get the operators to run the press under those settings because they just seemed wrong.

Once we found that the odd pressures seemed to help, we began to look at why that would make a difference. Also, we began to recognize the types of jobs that gave trouble. We found that these jobs often paired older engravings with new cutting dies. We began to form a theory that maybe the two did not fit.

Due to the nature of the printing process and the tightness of the tolerances there was not way to directly measure the two. Both the die makers and the engraving manufactures assured us the parts were made to specification. Also, the printer operators assured us that there was no way that they could be wrong.

Despite a lack of cooperation from the press operators, we designed s few experiments to see if the die was in fact the wrong size for the print. What we discovered was the print was actually shrinking slightly as it went through the press. By the time it reached the cutter, the die was too big and most of the tolerance was used up in the error. We were actually getting the press to hold a much tighter tolerance than it was designed to achieve but most of that tolerance was used up in the error of fit.

The weird pressures we had been running had actually been stretching the paper slightly to help it fit. Due to the cost of the engraving and dies involved, it took a while to get new parts made and mostly the corrections were only made to new jobs. We also had to fight the people who simply could not see the complicated concept of how the parts did not fit.

Eventually we were able to get the engravers and die makers to change their algorithms for making the parts so that they actually fit by the time the product exited the printing press. Our press speeds went up about 20% once the operators no longer had to fight to keep the alignment that simply did not exist.