Scott H. Taylor
Dimensional measurement and control systems have recently undergone significant technological advances. These systems now generally exceed what were simply the ambitious hopes of tubing manufacturers not so many years ago. These advances are partly the result of the demands of other markets --wire and cable foremost among them--also served by gauging suppliers.
In the wire industry, telecommunication markets demand a much higher level of precision than they did just a few years ago. Products that fall into Category 5, an Underwriters Laboratories rating for paired wires capable of transmitting to 100 MHz, are typical. A 24-gauge wire, approximately 0.020 in., receives an insulation of 0.0065 in. with outside diameter (OD) tolerances of ±0.0001 in. Specifications further require a minimum concentricity (minimum wall/ maximum wall) of >=95%. Almost any perceptible error in conductor centering causes unacceptable structural return losses once wires are paired. Gauging control systems for OD and wall thickness or eccentricity in wire extrusion must be able to meet or exceed these demands at production speeds to 6000 ft/min.
In many ways, medical tubing extrusion benefits directly from the wire industry's advances and is far less demanding on gauging suppliers. Even critical microbore tubing does not exceed the requirements described above and can be produced at a fraction of the speed under much better controlled circumstances.
Not long ago, resolution, the scanner light source, and the number of scans per second were key issues for tubing manufacturers. Suppliers claimed accuracy in thousandths or ten-thousandths of an inch, and manufacturers made calibration a routine maintenance or quality control function. Now, virtually all diameter scanners deliver repeatability in millionths of an inch and generally do not require recalibration.
The rotating mirror and scanning beam design of solid-state laser measurement devices produces a scanning beam velocity of 100200 ft/sec. The actual measurement time is directly proportional to the diameter of the product. The smaller the tubing OD, the faster the beam traverses the product. This high scan velocity ensures that neither product movement nor position influences the measurement as the product passes through the measurement field. Optional product guides can facilitate visual alignment and provide a place to set a calibration standard. However, using guides during extrusion is not recommended because contact surfaces can introduce drag on the tubing.
Dual-plane scanners, which were introduced in 1994, take up less than 5% of the space of earlier scanners (Figure 1). With a measurement field to 0.640 in., they can deliver a repeatability better than ±0.000006 in. at a rate of 480 scans per second per plane. For critical ovality-sensitive catheter tubing applications, their compact size allows operators to pair two dual-plane scanners offset at a 45° angle for a four-plane diameter measurement.
Measuring diameter in more than two planes significantly improves ovality measurement and monitoring. Achieving roundness is a major focus of current engineering efforts in medical tubing extrusion. The move from one to two planes decreased but did not eliminate the potential for error (Figure 2). The probability of small-diameter tubing twisting as it moves downstream to the point of diameter measurement further complicates the process. These systems generally control as well as monitor diameter. An accurate average OD measurement is critical to successful control. The transition to four planes can vastly improve the uniformity of a controlled OD.
Other processing enhancements have enabled single-scan evaluations or a form of in-line fault detection. Various averaging techniques are often used to derive an average OD for feedback control and general tolerance alarming. However, single-scan systems can now evaluate and compare each scan to the one before or to an average to identify short-term faults and activate appropriate alarms and data logging. Previous systems that only compared averages risked missing faults by averaging them in. In addition to short-term averages, single-scan measurements may be used as the data points for statistical quality reports to provide even greater accuracy.
As a result of these advances, manufacturers no longer purchase diameter measuring systems based on primary measurement functions, but rather choose based on price, reliability, and secondary functions such as quality reporting, integration with other systems, or fault detection capabilities.
Tubing manufacturers have used a number of technologies, including capacitance and backscatter gauging, for in-line measurement of wall thickness. Only ultrasonics has survived the test of time. Similarly, digital systems have replaced the early analog system designs, making what was once a tedious application routine.
A benefit of this technology for tubing extrusion is the use of multiple measurement points around a product's circumference, providing wall balance and uniformity data. Systems measure and display 4, 6, or 8 fixed points around a product's circumference. This enables an operator to quickly center the die with precision. Dedicated digital processors can measure wall thickness at rates exceeding 1000 measurements per second. High data rates allow the systems to implement sophisticated error-checking algorithms to maintain data integrity.
