What exactly is sigma value and can it be used to measure surface roughness?

[[Al...]]

Hello,
I have Duramax machine with standard rigid styli (non moveable head) and we haven't purchased surface roughness measurements (option locked in Zeiss Calypso), but I have physical measuring plates of known surface roughness value and I have tried scanning all four of them using 1 mm stylus and very slow scanning speed and 0.002mm point density (each 0.002mm it takes value of the surface point).

I know it is still a big stylus for such a fine measurement but values I have gotten under "sigma" are crazy close to what the surface roughness value are and it doesn't look to be a coincidence to me..

I would like to know what this sigma value actually means...

This is what I've gotten while measuring these plates:

First plate declared as Ra2.5um it measured Sigma 0.0026 (2.6um)
Second plate declared as Ra1.25um it measured Sigma 0.0015 (1.5um)
Third plate declared as Ra0.63um it measured Sigma 0.0007 (0.7um)
Fourth plate declared as Ra0.32um it measured Sigma 0.0004 (0.4um).

Those values are very very close to what those plates are declared to be in surface roughness... Is it a coincidence or there is something to it?

Also, I've got another question, when extrapolating this sigma value using formulas (PCM), I don't know which measuring unit sigma value should be, it extracts in degrees.

I've created another theoretical line and in one of fields inserted formula which extracts sigma value from one of my measured lines, but the sigma is 0,1491 decimal degrees instead of 0.0026 like it says in that first 2d Line window (on the bottom left).

How to convert this value of 0.1491 degrees to this value 0.0026?

I'm asking because I'd like to report those sigma values somehow...

Thanks in advance.

[[Al...]]

I actually found how to convert sigma from degrees, just add *pi/180 to the formula and then if you want to convert it to um just multiply by 1000 and you get that....

And the value for distance between points taken is 0.001mm not 0.002 like I said above.

I would still want an answer for other two questions...

Thanks again.

[[Da...]]

Please sign in to view this quote.

Sigma is the root of the mean squared error of a Gaussian element.

Please sign in to view this quote.

One hundred percent coincidental. Sigma and roughness have NO relationship whatsoever. The CMM doesn't have the accuracy to measure roughness without special equipment.

Please sign in to view this quote.

The sigma value is in the unit you're using (mm or inch). Definitely NOT degrees!

[[Al...]]

Please sign in to view this quote.

I would like a better explanation if possible in layman's terms.
I have found interesting stuff about this on internet, but how does CMM calculate this sigma value for each feature?
How do you get those values of 0.0026 etc..?

Please sign in to view this quote.

I've measured those plates like 10 times now and every time it is close to those values, are you absolutely sure it has nothing to do with it???

[[Ke...]]

Please sign in to view this quote.

Sigma 1000% has NOTHING to do with roughness...

[[Ke...]]

Please sign in to view this quote.

Sigma or standard deviation is a measure of the variation in a set of data. Simply stated, it is the average distance that each point is away from the mean of that data set.

[[Al...]]

Please sign in to view this quote.

Thanks for the explanation, that was something I was imagining what it was while thinking about it during weekend.
But then again, doesn't it remind you of the surface roughness?
Average distance that each point is away from the mean?

[[Ma...]]

You can not differentiate between Ra and others ( Rz, ... ) Probe radius will affect those numbers as well.

[[Ow...]]

Yeah, you're not going to be able to measure surface roughness on a CMM unless you have a Zeiss Rotos attachment for the Vast head.
It has a 5 micron diamond radius that replicates a portable roughness tester like the Mahr portable Pocket Surface IV tester.
Good for measuring surface roughness on the CMM and putting the results on the same report as other measured characteristics but, not cheap.

https://www.zeiss.com/metrology/solutio ... tions.html

[[Al...]]

Okay guys, thanks for all your inputs, but I'm still not convinced you're right...
I've just measured 3 test parts with surface roughness comparable to those test plates and on all three parts I have gotten very convincing values.

