roughness filter on Labview

[sylvainDCU]

Hello,

I have a surface profile and I would like to find the roughness. I need to use a filter to delete the waviness and keep only the roughness profile. How I can do it on Labview, because I don't arrive to use the filters to delete the waviness.

Thank you,

Sylvain

[johnsold]

Sylvain,

Do you have clear definitions of what constitutes "roughness" and "waviness?"  Without that information you cannot specify a filter in a meaningful way.  Is the waviness predictable or repeatable?  Do you have a good mathematical model which describes the waviness?

The more information you can provide, the more likely someone will be able to help you.   If you can post an image or VI with the roughness and waviness marked, that would be helpful.

Lynn 

[Coq_Rouge]

It sounds like sylvainDCU want to make a high-pass filter. This way the signal will fluctuate around zero and not some offset value

[sylvainDCU]

Hello,

It's links to explain better what is the waviness and the roughness.

I would like only keep the roughness>

Thank you

[sylvainDCU]

Hello,

It's links to explain better what is the waviness and the roughness.

I would like only keep the roughness>

Thank you

[johnsold]

The documents you posted indicate that low pass filtering or windowed moving averages can be used to separate the roughness.  LabVIEW can do either of those.

What have you tried? What did or did not work?  What version of LV are you using?

Lynn 

[F._Schubert]

All you need to use a high_pass filter, as Le Coq rouge wrote in his reply (that is what roughness.doc says). You will need to have the Full Developement Version for that.

I would try playing around with FIR filters.

Please post code/screenshot as far as you come...

Felix

[sylvainDCU]

Hello,

I have Labview 8.6.1

I tried all the type of filter in Labview, but it was not good. According the norm ISO 4287 about the roughness, I have to use a Gaussian filter but it doesn't work, I don't how I can do. I tried to plot the spectrum  ( magnitude-freqeuncy ) of my profile in order to find the good frequeny I have to keep, but I don't arrive to plot the spectrum ( when I send a sinusoidal wave in my spectrum, I haven't peaks of frequency, do you know how I can do a spectrum ?

I would like to find the same results than the software I sent you the last time.

Regards,

Sylvain

[F._Schubert]

Your FFT has no frequency information. Therefore the X-axis has no physical meaning.

Also try to filter the signal without the offset compensation (substracting the mean). You will see that the filter is doing it for you.

Felix

[sylvainDCU]

Hello,

How can I do for give a physical meaning to my frequency ?

The filter don't do the offset compensation

Sylvain