We are measuring perpendicularity and concentricity on a Zeiss CMM with Calypso software. What filter settings should we apply? Right now we are using Least Squares, Gauss, low-pass, outlier elimination on. Would you change any of these? Our supplier is recommending these and I want to make sure these are correct. My fear is that these are the filter settings that will make the parts appear "in spec", not necessarily the technically correct settings.

Are you scanning the features? How many points? What type of probing path? Circles, lines, helix?

What is the surface finish of the areas being measured?

The filters help eliminate outliers, but the proper settings are important, and will affect the size/form/location of the measured features.

Switching to "outer tangential" from the least squares will give you a more gage like fit as it uses the three high points. But with a rougher surface or mre irregular surface least squares offers better repeatability.

Without a little more detail i can't offer a more definitive opinion.

The best way to describe the parts an engine cylinder bore and a piston that has to fit inside. We want the sides of the piston to be perpendicular to the top of the piston. The surface finish is better than Rz 6.3. We are scanning three diameters so our point count is well over a thousand. We construct a cylinder based upon these three circles. We establish the reference plane from the top of the piston by scanning also.

Mitch

Are you scanning the features? How many points? What type of probing path? Circles, lines, helix?

What is the surface finish of the areas being measured?

The filters help eliminate outliers, but the proper settings are important, and will affect the size/form/location of the measured features.

Switching to "outer tangential" from the least squares will give you a more gage like fit as it uses the three high points. But with a rougher surface or mre irregular surface least squares offers better repeatability.

Without a little more detail i can't offer a more definitive opinion.

Try it with and without filtering, you most likely will not see much of a difference and depending on tolerance may not be a concern.

If it does change alot then something else is happening.

How are the form of the cylinder and the plane.

The Gaussian filter was until recently the industry standard for the measurement of roundness and surface geometry. This is now surpassed by the Spline Filter. It is a digital, phase corrected filter.

The spline filter fixes some problems with the Gaussian filter. Compared to the other filters it has the steepest gradient thus the best selectivity.

A Lowpass filter type reduces the short-wave surface phenomena (High frequency) and does not disrupt the longwave (Low frequency) effects of waviness and geometric deviation. So the Low pass filter will filter out surface irregularities (high end noise) but keeps the lower end noise, such as machine noise. However, Zeiss leaves Low pass as the default.

As with all collected data you may not know exactly what you have without supporting data. So surface finish results, form, and other aspects of the feature will play a key role in determining the results of the scanned data.

There is a more detailed description here.
http://www.cmmquarterly.com/site/index.php?option=com_content&task=view&id=66&Itemid=69

The Gaussian filter was until recently the industry standard for the measurement of roundness and surface geometry. This is now surpassed by the Spline Filter. It is a digital, phase corrected filter.

The spline filter fixes some problems with the Gaussian filter. Compared to the other filters it has the steepest gradient thus the best selectivity.

A Lowpass filter type reduces the short-wave surface phenomena (High frequency) and does not disrupt the longwave (Low frequency) effects of waviness and geometric deviation. So the Low pass filter will filter out surface irregularities (high end noise) but keeps the lower end noise, such as machine noise. However, Zeiss leaves Low pass as the default.

As with all collected data you may not know exactly what you have without supporting data. So surface finish results, form, and other aspects of the feature will play a key role in determining the results of the scanned data.

There is a more detailed description here.
http://www.cmmquarterly.com/site/index.php?option=com_content&task=view&id=66&Itemid=69This is a good response, and the link to the article is welcome.

Here at the Cove, we do however ask for a note acknowledging when we are referring to our own companies for information or help - and it also helps to note when we are not posting a link to an affiliated organization. Please refer to the TOS (http://elsmar.com/Forums/showthread.php?t=8831) for more information.