Thread: Scanning a hole with a CMM

We have made a gage 1 inch thick with a .8240 diameter hole dimensioned +0, -.0003" which has a 10-32 tapped hole breaking into the side of the hole at 12 oclock in the center of the 1 inch thick block. We wire EDM the hole with 4 skim passes and find that in the area where the 10 -32 hole breaks into the .8240 hole it is .0001 large (.8241) measuring with Sunnen dial bore gages, and confirmed with a 10 millionths resolution digital height gage like this:

http://www.fvfowler.com/onlinecatalog.html?page=shop.product_details&produ ct_id=1025&flypage=flypage.tpl&pop=0&keyword=motor ized+height+gage

If it is inspected with a scanning CMM, what do you suppose the diameter will be measured? Certainly they can't scan where the 10-32 side hole breaks in, but will it average the entire scan? I don't know what the scanning routine is , or if there is a standard? If a hole is 80% in tolerance will it check good?

We are probably going to remake the gage, but I wonder if it would pass a CMM scan?

Whether it passes a scan or not will depend on how much data you collect, and how you evaluate it, as far as which fitting algorithm you choose to analyze and fit the data. How are outliers to be handled? Often it is desirable to filter data points to eliminate "bad" pickups, or points in areas that will not effect the function of the part. Do you want to know the theoretical maximum size cylinder you can fit through the gage? Or perhaps you are trying to create a gage that would be used on a shadowgraph as a template, and its purpose is to allow light to pass where the part is out-of-round? I guess what I am getting at is, many parts that are nonconforming can pass CMM inspection depending on how the data is analyzed. The opposite is true as well. Many good functional parts can be rejected because a bad inspection declares them out of spec. And then there's those parts that machinists make and remake until they pass inspection, even when the inspection has no bearing on the reality of the function of the part.

It's a true position gage that a .8235 +0, -.0003 pin fits through. Outliers? From what you are saying, it sounds like its a "it depends" kind of thing. My thinking is the pin will be positioned correctly, the part they are checking is a stamped part. Just wondering if scanning a hole is done helically, vertically around the perimiter, or some hybrid if there is an inturuption. Sounds like it's up to the operator depending on the function?

Originally Posted by EDM JOE

How are outliers to be handled?

You need to be really careful about outliers.

Simply rejecting any readings that don't "fit" can invalidate the entire measurement cycle.

- Leigh

I am quite careful about outlier analysis and evaluation. Outlier evaluation should consider producing the part you are required to, as well as not tying your hands by spending more effort or money producing a part that may be scrapped due to an incorrect inspection. Obviously, this does not apply if you are required to produce a part to pass an entirely defined inspection process. My point is to consider all aspects of a parts analysis, its function, and the possibility if an incorrect inspection.

RJT,

Was the 10-32 hole plugged before being wire cut? Just curious.

Great idea toolmaker35! Secondary discharge (erosion) caused by a difference in flushing conditions is likely the culprit.

No they were not plugged, and yes I'm sure that was the problem. Plugging the hole is a great idea, whether we wire EDM or jig grind. Still wondering about the path the CMM takes to scan. Helical, random points, circular?

The cmm we use here takes the average of all the points taken.

On a side note it still depends on how it's measured if the inspector does not take a point around the threaded hole no worrys. but if the hole measures .824 plus the one bigger area the average could come out to ".82401" or something like that. So yes it could show the rest of the hole bigger. If the rest of the hole is on the smaller side .8237 and the one area bigger as stated then the average would still be under the .824

The best thing I can tell you is to take 5 or 6 measurments using your sunnen bore gages at randon places then average all your measurments. That will give you a good indication if your average is going to be over the .824 dim.

Shouldn't the customer for whom this gage is being built be involved in this discussion? It would appear to me that the amount of the hole that is out of tolerance is a small area adjacent to the intersection of the tapped hole. Probably caused by inconsistence in pressure of the wire or flushing when the wire crossed the tapped hole.

Lord Byron

Inconsistent wire pressure (tension) is highly unlikely, judging from the description of the remainder of the holes' accuracy. Involving the customer, if they are reasonable, is a fine idea. We may be discussing a gauge that is acceptable in its' current condition. Do you think the customer would favorably consider your current inspection of the gauge, RJT?

