Data from Scanning

The data that is collected from scanned data has two characteristics Wavelength and Undulations per Revolution. Wavelength is the distance between identical points in the adjacent cycles of a waveform. (See Figure 1)

Figure 1

Waveform is related to frequency. The higher the frequency the shorter, or more compact, the wavelength. Filtering along a set wavelength is advisable for straightness or roundness filtering.

Undulations per Revolutions set the number of undulations per revolution. This is typically used on roundness machines as the part rotates under a stylus. The ANSI Standard has set the 50 UPR as the default standard for out of roundness measurements. The lower you set the UPR, i.e. 0 to 15 UPR the more low frequency errors it reveals.

Filtering – What is it?

Filtering is the ‘smoothing’ of scanned data. There is some inadvertent smoothing of data based of the size of the stylus you are using. The larger the diameter of the stylus the greater the risk of not contacting all the highs and lows spots of the machined surface. When it comes to filtering scanned data the options can be staggering. We will try and sort out some of the confusion in this article. The first thing we need to discuss is; what exactly is data filtering?

Filtering is recommended only when you have a large number of points. The amount of points that are typically achieved when you scan a feature with today’s scanning heads. It is important to know the types of filtering available on your software. Some software have only a basic filtering option which on the whole is not bad but you must come to understand what is happening to your data.

Filter Methods

There are three filter methods that are generally used in data filtering. They are Gauss or Gaussian, Spline, or 2 RC. Each one of these offers a different filtering method and one is not necessarily better than the other.

GAUSS

This filter is newer than RC filters and introduces less distortion. Gaussian filtering is becoming a more preferred method than 2 RC filtering. It is considered by the ISO committee to replace the 2 RC filter as the standard filter for 2-D profile filtering. This profile filter separates profiles into long wave and short wave components. The λc separates the surface finish from the long wave components (waviness). 

SPLINE

This filter has a steeper gradient than the others and is considered to have best selectivity.

2 RC

RC stands for Resistors and Capacitors. This type of filtering is an older style of filtering used in electronics for filtering ‘noise’.

Figure 2

With Gaussian and Spline filtering the amplitude of the limit wavelength is attenuated in half (50%) and with RC it is attenuated by a quarter (25%). This means Gauss and Spline filtering are more robust filtering methods.

Filtering Types

There are three filtering types Low-pass, Band-pass, and High-pass.

LOW-PASS

With Low-pass low frequencies pass the filter and high frequencies do not. The filter largely eliminates surface roughness effects. This eliminates the short-wave surface effects and the long-wave effects of waviness are retained. This filter is also known as the Treble Cut Filter. In the audio world this filter removes the hiss from any audio recording. (See Figure 3)

Figure 3

 HIGH-PASS

This is the opposite of Low-pass, the long-wave effects of waviness are reduced and the short-wave effects of surface roughness are retained. This filter is also known as the Bass Cut Filter. In the audio world this filter removes the bass, or low end, from any audio recording while maintaining the hiss. (See Figure 4)

Figure 4

BAND-PASS

This filter type is a combination of the low-pass and high-pass filters. The band-pass is between the two filters and a UPR of wavelength distance must be established. This filter suppresses both wavelength and surface roughness. (See Figure 5)

Figure 5

Practical Application

Changing the filtering types will have a definite effect on the results of your scan. In my case, I have a sealing surface that is scanned. Below are the effects of changing the filtering.

1st Test - .034 micron Flatness
Wavelength 2.5mm
Spline Method
Low-pass

2nd Test - .0091 microns Flatness
Wavelength – from 2.5 to 8mm
Spline Method
Band-pass

3rd Test – .007 microns Flatness
Wavelength – 2.5mm
Spline Method
High-Pass

4th Test – .0371 microns Flatness
50 to 150 UPR
Spline Method
Band-pass

Each filtering method is neither right nor wrong and all the tests are within their individual ISO standards; however, they have varying results. What must be investigated is how this part is used and how the customer will check it. If you high end scanning and your customer uses an indicator the chances of correlation become nil or close to it.

While this article doesn’t intend to be the definitive answer on scan filtering I hope that it brings some insight to the filtering options. Care must be given to supply our customer with the correct information but understanding what our software is doing with the data becomes the first step in supplying that information