Developments in Dimensional Metrology in X-ray Computed Tomography at NPL Wenjuan Sun and Stephen Brown 10 th May 2016 1
Possible factors influencing XCT measurements Components Influencing variables Possible effects X-ray source Detector voltage current focal spot Inherent filtration (materials thickness) Stability of focal spot exposure, Image lag pixel size, resolution linearity and efficiency Others(dynamic range etc.) spectrum shape stability of intensity affect the resolution affect the X-ray spectrum affect projection image affect image quality Mechanical parts Environmental alignment of the x-ray source, the detector and the sample stage. scale (for example, source to detector distance, source to object distance) temperature, humility, etc. deviations cause partially blurred reconstructions cause error in pixel value Software reconstruction algorithm surface determination algorithm correction algorithms: beam hardening, scattering corrections. affect dimensional measurements Operators measurement and analyses approach can significantly affect the results 2
Most frequent questions: What are the best settings? (voltage, current, average of images, etc.) Should we use physical filters or not? What are the differences of having different thickness of filters? What are the influences of sample orientation and position? What is the influence of shading correction? Linear or nonlinear correction? How should we evaluate the quality of measurements? What is the smallest feature that can be measured? 3
Influence factors: X-ray source (focal spot) Stability during scan Part of X-ray tube that consists of X-Y coils and focus coil target 20 4
Influence factors: X-ray source (focal spot) Stability during scan Material: Tungsten carbide Two-sphere reference sample Projection images Sphere diameters: Sphere 1: 0.7954 mm Sphere 2: 0.7951 mm Centre to centre distance: 0.9795 mm (k = 2) Reference sample calibrated by traceable contact CMM at NPL. Boundary
Influence factors: X-ray source (focal spot) Sphere centre drift along x-axis Cold start (without system warm up) Hot start (with system warm up) Sphere centre drift along y-axis Sphere separation (reflect variation of X-ray source along beam axis) Reference: Sun W, et al. 2015 A preliminary study of characteristics of X-ray computed tomography, Proc. of Digital Industrial Radiology and Computed Tomography (Ghent, Belgium) Flay N, et al. 2015 Investigation of the focal spot drift in cone-beam industrial X-ray computed tomography, Proc. of Digital Industrial Radiology and Computed Tomography (Ghent, Belgium) 6
Influence factors: detector Detective quantum efficiency (DQE): describes the ability of an X-ray detector to transfer the signal-to-noise ratio (SNR) from the radiation field to the resulting digital image DQE results of the scintillation detector at 70 kv (top-left), 90 kv (top-right), 120 kv (bottom-left) and 150 kv (bottom-right). lp/mm is line pair per millimetre Reference: Sun W, et al. 2016 Characterisation of a scintillation flat panel detector employed by an X-ray computed tomography system: measurements of MTF, NPS and DQE, Proc. of EUSPEN (Nottingham) 7
Spatial resolution: µm Spatial resolution: µm Spatial resolution: µm gray value y Spatial resolution Edge response function Slanted edge sample 5 0 x 10 4 50 100 150 200 pixel 1.2 1 0.8 0.6 0.4 0.2 0 data fitted curve -0.2-30 -20-10 0 10 20 30 x Line spread function 1 0.8 0.6 0.4 0.2 0 0 5 10 15 20 25 30 m Modulation transfer function (MTF) 1000 15 20 100 10 10 5 15 10 5 1 1 10 100 Magnification 0 0 200 400 600 Current:µA 0 0 100 200 300 Voltage: kv Influence of magnification Influence of current Influence of voltage Reference: Sun W, et al. 2015 A preliminary study of characteristics of X-ray computed tomography, Proc. of Digital Industrial Radiology and Computed Tomography (Ghent, Belgium) 8
Geometrical error modeling for coordinate measurements by XCT Modelling radiographic binning errors due to detector angular misalignments Modelling volumetric errors due to detector angular misalignments Ref: Ferrucci M, et al. 2016 Evaluating the effects of detector angular misalignments on simulated computed tomography data Precis. Eng. 9
