To illustrate an example application of the CAItech's inspection machines, measurements were made on a small test part. This part has the typical kind of dimensions found in many inspection tasks: linear dimensions, angular dimensions, hole diameter and location, 2D true position and 3D true position. The machine used for this inspection was the Model 50L, a 2-axis model using laser triangulation to determine the location of points on the part's surfaces. Other CAItech machines capture 3D data points using confocal optics sensors. The process for extracting dimensions from a datacloud is the same regardless of the sensing method used.
Figure 1. Test Part
The part shown above is about an inch (25 mm) square and half an inch (12 mm) high. It has several lands milled in it and several holes drilled through it. One of the lands was milled at an angle, forming a vertical angled surface.
SETUP PHASE:
In inspecting a part, the operator must first indicate which dimensions are to be inspected. This involves three steps: scanning in the part's 3D data cloud, identifying the surfaces of interest on the data cloud and then relating those surfaces to the dimensions to be measured.
The steel test part was placed in a magnetic fixture that held the part accurately against mechanical stops. The fixture is mounted on a turntable that rotates such that all four sides of the part can be scanned by the machine's sensor. The part is scanned under computer control by moving the sensor's illuminator over the part, each time recording the 3D locations of surface points. Once one view of the part is scanned, the turntable rotates 90 degrees and the next view of the part is scanned.
When all four sides of the part are scanned, the 3D surface points are merged into a single composite called the datacloud. The datacloud for the test part is shown below in isometric view. Here the four views have been merged together.
Figure 2. Data Cloud
During the setup phase of CAItech machines, the raw datacloud is grouped into surface primitives such as planes and cylinders by our patented software. Later, dimensions are found from the location of these surface primitives. Shown below is the screen image of the test part after planes and holes have been converted to solid model surfaces. Just as in a Coordinate Measuring Machine (CMM), the best datapoints are near the middle of surfaces, not near the edges. Notice that the intersections where two surfaces meet have been excluded from the surfaces because datapoints taken close to intersections give poor measurements.
Figure 3. Solid Model Surfaces
The solid model can be zoomed, panned and rotated in three dimensions like the solid model images in a CAD file as shown in the images below.
Figure 4. Bigger
Figure 5. Smaller
Figure 6. Rotate
Figure 7. Adding References
Once the surfaces of interest and reference surfaces are identified, dimensioning is done by setting up relationships between the surfaces. For our test part, we'll pick the part's width as our first dimension. We click on the "distance" tool in the lower right corner of the screen, bringing up a dimension dialog box like the one below.
Figure 8. Distance Dimension
Figure 9. Width Surfaces
Figure 10. Dimension Box
Other dimensions are set up the same way. First the type of dimension is set by clicking on the desired dimension tool, bringing up a dimension dialog box. Then the surfaces associated with that dimension are selected by clicking on them, adding print values or tolerances where needed.
When we're done selecting dimensions, we get a summation of the dimensions that we want to inspect, as shown below. We have set up the Model 50L to simultaneously inspect six different dimensions on this part: four height measurements plus width and length. In general, the values of hundreds of setup dimensions can be measured with CAItechsoftware. Setup for this part took about 15 minutes.
Figure 11. Setup Summary
INSPECTION PHASE:
Once the setup phase is complete, we can inspect similarly shaped parts whose dimensions are not known. First the part is placed in the same fixture and the same orientation as the setup part. When we click on the "Inspect" button, the machine collects 3D data cloud points in much the same way as it did for the setup part. The datapoints collected for each surface are spatially averaged to find that surface's location very accurately. The surface locations, in turn, determine the part's dimensions.
The figure below shows the results of nine separate inspections taken one after the other. The black graph shown in the middle gives graphically the results of the nine inspections. Between each set of vertical lines is the data for one inspection. Each of the six dimensions is reported as a color-coded "delta" showing the difference between the part's measured dimension and nominal. The rightmost vertical band (in light gray) shows the deltas of the most recent inspection.
The vertical scale of the drawing is the percent of allowed tolerance, from +100% to -100%. Since the tolerances in this case are set at +2 thou (+50 microns) to -2 thou (-50 microns), each horizontal line represents 2 tenths (5 microns). Notice that the values for all dimensions are inside a +/- 2 tenths band from nominal. For ISO 9000 specifications, a repeatability of +/- 2 tenths is sufficient to measure a part to +/-2 thou, the required 10:1 ratio between tolerance and repeatability.
Figure 12. Inspection Results
Above the black graph is a graphical indication of the most recent inspection. Each row is a different dimension of the part. The machine operator can quickly check the latest values or deltas from nominal for any dimension. To the right of the delta column is a colored column showing the percent of the tolerance being used. If any delta exceeds its tolerance, its color changes from green to red, alerting the operator.
The time required to scan all four sides of this part and report its dimensions was 40 seconds. Most of the inspection time is spent scanning -- getting the datacloud. Once scanned, dimension values are calculated in a second or two regardless of how many there are.
No hay comentarios:
Publicar un comentario