Richard Leach, professor of metrology, manufacturing metrology team, Faculty of Engineering, University of Nottingham
The adoption of optical methods for measuring surface geometry, both form and texture, in advanced manufacturing is accelerating fast. Slow and often damaging contact measurement instruments are rapidly being replaced on the shop floor by robot-mounted optical sensors that can whizz through a programmed measurement routine at alarmingly high speeds and produce vast amounts of useful data. New optical measurement instruments are added to the catalogue almost every day. Therefore, it seems incredible that there is still no specification standards infrastructure in place that allows these instruments to be calibrated, or at least only in some specialised cases. In this piece, I intend to discuss why this is the case and take a future-gazing look at how we might go about introducing a complete calibration framework.
The conventional optics and semiconductor manufacturing industries are exceptions, as they already have well-established calibration infrastructures for optical measurements of surface geometry. However, these infrastructures are less developed for many high-precision manufacturers involved in complex surface machining. Highly complex freeform surface geometries, as found in the automotive, aerospace and medical industries, mean that many of the established calibration techniques for optical surface measurements may not be directly relevant. In addition, the increased industrial uptake of additive manufacturing technologies, and correspondingly higher complexity of geometries, both form and texture, has led to new measurement challenges.
It is commonplace in many manufacturing industries to hear users expressing alarm at the comparability of optical and contact instruments for measuring surface form and texture, and concerns are often borne out in formal comparisons. In many cases, the difference between the results of optical and contact instruments can be explained after critical assessment of the measurement conditions and sample geometries; but the damage has already been done, since takeup of optical measurement instruments in many manufacturing industries has been slowed, despite the apparent growth and huge potential.
A primary reason for this lack of confidence in measuring complex surfaces is the absence of a calibration framework for optical measurement instruments, where calibration is the process of comparing a measurement result to a reference result in order to establish traceability. It is relatively simple to understand and model the physical interaction of a contact probe tip with a surface, but it is not so simple to model the equivalent optical interaction; it is a more complex physics hurdle and this is the crux of the issue.
To try to address this issue in surface texture measurement, the International Organization for Standardization (ISO) is developing a framework that attempts to simplify the problem by introducing a number of common or instrument-independent metrological characteristics, namely instrument parameters that can be determined using a suitable material measure (artefact) and procedure, that are then propagated through a measurement model to provide an estimate of measurement uncertainty. The framework only applies if certain well-defined assumptions of the measurement scenario are adhered to, but it is a solid start and will significantly enhance the kudos of optical measurement instruments in manufacturing.
So far, there is the ISO 25178 part 600 standard, which lists and defines the metrological characteristics of instruments for measuring surface topography, and the ISO technical committee 213 working group 16 is currently drafting the ISO 25178 part 700 standard, which will inform instrument users how to determine the metrological characteristics and therefore establish traceability (through calibration) of their measurements. However, this infrastructure only applies to surfaces with relatively simple geometry, so if the optical interaction with the surface is too complex, then non-linear affects mean that the simple metrological characteristics are not enough to allow a rigorous estimation of uncertainty. Part 700 will mention these non-linear effects in passing but will not address them normatively.
At the University of Nottingham, in collaboration with a number of instrument manufacturers, we are developing rigorous models that will enable prediction of the interaction of any optical measurement instrument with any surface geometry, at least in the first instance, when we assume that material effects such as translucency are negligible. We have had to develop a rigorous model of scattering from a surface (which includes effects such as multiple scattering, shadowing and surface plasmons) for combining with models of the illumination, propagation and detection processes. We will be able to use these models as virtual instruments, an approach used in contact coordinate measurement. A virtual instrument considers the various influence factors and simulates the measurement using an accurate model that mimics the real measurement process. The influence factors can be varied based on appropriate stochastic models using a Monte Carlo method, and a large number of simulated measurements can be generated for estimating the final measurement uncertainty. Early work with coherence scanning interferometry is showing promise, and we intend to move towards other instrument modalities such as imaging confocal and focus variation.
In the area of optical coordinate measurement—encompassing, for example, laser triangulation and fringe projection systems—the ISO technical committee 213 working group 16 intends to bring optical coordinate measurement instruments into the performance verification framework that has been developed for contact coordinate measurement instruments. However, performance verification—in other words, determining whether an instrument is operating according to a technical specification—is not calibration. With the exception, once again, of the optics industry, there seems to be little research into how to apply the same equivalence to calibration of such instruments. Calibration of optical coordinate measurement instruments is not being addressed by the technical committee but is clearly needed in the manufacturing industries. In contact coordinate measurement, substitution (using the instrument as a comparator and usually only in a single axis) can be applied in simple cases, and virtual instruments can be used in more complex measurement scenarios. Virtual instruments are not available for optical coordinate measurement instruments, and it is not clear how to develop them.
Again, at the University of Nottingham, we are setting out to establish a framework for optical coordinate measurement instruments. We will attempt to define a set of metrological characteristics that can, at least, be used in simple measurement scenarios. At the same time, we will look to develop virtual optical coordinate measurement instruments that can be used to estimate uncertainty with any geometry, although this step is challenging due to the large affect that surface texture and material properties have on the measurement as well as the typical optical configurations used in commercial instruments (essentially, many of them are not shift invariant, which precludes the use of conventional transfer function approaches).
Lastly, we are developing a primary optical coordinate measurement instrument that measures coordinates on a surface with direct traceability to the metre and for which the magnitude of the various influence factors can be estimated. If it proves possible to develop this instrument as well as the aforementioned virtual optical instruments, we may be able to dispense with contact measurement instruments as a necessary part of the traceability chain; indeed, at least one of the university’s collaborators has done this for simple texture measurements using coherence scanning interferometry.
If the primary instruments can be made simple enough, then it would be possible for each company to have such an instrument in the gauge room (where they now have contact instruments), then apply the metrological characteristics framework, or virtual instruments, to estimate uncertainty for their shop-floor instruments. This all-optical dystopian world may be some way off or even science fiction, but we are making progress in this direction. Of course, the sceptics, myself included, are not going to be happy until almost every step along this path has been verified by comparison to traceable contact measurements and there needs to be a community effort to make this happen. It therefore seems apt to end with a call to arms: “I invite you to share this dream and join us on this bumpy topographic ride.”
Richard Leach
Manufacturing metrology team, Faculty of Engineering, University of Nottingham
https://www.nottingham.ac.uk/research/manufacturing-metrology
An additively manufactured car roof bracket (image courtesy of BMW Group).
An example of a complex additively manufactured object.