What You Can Learn from Shot Profiles

Producing good parts from every shot while keeping cycle time to a bare minimum is the goal for most injection molders. To get there, you have to master all the complex interactions between the many process variables that influence part quality. When these interactions aren't well understood, machine operators may become locked into an endless cycle of incremental control adjustments, accompanied by an unacceptably high proportion of bad parts. If the process interactions are understood, then logical, reproducible, and predictable results will follow. This is where process-monitoring systems come in.

With proper sensors and control signals, a rich variety of revealing process data can be acquired automatically from any type of molding machine. To maintain the highest part quality, it is important to immediately recognize when any key process variable, such as area under the pressure curve, goes out of control. But that knowledge is not enough. Finding root causes and making corrections requires knowing as much as possible about all the other factors that influence that key variable.

The charts and graphs reviewed here illustrate a few of the problems revealed by shot profiles--which consist of screw position and velocity along with hydraulic and cavity pressure stacked over a common true axis. All are real-world examples furnished by Vision Plastics of Wilsonville, Ore.; Triquest Precision Plastics of Vancouver, Wash.; and Statistical Plastics Corp. of Germantown, Wis. The profiles were acquired with Shotscope monitoring systems. This system can record as many as eight dynamic channels of information on each shot, in addition to 16 thermocouple inputs and 12 digital input "events." From analysis of these inputs, as many as 50 process parameters can be stored automatically per shot.

1 Correcting Cold Slugs

A common processing problem easily spotted in a shot profile is a cold slug in the nozzle. When a nozzle gets cold, process variations follow, and part quality suffers. Figure 1A and Figure 1B show "before" and "after" profiles.

Nozzles can freeze because of improper control settings, poor or malfunctioning temperature controllers, bad thermocouples, prolonged delays by operators removing parts, stuck parts, cold ambient air, or any process interruption that keeps the machine from cycling at a consistent rate.

Figure 1A overlays profiles from two different shots. The blue trace represents a shot where a large cold slug was present. The yellow trace shows a shot with a smaller cold slug. The identifying characteristics of cold slugs are flat spots on the screw-position curve, near-zero velocity on the velocity curve well before transfer, and high sustained hydraulic pressure well in advance of transfer. The hydraulic pressure will usually be "maxed-out"--that is, up against the permissible machine pressure settings--while the screw is trying to break through the frozen nozzle. After breakthrough is achieved, the screw again moves quickly. But by this time the part quality will be compromised by over-packing or under-packing. Note that between these two traces there is a large difference in transfer position (red vertical lines), which underscores the magnitude of the undesirable process variability.

Figure 1B shows the same machine running the same mold after the nozzle temperature was increased to eliminate the cold slug. Again, two shot traces overlaid, but it is almost impossible to distinguish them. There is now uniform screw movement and velocity, a lower peak pressure, and a repeatable transfer position. Parts made under this set-up are all identical.

Another way to look at process phenomena is with a "run chart," which plots the value of a particular process variable over numerous shots. Such charts help to spot trends and observe the results of process changes. For our cold-slug example, Figure 1C shows peak hydraulic pressure that starts out high and wildly variable. This pressure behavior was due to the high force required to break through the cold slugs. The plot shows about 600 shots. At about the mid-point in the chart, the pressure drops and the variation is greatly reduced, indicating the absence of cold slugs. In this case, the culprit causing the frozen nozzle was a faulty thermocouple near the nozzle. It provided incorrect information to the temperature controller, which in turn wasn't heating the nozzle when needed. After the thermocouple was replaced, the problem was immediately cured.

2 Hard-to-hit transfer position

The shot profiles shown Figure 2A and Figure 2B are classic examples of the usefulness of profile anal- ysis. Figure 2A shows a problem when molding acetal on a 110-ton machine. Profiles of two shots are overlaid. Variations in hold times are evident, and a little more analysis reveals further problems.