My "feeler gauge" for those part surfaces was my nail (same as you would feel the difference on those surfaces of test plates) and it seems sigma "nailed it"...

First part was very rough (saw cut) but sanded on fine grain sandpaper and it's value was Ra1.47 (comparable to test plate of Ra1.25 using my nail).

Second part was very good and shiny, like it was polished (but using facemill on a CNC machine) and it's value was Ra0.31 (feels a bit better than surface of the test plate of Ra0.32).

Third part was milled (end mill) but mill was dull and it left bad surface, it was measured as Ra 4.5 (and my thumb confirmed that it was pretty much rougher than test plate of Ra2.5).

So I still believe I can roughly but still precise enough, measure surface roughness on my CMM using just this 1mm ball stylus and a scanning pattern for a line with 1um point taking method.

We had that machine that measured surface roughness but unfortunately it is broken so I can't compare those two, but still it feels pretty much right on the money.

I will keep trying and testing my theory until I feel I'm wrong, you should try it too and see for yourself...

[[Je...]]

Please sign in to view this quote.

Alen,

Although I truly respect the experience and wisdom of our amazing forum contributors, I will dare to disagree here.

I believe surface roughness can plausibly be evaluated with a tactile scanning CMM in combination with the proper sensor and post-capture data treatment. I believe this task is more achievable when utilizing a high-end dynamic sensor such as a VAST as well as other hardware components that mitigate inertia, vibration and deflection error. However, it may be possible to get a glimpse at surface texture when using your Duramax. It comes down to your uncertainty budget for the task at hand.

Some of the application engineers at Zeiss have suggested as much. I'd be interested to hear how further testing goes for you.

While my knowledge of surface texture is mediocre at best, these folks are pros, and I like watching their Notepad Series on Surface Texture:

https://digitalmetrology.com/resource/notepad-series/

[[Ju...]]

It's incredible that the deflection and positional accuracy of your stylus system compared to the master probe accurately measures the surface roughness of your parts being checked. Zeiss should really get into this market! Or maybe they already have...

[[Je...]]

Please sign in to view this quote.

I will confirm that they are, for the vast majority of the statements, spot on. You cannot measure and accurately report surface finish with a CMM without a ROTOS sensor.

The ONLY exception is a very specific routine used to measure waviness ONLY which utilizes the Roughness characteristic.

NO ROTOS = No roughness/finish. Any value you are seeing is 100% coincidental. The XXT is not capable of the sub-micron ACTIVE scanning requirements for Surface finish. The mechanical filtering alone seen from using a spherical stylus negate all results seen.

[[An...]]

_____

[[Ju...]]

Please sign in to view this quote.

Can't argue with this logic.

[[Ke...]]

Please sign in to view this quote.

Agreed: Waviness - Yes. Ra - No.
...unless someone finds a way to build & calibrate a 2-5 micron CMM stylus 🤣

[[Ma...]]

I would think everyone would agree that there is no way you can give legitimate roughness results with a 0.5mm radius stylus, even if the rest of the measuring system was capable.

That said, the formula for Ra is quite similar to the formula for "s". The Ra parameter replaced an older parameter called RMS (current parameter Rq). Rq/RMS represents Root Mean Square, pretty much exactly the same formula as "s".

So, I would say that while it is likely that there would be some correlation between "s" and Ra, it is still by no means an acceptable measurement method.

[[Je...]]

Please sign in to view this quote.

🤣

By all means, I encourage you to support your accusation with presentation of relevant data and methodology. Please tell me how Sigma correlates to relevant Ra during measurement of surface using a spherical stylus (of 0.5mm or larger as is required for an XXT TL3 on a DURAMAX) on a passive sensor on a machine that has an accuracy spec that exceeds amplitude in standard Ra or Rz requirements.

