Meaning
Surface metrology algorithms use a weighted moving average to separate rough textures from long-wave form errors on measured polymer parts. This process, which applies a Gaussian filter to the raw height data, employs a bell-shaped probability density function as the weighting function. It provides a smooth reference line that avoids the phase-shifting errors associated with traditional RC filters.
The calculated surface profile represents the deviations of the physical surface from this reference line.
Transmission Characteristic
Sensing frequencies are attenuated depending on the selected cutoff wavelength of the filter. When applying a Gaussian filter, the attenuation of the sinusoidal profile at the cutoff wavelength is exactly fifty percent. This characteristic ensures a clear, predictable boundary between high-frequency noise and low-frequency waviness.
The mathematical consistency allows different metrology laboratories to achieve comparable results on the same moulded component.
Roughness Separation
Micro-fluidic channel depth and optical lens curvature are isolated from microscopic surface rough spots by adjusting the cutoff limit. Low-frequency components are subtracted from the total profile, leaving the primary roughness evaluation data. This separation is necessary because high-frequency roughness affects optical clarity, while low-frequency waviness dictates sealing performance.
Choosing the wrong cutoff can falsely indicate that a rejected part meets specifications.
Filtering Limitation
Edge distortion occurs at the start and end of the measured profile because the weighting window extends beyond the physical boundaries of the data. Moulders must collect excess scan length to ensure that the active measurement zone remains unaffected by this edge effect.