Checking runout from cylinder to cylinder

[[Ju...]]

Is there a way to make an end point of a cylinder when checking runout? It seems to be projecting the cylinder, my datum, to the next cylinder. Thanks in advance for any help.

[[To...]]

Your datum cylinder axis is infinite in length. The runout is only evaluated within the length of the feature cylinder's depth.

[[Ju...]]

Is there a way to stop the datum cylinder from being infinite?

[[To...]]

It would not change the results, even if you could do that to the datum cylinder. The length (depth) of the feature cylinder can be edited but then you would be cheating. Also, if you didn't create the cylinder from the model, you're likely already cheating if you didn't correct the nominal depth. Sorry, I don't mean to sound harsh.

Look at the attached sketch. The position of the feature along the axis of the datum cylinder does not change the runout results.

runout cyl to cyl.pdf

[[De...]]

Please sign in to view this quote.

This is true, however an important caveat is that any inaccuracy in measurement of the datum feature that causes errors in location the Datum Axis will be multiplied as you get further from the Datum feature.

In my experience, most error in runout checks comes from Calypso not being able to find the actual Datum Axis due to Taper and or Out of Round condition in the Datum Feature.

I have attached an image below to show this..

[[To...]]

Please sign in to view this quote.

I agree but this doesn't change the fact that the datum axis is the datum axis, correct? That's why the creation of an A-B datum is a preferred callout on a part.

[[De...]]

No it doesn't, the datum axis is the datum axis. That is why I show the 'actual datum axis' on my picture as well as the 'measured datum axis'.

And Yes, the reason a A-B datum is so helpful is that it usually allows for a Datum Axis to be established over the length of the part and therefore any error caused by measurement of the datum axis is reduced due to measurement error being in relation to the entire length of the datum rather than being projected out past the end of the Datum.

I just figured I would point out the high probability of bad data from the Datum being the source of the problem.

If a Datum has even a very limited amount of form error, an OTE evaluation can show excessive runout when a LSQ check shows the runout that matches a good manual check. Unfortunately, if cylindrical Datum features are not created so they have good form it becomes virtually impossible to check runout on the CMM and attain accurate results.

[[Ju...]]

So if a measured Datum axis is already out of round due to over-tightening the part the further it is from the original measurement the more it is going to show my runout to be out of tolerance. Is there a way to stop this or a way to manipulate this to get a more accurate reading?
I get what everyone is saying about "cheating" the system and that is not what me nor my company is about. I can check the part in a v-block with two indicators, one on the datum (OD) and one on the runout bore, and get that the part is nowhere near what my CMM readings are giving me.
There's got to be a way around this???

[[To...]]

So, it sounds like you're willing to do just about anything to make your parts pass, right? Here's one idea. Instead of measuring only Datum A, you measure the other end. Then, create a 3D line feature and under options, select Recall and select both end features. Instead of using the real datum A, use the 3D line. That should help your results similar to sitting in a v-block.

Disclaimer: The previous statement does not reflect the opinion of the author. Just trying to help a brother out. A discussion with the design engineer should follow.

[[De...]]

Best method, get manufacturing to make better Datum features.

As Tom said, making a combined Datum A-B that uses both ends of the part will give more consistent measurements because the Datum Axis is comprised of the entire axis of the part. This works best on parts where there are mounting journals on either end of the part. Unfortunately sometimes part geometry does not allow for this option.

As I said previously, the most amount of error comes from the evaluation method, using outer tangential is correct, however it is severely impacted by bad data / bad form. Depending on how the form is on the actual parts being measured LSQ may be acceptable in cases where averaging the measured points would not adversely effect the CMM's ability to determine part centerline.

For instance, if you have a part with taper on the Datum, Least Squares will likely give a more accurate result than Outer Tangential will. In some cases where the out of round is consistent like a tri-lobe shape the same generally holds true since averaging the data does not seriously impact the location of the datum axis. The problem lies in the fact that if the program uses least squares, then an informed decision must be made as to whether the problems with the data will detract from the accuracy of the measurement results or not.

I would suggest the following.

Make sure to get as much data as possible on your datum, 4 circles and 4 lines so you can evaluate form and make a valid determination. This helps for visual evaluation as well as for the algorithm to get a better result for the centerline of the feature.

Add a Cylindricity characteristic and use it with Cad evaluation to get a visual result so you can see what is going on with the form of the Datum.

Then assuming after review, it looks like taking the average of the form would still give the center axis of the datum, then try to evaluate the runout with your datum set to least squares. (minimum zone should be used on the feature)

If your form is not concentric, then using LSQ is not advisable and will likely be of no help.

[[To...]]

Derek is 100% correct. I would first try to see if I can make my program produce more consistent, repeatable features, which should make for better results. If all else fails, then you do plan b.

[[Ju...]]

I do appreciate the advice however not the low blow on my integrity. It's the engineer that I am needing to convince which way is giving the most accurate measurement. He is the one that checks in the v-block and tells me the CMM is wrong.
I will give the cylinder with the 4 circle paths and 4 line forms and the 3D line method a shot.

Thank you again.

[[An...]]

Please sign in to view this quote.

The form deviation of the datum makes the difference.
See attached.

Radial_Runout_1.pdf