CNC Milling Surface Waviness vs Roughness Separation: Measurement Standards for Bearing Races and Seal Surfaces

Content Menu

● The Evolution of Surface Integrity in Precision Machining

● The Technical Anatomy of Surface Texture

● Measurement Standards and the ISO 21920 Transition

● Bearing Races: When Waviness Becomes Audible

● Seal Surfaces: The Hidden Pumping Effect

● Detailed Analysis of Tooling and Parameters

● The Metrology Lab vs. The Shop Floor

● Practical Examples: Troubleshooting the Surface

● The Future: AI and Real-Time Surface Control

● Detailed Conclusion

 

The Evolution of Surface Integrity in Precision Machining

When you stand on a shop floor listening to the rhythmic hum of a three-axis CNC mill, it is easy to focus on the visible aspects of production: the cycle time, the chip color, or the overall dimensions of the part. However, for those of us working in the high-stakes world of bearing manufacturing and fluid power seals, the real story is written in the invisible peaks and valleys of the surface finish. For decades, the industry relied heavily on a single number, Ra, to tell that story. We now know that Ra is a blunt instrument. It tells you the average height of the roughness, but it stays silent about the underlying waves that can cause a high-speed bearing to howl or a hydraulic seal to weep oil.

Understanding the separation between surface waviness and roughness is no longer just a laboratory exercise; it is a fundamental requirement for manufacturing engineering. As we push milling machines to operate at higher spindle speeds and use advanced tool geometries, the signature left on the workpiece becomes more complex. We are seeing a shift where CNC milling is increasingly used to reach near-final dimensions on bearing races and seal seats, tasks that were historically reserved for grinding. This transition brings a unique set of challenges. A milled surface has a directional texture and periodic irregularities that differ significantly from the stochastic, random nature of a ground surface.

The distinction between these two components of surface texture—roughness and waviness—is a matter of spatial frequency. Roughness is the high-frequency “noise” caused by the cutting edge interacting with the material. Waviness is the lower-frequency “signal” caused by machine tool imperfections, such as spindle runout, vibrations, or thermal instability. If you cannot separate them, you cannot fix the root cause of a quality failure. In this deep dive, we will explore the technical nuances of these two phenomena, how they are governed by the latest ISO standards, and why failing to distinguish between them is a recipe for catastrophic component failure in the field.

The Technical Anatomy of Surface Texture

To talk about surface separation, we first need to define what a surface actually is in the eyes of a metrologist. Every surface profile you measure is a composite. It is a “total profile” that contains everything from the microscopic irregularities of the metal grain structure to the macroscopic curvature of the part itself.

Roughness: The Micro-Texture of the Cut

Roughness is what we typically think of when we talk about surface finish. In a CNC milling context, roughness is the direct result of the tool’s path and the physics of chip formation. Think about the last time you ran a face mill across a block of 4140 steel. The little ridges you see are the feed marks. These represent the high-frequency component of the surface.

Roughness is primarily influenced by the tool nose radius, the feed rate per tooth, and the ductility of the workpiece. When the tool shears the metal, it leaves behind a specific “fingerprint.” If the tool is slightly dull, you get “plowing” instead of clean shearing, which increases the roughness. From a functional standpoint, roughness is what holds oil on a surface. In a seal application, a little bit of roughness is actually a good thing because it creates a micro-reservoir for lubrication. However, if the peaks are too sharp, they will shred the elastomer of the seal.

Waviness: The Ghost in the Machine

Waviness is the more insidious brother of roughness. It is defined by longer wavelengths and is usually not visible to the naked eye under standard shop lighting. If you take that same milled block and look at it under a profilometer, you might see a gentle, repeating undulation that spans several tool paths. This is waviness.

Waviness is almost always a reflection of the machine tool’s health or the setup’s rigidity. For example, if your spindle has a slight imbalance, it will create a harmonic vibration. That vibration causes the tool to move toward and away from the workpiece at a frequency different from the tooth pass frequency. The result is a wave. Other sources include “form error” from a worn lead screw or the elastic deformation of the part under cutting forces. In bearing races, waviness is the primary cause of vibration and noise. A bearing can have a perfect Ra value and still fail a noise test because the waviness of the race acts like a miniature speed bump for the rolling elements.

