Finding and nailing true position

Sirs,
I am posting this issue to you because as business owners you are without question experts if faced with tight true position features on parts sometimes having only a few tenths given for it based on so called bonus tolerances.


Using new and improved CMM machines brand new we have many situations where how we have done a part for years suddenly is not right. We have always noticed that arch shaped parts have caused the CMM guys to wish to pull their hair out or return to drinking and they have good company for the dismissed temptation.



Are the machines part of the problem. Are the ruby or carbide ball tips wearing too much needing replacement? Is it the way a program is written having as it's X0Y0 referenced and programmed from the X center point of cut jaws to hold the shape machined on open one while the Y zero is not like on the print referenced at a point where everything is referenced from?


I understand making more complex parts I would like it best if my brain could figure out what is wrong...process,torque pressure,tool sharpness, depth of cut ,relieve stress, or is some of it software or basic CMM issues perhaps a parameter in the CMM is wrong or the part should be secured better.

Since I live in Texas I remember the Astronaunts asking Houston do you copy or Houston we have a problem. I am calling out liked them to you trusting in your experience as owners and including trusted and successful managers any feedback to shine some light on this issue .

Not understanding how to trap or fix this problem I humbly ask for any enlightenment you may graciously provide.

Thanks in advance,
Sincerely asked,
Spinit

So you're being tasked with holding tenths on true position on contoured features, and sometimes it goes sideways, and you're hoping we know why? Not on any specific part, just in general?

The most likely cause is your operators are not eating enough. If you were to buy a hotdog cart and wheel it around to your workstations to keep the operators well fed, that would be an excellent start.

Sounds familiar.

Does this have anything to do with a company who's initials are Schlumberger?

First you have to ask the tough question:

While the measurements show the part is out of tolerance - why assume the CMM wrong?

This begins a process of double checking or verifying other measurements with alternate means, if available. If you measure something 3 different ways and the results agree w/in a reasonable number, then you maybe just have a machinist/material/tooling/workholding problem.

What do you mean "so called bonus tolerances" ? I love me some MMC !!!

Curves are about the worst thing to locate from. Especially arcs that don't have a consistent radius.

When faced with a problem like this, I usually look for the inconsistencies first. Was there a new operator? Work shift? Program edit? Try to see if anything changed and if that change would give you a positioning problem.

I usually don't consider the probe tips to be wear items. Now if you just got a new CMM and are all of a sudden having problems, then I would see if the new CMM checks the part the way the old CMM did, or if you were checking the part on a CMM in the past at all.

I've been around several parts that ran fine for years. Send one to the CMM to be checked and find out what it's out of tolerance. Then that leads to the management decision of going off the CMM results or going off of Joe Blow's test jig.

I will chime in here and add that a cmm, much like a cnc, is only as good as it's operator. I second the suggestion to check with alternate methods when possible. I think I shared this, but I consistently got 'bad' numbers from the cmm (really the guy running it), that time after time I had to show the guy how the hole checked with gage pins*, or a micrometer measurement, or indicator and gage blocks, etc.

*Not talking tenths, although we do have some sizes in tenth increment pins. One time it was .005"!

Can you not verify the CMM with knowns.

.

Few things in a machine shop illustrate the old axiom "Figures don't lie, but liars figure" like a CMM does. If you are proving a part on a surface plate you can have a half a dozen people witness it and get buy-off or rejection on questionable features on the spot. If you witness a program on a CMM you know nothing about what actually happened unless you have a complete copy of the program, not just the output.

One thing my head of QC taught me was always print out the errors of form. That way if you measured a chip, shanked a probe, or didn't catch the tip on centerline the error of form would often show the feature wasn't established properly.

"Sirs,
. We have always noticed that arch shaped parts have caused the CMM guys to wish to pull their hair out or return to drinking and they have good company for the dismissed temptation.

Are the machines part of the problem. Are the ruby or carbide ball tips wearing too much needing replacement? Is it the way a program is written having as it's X0Y0 referenced and programmed from the X center point of cut jaws to hold the shape machined on open one while the Y zero is not like on the print referenced at a point where everything is referenced from?