A major advance in ultrasonic thickness measurement is the ability to measure multilayer extrusions, which are becoming more common in medical applications. High measurement rates and sophisticated processing allow these systems to isolate specific layers to measure, display, and control. This can often be done on a per-product recipe basis.
These two mature measurement capabilities define an outside diameter/inside diameter (ID)/wall measurement, control, and data-acquisition system. A multipoint wall measurement provides the data to calculate an average wall thickness or calculated ID.
Although manufacturers continue to make gauging systems more user-friendly, the work here is largely finished. The necessary precision, repeatability, and relative ease of use already make these systems an everyday production tool, requiring no more operator effort than any other extrusion line device. The tubing extrusion environment presents some opportunities to close the loop between production and the final product to achieve more consistent results. These in-line gauging systems provide valuable diagnostic data to allow users to further fine-tune the process. More sophisticated control algorithms and integration into a paperless world present additional opportunities. The remainder of this article looks at how manufacturers can meet a few of these challenges.
Unlike those for other extruded products, tubing measurements made during the extrusion process do not accurately reflect final dimensions. Although in-line gauging precision and sophisticated controls improve dimensional repeatability during extrusion, such repeatability is only meaningful if the end product is also repeatable. Herein lies the challenge not only for gauging-system manufacturers, but more directly for tubing producers' techniques.
Depending on the material, final tubing dimensions change once the tubing relaxes after the cooling process. Polyvinyl chloride (PVC) tubing, extruded under tension between a die and puller, relaxes and grows from its actual in-line measured diameter. A 2% increase in outside diameter corresponds to an approximately 10% shrinkage in length. This manifestation of Poisson's ratio is applicable to PVC tubing that has been stressed and oriented as part of the extrusion process. Some tubing extruders relax internal stress by coiling freshly extruded tubing loosely in a rotating drum before coiling it on reels. A polyurethane catheter extruded over a mandrel may shrink during postextrusion cooling. Cooling primarily affects OD; however, a smaller change also occurs in wall thickness and therefore also affects ID.
All in-line gauging systems provide some means of adjusting the in-line measurements to account for a consistent change in the end product dimension. This adjustment, or offset, might be equivalent to a 2% increase in postextrusion OD or ID, for instance.
The method of establishing offset values therefore directly affects final product variability. Unfortunately, manufacturers often take little care in the final stage of off-line measurement. Many off-line tools, including calipers, pin gages, go/no-go pins, and machined holes and tapered pins, are subject to operator feel. Significant variation in manual operator measurement is often inadvertently added to a process that had been in control before human intervention.
Medical tubing producers have run exhaustive statistical analyses of finished tubing dimensions over periods of up to five 24-hour days to evaluate gauging system performance. They discovered that the substantial dimensional variation due to shift changes of operators or QC personnel was greater than the inherent process variability that the gauging systems were designed to improve.
It is senseless for manufacturers to invest in precision tubing line equipment and in gauging and control systems that are accurate to millionths, and then leave the end product measurement subject to a human-derived factor that is accurate only to within a few thousandths.
Fortunately, there are alternatives. A number of gage suppliers offer benchtop gages to replace traditional manual methods of off-line measurements used by both line operators and quality control. A majority of tubing extrusion operations probably have one or more of these gages available. What is generally missing is the hardware and procedural integration of bench gages into a quality control system.
Some gage suppliers have integrated off-line bench gauging into their in-line measurement and control systems both for OD alone and for complete OD/ID/wall systems. Routine off-line measurements dynamically change these offsets during production as processing conditions change. In this way, the gauging process will be self-calibrating during the extrusion process. More important, self-calibration can be done without human variability, substantially reducing product variability. Achieving this goal will require improved systems hardware and documented procedures to eliminate traditional manual measurement. Manufacturers need to approach the extrusion process as a self-contained system rather than as diverse technologies.