Much less without considering the mechanical noise of a drive motor or mechanical filtering done by stylus radius (ROTOS does not move the CMM drive motors/axes and therefore does not require an algorithm to compensate for drive motor noise. This noise is present in an active head measurement and completely destroys the data due to high frequency input).

This is why only waviness can be reported from an active scanning tactile sensor (VAST/XT Only) or, theoretically, DotScan as high-frequency static can be digitally filtered from the data set.

While the formula for Sigma and for Ra are similar, when the data set inputs get to Sigma values relevant for Ra use, the data set is no longer applicable due to added "noise" induced by the CMM and sensor. IE - once the Sigma value is decreased to a point where it would be within a normal Ra requirement, the data is no longer valid for that calculation and correlation is 100% coincidental.

If you are going to accuse someone of being wrong, I hope you come prepared with an elegant argument to support your accusation. I am looking forward to discussing this further.

Additionally. Please do so without a PDF attachment. Consider this a pre-emptive circumventing of the infamous Andreas Binder "See Attached" 😉

[[Ju...]]

Andreas has now redacted his response, as custom dictates.

[[DW...]]

I am having a hard time believing that OP isn't trolling us. If that is not the case, it would appear they know just enough to be dangerous but not enough to know they don't know.

Please sign in to view this quote.

Tell me you don't know how to qualify probes without telling me you don't know how to qualify probes.

Please sign in to view this quote.

There are literally responses from senior managers who work for the company whose name is on the side of the machine you are using telling you what you are doing is completely wrong.

Please sign in to view this quote.

It concerns me that OP is willing to report this as fact to the internet, his employer, or possibly a customer.

Please sign in to view this quote.

It's called a profilometer. You should get it repaired, or buy a new one.

Please sign in to view this quote.

How do you calibrate your nail? Your nail is a guide, not a measurement tool.

I digress. OP if you read this please get your employer to invest money into training so you don't convince yourself of things that are blatantly false or make costly mistakes.

[[Al...]]

Please sign in to view this quote.

Yes! That was my feeling exactly after I've seen results first time. This was too much of a coincidence which kept bugging me and thinking of sigma which I don't quite got a feel for, but it is much clearer to me now what it represents than it was before.

This is what I've been trying to express here, my feeling that it is "same formula" as Ra.

I can agree with all of you guys that it is not an acceptable measurement method and I will not be using it for any kind of report for our parts which would prove to our customers that our surface roughness is in tolerance, but I will be still using this method to see how much I can get from it (just for the sake of learning more) and for roughly estimating parts in question which we cannot measure differently (like I said our surface roughness machine is broken and it is not likely it will be fixed or that we will get a new one).
So when I'm in doubt about some part surface roughness, we will check it like we did before (rule of thumb) but again, I will try and measure it like I did on CMM and see if the results agree..

Oh I like a good discussion...

[[Al...]]

Please sign in to view this quote.

But what if the data used from measuring sigma value is not converted, but sigma value is used as Ra?

That is, sigma=Ra?
What happens then?

[[Al...]]

Please sign in to view this quote.

Mother nature will show you how if you're willing to learn.

You can laugh all you want but read something like this:
https://www.researchgate.net/publicatio ... _tribology

If you don't believe me, try it for yourself, buy test plates (gauges) similar to these (but I believe you already have those):
https://www.messmittel.tools/Measuring- ... 5---m.html

And then feel them using your thumb and nail (both skin and nail) and see if you'll feel the difference between them.

Then do the same on some part, try to compare one of those gauges to your part and decide which Ra plate it is most similar too.

Then measure that Ra on your CMM and see if you'll be close enough.
If you were off a little, try the feeling again on that gauge of which Ra value is closest to what your CMM has measured as Ra value. Then repeat this procedure over and over again using different parts and train yourself and your thumb feeling.

Tell me I'm wrong then.

This is how you calibrate your nail. Years of experience doing the same thing.

[[DW...]]

Please sign in to view this quote.

Please show us your nail calibration certificate traceable to NIST 🤣