Question Calculaton for Flatness of Planes

[[email protected]]

I measured a plane on the surface plate using a 50 millionths indicator. I leveled the three highest points on the plane using jacks.

I found over .011" error in flatness.

PC-DMIS reported only .006" of form error when taking over 400 points measuring the same surface.

Can anyone explain why I would not see the .011" error found on the surface plate?

Thanks,

Daniel

[bob mappes]

1) a CMM is not a good tool for checking flatness....indicators are the best method
2) is the part "sagging" while on the jacks?...you may need to support the entire part with another thick plate to retain "form"

good luck
bob

[Mike N.]

A couple things to consider when using flatness on any CMM

1. a three pt plane should be used to level

2. analog scanning will provide a more similar result to an indicator ( 400pts just won't compare to an indicator sliding continuously over the surface ) using touch pts on a cmm wont cut it the same as an indicator EVER

3. Did you level those same 3 high pts on the cmm? I'll bet they weren't exactly the same....its nearly impossible to capture unless you marked them with a small dot, and then its only "close"

[[email protected]]

The part is .30" steel, thus rigid. No sagging.

What do you mean the CMM is not a good method. You're gonna blow every bodies mind, letting that cat out of the bag.

[[email protected]]

Thanks Mike and Bob!

I did mark the three high points on the surface, then leveled it on the CMM.

I was able to find the three points were zero when remeasuring them. I then went to the .011" area and found it to be .010". This is certainly acceptable in my mind, but when the form of this surface only shows .006"--it makes me wonder what's going on.

[#2#]

Was your .011 measured point encompased in the triangle created by the three high points? If not then that may be your problem.

Is flatness dependent on alignment in PC-DMIS?

It definitely is on a surface plate. For example look at the letter "L" Imagine the L is a profile view of a very bad plane. If you level along the vertical leg the tip of the horizontal leg is off .xxx. If you level along the horizontal leg the tip of the vertical leg is off .yyy. If you level using the tips of each leg the intersection point of the vertical and horizontal legs is off .xxx(sin(tan^-1(.yyy/.xxx)))

[bob mappes]

The part is .30" steel, thus rigid. No sagging.

What do you mean the CMM is not a good method. You're gonna blow every bodies mind, letting that cat out of the bag.

depending on how how is "hanging" .30" steel is still able to sag

think about how thick your granite plate is......there is a really good reason for it being thick

[Roger Kilpatrick]

I will NEVER check flatness with a CMM, I bought a 2" thick surface plate, had a hole cut through it and fabricated a holder for a digital gage sender unit that sticks up about .05 through the plate. I lay the part down on the plate and run it over the indicator sending unit tip checking flatness of the surface while the part rests on its high points. We never had a way of accuraely checking flatness per ASME Y14.5M prior to that. This method works great.

[bboyd]

CMM's are not as accurate as indicators for "any" form tolerances.

[rebeldude]

I agree that some checks a cmm is not as accurate, but they are good enough for sheet metal and accepted as final say by most companies today

[Jeffrey Lovely]

Do the 400 points include the 3 level points when creating this plane
if not add them
then create the plane with 403 points
report flatness of that plane

[Simon_Fuller]

I was always told that the flatness is calculated by finding the average point and then finding the difference between the average and the max variation and that is what the figure outputed is.
If anyone can tell me different I would love to know!!!!

[Mike N.]

I was always told that the flatness is calculated by finding the average point and then finding the difference between the average and the max variation and that is what the figure outputed is.
If anyone can tell me different I would love to know!!!!

As written in a .pdf from wilcox

"Flattness is based on the least squares solution of the plane. The flatness of a plane is calculated as the deviation of each measured point from the measured plane. The difference between the max and min deviation is the flatness tolerance."

[Jan d.]

As written in a .pdf from wilcox

"Flattness is based on the least squares solution of the plane. The flatness of a plane is calculated as the deviation of each measured point from the measured plane. The difference between the max and min deviation is the flatness tolerance."

This is typically how all CMM companies do this.

I do measure flatness on my CMM's. However, you MUST PERFECTLY adhere to the datuming scheme. As you found out, being off, and you can get all kinds of strange results. However way you datum on the surface plate, you MUST use the same scheme on the CMM. If you do that, I find decent (read: usable) correlation between the hard gauge and CMM.

I disagree that the CMM is not a good way to measure form. Yes, it samples, but still it can be a good tool, if (and only if) you perfectly adhere to the datuming scheme on the print. I do agree that if you need ultimate and perfect numbers (like flatness in the 0.0001" range), you may run into trouble on a CMM.


Jan.

[cmmguy]

As written in a .pdf from wilcox

"Flattness is based on the least squares solution of the plane. The flatness of a plane is calculated as the deviation of each measured point from the measured plane. The difference between the max and min deviation is the flatness tolerance."

And that is how most if not all CMM softwares calc flatness.

The problem could be either:
- The three highest points used are actually not the highest points.
- The Least Squares plane is significantly different from the plane created by the highest three points.

With the CMM, probe the three highest points(as found with the indicator), create a plane, measure the remaining 400 points and report the difference between the highest(should be the plane, right?) and the lowest. This is perfectly acceptable. These differences are not deficiencies in the methods, just different methods. It is just like any other measurement methods - you will have differences.

You need to determine what method is acceptable for the manufacturing process(how irregular will your surface be) and decide which method is best.

You could probe the 400 points, create a function that iteratively determines the three highest points, creates the plane and does the deviation calculation for you. I don't think PCDMIS has that feature. Maybe that should be added to the 5.0 wish list. All right I will add it to the list....