QUALITY MEASUREMENT: A Question Of Wavelength
While it is true there are differences between instruments that measure form and those that measure surface finish, both measuring processes are fundamentally the same. In both processes, a probe or stylus is moved along some ideal circular or linear path to gather data; that data is filtered or sorted in some way; and certain mathematical operations are performed on it to ascertain the desired results. The only real difference is the frequency or wavelength of the data set used in the measurement.
Understanding the role frequency plays in the measurement process and why it needs to be specified on part prints is critical to making reliable quality checks on manufactured parts. It also is important in the design process to assess the functionality of certain part parameters, such as the ability of a surface to retain oil or to resist long-term wear. Misunderstanding the role frequency plays in form and surface measurement, not knowing what is being looked for-or worse, ignoring frequency altogether-can lead to disastrous consequences.
Filtering BasicsData is filtered by wavelength or cutoff. Within the limitations of a probe or stylus tip and the device used, any part trace includes an amalgam of a nearly infinite number of different frequencies present in the path being traced. These, in turn, reflect characteristics imposed by the manufacturing process. Traditionally, this data has been divided into three categories, roughness, waviness and form.
Shorter wavelength data tends to reflect surface roughness characteristics imposed by machining operations such as turning, grinding or polishing. Waviness involves longer wavelength data and reflects conditions imposed by instabilities in the machining process, such as imbalance in a grinding wheel or worn spindle bearings. Long wavelength data tends to reflect errors such as lack of straightness in the guideways of a machine or misalignment of machine axes. These long wavelength errors are usually thought of as form characteristics such as roundness, straightness or flatness. Perfect straightness, for example, could be described as a line, or wave, with an amplitude of zero.
Filtering is simply a means by which specific wavelengths are selected for analysis, usually by excluding those wavelengths above or below the desired range. In the predigital era, analog instruments used simple electronic RC circuits for this purpose and collected only the data of interest. These days, filtering is done digitally via mathematical algorithms that preserve a complete data set for the subsequent mathematical analysis. This not only allows more sophisticated analysis and a consequent proliferation of measurement parameters, but also the retention of unselected data from the trace, making additional analysis of other wavelengths possible without having to remeasure the part.
Cutoff length is the specification of a particular wavelength or filter. Sometimes confused with the traverse length, which is the actual distance the probe moves along the part, cutoff is expressed as a numerical value, above or below fall the frequencies of interest. These, in turn, depend on the parameter to be calculated. For example, with roughness, a cutoff value of 0.8 millimeter selects wavelengths below 0.8 millimeter, while with a waviness parameter, the same cutoff or filter would select wavelengths above 0.8 millimeter.
Fundamentally, knowing which cutoff to use is as important as knowing whether a measurement is expressed in inches or millimeters.
Frequency and Surface FinishPrior to 1995, the ASME B46 Standard for surface finish stated that if a cutoff length was not specified for a surface measurement on a part print, a default cutoff of 0.8 millimeter or 0.03 inch was to be used. That, however, is no longer the case. Under the last two editions of the standard, there is no default. Cutoff length must be specified on the part print for every surface measurement.
The reason behind this change was simple: the default setting was too general to apply to many situations, yet because it was listed, designers too often took the easy route and did not bother to specify a more appropriate filter. Thus, the standards committee thought asking designers to put a cutoff on that specification would be a good way to draw attention to the need.
Unfortunately, the message has yet to sink in. Too many part prints have no cutoff specification, and rather than ask the question, too many quality engineers simply go with the default. Many do not realize the default no longer exists. Worse, rather than take the time to figure out which value they need to specify for a certain parameter-such as Ra roughness-too many designers simply look at the value specified on an earlier part, rationalize that this new part needs to be better, and simply put down a smaller number, not realizing that it may take an entirely new and costly process to create that surface.
Most often, misunderstanding the two variables of frequency and value results in parts that are over-engineered for their applications, or are more costly than they need to be. Occasionally, however, consequences are more serious.
One case study that dramatically demonstrates how costly incorrect specifications can be involves an automotive manufacturer who was producing a shaft with bearing surfaces having a surface finish specification of 10 microinches.
When production started, the shafts typically had a surface roughness of 4 to 6 microinches. Things were going fine for a year until a few shafts started seizing up on the bearings. The engineers double-checked the problem shafts and found the offending parts all had Ra values of approximately 8 to 10 microinches. Thinking that the Ra spec was even more critical than previously thought and that their manufacturing process was obviously not capable of holding it over the long term, they changed the specification to 6 microinches Ra. They also decided to add a superfinishing operation to their process to ensure that the Ra value would be brought down well below the required specification.
