Requirements for product testing vary widely within the market. Clearly, a variety of means can accomplish this task, from simple photoelectric sensors capable of evaluating a single feature to expensive custom vision systems with nearly endless capabilities, limited only by the size of one’s checkbook. In the end, however, the questions are: how much testing is necessary? And how can one minimize the costs of such testing? Let’s look at some decision points of vision sensors, and how they might relate to this continuum of testing needs from simple to complicated.
The vision sensor emerged on the market several years ago. Recently, with the participation of many vendors and an array of products in their offering, the number quickly escalated so that today there are myriad products to choose from.
From a hardware standpoint, vision sensors share common capabilities with vision systems and smart cameras. Generally speaking they are not so different and consist of imaging optics, imager, processor, I/O, firmware or software. Where they do differ is in the flexibility of the firmware or software to accomplish certain tasks.
A vision sensor has a much-abbreviated set of methods that are specific to its testing capabilities. If the function is not included from the manufacturer it is generally not possible to add the capability. On the other hand, the smart camera has a large array of functions from which to build an application with the possibility of writing external routines and hooking them into the program environment to semi-customize tasks. Finally, a vision system may be completely free-form, allowing it to be configured to specialized test and inspection needs.
Flexibility of PositionUnlike a traditional sensor, an important feature of the vision sensor is its ability to evaluate pixels; these pixels are in a region of interest rather than at a specific spatial point. Therefore, an evaluation can be made at any location within the field of view of the imager.
In this example, the component is free to move vertically or horizontally so long as the attribute we are inspecting (the c-ring and its opening) remains in the region of interest (the yellow box). Making a presence measurement of the c-ring with traditional sensors would likely be quite easy, but the ability to make that evaluation over a dimensional variation that exceeds the thickness of the c-ring would be impossible; it simply would not remain in the field of view of the sensor.
Flexibility of Number of TestsTraditional sensors do a single job very well; however, they do not typically have the latitude to make several tests, for instance, in Figure 2. Inspecting the presence of any one of the products in the wrapper is easy, but inspecting all six using traditional sensors either requires six sensors connected in some logic and configuration, or an elaborate indexing and control mechanism to carry out six separate tests and some storage mechanism to tabulate the index vs. result for each position and logically connect them together at the end of the steps.
For the vision sensor, there also is no requirement that the method used in the first evaluation be repeated. One could just as easily asked for an inspection of printing and sealing components in the same field of view as the count. All can be accomplished in a single test so long as the vision sensor can see all the attributes that it needs to inspect.
A great example of multiple checks is a milk carton. During production, several attributes need to be checked, including seal integrity, printed date and lot coding, and cap integrity.
It is easy for the vision sensor to locate the part feature of a specific check, such as the cap, and test for its presence then move to the date code and the sealed edge.
Flexibility of Inspection TechniqueAs noted earlier, in most cases, a photoelectric sensor comes out of the box ready to use. There may be some set up, certainly alignment, and also perhaps some sensitivity adjustment, which may be mechanical or programmed via pushbutton. Outside of minor adjustments, when it comes to detecting a single aspect of an object, the capability of a photoelectric sensor cannot be matched by vision sensors.
In the case of the vision sensor, flexibility lies in its ease of use. The ability to vary the test method is generally accomplished via a program, a graphical user interface that facilitates customizing the application to achieve the test result. Once the operator has set up the program, it is downloaded to the vision sensor and the vision sensor carries out the set of instructions.
There are a variety of ways to accomplish this. Figure 3 is an example of the programming environment for one type of sensor. It is a single screen divided into sections for operation, configuration, results, setup and display. Some manufacturers of vision sensors compartmentalize the user interface into a step-by-step wizard. In either case, under the umbrella of ease of use, the functionality of the program environment is huge, albeit fixed to certain methods.
Ancillary EquipmentAs with any vision system, the need to appropriately illuminate the subject matter in order to achieve unambiguous results is paramount in the setup. Considerations for interchangeable optics, if available, and external lighting to maximize image contrast and minimizing ambient lighting variations are real considerations that need to be factored into one’s decision and budget.
The vision sensor is a great addition to the tool box of industrial process engineers and technicians; it brings new capability to error-proofing in a variety of production environments from automotive to consumer packaged goods.
Pepperl + Fuchs Inc.