Generally, the transfer position is set to occur when the cavities are nearly full and when the screw's forward movement nears zero. Final cushion position is normally achieved a short time later, after hold pressure has had a chance to pack the part. Hydraulic pressure is typically lowered slightly after transfer to reduce the possibility of flashing.

Figure 2A shows things weren't working this way. The left red vertical line marks the point where transfer is expected, about 1 sec into the shot. The screw has almost stopped its forward movement and the velocity is near zero. The hydraulic-pressure trace should have dropped to approximately 500 psi, but it remained at about 650 psi, which was the maximum pressure setting. This situation indicates that the machine hasn't transferred to pressure control and is still trying to control by velocity. (By the way, the twin peaks in the pressure curve at the start of the shot are characteristics of this particular machine when very rapid screw acceleration occurs. The hydraulic pressure overshoots the specified 650-psi upper pressure limit.)

The other two vertical red lines indicate where actual transfer is occurring for the two shots recorded. The difference between expected transfer time and the shot shown in blue is about 1 sec, roughly twice what it should have been. During this extra fill time, elevated pressure was exerted on the melt for a prolonged period, causing flash and over-packing. The shot represented by the yellow trace wasn't quite so bad but still suffered from an unacceptably long fill time that caused inconsistent part sizes.

What caused all this process variation? The transfer wasn't happening when it should, suggesting that the machine was "fighting" to achieve transfer. It kept applying all the pressure it could until the screw position finally reached the pre-set transfer position. In essence, the machine was packing the part under velocity control rather than pressure control. The cause was that the transfer point was set too close to the cushion position.

The problem was corrected by slightly increasing the transfer position. This strategy allowed transfer to be reached just as the screw slowed down, thus avoiding prolonged high pressure under velocity control. Figure 2B shows two shot traces after the process adjustment was made. Transfer always happens at the same time, and there is no evidence of inconsistent hold times.

3 An 'incapable' machine

A machine is not "process capable" if it cannot repeatedly produce good parts, even though the settings are "correct." Some machines may be incapable under some circumstances and capable under others.

Figure 3A shows two shots made on a 90-ton press running glass-filled polycarbonate. Gate blush resulted in an unacceptably high percentage of rejects. Significant indicators in this profile include the "crook" in the position trace and the corresponding low velocity, as well as the variability between the two shots, as shown by the blue and yellow traces. Low velocity during the early fill stage is an intentional process setting, not the result of a cold slug as might be suspected. Rather, the gate blush appeared to be related to the process variation. After some experimentation and observation of shot profiles, the molder discovered that the variability could be eliminated by increasing the average velocity during initial fill from about 1.0 in./sec to about 2.0 in./sec.

Figure 3B shows the profile for two shots after the fill-speed increase. Shot-to-shot variability disappeared, and so did the gate blush. This machine wasn't able to maintain the velocity setting when set below 2 in./sec. However, above that setpoint it proved capable.

4 Case of the vanishing cushion

Short shots, sinks, part-weight variation, and war- page can be caused by insufficient cushion length.

The hold phase can't do its work without an ample cushion, and if cushion length reaches zero you are guaranteed to make bad parts. So whenever cushion length changes, it's time to take action.

Figure 4A shows an alarming decline in cushion length over the hold period. This parameter should remain steady and level. A simple fix might be just to increase shot size to compensate for the diminished cushion, but this strategy amounts to treating the symptom and not the cause.

As the cushion gets smaller during hold, where is the material going? The cavity is already filled and packed, and the gates are frozen. The probable answer is that material is going back into the barrel. The check ring isn't doing its job and has permitted the melt to re-enter the barrel, allowing the screw tip to continue to move forward. While hydraulic pressure is maintained on the screw, constant pressure is not maintained on the melt and the cavity. Undersized and under-packed parts result. After replacing the bad check ring, Figure 4B shows that cushion length stays constant throughout the hold period.

