.1 mm undersize reading

[terry]

I had to go in to work last Saturday night to run a hot job for a customer 9pm to 5 am. To get the job done in 8 hours I had to run the same part on two machines at the same time B&S 7-10-7 and a B&S12-22-10 both running 3.7 mr3. After I had completed the run I was compiling a report for the customer from Datapage and noticed a saw tooth in the run chart for the diameter I was checking but the true position looked normal. After identifying the 12-22-10 diameter checks was .1 mm smaller I recalibrated, same result very repeatable .006 mm over 3 parts. Ran the same part 3 times, after recalibrating the 7-10-7, and the turn measured .1 mm larger then the 12-22-10 still repeated with in .006. I confirmed that the 12-22-10 was measuring undersize with a snap gage. I switched tp20 probe body’s, 4mm styli, every thing I could think of that could have possibly made a difference but still had the same result. Both machines were running the same program at the same speed. I had been up for 24 hours so I was getting punchy and figured I would figure out what was wrong with the 12-22-10 this morning. When I got in this morning I put a set disk on the machine that was about the same size as the turn, guess what, it measured with in .004mm of the stamped size on both machines. I did a few more tests before getting pulled to do the next hot job but I could not repeat the .1mm undersize reading. The only thing I can think of that was different was I had the roll up door open more that usual and the temp may have rose a little bit but I cannot believe that it would have made that much difference and if it did it should have effected my true position readings. Any ideas where to look?
Regards,
Terry

[Hilton Roberts]

There are several posts on the subject that point to the same thing. Temperature and more specifically temperature variation is the issue most responsible for measurement variation when the same part is being measured with the same program on two different machines. If both machines show a continued bias after temperature variation has been corrected, then look for a temperature differential at the time of calibration.

H

[cmmguy]

Did you set the disk up in the same plane as the part was setup?

What makes it odd is the TP was correlating but not the diameter. Was the TP over a distance or on axis with another feature?

[Lee Johnson]

temp

How big was the part?
What was the material ?
What sort of temperature change occurred 10Deg ,30deg ?
With that info temp variation can be figured out.
If it was exactly 0.1 off I would look elsewhere. Errors dn't usually round off to nice even numbers

[terry]

I do not believe that the root cause was temp variation; the .1 seems to be pretty close because the smaller measurement was within .008 mm of the measurement of the correctly measured size on the other machine. I am leaning toward something not functioning correctly with the probe calibration. Seems like the machine thought the ball was .1 mm bigger than it was.
As for part info.. Planetary gear housing weighing about 125 lbs. Cast iron (heat treated in the spline). Turn is 180.055 +/- .013 mm. Alignment is a bit strange, part is leveled to a cylinder that was constructed from a best fit of six 85.50 mm circles (three on each side of the part that holds the gear bearings). The bearing bores are located about 140 mm from the center of the part.
Temp change was about 10 deg F, door was open about 10 min in to a 80 ish environment. Both machines were calibrated at the same time.
As for the axis of the set disk I measured it in two axis (one the same as the part) and got the exact measurement for both measurements.
The machines are with in spitting distance from each other.
Regards,
Terry

[terry]

Figured it out

I figured out how to get correct diameter measurement. I first thought it might have something to do with the hits so I changed the number of hits from 19 to 5 just to see what would change. The cmm would not measure the turn with out contacting ruby between hits when moving to the next probing position. So I changed the prehit retract from 1 mm to 2 mm and the cmm then was able to move from probing position to probing position without bumping the ruby. ( I have FLY MODE on). Guess what 5 bumps showed correct diameter measurement. I then change the hit number back to 19 and I still get the correct diameter measurement. I know the ruby was not making contact with the part at 19 hits but I think the machine was having trouble making the calculation because it was unable to correctly position the probe at 80 % move speed and a small prehit retract.
Regards,
Terry

[tking]

Good catch, Terry!

I like the name by the by,

Terry King

[Hilton Roberts]

I gotta get in my 2 cents worth. If you are measuring anything at 80% move speed and 1 mm prehit.....you are asking for trouble.......no way will your machine settle down before taking a hit. Your machine was not having a problem crunching the numbers........it was the numbers that were the problem.

H