Cmm Deviation

[baylorbear]

Here is a great questions for you gurus. Should there be deviation from one CMM to another, if so how much? If someone certified a gage on a manual CMM and then I meausure it in DCC mode should there be a difference? I am talking a half MM or smaller. We have two BNS machines of different sizes and I have measured things with the same programs and gotton different results. What are your thoughts on the matter? We have the same probe setups on both machines.

[Matthew D. Hoedeman]

Here is a great questions for you gurus. Should there be deviation from one CMM to another, if so how much? If someone certified a gage on a manual CMM and then I meausure it in DCC mode should there be a difference? I am talking a half MM or smaller. We have two BNS machines of different sizes and I have measured things with the same programs and gotton different results. What are your thoughts on the matter? We have the same probe setups on both machines.

1) Are you using probe files from a drive that is NOT on the CMM PC? If so, that is BAD.
2) How are your machines calibrated? If they ONLY get the ball-bar for calibration, then the linear calibration of BOTH machine has NOT been done and it is VERY suspect.

From machine to machine, I would expect NO MORE THAN 0.001 to 0.002" difference, if BOTH were properly calibrated, even going from manual to DCC. HOWEVER, if the manual hits are not taken properly, you could expect to see big differences.

[Paul Sarrach]

.05mm is what is acceptalbe from CMM to CMM, within my circle. I have not seen anything in writing. Its an unwriten law.

[#2#]

That seems like a pretty big difference to me. Especially since you are running the same program on both machines.

As far as things to correct the issue???

Is it possible that your manual alignment affects the program this much. Try running it on one cmm manually aligning as different as you would expect to do normally and run it a few times.

Is it possible that the squareness to the machine's axes affects your results this much (shanking?)? Try squaring up the part and measuring, then turn a little one way and measure, then turn a little the other way and measure.

[Matthew D. Hoedeman]

.05mm is what is acceptalbe from CMM to CMM, within my circle. I have not seen anything in writing. Its an unwriten law.

That would be 0.002", DUH!

Gee, read my post, eh Paul(a)!

[bw_bob]

Thats us too....05mm or something is up....dcc vs manual i would think it would be closer then 1/2 millimeter

[Wingman]

Here is a great questions for you gurus. Should there be deviation from one CMM to another, if so how much? If someone certified a gage on a manual CMM and then I meausure it in DCC mode should there be a difference? I am talking a half MM or smaller. We have two BNS machines of different sizes and I have measured things with the same programs and gotton different results. What are your thoughts on the matter? We have the same probe setups on both machines.

There will always be differences. How much depends on several factors.

Manual to DCC will usually return a bigger difference than DCC to DCC.

A part ran on two machines with same program (assuming both DCC) should not vary anywhere near 0.5mm. Try running the part 5 or 6 times (10 is good) on each machine and see what the Spread is on each machine. Sounds to me like there is a problem with setup / program. I would be surprised to see a diff of 0.1mm from two calibrated machines running the same program on the same part. Of course, running it several times on the same machine (pulling the part off and putting it back on) will give you a good indication of how well your measurement method is performing. I would try an R&R.

As an indicator, if we run a part here in Ohio with my program on PC DMIS and run the same part in NY on a diff machine with diff program in MMax we expect to get closer results than 0.5mm and we are shipping them from plant to plant and using different programs.

[Paul Sarrach]

That would be 0.002", DUH!

Gee, read my post, eh Paul(a)!

Matt you just type faster than me, you beat me to the enter button agian. You have fast hand and wrist action. How do you keep them so limber?

[Matthew D. Hoedeman]

Matt you just type faster than me, you beat me to the enter button agian. You have fast hand and wrist action. How do you keep them so limber?

LOT and LOT of practice!

[baylorbear]

Our Probe files are on the each computer. We have a 15-26-14 and a 12-22-10. Does size of a cmm have anything to do with it? One is made in the US and one came from Italy. I got to go to lunch now but I will get back to this ASAP.

[Wingman]

Our Probe files are on the each computer. We have a 15-26-14 and a 12-22-10. Does size of a cmm have anything to do with it? One is made in the US and one came from Italy. I got to go to lunch now but I will get back to this ASAP.

If probe setups are the same and both are calibrated it won't be an issue. I would suspect setup or programming issues. You need to run the program multiple times on each machine to see if you really have a diff of measurement or a repeatability issue.

Have you checked a gage block (the longest one you can find) on both machines? If it checks out (I suspect it will) then you know you have a setup or programming issue.

[Matthew D. Hoedeman]

Our Probe files are on the each computer. We have a 15-26-14 and a 12-22-10. Does size of a cmm have anything to do with it? One is made in the US and one came from Italy. I got to go to lunch now but I will get back to this ASAP.

Another thing you have to watch out for with Pcdmis in MANUAL mode, if you are taking SINGLE points, Pcdmis will ONLY comp in the direction of the major machine axis you are moving in.

For example, if your fixture is sitting at a 30 degree angle to the machine axis and you align it using a plane and 2 lines., it will work out fine. Level and origin to the plane, rotate and origin to the line and last origin to the other line, it will work fine. However, if you then go and measure a manual hit on the edge of the fixture, you will get a vector for that point that is incorrect, it will give it the aligned vector that matches the machine axis of major motion direction, if ya follow me. So, even if YOU move the machine perpendicual to the edge of the fixture (that is 30 degrees off from machine axis), the point will end up with a vector of 0.5, 0.866, 0, NOT with a vector of 0,1,0, which it SHOULD have. So, at that point, the probe comp for the point will NOT be in the correct direction and the result will be wrong.

[dwade]

I have 3 MicroXL Pfx's and 1 MicroVal Pfx and run the B&S test block across them and each CMM will repeat it's measurment within .0002" and then between the 4 of them my worse case variation is .0005". That checks diameters, depths, lengths, slot widths/lengths, distances between holes and distances between a holes and an adges.

[baylorbear]

We only do the Bar check when they are qualified. We struggle with the temperature during the summer too. We are trying to control as many things as possible.

We use the same type of alignment with all programs. We use an DCC iterative using tooling balls.

I am certifying a gage that has been changed recently. My SPC ports were off .2 and .4mm. He has an old manual cmm. I didn't know if it would be different.

If your programs are on a server would it make a difference? Our probe files are on the hard drives.

[baylorbear]

I will make a program for the BNS block and do it on both machines. Thanks.