5 Inconsistent screw starting position

Too large or too small a cushion both threaten optimal process control. So do shot-to-shot variations in cushion length. Figure 5 shows profiles of two different shots. The blue and yellow traces illustrate inconsistent screw starting positions and variations in cushion length from shot to shot. The profiles come from a year-old 100-ton machine running a hot-runner tool and producing 4.7-g polycarbonate parts. Note the time and position scales. This whole shot profile lasts only 1.2 sec and the screw moves only about 0.70 in., quite small for a machine this size. The complete cycle takes about 12 sec, with a relatively long recovery and cooling phase compared with the injection time.

What causes the 0.10-in. differences in starting positions? At first, the molder suspected that it was simply a matter of inconsistent decompression--or "suck-back." Yet, if variable suck-back were the cause, then the cushion length would have been the same from shot to shot. When the shot begins, a shot with the excessive suck-back pushes only air initially. But, in the end it should purge the air and ultimately reach the same cushion length as the shot with less suck-back.

Because the cushion length is too small for the shot with the lower starting position and too large for the shot with the higher starting position, the profile suggests that there is additional material in front of the screw at the higher starting position. Where did the material come from and how did it get there? During suck-back, a variable amount of melt got in front of the screw by passing through the check ring. The variations in starting position and cushion length were eliminated after the decompression setting was reduced.

6 Controlling cavity pressure with hydraulic pressure

In this case, a large computer OEM required the molder to continuously monitor mold A and B temperatures and peak cavity pressure on a reinforced PET part. Cavity-pressure measurements from a sensor mounted behind an ejector pin had to fall within prescribed lower and upper limits in order to ensure consistent part strength.

While the molder's process-monitoring system could easily obtain the desired data, the machine controller on the five-year-old 120-ton press could not control injection directly from cavity sensor input. So the molder decided to use hydraulic pressure to control cavity pressure.

Figure 6 shows the relationship between peak hydraulic pressure and peak cavity pressure, and the custom molder's experience in maintaining acceptable peak cavity pressure. The left two-thirds of the run chart shows wide fluctuations in both hydraulic and cavity pressures. Acceptable peak cavity pressure could not be maintained by controlling peak hydraulic pressure.

Realizing that more precise pressure control could be accomplished by using hold-stage, or "post transfer," pressure control, rather than "pre-transfer" hydraulic pressure during fill (which is determined by the velocity settings), the molder decided to bump up the transfer point from 4 mm to 12 mm. Transfer from velocity control to pressure control thus occurred much earlier in the shot and at a lower peak hydraulic pressure.

The right third of the run charts shows the pressures after the transfer-point was changed to occur earlier in the cycle. Peak cavity pressure became significantly less erratic and stayed within its limits.

7 The case for robots

In a variation on the frozen-nozzle theme, consi-der the use of an operator versus a robot for extracting and loading parts during insert molding. Figure 7A and Figure 7B illustrate a case in which an operator was replaced by a robot, not for labor-cost savings, but for the sake of quality improvement and scrap reduction.

When an operator manually loaded inserts and unloaded parts, minor delays routinely caused the press to stay open too long and the nozzle to freeze. The hydraulic-pressure traces in Figure 7A show a rapid increase in pressure as the controller tries to achieve the proper shot velocity. This figure includes two shot traces, indicated in blue and yellow. A breakthrough is finally achieved at about 1.200 sec into the shot. At this time, shot velocity shoots up well past its desired limit. At no time during the shot does a uniform filling velocity occur. As the cavity fills, pressure again builds up until transfer is reached.

The molder then installed a robot (Figure 7B). Two shot two traces are shown, but it's difficult to discern them because they are virtually on top of one another, showing that the process is quite stable. The velocity trace shows a nice flat spot of uniform velocity at a satisfactory level. The double spikes on the hydraulic-pressure trace are gone, replaced by a single increasing pressure trace that builds uniformly as the cavities fill and transfer is reached. The robot's unvarying speed and high degree of accuracy eliminated processing inconsistencies that compromised quality.