Measurement Standards and the ISO 21920 Transition

For years, we lived in the world of ISO 4287 and ISO 4288. These were the bibles of surface metrology, defining parameters like Ra, Rz, and Rq. However, as machining technology evolved, these standards started to show their age. They were designed for 2D profile measurements, which often miss the “big picture” of a 3D surface.

The Role of the Cut-Off Filter

The most critical concept in separating roughness from waviness is the filter, specifically the $\lambda c$ (lambda-c) cutoff. This is a mathematical function—usually a Gaussian filter—that decides which frequencies belong to roughness and which belong to waviness.

Imagine you are at a concert. The high-pitched cymbals are like roughness, while the deep, thumping bass is like waviness. If you turn up the “treble” filter, you hear the roughness. If you turn up the “bass” filter, you hear the waviness. In metrology, the cutoff length is the threshold. If an irregularity happens over a distance shorter than the cutoff, it is counted as roughness. If it happens over a distance longer than the cutoff, it is waviness.

A common mistake on the shop floor is using the “Auto” setting on a portable profilometer. If the operator doesn’t manually set the cutoff based on the expected waviness of the CNC mill, the instrument might accidentally lump waviness into the roughness calculation. This inflates the Ra value and leads the engineer to believe the tool is worn out, when in reality, the spindle bearings are the ones failing.

Moving Toward ISO 21920

The new ISO 21920 standard, which effectively replaces the older standards, aims to simplify this. It moves away from the confusing “sampling length” vs. “evaluation length” terminology and focuses more on the actual functional requirements of the surface. For manufacturing engineers, the takeaway is that we must now specify our filters more clearly on technical drawings. You cannot just put “Ra 0.8″ on a print for a bearing race anymore. You need to specify the filter and the waviness limits (Wa or Wt) to ensure the part actually works in the assembly.

Bearing Races: When Waviness Becomes Audible

In the world of high-precision bearings, the race is the stage upon which the performance is played out. Whether it is a deep-groove ball bearing or a tapered roller bearing, the smoothness of that race is paramount. When we transition from grinding these races to hard-milling them, the waviness profile changes dramatically.

The Impact of Harmonic Vibrations

Consider a case where a manufacturer is milling the inner race of a large-diameter wind turbine bearing. Because of the size, grinding is slow and expensive. Hard milling with cubic boron nitride (CBN) inserts is faster. However, during the process, the long overhang of the boring bar creates a subtle chatter.

This chatter might not be loud enough to hear, but it leaves a waviness pattern on the race. When the bearing is assembled and put under load, the balls or rollers pass over these waves. This creates a specific frequency of vibration known as a “bearing pass frequency.” If the waviness is severe enough, it can lead to “brinelling,” where the rollers actually dent the race over time due to the repeated impacts. Here, the roughness (Ra) might be a pristine 0.2 microns, but the waviness (Wa) might be 2.0 microns. The bearing will fail prematurely, and the failure analysis will point to “surface irregularities” that the standard roughness check missed.

Real-World Example: Spindle Runout in Bearing Milling

In a recent production run of automotive wheel bearings, a Tier 1 supplier noticed a spike in noise, vibration, and harshness (NVH) testing. The surface roughness of the races was well within the 0.4 Ra specification. However, when they performed a full profile analysis, they found a periodic waviness that matched the spindle speed of the CNC lathe used for the finish cut.

The culprit was a worn spindle bearing in the machine tool itself. The spindle was “orbiting” by just a few microns. Because the wavelength of this error was much longer than the standard 0.8 mm cutoff used for roughness, the Ra measurement completely ignored it. By switching to a waviness measurement (Wt) and identifying the period of the wave, the engineers were able to trace the fault back to the machine’s headstock. Replacing the spindle bearings reduced the waviness by 70%, and the NVH issues vanished.

Seal Surfaces: The Hidden Pumping Effect

While bearing races suffer from vibration due to waviness, seal surfaces suffer from leaks. This is especially true for rotary shaft seals, where a rubber lip sits against a rotating metal surface. In these applications, the “texture” of the surface determines whether the oil stays in or gets pushed out.