An accurate dimensional measurement system with a reasonably high data acquisition rate should enable manufacturers to implement advanced and automated process diagnostic techniques. Some systems already provide visual displays trending selected dimensions over time. Borrowing again from wire extrusion, it is possible that manufacturers could use such existing dimensional data for a more sophisticated diagnostic advantage. Implementing spectrum analysis of sampled dimensional data can isolate and quantify more subtle periodic variations not only from simple screw ro-tation but also from more unpredictable sources, including extruder surging, automatic speed and voltage controls, melt-flow variables, eccentric sheaves and rollers, cycling air pressure sources, periodic tension changes, and temperature cycling. Identifying the frequency and length of time of
these periodic variations will enable process engineers to track the variation source and eliminate it. The tools to accomplish this are available but have not been used for tubing extrusion. A process engineer can access an analog output, typically from 0 to 10 V dc, appropriately scaled and representative of diameter, average wall thickness, or, in some cases, a single wall-thickness point. This analog value, using an analog-to-digital (A/D) interface, can be imported into a PC with FFT (fast Fourier transform) and spectral analysis capabilities. Dimensional variation in the final product reflects the sum of periodic and random variations. In-line dimensional data can provide access to this variation.
Early measurement and control systems used fairly simplistic control strategies because of uncertainty of data quality, limited data quantity, and processor power limitations. Manufacturers typically would establish a dead zone within which the product could wander. Some algorithms were written so that the controlled dimension had to violate this dead zone for a fixed time before a corrective action was taken, in order to ensure that it was not a singular occurrence, but a trend. The goal was to avoid worsening a short-term variation. The magnitude of the short-term variation and the size of the dead zone dictated the variation range integrated into the final product.
Newer processors that use statistical process control (SPC) distinguish between and quantify short- and long-term variations. Their control algorithms ensure that the control applied continuously reduces the variation. These systems monitor the mean or average value and the standard deviation for a dimension and recognize when to correct it.
On an extrusion line, the gage's distance from the extruder limits how well a controller can perform. The gage sees a dimension that may differ from the one the extruder is currently producing. Figure 3 illustrates a worst-case scenario in which a fluctuation equals two times the travel time from the extruder to the gage. When the gage sees the dimension as too small, the extruder is producing tubing that is actually too large. Figure 4, however, demonstrates how a statistics-based controller uses algorithms to control the initial error but not the periodic fluctuations. It will continue to correct shifts in the average dimension even if they are much smaller than changes caused by periodic disturbances. Monitoring corrective actions enables these systems to self-tune, optimizing the control algorithm as the process changes.
The result is that control chart and statistical quality control (SQC) documentation will reflect a final product whose mean value consistently matches the nominal dimension. This result will be reinforced by industry's gradual replacement of conventional mechanical motorized potentiometers with internal static voltage regulators. A precise digital device that eliminates the mechanics, hysteresis, and need for a gain setting will greatly improve the speed and precision of the actual control response.
Gauging systems can play a key role in local or plantwide data collection and quality documentation. These systems have access to pertinent product parameters, including nominal dimensions and tolerances, and are the source of critical finished product data, control charting data, and SQC production reports. Gauging suppliers are pursuing other strategies to make access to such data simple and compatible with existing system architectures.
These gauging systems can archive local quality reports and control chart data automatically for storage or postprocessing. Product recipe libraries allow process engineers to define the dimensions for a given product for every production line, eliminating operator intervention.
Gauging data collection software often runs in the background in a Windows environment and complies both with dynamic data exchange and open database connectivity. This enables tubing producers to import dimensional data into other commercial spreadsheet or database programs for custom processing and reporting.
Gauging systems now provide reliable measurement, process control, and data acquisition. The challenge remains for systems to operate as part of a seamlessly integrated extrusion system. Closing the loop with benchtop quality control gauging, efficient control strategies, and transparent integration with plantwide networking will be integral parts of achieving that result.
Scott H. Taylor is regional sales manager for Zumbach Electronics Corp. (Mount Kisco, NY).