Mechanical cams, once a mainstay of automation, have been largely replaced by servomotors. The reasons are easy to understand: Mechanical simplicity and system configurability are the two most common motives, but improved synchronization, easier machine maintenance, and elimination of wear parts are also frequent objectives.

This setup is typical of a VFFS machine. During the sealing operation, the mechanism travels together with the film as it is fed downward. After the sealing operation is complete, the jaws open and the mechanism returns upward to its starting position for the next cycle.

That said, even when not physically present, the influence of mechanical cams is still there. Cams continue to shape the way design engineers approach machine building and the way automation systems are programmed; designers tend to think in terms of “electronic cams” and their associated cam tables. To some extent, this makes sense. Packaging machines, for example, often require repetitive motions and in these cases, the cam paradigm works well. However, in many situations, engineers fail to take full advantage of process information already present in the system. The result? Machines are repetitive when they could be adaptive.

Cam model versus event-driven model

We will now compare the traditional cam-based model to the event-driven model with a common example from the packaging world — the vertical form-fill-seal machine (VFFS). We will then look at one new approach to achieving adaptive control, called Flex Profile.

Cam profiles based on an assumed master speed of 100 cylces/min. In this case, to achieve a sealing time of t=150 msec, sealing is defined to occur between 90° and 180° of the master cycle.

Cam model: Following the cam paradigm, the programming task might be approached by predefining the x and y motions of the mechanism via cam tables — as shown in Fig. 2. Here, the motions are defined according to our knowledge or assumptions of the process requirements. For example, we could predefine a sealing time of 150 msec and define cam profiles based on the current machine speed accordingly.

This method might be acceptable in situations where the sealing process is fairly consistent. In the case of format or process changes (such as longer bag length or increased sealing time), cam tables could be recalculated to accommodate the new requirements, although these changes could not be realized during the current machine cycle. At best, the new motions would be loaded into a memory buffer and would be available for activation only in later machine cycles. Even so, abrupt changes in film-feed speed could result in over-sealing or under-sealing.

Event-driven model: Suppose instead that the machine could monitor the sealing process — using a thermocouple mounted to the sealing surface or an external timer — and send an alert when the process is complete. The machine would then immediately open its sealing jaws and return to the start position in preparation for the next cycle. Doing so minimizes total travel, resulting in lower average velocities over the machine cycle at a given machine cycle rate. The general idea is illustrated in Fig. 3.

This type of setup requires the motion to be defined with a certain amount of flexibility built in. First, the motion profile must be able to respond to the external event itself.

Additionally, it must be able to compensate in-process to the variability implicit in any event-driven scenario. In other words, the motion plan must be able to adjust automatically regardless of when the trigger event occurs. Note that the cam approach described previously cannot realize this optimization because changes to the cam motions will only take effect in later machine cycles.