The Phenomenon of Surface Lead (Drall)

When we mill or turn a seal surface, the tool follows a helical path. This creates a “lead” or “twist” (often called Drall in German engineering). If you look at it under a microscope, the surface looks like a very fine screw thread.

This lead is a form of waviness. If the “thread” is oriented in a certain direction, the rotation of the shaft will act like a microscopic Archimedes’ screw. It will literally pump the oil past the seal lip and onto the floor. This is a classic case where “smoothness” isn’t the goal—”neutrality” is.

Roughness as a Lubricant Reservoir

In seal applications, we actually want a specific amount of roughness. If the surface is too smooth (like a mirror), the seal lip will create a vacuum, wipe away all the oil, and then burn up due to friction. We need the “micro-valleys” of roughness to hold a thin film of oil.

The engineering challenge is that the waviness must be zero (no pumping effect), but the roughness must be controlled. If a CNC milling process is used, the feed rate must be carefully managed to ensure the lead angle is within the “non-lead” specifications (often less than 0.05 degrees). Measuring this requires specialized equipment that can trace the surface in a spiral or take multiple longitudinal traces to calculate the “twist” parameters.

Detailed Analysis of Tooling and Parameters

How do we control these two distinct animals in a CNC milling environment? It comes down to understanding how your machining parameters feed into the surface profile.

Feed Rate vs. Roughness

The relationship between feed rate and roughness is geometric. For a ball-end mill, the theoretical roughness (Rt) is roughly the feed per tooth squared, divided by eight times the tool radius. This is the “high-frequency” part of our profile. If you want lower roughness, you decrease the feed or increase the tool radius. This is straightforward manufacturing logic.

Machine Dynamics vs. Waviness

Waviness, however, does not follow a simple geometric formula. It follows the laws of dynamics. If your CNC mill has a resonance at 150 Hz, and you are running a 4-flute end mill at 2,250 RPM, your tooth pass frequency is exactly 150 Hz ($2250 / 60 \times 4 = 150$). You are hitting the resonant frequency of the machine with every single tooth strike.

This will create massive waviness. The surface might still feel “smooth” to the touch, but the profilometer will show a large, undulating wave. To fix this, you don’t change the feed rate; you change the spindle speed. Moving to 2,100 RPM or 2,400 RPM “de-tunes” the system, breaking the harmonic and flattening the waviness profile. This is why high-end CNC controllers now offer “vibration control” or “active damping” modules—they are essentially “waviness killers.”

Tool Wear and Surface Degradation

As a tool wears, both roughness and waviness change, but in different ways. Roughness usually increases as the cutting edge loses its sharpness and starts to tear the material. However, tool wear can also increase waviness. A dull tool requires higher cutting forces. These higher forces cause the tool holder or the workpiece setup to deflect. If the material has varying hardness (like a casting with hard spots), the tool will deflect more or less as it moves, creating a wavy surface.

The Metrology Lab vs. The Shop Floor

One of the biggest hurdles in manufacturing engineering is the “metrology gap.” The lab has a $100,000 stylus profilometer or a white-light interferometer that can separate roughness and waviness with extreme precision. The shop floor has a $1,500 handheld unit.

Bridging the Gap

To maintain quality in bearing races and seal surfaces, the shop floor needs better tools. Modern handheld profilometers are getting better, but they still rely on the operator to select the right filter. A “standard” 0.8 mm cutoff is the most common setting, but for large bearing races, a 2.5 mm or even an 8 mm cutoff might be necessary to capture the waviness that causes noise.

Engineers should create “Measurement Standard Operating Procedures” (MSOPs) that specify exactly which filter to use for which part. For a seal surface, the MSOP might require three measurements around the circumference to check for lead. For a bearing race, it might require a “long-trace” measurement to capture the waviness that a standard 5.6 mm evaluation length would miss.

