This is a great question!
It seems that there is a prevailing attitude out there that the implementation of shot meters or other flow control systems eliminate the need for any other kind of control. And it isn't without some basis in logic. After all, shot meters and flow control systems are designed to assure that exactly the same volume of material will be applied, over the same time period, on every dispense, time-after-time-after-time…
What Happens After the Dispense Matters
This kind of control system is prevalent in sealer, adhesive, potting and encapsulation applications. What’s interesting about this type of application is the importance of what happens after the dispense. And what happens after the dispense matters – a lot.
Consider a gluing operation. Assuring that the glue stays in the right place until the parts can be brought together is just as important as the dispense itself to the outcome of the assembly process. The same applies to a hem seal process. In the case of potting and encapsulation processes, how the compound flows and its ability to fill the voids and drive out the air is perhaps even more critical than the actual dispense.
The key in each of these operations is that the viscosity of the material when it is dispensed is just as important as the dispensing process itself. And the primary determining factor of fluid viscosity is temperature.
A Good Day Gone Bad
Let’s return to that gluing operation. The results are critical to our customer, so we’ve spared no expense to assure that we have total control over our dispensing process. Our gluing system pumps from a 5 gallon pail over to a shot meter where it is metered to the gun to dispense exactly the same volume on every cycle. The variation is below our ability to measure. The gun is mounted on a small robot to assure that the dispense path, nozzle speed, and distance to the part is identical from part-to-part.
We’ve carefully selected our glue supplier for their ability to provide consistency from batch-to-batch. The glue is the color and consistency of honey. And like honey, when it is cool, it is stiff and flows very slowly. On the other hand, as it gets warm, it gets thinner and flows more freely.
It’s a beautiful morning – not a cloud in the sky! According to our night superintendent, the PM’s we did over the weekend went smoothly, and the line has been running like a watch since she came on at 11:00P. Our throughput is at 100% of rated capacity and we haven’t seen a reject all night! All is well with the world and it looks like it’s going to be a good day!
But at 1:30P we get a call from the maintenance supervisor. The line is down. It’s the glue cell again. Just before lunch, QC started to notice glue outside the bond line. It continued to get worse and then, at about 1:00P, the parts started failing for bond strength. As the reject rate skyrocketed, the only solution was to shut the line down. The weekend PM’s were suspected, but our best tech went out there and tested the robot for path tolerance, and the shot meter system for dispense volume, and both were spot-on. It’s the same batch of glue we were running last night, but QC took a sample to the lab anyway, and it tested right down the middle of the specification.
Perfect Path, Perfect Volume – 100% Scrap
What could have changed?
The reality is that the glue changed. As the sun rose in the sky and the day got warmer, the viscosity of the glue dropped. In the time between the dispensing of the glue and the joining of the parts, this lower viscosity allowed it to slump out of the glue line where it became an aesthetic defect with glue appearing on the outside of the seam. The slumped bead profile was too low to make good contact with the top half of the shell and bond strength suffered.
So, why didn’t this show up when QC tested the glue?
The fact is that most QC labs are temperature controlled. This is critical for consistent, reliable operation of their sensitive measurement tools. By the time the sample was tested, its temperature was the same as the lab – and all measurements showed up nominal.
And our nagging problem with the glue cell continues because we’re off looking for the problem in all the wrong places…