System and method for sensing variations in a strand

The invention is an inspection method and system for detecting flaws in a cylindrical strand.

Systems for optical inspection of strands, such as thread, optical fiber and polymer filaments are known. One such sensing head utilizing three noncolinear light beams is marketed by Takikawa Engineering of Tokyo, Japan. In this system the three light beams are detected after being incident on the strand being monitored and each signal is monitored to produce an alarm signal if the monitored signal exceeds a threshold indicating the presence of an unacceptably large fault. For demanding, high tolerance applications, such as sutures to be used in heart operations, systems of greater sensitivity are desired.

The disclosed invention is an optical inspection system for effectively detecting defects in cylindrical strands, such as monofilament surgical suture materials. The system described is capable of detecting defects such as: nicks, bumps, scrapes, abrupt neck-downs, splits, frays, contamination (dust), as well as internal voids and color variation.

One objective of this system is to provide a highly reliable method of high-speed continuous detection of micron-sized defects in monofilament surgical suture material. Resulting benefits associated with this system include improved product consistency and quality, reduced product costs, and facilitating integration with collateral automation processes, such as in-line suture annealing and automated cutting. The exemplary optical inspection system discussed here utilizes a three-beam LED scanning head and digital signal processing systems.

The first subsystem is a three-beam infrared light emitting diode scanning head. For this exemplary application, this subsystem was set up to detect surface anomalies in the range of from about 10 uM (0.0004″) to about 25 uM (0.001″). The unit was mounted with the (3) optical beams normal to the thread path, approximately equally spaced circumferentially about the strand's axis.

The Digital Signal Processing (DSP) subsystem unit comprises application specific software and data collection hardware. The software functions as an operator interface, controls the application, collects and interprets data, digitally processes analog sensor signals through application of mathematical algorithms, and displays information. Data collection is accomplished through computer-mounted, commercially available data collection cards, such as Item #NI DAQ, available from Labcon, Corp. of San Diego, Calif. An operational overview of the DSP is as follows:

Data from each axis of the three-beam infrared light emitting diode scanning head is consolidated into a single composite signal, reducing signal noise. This results in tighter detection thresholds. Signal symmetry is adjusted about the zero axis by mathematically removing all DC components from the signal. This centers the signal around the zero axis and allows symmetrical positioning of positive and negative thresholds. Software selectable detection thresholds are established above the signal noise base. Signal peaks above these thresholds trigger defect outputs. A sensor failure detection function is incorporated to trigger shut down and alarm if sensor input is lost. Optional surface roughness tracking, as well as Statistical Process Control (SPC) charts can be incorporated. The system can be configured network-ready for remote data collection. Outputs are configured for Programmable Logic Controllers (PLC) and Robot control with a 24VDC logic interface.

The increased sensitivity and reproducibility is achieved, primarily, through synchronizing the outputs of the three axes and summing them to produce a single monitoring output. Synchronization is done mechanically, electronically, or by a combination of the techniques. The mechanical synchronization is accomplished through placement of one or more shims in the mounts of either the light emitting diodes or the sensors detecting the optical signals such that the three beams are illuminating the same section of the strand. In an exemplary system this adjustment produced beam coincidence to within approximately 0.002 inches.

Synchronization can be done electronically by placing an electronic time delay device in each signal leg to adjust the signal paths to within approximately 30 microseconds. A combination of these two techniques can be applied, for example, by utilizing the mechanical technique during initial system set up and the electronic technique to correct system drift detected during periodic alignment checks.

The system prototype was tested on a servo driven test strand, which re-circulates a continuous loop of suture material with known defects, through all sensor units. The purpose of the test strand set up was to validate repeatability of all system components.

By combining a system of sensors with sophisticated data collection and signal processing software, a broadened spectrum of potential suture defects are detectable, both in defect type and size. The system is capable of detecting external defects, typically, but not limited to: nicks, bumps, scrapes, abrupt neck-downs, splits, frays, contamination (dust), as well as internal defects, typically voids and color variation. The system has shown to meet or exceed sensitivity equivalence with human tactile and visual capabilities for critical suture inspections, while additionally providing statistical process control and repeatability. Variations of the system are applicable to the wire, textile and fiber optic industries.

