non-contact 3d surface roughness measurement 1

[minimotorsports]

I'm trying to find a company/lab to do some 3D surface roughness measurements for me. I'm looking for some specific parameters and having trouble finding someone who can measure what I need for a model that has been developed.

The tricky parameter is the standard deviation of peak distribution.

I have found several labs that use non-contact white light interferometers, but they do not know how to interpret the data to give me the standard deviation of peak distribution.

Can anyone give me any tips to find the standard deviation of peak distribution?

 

[EdStainless]

Can they give you the raw data and you calculate it yourself? Since you seem to know what this factor is and they don't.
Is there another surface texture factor that you can use as a proxy?

= = = = = = = = = = = = = = = = = = = =
Rust never sleeps
Neither should your protection

http://www.trent-tube.com/contact/Tech_Assist.cfm

 

[TVP]

Have you tried Michigan Metrology? I bet they would be able to assist you with this testing.

http://www.michmet.com/

 

[minimotorsports]

I did try Michigan Metrology. Don was very helpful, but he was unsure if he could get me the standard deviation of peak distribution. It's one of the really important characterizations of the surface roughness I need for my model. I also contacted Solarius, but they were unable to help.

 

[prost]

IMO the best option is still to get the data yourself and compute it yourself. What happens if you get the lab to compute it? You get the answer from the lab once, and you hope it was correct; plus you have no control over the data and no way to check if the lab did the computation correctly.

My advice to you is to _always_ get the raw or massaged data (say if your lab records voltages, but you want some distance recorded. the raw is the voltage, the massaged is the distance).

 

[minimotorsports]

I understand that calculating it myself from the raw data is an option, but I was hoping I could find a lab that would do the calculation for me. If you don't trust the lab to do a calculation from data, how can you trust their measurements in the first place? You can't do everything yourself.

If anyone knows a lab that has the capability to perform non-contact 3D surface roughness measurements and can give me standard deviation of peak distribution please advise me.

 

[prost]

You always have to trust the lab to do the things they are good at; in this case, they don't seem to have any idea how to compute this particular quantity. Personally I want the data because today I might want "slipnut coefficient" but tomorrow or next year I might want some quantity I don't know about today, the "doo-hickey value."

 

[IRstuff]

And what exactly is the "standard deviation of peak distribution?"

TTFN

Eng-Tips Policies FAQ731-376

 

[sreid]

There are well specified ways that surface roughness is specified. Surface roughness measuring machines have the math built in so measurements are calculated automatically. If you want something non-standard, you'll probably have to do the calculation yourself from the raw data.

 

[EdStainless]

Just curious, if you don't have a way to measure this value then how do you know that it is what you need?

= = = = = = = = = = = = = = = = = = = =
Rust never sleeps
Neither should your protection

http://www.trent-tube.com/contact/Tech_Assist.cfm

 

[minimotorsports]

A surface consists of peaks that are assumed to have a Gaussian distribution (normal). When two surfaces come together the highest peaks make contact and contact force develops. The magnitude of contact force is critical to our model. Greenwood and Williamson have a theory to relate the contact force to the separation of the two surfaces and the roughness parameters.

I am now looking for software to do the following:

Locate all peaks (points higher than their nearest neighbours (4 or 8))
Work out the curvature of each peak, from the heights at the peak and the 4 or 8 neighbours
Obtain the total number of peaks in the sampling area. Scale to unit area.
Produce a list of peaks and their heights, or a distribution (number of peaks in each height range ).
Calculate average curvature (Ssc parameter)

I believe most of these calculations are done internally by profilometer software in order to calculate Rpm etc, but may not be accessible to the user.

Peaks, summits and asperities mean the same or similar. There may be differences due to an exclusion zone etc.
There should not be a lower cut-off (ie a peak whose height is less than zero is valid - but it is a peak, not a valley, ie like a mountain under the sea!)

I expect to have to do any further analysis using Excel or similar package.

The peak distribution would be fitted to a Gaussian curve. The mean value is the "mean peak height" - from all the peaks, not just the top 5 or 10. The standard deviation is a measure of the width of the distribution. The magnitude of the curve is related to the total number of peaks.

Contact pressures are low, so the true area of contact will be very small - I estimate around 0.001% of the nominal area. So very few peaks in the sampled area will come into contact. The statistics of most of the peaks are irrelevant as they do not come into contact.

A fit can always be forced, but the important thing is to get a close fit to the highest peaks in the sampling area, say the top 100. This distribution can then be used to extrapolate to the very highest peaks on the surface, which are unlikely to appear in the sampling area.

I would therefore use the goodness of fit to the top 100 peaks only in fixing the parameters of the Gaussian distribution, even if the fit elsewhere is poor.

As a final point, it would be useful to have the curvature of each peak listed, or some way to link the curvature to the height. I think there is often a correlation between the two - higher peaks have greater curvature.

 

[TVP]

I do not think that you will find any software that will automatically output this type of information. Your best option will be to use a good company like Michigan Metrology and have them provide as much raw data as possible. You may be able to contact Mitutoyo or similar contact metrology equipment manufacturer (or Zeiss, Wyko, etc. non-contact metrology equipment manufacturer) to discuss the best method for obtaining raw data that can be exported to Matlab, Excel, etc. for further manipulation. It sounds like an interesting project; good luck.

 

[unclesyd]

Here is another company that may be of interest. They have some software, OmniSurf, that looks like it can resolved the biggest part of your problem. They have both a demo and online version.

http://www.digitalmetrology.com/

We had a PhD intern that work on some wear problems we were having with out high speed equipment and about the only thin I remember was the he considered the peaks to conical in most cases.

Another approach might be Fractal Geometry. Sometime back I ran across a paper extolling the virtues of describing surface roughness and topography by means of fractal Geometry. For some reason I didn't mark the site.

 

[EdStainless]

One thing that you will find is that if you give Mich Met a few samples the textures will vary a lot depending on the finishing methods. Some finishes will have conical points, other will have a direction (lay) to the finish. Some may have no points, only plateaus and valleys.

I get an idea of what you are after, I wasn't thinking of low contact pressures. You are going to need to do this optically. Otherwise stylus geometry will limit you data.

= = = = = = = = = = = = = = = = = = = =
Rust never sleeps
Neither should your protection

http://www.trent-tube.com/contact/Tech_Assist.cfm

 

[cocobolorose]

http://www.alicona.com/

Their Infinite Focus Instrument is capable. I use one regularly.

 

[KENAT]

FYI Wyko is now part of Veeco so they can offer AFM as & stylus profilers as well as optical profilers. However they don't offer measurement services as such.

http://www.veeco.com

 

[minimotorsports]

cocobolorose:

Thanks for the tip, I hadn't heard of them. Does anyone know a metrology service that has an Alicona Infinite Focus? I found one company in Manchester (ISS Group), but their email address doesn't work and I'd like to find a company in the states. Thanks.

 

[cocobolorose]

minimotorsports:

Get in touch with Paul Beaulieu or Chris Rinaldi at Excel Technologies (Enfield CT;

http://www.extec.com

)
They should be able to place you with someone.

good luck.

 

[minimotorsports]

Is Extec a lab or just a lab supplier?

 

[material347]

We can do this using our MicroXAM 3D non-contact surface profiler. We export the ASCI file to Excel. If you need this we can supply it. Mail me at chrispickles@csma.ltd.uk.