I'm sure the difference in flushing pressure caused the problem.

My customer is a very large company, the engineer who ordered it is aware of the circumstances, but it must pass their gage inspection to be accepted. It's a different department (inspection) and we are unfamiliar with how the CMM scans. We provide an inspection report that they have to verify. Our inspection report has an asterick and explination, the engineer says the gage will be functional, but it still has to pass their gage inspection. We are awaiting the results of their inspection.

Since I'm not current on cmm abilities, perhaps someone can fill me in. Back when I worked the field, about a decade ago, we wouldn't trust the cmm for any close hole sizes; in a contest between the cmm and a dialbore, the dialbore always won. If it was an LVDT gage, there was no contest. Note that everything was cal'ed, the gages were set to a ring within .002 of size (part of the milspec heritage) and most of the inspectors running the cmm's knew their stuff.

The CMM's here are all check least squared but you can change it to max inscribed. The ones ive seen didn't check helical. Dial bore will be off on a odd number lobed hole. I sat in a class with one of the board members for AMSE Y14 was teaching he showed me how a bore gauge could be wrong. Changed my mind on the bore gauge. I'm very much still a rookie on the cmm's.

Originally Posted by machinistrrt

Since I'm not current on cmm abilities, perhaps someone can fill me in. Back when I worked the field, about a decade ago, we wouldn't trust the cmm for any close hole sizes; in a contest between the cmm and a dialbore, the dialbore always won. If it was an LVDT gage, there was no contest. Note that everything was cal'ed, the gages were set to a ring within .002 of size (part of the milspec heritage) and most of the inspectors running the cmm's knew their stuff.

I'm the same way of thinking as you. Our CMM would give an average size, and would show the form (roundness), but I wouldn't trust it at all on this job. The machine *only* measures to a tenth resolution, and that is what you are saying your deviation is.
I bet the gauge is more than fit for function-just hope the inspector has common sense...

BTW - what is the general rule of thumb for checking a bore with a cmm? 4 points minimum I asume, but I was once told that the more points you take, the more potential error you have (talking manual so inconsistent pressure perhaps compared to motorised)? The rensihaws are fantastic nowadays, so again perhaps back in the day there was more trigger error on the older heads?
Just curious as it maybe an old wives tale...

RJT - if you check 4 points (11/3/5/8 oclock ie either side of the 12 o'clock posn) you'll obviously be ok

Originally Posted by barbter

I'm the same way of thinking as you. Our CMM would give an average size, and would show the form (roundness), but I wouldn't trust it at all on this job. The machine *only* measures to a tenth resolution, and that is what you are saying your deviation is.
I bet the gauge is more than fit for function-just hope the inspector has common sense...

BTW - what is the general rule of thumb for checking a bore with a cmm? 4 points minimum I asume, but I was once told that the more points you take, the more potential error you have (talking manual so inconsistent pressure perhaps compared to motorised)? The rensihaws are fantastic nowadays, so again perhaps back in the day there was more trigger error on the older heads?
Just curious as it maybe an old wives tale...

RJT - if you check 4 points (11/3/5/8 oclock ie either side of the 12 o'clock posn) you'll obviously be ok

Depends on the size of the hole. When scanning you don't want to overpopulate the data points. Scanning speed and points have a lot of variance and you have to know how to use both when using cmm data...something they don't teach to incoming receiving inspection workers..."Oh I'll just take 2000 points when scanning at 50% on a hole that is .250dia and my probe is .200 diameter." Our zeiss has proven to be repeatable and accurate to within .0001-.0002. I wouldn't trust it for that measurement IMHO.

Update. The part passed inspection, no word on any problems.

Very good. Congratulations.

- Leigh

To the OP: Regarding the scan type, typically it is a helical scan down through the bore for whatever specified length is required. We also have a Zeiss machine, and ir does a nice job, but like dstyr said, "I wouldn't trust it for that measurement". Many put their faith in CMMs for complete answers to metrology issues, but at this level of precision, a general-purpose CMM is not the tool to use for absolute assurance, IMHO. Expectations of absolute accuracy on tolerances of a few tenths are misplaced when it comes to CMM technology, as the great majority of CMMs are working at the edge of their own precision envelopes, regardless of those displays of 10-millionth increment resolution.