The CT Solution
Combining X-ray computed tomography (CT) with the design and components of industrial coordinate measuring machines (CMMs) makes it possible to achieve an accuracy that enables CT to be used in industrial coordinate metrology. The use of CT in multisensor CMMs creates additional possibilities, particularly the capability of measuring parts made from composite materials and improved accuracy of measurements on real parts. This combination makes it possible to achieve specifications in accordance with DIN EN ISO 10360. The measurement uncertainty of CMMs, with and without the use of multisensor concepts, will be compared to CT measurement by means of practical results.
By using modern multisensor CMMs, it is possible to cover almost all applications and measure work pieces with high precision. However, measuring the complete surface of parts with high point density means that, despite relatively rapid scanning processes, the acquisition of measurement points can be time consuming. For these types of applications, the use of tomography can be advantageous.
The first forays in the use of CT in industrial metrology were based on systems designed for nondestructive testing and material inspection but not for metrology. Because of the mechanical design of these systems, the X-ray components that were used, and the lack of metrology software and temperature compensation, a limited accuracy in the range of some 10 microns (µm) or more was attainable. With the limited accuracy and capabilities, only a few applications in the field of industrial metrology were possible. But it proved that tomography was a powerful tool that could lead to a new kind of CMM for rapid and complete geometrical measurement of a variety of work pieces.
X-ray Computed TomographyBy combining the CT principle with proven CMM technology, a new generation of accurate, fast CMMs were created. Systems that are optimized for different applications require the use of different X-ray components, such as an X-ray source and detector. Using X-ray tubes with relatively low voltage, for example, 130 kilovolts (kV), in combination with compact, high-resolution detectors, such as pixel size 50µm, provides systems that are well suited for measuring plastic parts up to maximum dimensions of around 200 millimeters.
By using more powerful X-ray tubes and larger detectors, one can create CMMs with extended measurement range and the capability of measuring metal parts as well.
Measuring parts that are larger than the detector requires expansion of the measurement range by raster tomography. Raster tomography means that several partial images of the object captured in each projection are combined into virtual images that are fed into the reconstruction algorithm. Also in this way, smaller objects can be measured with higher magnification, resulting in increased resolution and accuracy.
Multisensor CMMs with CTThe use of CT in multisensor CMMs extends the capability of CT significantly. Measurements of parts made from more than one material can be performed. Additionally, combining CT with other sensors can increase the achievable accuracy.
Previously, automated measurements with CT were limited to parts made from one material or from materials with similar absorption coefficients. Tomography of parts made from materials with significantly different absorption coefficients results in volume data from which the surface of the less absorbent material cannot be extracted automatically.
CT measurements on multi-material parts such as assembled plugs or connectors with metal pins in plastic housings could only be performed on the metallic components. The outer surface of the plastic housing would not be visible within the volume data sets. Therefore, a complete measurement of multi-material parts required measurements with different CMMs and a realignment of the part for the different measurements requiring multiple setups on multiple machines.
One way to solve this problem is to combine the CT sensor with other sensors on powerful multisensor CMMs with a CT sensor. Depending on the operator’s typical applications, the combination of various sensors can be helpful. If, for example, the complete shape of the housing must be measured and compared to a computer-aided design (CAD) model, the combination of CT with a scanning laser line sensor, capable of capturing surface data rapidly and with acceptable accuracy, is most useful. If some features must be measured with higher accuracy than that of the CT sensor, then the CT sensor can be combined with a tactile probe, a laser distance sensor or image-processing sensor. Additional measurements on micro parts or micro geometries may require the combination of the CT sensor with a fiber probe.
For the greatest flexibility with X-ray detection, the other sensors should be mounted on a separate axis. Thus, independent moves without interference or the risk of collision are possible. In addition, rotary tilt axes are helpful for measuring in all orientations.
Increasing Absolute AccuracyThe multisensor concept also can be used to improve the accuracy of the CT measurements by correcting systematic deviations of the CT data caused by different physical effects, for example, by interference between X-ray radiation and the work piece. This strategy can be used to improve the accuracy for tight-tolerance features while the majority of dimensions can be measured without such optimization, thereby reducing cycle time.
The correction is split into two phases:
- the calculation of a correction data set which is necessary only once for each type of part and
- the application of the correction data set to each following measurement of the same type of part. Generally, the correction of CT point clouds can be done with any sensor that is capable of measuring points on the part’s surface with a better MPE than the CT sensor.
Generation of a correction data set consists of four steps:
- Measure reference points with a precise sensor-for the first part only.
- Generate a CT point cloud from the first part.
- Calculate deviations between the reference points and CT point cloud.
- Generate and store the correction data set (automated process).