For a more complete understanding of the present invention, reference is made to the following detailed description of exemplary embodiments considered in conjunction with the accompanying drawings, in which:

FIGS. 1 a and 1 b are elevational views which schematically illustrate the passage of a fault in front of the two slits in the mask covering a sensor.

FIGS. 2 a and 2 b are top views, in section, showing the two sensing elements in a sensor and how the illumination of each of element changes as a fault passes in front of the two slits.

FIG. 3 is a perspective view of a portion of a production system, showing a strand passing through a sensing head.

FIG. 4 is an elevational view of a sensing head showing three non-colinear light beams and details of an exemplary sensor and an exemplary emitter.

FIGS. 5 a and 5 b are sets of curves showing the inspection signals from the three sensors before ( 5 a ) and after ( 5 b ) synchronization.

FIG. 6 is a schematic representation of the signal processing portion of an exemplary system of the invention.

FIGS. 1 a, 1 b, 2 a and 2 b illustrate how defects can be detected in a continuous strand ( 10 ) by interposing the strand ( 10 ) between a light source, such as an LED and a pair of sensing elements ( 5 , 6 ) shielded by mask ( 7 ). The mask ( 7 ) has a pair of parallel slits ( 8 , 9 ), e.g., about 100 micrometers wide, delimiting the light ( 13 ) projected therethrough.

When a uniform portion ( 11 ) of the strand ( 10 ) is passed in front of the parallel slits ( 8 , 9 ) covering light sensing elements ( 5 , 6 ), as illustrated in FIGS. 1 a, 1 b, the difference in the amount of light received by sensing elements ( 5 , 6 ) is null. When a lump or bump ( 12 ), i.e., a flaw, is passed across the slits ( 8 , 9 ), the amount of light received by sensing elements ( 5 , 6 ) becomes unbalanced. The difference is in proportion to the size of the flaw 12 , i.e, lump/bump, in the strand ( 10 ). When this difference exceeds a preset value, an alarm signal is output.

If this type of detection apparatus is used for primary or secondary glass fiber or other translucent or transparent materials, the system can also detect small internal inclusions, bubbles and cracks. Light ( 13 ) penetrates through the material, but internal defects interrupt the light pattern due to refractions in the glass. This change causes a difference in the amount of light received by the sensing elements ( 5 , 6 ) which causes an alarm output.

FIG. 2 a shows a sensor ( 4 ) with two sensing elements ( 5 , 6 ) behind a mask ( 7 ) defining two slits ( 8 , 9 ). The strand being inspected ( 10 ) has a uniform section ( 11 ) and a bump fault ( 12 ). A light beam ( 13 ) illuminates the strand ( 10 ) and the slits ( 8 , 9 ) distal of the strand ( 10 ). In the figure, the uniform section ( 11 ) of the strand is in front of both slits ( 8 , 9 ) intercepting equal amounts of light from the light beam ( 13 ), producing equal signals in the sensing elements ( 5 , 6 ) and a null in the external circuitry. In FIG. 2 b, the fault ( 12 ) has progressed in the direction of the arrow ( 14 ) and changes the amount of light falling on sensing element ( 5 ) through slit ( 8 ), producing an unbalance signal in the external circuitry and a fault signal.

FIG. 3 shows a portion ( 15 ) of a production system in which a strand ( 10 ) being produced and monitored is fed from a feed head ( 17 ) and passed through a sensing head ( 18 ) as it proceeds in the direction indicated by the arrow ( 19 ) through the remainder of the production system. An automated production system might include a counting wheel to correlate a fault signal with a position on the strand so that the fault could be excised by automated cutting equipment. If the cutting equipment is set to cut the strand to preset lengths, data from the counting wheel would identify the particular length possessing the fault, which could be automatically rejected. The production system could also store the location of the fault in a memory for later use in excising the fault.

The exemplary sensing head ( 18 ) illustrated in more detail in FIG. 4 incorporates three non-colinear light beams generated by three emitters ( 21 ) and detected by three sensors ( 22 ). The light beams are disposed with approximately equal angular separation. In an exemplary device, near infrared LEDs operating at a wave length of approximately 720 nm were used. In order to accomplish mechanical synchronization, a spacer or “shim” ( 23 ) is used to shift the sensor ( 21 ) in the axial direction. In this exemplary system, it was determined that a shift of 0.001 inches (one mil) produced approximately a twenty microsecond synchronization shift during the calibration procedure. The arrows ( 19 ) indicate the direction of motion of the strand ( 16 ), as illustrated in FIG. 3 .