When engineers tested prototype parts from this process, all with an Ra of 3 to 4 microinches, almost everyone of them seized. A more thorough analysis of the parts finally revealed a waviness problem. It was not a roughness issue at all; they had simply been looking at the incorrect frequency. All the time and money invested to make the parts smoother only exacerbated the problem.
The bottom line is that there are an infinite number of frequencies on the surface of a part, and by setting the cutoff, an engineer decides what data will be looked at and what data will be dismissed. So look at the correct parameters, not just the usual parameters.
By the same token, there are no hard and fast rules about what frequency, what value, or even what parameters are right for every application. For example, both a train wheel and a tiny bone-suturing implant may specify Ra, but they probably require different cutoffs and different values. Determining what those are is part of the design and prototyping process, and to ignore or misunderstand the role of frequency in this arena is to miss an opportunity.
That tiny bone suture part, for example, may require more roughness, or even a special microscopically threaded surface pattern to help the bone fibers attach. Functionality such as this can be assessed by measuring certain characteristics with different surface finish parameters. Perhaps the best known functional parameter is Rvk, or valley depth, which is used to assess the ability of cylinder bores to retain oil for lubricity. Other functional parameters measure the potential for wear in cylinder bores by assessing reduced peak height (Rpk), kernel or core roughness depth (Rk).
There are more than 100 different surface finish parameters referenced by ASME, ISO and JIS standards, as well as a number of others particular to specific applications, industries or companies. For each, an appropriate value and cutoff must be specified.
Frequency and FormOn the form side of the measurement equation, looking at the correct frequencies is as important as it is with surface finish. Even though the process is the same-moving a probe along an ideal path-and the math and the logic are the same, the results are not yet universally standardized to the extent that they are with surface finish. There really are no fully implemented standards for things such as straightness, flatness and cylindricity. In the case of straightness, for example, there is a standards number, but no paper has yet been published.
Further exacerbating the problem is that the standards for drawings do not provide a methodology for the designer to specify which wavelengths are to be examined in a straightness measurement. In contrast to surface finish measurements where it is well defined how to specify a cutoff and which type of probe tip to be used for measurement, this is left unspecified in the realm of form measurements.
The reason is tradition. The scope of form measurement-for example, the vast difference in wavelengths considered-is much broader than for surface finish. In trying to compile uniform standards, committee members had to contend with historical or industry specific definitions for things such as straightness that, while certainly not contradictory to the logic of frequency, were very different. Machine tool makers, for example, traditionally defined straightness in terms of the process used to measure it. Things were straight when the laser interferometer and a mirror used to traverse the guideways showed them to be straight. This methodology looks at different wavelengths than the manufacturer of tiny valve parts that wants to measure straightness to be sure the valve spools do not seize in the housings they are used in.
When it came to flatness, the people who made large items, such as granite plates, also had different ideas from those who made tiny parts such as those that fit into antilock braking systems and have to have flat-sealing surfaces. That the two measurement processes are essentially identical and that only the wavelengths of interest are different-one dealing in feet, the other in fractions of an inch-has been a hard argument to make. What the tiny part guys consider long wavelength, form frequencies, the granite plate guys consider short wavelength, surface finish frequencies.
Future ImplicationsThere are de facto standards used by makers of form measuring equipment that analyze based on wavelength. And at some point, the various standards committees will achieve unanimity and provide clear guidelines. Until then, manufacturers must be careful to always specify what frequencies are important for the measurement being made.
However, after the concept is accepted that form-and-surface finish are but two parts of a whole, and that frequency is central to both, interesting possibilities arise, some of which are already in the works. A number of manufacturers are working to create measurement platforms, both hardware and software, that are common across all types of instruments. Common software, measurement methods and operational logic will allow operators to move easily from form to surface instruments without additional training, to move data from instrument to instrument, or after a single scan is done, to perform a number of different analyses on it, irrespective of instrument.
Developments such as this will reduce measurement costs, improve measurement efficiency and ultimately, improve the efficacy of the entire measurement process. Q
Tech Tips• Cutoff length must be specified on the part print for every surface measurement.
• There are an infinite number of frequencies on the surface of a part, and by setting the cutoff, an engineer decides what data will be looked at and what data will be dismissed.
• On the form side of the measurement equation, looking at the correct frequencies is as important as it is with surface finish. Even though the process is the same, and the math and the logic are the same, the results are not yet universally standardized to the extent that they are with surface finish.