I understand making more complex parts I would like it best if my brain could figure out what is wrong..."


Hmmmmmm... CMM - I will take a guess and figure the old way of checking has shown parts to be in tolerance using what-some kind of functional gauging? Try measuring with the CMM in a linear pattern (one axis direction) rather than a random pattern. Sometimes when measuring spherical radii or even linear radii you can pick up a profile that seems is out of tolerance but really is within the tolerance band. To check you can layout points around or imposed in a 3D model of the part. Using the linear technique you can layout on a paper and calculate at certain sections of the part. UOS - Beware of non-existent profile tolerance.

I watched a cmm guy "fix" an out of spec part.
The bore was measured by touching 12 places around the bore.
He deleted first point, ran the cmm ,still out of spec, restored that point, deleted the second, and so on until the point deleted was the one that caused the bore to measure out of spec.
So was the part good, or not?
.

Toolroomguy said:

I watched a cmm guy "fix" an out of spec part.
The bore was measured by touching 12 places around the bore.
He deleted first point, ran the cmm ,still out of spec, restored that point, deleted the second, and so on until the point deleted was the one that caused the bore to measure out of spec.
So was the part good, or not?
.

Click to expand...

It's no different than mic'ing different locations of the bore until it reads correctly. This isn't special to CMMs. They are no more and no less honest than any other gage.

It's the Indian; not the arrow.

I once had a customer tell me there cmm told them my 1/4 slot was a 1/4 to wide. I replied my eyes say your cmm and it's operator are idiots. I grabbed their calipers off their table and they agreed with my eyes as well, however the qc guy still said parts where bad. I replied which is it your cmm is wrong or the qc guy who calibrated your calipers is an idiot. The guy I was talking to was the caliper man. I won that fight but lost a customer. As mentioned above could be a lot things, but it could be a bad qc man. Later Jason

Edit: By the way, this was a tier one automotive company. I can say the headlights in your car my be out of spec. Also note I fired this customer after this meeting not the other way around. It sure felt good.

IME, it's more about how you align the part on the CMM. I've seen perfectly good parts check bad because the CMM operator didn't establish -A-, -B- and -C- correctly.

WARNING, wall of text incoming. There are CMM's involved though.

Some years ago now we were contacted by the above mentioned company to make part of a saddle valve assembly for them.

They wanted us to make the "flapper" for several different sizes.

The part shape consisted of several intersecting concave and convex surfaces.

Supposedly they were so busy they didn't have any spare capacity to make them.

Their process involved wedming the initial profile from bar stock and then a combination of milling, edm, and wedm to finish the other profiles and details.

The customer was making these from X-750 Inconel and getting 4 pieces from each 9" section of 12" diameter bar stock.

We managed to get 9 ...

Because of the complexity of the parts we even bought a CMM to check them with.

It was during this "project" that we found out just how much inconel, especially inconel bar stock, can move when you wedm it. We had to press the first piece out of the bar with the hydraulic press in the tool room.

It took two more pieces before we managed to get one that didn't move so much the first profile was out of tolerance.

It wasn't until we sent a bar to Bonal Technologies for a trial of their Meta-Lax process that we got really good first profiles.

Then the fun started.

The second wedm cut was @ 90 degrees to the first and produced a curved seating surface and the hinge pin hole, both of which were toleranced to +/-.0005 true position.

The first part was ok, the second was much better. Our QC manager and I checked and verified each other's findings on both the CMM and surface plate and it was this second part that we took to the customer.

Some interesting things came out in the initial meeting there before we got down the checking the part.

At one point early on I asked how it was that they came to ask us to do this work for them. They replied that our name came up in the vendor's list of the company they had bought at this location.

It was at this point that we found out what was really going on with these parts as one of them announced that he "didn't understand what was going on" with them because the previous manufacturer, whom they had bought, "never had a problem with them but now that we've brought in CMM's to check them we can't get a good one!"

That raised my eyebrows ...

Then we took a walk through the facility to the QC area, where they had several large CMM's.