The sensor ( 22 ) is covered by a mask ( 24 ) defining two slits ( 25 ). As illustrated, the slits ( 25 ) are perpendicular to the direction of motion ( 19 ) of the strand, perpendicular to the corresponding light beam ( 20 ) and parallel to the measurement plane, defined by the broad surface ( 26 ) of the sensing head ( 18 ). An exemplary system employed 100 micrometer wide slits separated by approximately three millimeters.

FIG. 5 shows oscilloscope traces of inspection signals produced by the three sensors ( 22 ), labeled channels 1 , 2 , and 3 observing a sixteen micrometer deep groove in a test pin. FIG. 5 a, taken before mechanical synchronization, shows a maximum offset of approximately 200 microseconds in the three fault signals. After placement of appropriate shims, FIG. 5 b shows the channels synchronized to within 30 microseconds. With this degree of synchronization, addition of the three inspection signals produces reinforcement of the fault signals and averaging of the surrounding noise signals.

FIG. 6 is a block diagram schematically representing the signal processing elements of an exemplary system of the invention. Each sensor (e.g., 4 of FIG. 2 ) produces two sensing signals ( 26 ) that are passed into a comparator ( 27 ). Each comparator ( 27 ) is balanced to a null at the noise level when a uniform section of the strand passes across the two slits of the sensor. Each sensing signal is produced by one of the sensing elements ( 5 , 6 ) behind one of the slits ( 8 , 9 ). When a fault ( 12 ) in the strand passes in front of one of the slits ( 8 ) the fault intercepts a different amount of light than the uniform part of the strand and an unbalance is produced in the comparator ( 27 ), resulting in a fault condition in the inspection signal at the output ( 28 ) of each comparator ( 27 ).

If the fault is asymmetric, such as a lump or nick on one side of the strand, the inspection signals may not be equal. However, synchronization of the inspection signals assures that when combined in the adder ( 29 ), the resulting monitoring signal ( 30 ) accurately reflects the magnitude of the fault. It should be appreciated that many flaws will cause a signal variation in at least two of the sensing elements ( 5 , 6 ). For example, a necked-down portion of the strand ( 16 ) that is directly sensed (in profile) by a first sensing element, e.g., ( 5 ), may be indirectly sensed by a second sensing element, e.g., ( 6 ), due to a greater eight transmissivity of the necked-down portion. This cumulative effect enhances the sensitivity of the present invention due to the aforesaid sensing upon a uniform section ( 11 ) of the strand ( 10 ) and adding the individual signals. The monitor ( 31 ) then compares the monitoring signal ( 30 ) to a preselected fault threshold, which produces a fault signal ( 32 ). The fault signal ( 32 ) is carried by a fault signal transmitter ( 33 ) to the strand production system ( 34 ).

The production system ( 34 ) either includes an automated cutter adapted to excise and discard the section of strand that includes the fault or a counting wheel (or other mechanical locater device) with an electronic output that feeds memory that records the location of the fault for later processing to excise the fault.

As an alternative, or in addition to, mechanical synchronization, the inspection signals can be synchronized by insertion of a time delay device ( 35 ) in each channel to synchronize the signals passing into the adder ( 29 ). It may be efficient, for example, to mechanically synchronize the sensor head ( 18 ) during initial system set up or periodic major overhaul and trim the synchronization electronically during daily or weekly recalibration.

It must be realized that objectives of this invention can be accomplished in many ways employing the fundamental synchronization and addition teaching disclosed herein. Further, that as used herein, the term “cylindrical” is used in the broadest sense and includes the linear translation of any regular closed geometric figure, such as a circle, square or hexagon. The individual functional elements are all well known in the art. The signal processing and logic can be accomplished through analog or digital methods, as desired by the system developer.

It will be understood that the embodiments described herein are merely exemplary and that a person skilled in the art may make many variations and modifications without departing from the spirit and scope of the invention. All such variations and modifications are intended to be included within the scope of the invention as defined in the appended claims.