Other factors considered Scattering Beam hardening Cone beam error Reconstruction algorithm Surface determination algorithm More 10
Sphere seperation/mm Scale check 1D distance verification Ball separation vs height 85.585 85.58 85.575 85.57 85.565 85.56 85.555 85.55 180 190 200 210 220 230 240 250 260 270 280 290 300 Y position/mm Traceability: Calibrated Ball bar Scale: Distance is independent on threshold Downfall: measured at a fixed height, therefore alignment issues not highlighted d 11
Ball plate 2D distance verification Ball plate manufactured from 6 mm ruby spheres, 25 in total Allows measurements between spheres in multiple directions Verification of system at different scales Alignment issues 12
Holeplate investigation Multi direction distances Bi-directional and unidirectional measurements of hole plate, immersed inside NPL cam. Multi-machine measurements Bi-directional Uni-directional 13
XCT vs Other measurement techniques Advantages Disadvantages Tactile Optical XCT Profile/surface data Can be traceable Surface data Noncontact Volumetric data (both external and internal) Noncontact Non-invasive Contact method, can damage surface External surface only Angle limitation Time consuming for large area External surface only Angle limitation Time consuming for large area Influence of some systematic errors on measurement not fully understood Systematic errors Large amount of data Influence of systematic errors on measurement not understood 14
Dimensional summary - Verification Standards Tactile CMMs ISO 10360 Optical VDI/VDE 2634 XCT VDI/VDE 2630, (ISO 10360 Part XCT under development) Performance verification External dimensions Surface roughness may affects all measurement systems Young s modulus affects tactile but not optical or XCT Optical properties may affect optical techniques but do not affect tactile or XCT Internal dimensions Surface has to be accessible for tactile and optical XCT has the ability to measure both surfaces and material interfaces, through threshold values, however, not traceable. The surface roughness of the AM parts, limit the ability to make confident dimensional measurements. 15
An example of NPL designed test artefact Corkscrew distortion during build Prismatic simple geometric structures Equivalent internal structure Allow traceability Effects of beam hardening AM capability Parallel flat side faces Include cube and cylinder Internal cube and cylinder Tooling balls External cylinders of a variety of material Conventional manufactured Manufactured from XCT compliant materials: AlSi10Mg and Al for conventional 16
Overview of dimensional tests carried out using contact CMM A Measurements of large flat areas on the AM surface. Evaluation of best fit plane Corkscrew distortion Measurement of cylinder at different heights, above the plane A : (18.5, 21, 23.5, 26, 28.5) mm Best fit circle and standard deviation of data External (and internal proposed work) comparison measurements. AM allows easy to build but, unfortunately, sometimes difficult to measure. 17
Dimensional measurements - Summary C B A Item CMM CMM- Optical CMM- XCT A(Dia) 9.991 0.079 0.005 B(Dia) 9.991 0.188 0.006 C(Dia) 9.992 0.064 0.006 A-B 68.381-0.063 0.014 A-C 50.769-0.026 0.013 Comparison of data from CMM, XCT and optical. Sphere diameter and distance Circle diameters and standard deviation (Std) B-C 57.820-0.025 0.008 All units are in mm 18.5 21.0 28.5 26.0 23.5 Position of scans CMM Std CMM - Optical Std CMM - XCT Std 18.5 14.070 0.037 0.206 0.030 0.165 0.012 21.0 14.051 0.036 0.232 0.029 0.177 0.012 23.5 14.066 0.043 0.246 0.030 0.192 0.013 26.0 14.052 0.040 0.202 0.035 0.163 0.012 28.5 14.059 0.034 0.201 0.034 0.177 0.013 18
Questions. if time permits Contact: wenjuan.sun@npl.co.uk stephen.brown@npl.co.uk
The National Measurement System delivers world-class measurement science & technology through these organisations The National Measurement System is the UK s national infrastructure of measurement Laboratories, which deliver world-class measurement science and technology through four National Measurement Institutes (NMIs): LGC, NPL the National Physical Laboratory, TUV NEL The former National Engineering Laboratory, and the National Measurement Office (NMO). 20