The QC Group has taken another step to provide the highest level of customer satisfaction for scanning services. Now QC has expanded its capabilities to include chromatic white light interferometry. This technology is based on the confocal measurement principle and works using chromatic depth scanning. White light is refracted differentially depending on its wavelength in order to carry out distance measurements and to capture 3D geometry.
The technology greatly enhances QC’s commitment to serving precision tooling designers and manufacturers. Now the quality of a micro-scan can be produced for scanning highly reflective tooling inserts, transparent materials, and microstructures without prepping the part surface in any way. This improves the accuracy of the CAD model and CAD comparisons that result from the scan.
The QC Group will be providing this service at the start of 2012. |
QC Invests in Chromatic White Light Technology
Scanning Precision or Artwork – Does it Matter?
3D scanning technologies have matured to the point where virtually any object can be scanned and converted to an electronic image. With the right equipment, the right software, and some scanning experience, does it matter what type of objects a company scans? After all, much the same base knowledge is required to scan a syringe or an architectual sculpture. Then why should it matter?
Let’s assume we’re talking about the same equipment, the same accuracy, the same post-processing software, and the same modeling capabilities.
The difference in results (and it is tremendous) lies in the foundational experience the engineering group has in your product, the processes by which it was produced, and the next steps you intend to take with the 3D file. This foundational experience is far more important in a precision-manufactured component than in a scan of a human face or a piece of artwork. In fact, those experiences set up the conditions for success in the development of a precision 3D file.
That’s why all of The QC Group’s scanning investments are aimed at supporting precision manufacturing. That’s why QC’s team is selected based on industry experience and knowledge - years of tooling design, casting, molding, machining, leak testing, plating, finishing, heat treating, among other manufacturing processes.
Yes, we may occasionally scan a sculpture. But our focus is where accurate 3D scanning has the greatest positive impact – in the development of tomorrow’s precision products.
How Accurate Will the Scan Be?
That’s an important question to ask. The answer differs widely from an automotive seat to a syringe needle. Some have stated that they are receiving fully parametric models accurate to +/-.0002″ from 3D scans of plastic parts. That statement tells me just one thing. The customer didn’t receive a live 3D review of the CAD model in comparison to the scanned data at the close of his project.
Remarkable accuracies can be achieved through the scanning and modeling process – from a few tenths (.0001) on miniature machined parts to a couple thousands on some plastic parts. Accuracy becomes less with the size of the part, flexibility, methods and equipment used.
Let’s look at the steps in the scanning and modeling process to develop a realistic expectation of just how accurate the “scan” will be. Each of the following scanning and modeling steps is a contributor to accuracy or inaccuracy of the end result.
Point Capture
This step in the process is very much machine dependent. Points can be captured to +/-.003 ” (or more), or points can be captured to +/-.00004, depending on the size of the part. Sometimes these numbers are quoted as the accuracy of the resultant CAD file, but there are several steps that must take place before a CAD model is completed. It is in these post-processing steps that most inaccuracies are introduced.
Once the points are captured, post processing steps such as outlier filters and noise reduction can be performed to refine the data set. Noise reduction is generally minimized or not used – particularly in the most accurate applications.
Polygonal Model
This is a necessary step of connecting the points with polygonal surfaces. Many options are available at this stage to improve the polygons before going to surfacing. Each of them will also affect the accuracy to the captured points. Some of these processes are smoothing, spike filtering, hole filling, decimation, and noise reduction. The CAD technician must be cognizant of the end goal at all times to optimize the file while minimizing any change in accuracy. In many cases, no alterations are necessary at this stage, maintaining the accuracy of the point cloud.
Surfacing
When a NURBS surface is developed from the polygonal model, grid layout, edge positioning, and grid density become important. NURBS surfacing is a process of laying small surface “patches” onto the polygonal file. Each step is performed according to the accuracy required for the final product. NURBS surfacing is a very accurate method of electronically replicating a part. Done well, the process should affect accuracy by only a few tenths (.0001″).
Modeling the part parametrically means fitting parametric features such as planes, lines, and arcs to the polygonal model. Here, it’s important to consider design intent and manufacturing deviation to create a model that most accurately represents the ideal CAD. Parametrics don’t fit organic shapes well (such as sinks, hand-ground features in a mold, etc). These are typically the areas which add the most “inaccuracy” to the model. Most features of a precision component should be modeled “right on” except where the part feature deviates from perfect shape. Much of the mismatch of a polygonal model is attributed to how imperfect the finished part was.
As you can see, accuracy can be significantly affected by post processing steps. Knowing the accuracy of the scan is only part of the equation. In general, models for small machined components should be accurate to tenths; small plastic parts to a few thousandths, and larger castings or flexible plastic components may be accurate to .010-.050 when modeled parametrically.
In the end, expect a live 3D review of the CAD model in comparison to the raw point cloud. It will bring you to a full understanding of just how accurate your model is.
Measuring Objects in Motion
Have any of you had experience measuring x,y,z coordinates of objects in a dynamic state? This request has come up a few times: How do you capture the physical 3D coordinates of an object in motion – throughout its dynamic range such that the dimensional positions affected by mass and acceleration are taken into account?
