repeatability

[Kevin Watts]

I have a part program that i have ran six times. the first measured a diminsion diameter at 2.0062" the second at 2.0055" the third at 2.0047" fourth fifth and sixth at 2.0048" and 2.0049". Same part same program no climate variations or movement of the part. program was ran back to back. i wiped the part off inbetween runs to ensure no dust had settled. the diminsion has a .001" tolerance and is a semicircle so a gage pin wont work. i use a manual alignment and dcc alignment. why would the diminsion be different on the first three and then repeat on the last three. nothing has changed. any help would be appreciated. also would after the measurement repeating the alignment and remeasuring in one program help to check for repeatability.
thanks.

[Paul Sarrach]

The CMM is not a good tool for checking size of holes,slots. Why cant you use a pin?

hole size is not a good feature for repeatablity either.

[Matthew D. Hoedeman]

What are the specs of your machine? You are showing 0.0015" of variation, max to min, which would be a spec of +/-0.00075", is your machine at least THAT good? IF you follow the 10% rule, your machine SHOULD have a spec of +/-0.000075" to accurately measure that amount, and if you ahve a tolerance of 0.001" (or +/-0.0005") then your machine should have a spec of +/-0.00005". That's pretty tight!

[kbotta]

He can't use a pin, because it is an incomplete negative feature (hole...or part there-of..
Kb

[craiger_ny]

.......... is a semicircle so a gage pin wont work. .....

He can't use a pin, because it is an incomplete negative feature (hole...or part there-of..
Kb

Makes it even worse (size accuracy/repeatability wise)

[Paul Sarrach]

He can't use a pin, because it is an incomplete negative feature (hole...or part there-of..
Kb

Still dont see why you cant use a pin ? Can I see a pic or a print? Enliten me Kev. Espeically if it that tight of a tolerance. I am not trying to be funny. I just cant seem to get it in my mind.

[Underspec]

I would make sure to calibrate your probe again then measure the part "without" touching the part. Any movement of the part will throw it off the next time it's measured.

The reason Paul says it's not a good tool for measuring holes etc, is because your probe is not touching all of the diameter (unless you have a scanning probe) and gives you a "best fit" diameter.

[craiger_ny]

Perhaps if he has more than 180º of the feature he could use a pin, perhaps (I'd want to see it too). But less than 180º I'd be hesitant with 0.001 tol.

Craig

[Kevin Watts]

we can't use a pin because this is a radius with 180 degree arc. The machine was calibrated last month and certified to .0001 accuracy and repeatability. What's odd here is we loaded the part and ran the program six times in a row. the results are as follows
1st 2.0062
2nd 2.0055
3rd 2.0047
4th 2.0048
5th 2.0049
6th 2.oo49
My question is why would the machine vary on the 1st two runs and then "settle in" and repeat within .0002 on subsequent runs back to back? Would it possiblly help to do 2 duplicate DCC alignments? I can live with not being accurate within .0007 but I'd like to improve repeatability so I can make a decision on disposition of the part.

[kbotta]

Perhaps if he has more than 180º of the feature he could use a pin, perhaps (I'd want to see it too). But less than 180º I'd be hesitant with 0.001 tol.

Craig

Right there paul...and thats a tight tolerance, esp. for a diameter...on a cmm...without more than 180' of surface....= tough.

[kbotta]

Did you change your measurment speed? Time of day? Alignment of Moon and Stars??
Kev

[Paul Sarrach]

I would make sure to calibrate your probe again then measure the part "without" touching the part. Any movement of the part will throw it off the next time it's measured.

The reason Paul says it's not a good tool for measuring holes etc, is because your probe is not touching all of the diameter (unless you have a scanning probe) and gives you a "best fit" diameter.

Honey? Is that you? I swear the only other person who told people what I though was her.

Sorry I just had to do that. But Hey, I made myself laugh.

[bboyd]

Your picking up and replacing the part in between runs could point to the form of the part if you are not hitting it in the same places everytime.

Did you get the larger sizes initially with them gradually shrinking after a period of time could point to temperature change if your part came from the shop into the lab climate control environment vs none.

If you have a ring gauge of approx the same size then try a repeat of that

[VADER]

If you are checking radii as opposed to a full I.D. the variation you are seeing is quite nomal. My machine is in excellent condition with very tight specs. and my error is not even half of that spec. I see this all the time when measuring radii.

[Kevin Watts]

Basically we're trying to measure the radius at the bottom of a feature shaped like the capital letter "U" the multiple runs were back to back without moving the part, program cycle time is about 5 minutes so it was a total of 30 minutes with no temperature change.