Automating & Educating the World’s Leading Companies

The Challenge 

A global manufacturer of plastics experienced manufacturing defects in their web extrusion processes. During the  production, voids and/or unmixed batch material would eventually cause unacceptable quality in manufactured molded parts.  The existing inspection system for identifying such defects gave instantaneous and localized indication only with no logging  capability. 

The Solution 

Data Science Automation utilized the same test fixture for off-line inspection, but replaced the instrumentation package with  NI LabVIEW and NI DAQ with more sensitive displacement sensors. The PC-centric solution offered more functionality,  accuracy and expansion capability. Software trending and analysis readily identified defects and even exposed deficiencies  in the test fixture itself prompting a redesign. The first system has been deployed off-line with plans for eventual on-line  production testing. 

This contact measurement solution was considered more cost effective than automated visual inspection. The initial phase of  this project involved possible interaction with the existing sensor for measuring micrometer-sized defects in batch samples of  plastic. Originally, two-inch wide plastic ribbon samples were passed through a digital dial feeler gauge fixture to sense any  “bumps” in the material. There were two major draw backs to this type of sensor/indicator arrangement. First, the system did not provide historical data. If an operator happened to look away for just a moment while the sensor had come in contact  with a defect they would never know it. Secondly, trying to interface with the sensor through its proprietary bus would not  allow sufficient speed to dynamically acquire the data and display with sufficient resolution.  

With the use of the NI SCXI modules coupled with a NI-PCI-6052E data acquisition board we were able to quickly develop an application which not only gave immediate indication of a defect to the operator but also displayed a history and wrote the  data in an ASCII type format for easy review in a spread sheet application. 

We replaced the proprietary sensor with a linear variable differential transformer (LVDT) with the necessary accuracy and  could mount directly in the existing fixture without any modifications. The NI SCXI-LVDT module allowed quick  integration of the sensor into the LabVIEW environment. With its eight channel capabilities on that module alone, additional  channels may be added for enhancement in the future. The operator will mount the sample in the fixture and start the application for acquisition of data. Immediate visual indication of a defect is displayed on a chart  (Figure 1) as the sample passed under the  LVDT head. The sensor is mounted in such  that a way that constant pressure keeps the  sensor tip in contact with the sample but does  not damage or deform it. For this phase the  operator would then manually place an  indication on the location of the defect for  later visual review. As data is being  acquired, analyzed and presented, it is also  being stored to disk. The operator can also  play back existing files for sample comparison if desired.

  
_{Figure 1}

During initial testing of various samples of ribbon a sine wave type indication was noticed on the chart display. After some investigation and discussion with the customer we found that the new inspection system was actually showing the variance of  the test fixture. The rolling cylinder that brings the sample in contact with the LVDT sensor was actually out of round and  that was manifesting itself in the data set. (Figure 2) Although this could eventually be taken out of the system through  software analysis it was more cost effective with the customer’s onsite resources to make a minor modification to the fixture. 

Figure 2

The current system is set up in a laboratory environment for analyzing samples off line. The next phase of  development will incorporate more intense online  analysis of the data and eventual report generation for  batch quality records. This capability is being handled  by an old version of analysis software which is  cumbersome and labor intensive to get the data in a  acceptable format for input into the analysis package.  Once the analysis is completed the analyzed data set  then has to be exported to another application and  various reports generated. This proves very labor  intensive for the operator and is also very inefficient use  of time. Incorporating some of the many advanced  analysis functions that LabVIEW is capable of will  stream line the system and free up valuable time. To  further enhance the system capability the use of the Report Generation Toolkit can be added for easy generation of professional looking reports. 

The final configuration will have the system on the production line where samples are being inspected directly down stream  of the batch mixing process. The final goal will be to integrate the inspection system into the process control system to allow  the company to better control the quality of the product. 

In Conclusion 

By use of the just the basic NI data acquisition equipment coupled in a LabVIEW environment Data Science Automation was  able to quickly demonstrate the ability to greatly enhance the customer’s current application. In addition to providing a more  accurate inspection system we were able to structure the application in a way that the customer will have the ability to  incorporate additional capabilities in the future.