Non-Contact Measurement

We are also seeing a rise in non-contact, optical measurement systems on the shop floor. These systems take a “snapshot” of the surface in 3D. This is a game-changer for separating roughness and waviness. Instead of a single line, you get a topographic map. You can use software to “peel away” the roughness layer to see the waviness underneath, much like looking at the rolling hills of a landscape after the trees have been removed.

Practical Examples: Troubleshooting the Surface

Let’s look at three specific scenarios where understanding the separation between these two metrics saved a production run.

Scenario A: The Whining Bearing

A manufacturer of high-speed spindle bearings was getting reports of a high-pitched whine during the break-in period. The races were hard-turned and then super-finished. The Ra was 0.05 microns—virtually a mirror.

Upon investigation with a high-resolution profilometer, it was discovered that the hard-turning operation was leaving a periodic waviness with a wavelength of 0.15 mm. The super-finishing process, which uses a flexible stone, was “polishing” the peaks and valleys but was not removing the underlying wave because the stone followed the contour of the wave. The “roughness” was gone, but the “waviness” remained. The solution was to adjust the hard-turning parameters to eliminate the vibration before the part ever got to the super-finishing stage.

Scenario B: The Leaking Hydraulic Cylinder

A hydraulic cylinder rod was failing its leak test. The surface was milled using a specialized high-speed process. The Ra was well within the 0.4 micron spec.

The engineers used a “lead” (Drall) measurement tool and found a 0.1-degree twist on the surface. This was caused by the combination of the tool’s feed rate and a slight misalignment in the machine’s tailstock. Even though the surface was “smooth,” it was pumping hydraulic fluid out of the cylinder with every stroke. They corrected the tailstock alignment and changed to a “plunge” finishing method that eliminated the helical path, stopping the leak.

Scenario C: The Premature Wear of an Oil Seal

An engine crankshaft seal area was showing premature wear on the rubber seal lip. The surface roughness was actually too low. The milling process was so efficient that it was producing a “plateau” finish that was too flat.

Without enough roughness (valleys) to hold oil, the seal was running dry. However, the engineers couldn’t just increase the feed rate because that would introduce unwanted waviness. Instead, they used a “structured” milling approach where they intentionally introduced a specific, high-frequency roughness pattern while keeping the machine’s stability high to prevent waviness. This provided the necessary lubrication pockets without compromising the geometry of the seal seat.

The Future: AI and Real-Time Surface Control

As we look toward the future of manufacturing, the goal is to move from “measuring failure” to “predicting success.” We are now seeing CNC machines equipped with accelerometers and high-speed data logging that can predict the surface profile in real-time.

By analyzing the vibration signatures during the cut, the machine’s AI can estimate what the waviness and roughness will be. If it detects a harmonic developing that will cause waviness in a bearing race, it can automatically shift the spindle speed by a few RPM to break the resonance. This “closed-loop” surface control will eventually make the distinction between roughness and waviness something the machine manages internally, but until then, the burden remains on the manufacturing engineer to understand and specify these parameters.

Detailed Conclusion

The separation of surface waviness and roughness is not merely a technicality; it is the bridge between a part that looks good on paper and a part that performs in the real world. In the precision-driven sectors of bearing and seal manufacturing, the “total profile” is a deceptive metric. As we have seen, a low roughness value can mask a devastating waviness profile that leads to noise, vibration, and premature mechanical failure. Conversely, a lack of controlled roughness can lead to friction and seal degradation.

For the modern manufacturing engineer, the transition to standards like ISO 21920 represents an opportunity to bring more rigor to the shop floor. It requires a shift in mindset—from chasing a single Ra number to understanding the spatial frequencies of the machining process. We must recognize that the CNC mill is not just a tool for removing bulk material; it is a high-frequency signal generator. Every choice we make—from the rigidity of the workholding to the selection of the $\lambda c$ filter—determines the functional integrity of the final product.

By implementing strict measurement protocols, utilizing advanced filtering techniques, and understanding the physical root causes of waviness (machine dynamics) versus roughness (tool interaction), we can produce components that meet the rigorous demands of 21st-century engineering. Whether you are preventing a “pumping effect” in a high-pressure seal or silencing a high-speed bearing race, the secret to success lies in the ability to separate the waves from the noise.