How to measure flatness (of a plane)?

[James Doolittle Member Find all posts View Profile Close James Doolittle]

We've got some plates to check the flatness and parallel of and we've got a total tolerance of .00011 (yes, that's one tenth and ten millionths).

I've been asked checking it with a CMM, it's about a one square foot plate and I'm taking about 200 hits, but I'm a little fishy our CMM is capable of such tight tolerance and having trouble convincing everyone else. I mean, our volumetric error for the machine is .0002 (almost twice the tolerance!), but I keep getting "it can't all be within that square foot" and "it use to check good and now all of the sudden we can't get a good reading, that can't be it".

We've got a Sheffield Endeavor 9.12.7 and a B&S One, both running 3.7MR2.

I just want to know what you guys would use to check flatness ACCURATELY so I can prove or disprove the CMM's results. What do you think? The company would probably buy whatever it is we need if we don't have the capability right now, as long as it's not outrageously expensive...

Thanks,
Jim

[Matthew D. Hoedeman has received PC-DMIS total consciousness Find all posts View Profile Close Matthew D. Hoedeman]

I doubt you can get it cheap. I'll say it again, no matter what your tolerance is, your machine should be capable of 10% of that tolerance, in this case, the machine needs to be capable of 0.000011" and I don't think I have ever heard of such a beast. Heck, with that tight of a tolerance, just sitting it on 4 blocks on the corners will make it sag out of tolerance in the middle of the plate, I would bet. You might need a calibration laser to check that beast.

[craiger_ny PC-DMIS Guru Find all posts View Profile Close craiger_ny]

I looked at a machine designed for checking flatness a couple of years ago for checking armature plates we make for electric brakes. It is an optical piece of equipment made by some folks that split off from Corning and I think they are in Rochester NY but I am not positive. It is very slick and quite accurate, I forget the resolution but well within what you are talking. At any rate you'll need to dig deep. I think that one started around $50,000. Kind of pricey for something that only checks flatness. If your volumes don't justify that type of thing I'd consider sourcing it out. As far as the CMM goes, nope, not that kind of tolerance.

Craig

[kbotta Ultra Senior Member Find all posts View Profile Close kbotta]

WMS (what matt said...)

[James Doolittle Member Find all posts View Profile Close James Doolittle]

Hmm.. just for kicks, do you know this company name in Rochester? The customer this plate if for is based out of Rochester NY, and does optical metrology... it'd be riot if it were the same company

[Paul Sarrach PC-DMIS Guru Find all posts View Profile Close Paul Sarrach]

Do it the old school way. flip it over and use a indacator

[John Kingston PC-DMIS Guru Find all posts View Profile Close John Kingston]

WMS (what matt said...)

WKS (what Kevin said...)

[craiger_ny PC-DMIS Guru Find all posts View Profile Close craiger_ny]

Hmm.. just for kicks, do you know this company name in Rochester? The customer this plate if for is based out of Rochester NY, and does optical metrology... it'd be riot if it were the same company

I think Tropel is the name of the company, very pricey. But it sure did look like a nice piece of equipment. If you are making something for them that is supposed to be flat, it better be. They know how to check it!

[James Doolittle Member Find all posts View Profile Close James Doolittle]

Nope, we don't do business with Tropel (to my knowledge). The other engineer over at my place said he's been over to this other customer and claimed that they check it with a CMM when they recieve it.

Tropel looks like they have some nice stuff. The Flatmaster suppose to be accurate within .000002" (two millionths) and repeatable within a tenth of that. http://www.corning.com/semiconductor...y_instruments/

[george frick Senior Member Find all posts View Profile Close george frick]

flatness

Do it the old school way. flip it over and use a indacator

i agree with paul. i would set the plate up on 3 calibrated jackstands and check underneath with a bns .00005 indicator. works 4 me.

[Paul Sarrach PC-DMIS Guru Find all posts View Profile Close Paul Sarrach]

i agree with paul. i would set the plate up on 3 calibrated jackstands and check underneath with a bns .00005 indicator. works 4 me.

[Underspec PC-DMIS demi-Guru Find all posts View Profile Close Underspec]

If you have nothing else to check but the CMM I would do a scan if possible if not use an auto plane with say 1000 hits....might take awhile but what choice do you have?

[craiger_ny PC-DMIS Guru Find all posts View Profile Close craiger_ny]

Nope, we don't do business with Tropel (to my knowledge). The other engineer over at my place said he's been over to this other customer and claimed that they check it with a CMM when they recieve it.

Tropel looks like they have some nice stuff. The Flatmaster suppose to be accurate within .000002" (two millionths) and repeatable within a tenth of that. http://www.corning.com/semiconductor...y_instruments/

Are you making this for OGP? They are in Rochester. If that is the case it may be a Zies that they are checking it with but din't take that to the bank.

Craig

[John Kugler Ultra Senior Member Find all posts View Profile Close John Kugler]

At a recent GD&T class, the instructor said that the GD+T math committee came up with the number of hits for a CMM to check flatness accurately. He said it was like 1800 per square inch.

I'll repeat, 1800 PER SQUARE INCH. Obviously, that is not doable in the real world, but you get the idea. It has to be LOTS of hits. 200 is not enough...

[Paul Sarrach PC-DMIS Guru Find all posts View Profile Close Paul Sarrach]

At a recent GD&T class, the instructor said that the GD+T math committee came up with the number of hits for a CMM to check flatness accurately. He said it was like 1800 per square inch.

I'll repeat, 1800 PER SQUARE INCH. Obviously, that is not doable in the real world, but you get the idea. It has to be LOTS of hits. 200 is not enough...

Now with that many. sounds like your checking it for surface finish