After the correction data set has been created, it can be applied automatically to each CT measurement of parts with the same nominal shape. In this way, systematic deviations caused by any effect, including beam hardening, scattering and Feldkamp-effects, can be significantly reduced.
In contrast to other approaches for the correction of systematic deviations, the auto-correction directly uses the deviations between the CT measurement of the part and reference measurements on the same part, for example, the real deviations caused by CT, including influences of the material and part geometry, are corrected. Special bodies made from the same material as the work piece do not need to be inserted into the measurement volume nor is a CAD model with idealized geometry, which is necessary for simulation tools, required.
Colloquially speaking, the term accuracy refers to everything that characterizes the precision of measured results. In coordinate metrology a distinction between specification and measurement uncertainty is made. Specification is a definition of the characteristics of a CMM in order to specify its measuring performance and define processes for checking this. Measurement uncertainty relates to processes for determining the uncertainty of measured results when measuring part features, taking all influencing factors into account.
SpecificationUntil now there was no official concept for the specification of CMMs with CT sensors. One approach was to use specifications and standards-such as MPEE, MPEPS and MPEPF-that are clearly defined in DIN EN ISO 10360 [ISO 1] and to adapt them for use in CT technology. The practical use of this approach shows that the specifications and standards can indeed be helpful to give a binding specification for the CMM with a CT sensor. This can guarantee that the device meets the specifications stated by the manufacturer.
The verification of the specification is done with spheres for MPEPF and MPEPS and a specially designed standard comprised of spheres with calibrated spatial distances for MPEE in each magnification. It must be pointed out that the given specifications must be met without autocorrection unless explicitly stated by the manufacturer.
The specification in accordance with DIN EN ISO 10360 as described gives an impression of the system’s performance under ideal circumstances, such as using cooperative materials with conducive geometries. In contrast to conventional CMM sensors, the measurement results strongly depend on the geometry and material of the work piece. Therefore, the traceability of measurements of real work pieces cannot be guaranteed by only measuring geometrical standards like spheres and ball bars. Using more complex standards might ease this difficulty, but still does not lead to certifiably traceable measurement results of real work pieces.
Currently, the only solution is to check the measurement uncertainty for the specific part or to perform additional measurements of the work piece with traceable sensors. This task can be performed most advantageously by the use of multisensor CMMs with an integrated CT sensor. With this concept, the measurements can be performed in one setup on one machine without unclamping the work piece.
Measurement UncertaintyIn addition to the well-known influences on measurement results such as the environment, operator, measurement strategies, work piece characteristics and calibration procedures, the measurement uncertainty of CT measurements is strongly influenced by X-ray-specific sensor settings such as magnification, anode current, voltage and prefiltering; CT-specific software components (reconstruction algorithms, pre-processing of projection images, corrections); and interference between work piece and X-ray radiation.
Similar effects are known from tactile and optical sensors as well. With all things considered, when compared to CT, the influences of material and geometry are typically negligible for other sensors.
It has recently been shown that the measurement uncertainty of CT measurements can be improved to a range of 5 µm to 20 µm by using optimized, stable mechanical setups. These are based on the principles of CMMs with CAA and compensation of thermal influences in combination with modern X-ray components and advanced mathematical algorithms. When comparing results between CT measurements and measurements with a tactile probe, the differences between the measurements are less than 10 µm. It should be noted that the CT results were achieved by using all CT points that can be assigned to the measured feature.
Applying a pre-selection of the CT points that contribute to the calculated results can produce a better correlation between tactile results and CT results. For this purpose, only the points in the CT point cloud that are close to the point measured with the tactile probe are selected and used for the calculations. Thus, most of the CT points that compose the biggest part of the surface and, therefore, are a main advantage of the CT technology remain unused.
Further improvement of the accuracy requires the correction of systematic deviations of the CT process. This is most effectively accomplished by using the autocorrection feature.
CT is a powerful sensor, adequate for many applications, and can be used beneficially in industrial CMMs. By using raster tomography, the application of CT in coordinate metrology can be feasibly extended to measure large work pieces with high resolution and effectively used for high-precision measurement of micro components.
Initial approaches for specifications of CMMs with CT sensors were a good starting point based on existing methods. However, a solution for achieving traceable measuring results by the use of the multisensor concept and autocorrection provides a dramatic improvement. The measurement results of real parts show that a measurement uncertainty of better than 10 µm can be reached without a multisensor concept. For applications that require higher accuracy, the measurement uncertainty can be significantly reduced to values far below 5 µm with the use of autocorrection. ndt
Ralf Christoph is president of Werth Messtechnik GmbH (Gießen, Germany). Jeff Bibee is vice president of sales and marketing, Werth Inc. (Old Saybrook, CT). For more information, call (860) 399-2445, e-mail firstname.lastname@example.org or email@example.com or visit www.werthinc.com.