One thing I certainly noticed along the way was that they certainly had a lot of capacity that was not being used. So that part of the initial story was false.

Then they handed the part to the CMM operator and we stood and watched through what seemed to me a painfully overlong process of checking and, rechecking, and checking it again.

In the end everything came up green except for one dimension. A dimension that I knew well because it had come up wrong for me too ... until I realized I was checking from the wrong datum!

Well, dummy me who hadn't felt the need to bring a copy of the customer's own part-print with him back to their facility asks ... "Do you have a part-print handy?"

To which their reply was ... "No, we don't need one for the CMM. It tells us if the part is good or bad."

I was incredulous, and nothing ... but nothing ... would convince them that their precious CMM could possible have been programmed to check the part wrong.

It took two weeks, a lot of talking and hand waving, the print-outs from our CMM, and a part-print in my hand to convince them of it.

We eventually had to fire that customer because the parts we sent them would sit on a shelf for 2 weeks - 2 months before they would check them.

Then they would reject them because they were out of tolerance by +/-.0001" on one dimension or another, and send them back to us for rework. Whereupon we would check them and find nothing wrong.

Then there would be weeks of negotiation after which they would either accept the parts or we would rework them. Sometimes we would rework them and sometimes we would "rework" them.

Then they would go back to the customer and the process would begin again.

The last batch of parts we sent them sat on a shelf somewhere for 2-1/2 years, after we fired them, before they did that ...

KilrB said:

....
Then they would reject them because they were out of tolerance by +/-.0001" on one dimension or another, and send them back to us for rework. Whereupon we would check them and find nothing wrong.
.....

Click to expand...

As niffty as they may be there are not very many CMMs out in the world you can trust to a tenth.
That is the calling of very, very high dollar machines. You have to be a goodly way into the world of six-figures here.
I can make a $30,000 machine repeat down here in a demo and sell you on it. I know darn well it is not right and that a program change will give different numbers.
It amazed me many times just how much customers really wanted to trust these things.
Bob

KilrB said:

Sounds familiar.

Does this have anything to do with a company who's initials are Schlumberger?

Click to expand...

No sir yet I have made a lot of nice things for them and did it good.
Also really my company has concerns about sharing too much and take seriously proprietary information so I respect and obey I wish I could be more specific guys.

Stuff

I had a CMM fellow tell me a 1st piece was wrong. I showed him the hard checks using machinist measuring micrometers and so on to explain patiently why I found different and to find out his input to that.

I also explained what I was using to check ,why,and also how to read them. After this he came to me (a great guy who's personality gets along with maverick machinists even the rude ones whose professionalism is non existent)

Well he took a square gage and measured it on comm and it was measuring -.01 undersized! Being on nights we told days who have resources to check the machine calibrating it like our guy did to no avail.

That day we all helped him confirm by hardcover the accuracy of print tolerances. They called out the man who fixes those machines. It is older not complicated stuff like we have moved to. These machines seem to say everything at times is wrong and different readings on same machine's twin.

Go. figure! It is our process and procedure to use them this can not go away it is required by our customer who have similar issues we are told. It slows everything down on parts accepted with proven programs we have used for years.

It is frustrating and like any problem we just begin to seek solutions as complaining is only useful in finding a problem and should stop while we turn to finding out the solution.

Responding to OP here,

Your trying to solve too many problems at once. In the end it's what your customer measures that really matters. So, you need to make sure your on the same page. I'd recommend a sample part that you both agree is correct, so you have some kind of standard. Once, you have that nailed down you can decide whether you need to tweak your processes or calibrate your CMM. I believe doing both at the same time results in a technical paradox known as 'chasing your ass'.

Short Answer : air gauge

Edit: Did you say your CMM is off by .010 and parts tolerance is given in .0001? Why bother turning it on? Just use a height gauge and a ball bearing. Or really barley corns would be a better method if the CMM is that far off.

Cheshire said:

.....
Short Answer : air gauge....

Click to expand...

I've designed and built many air gauges but admit to being confused how this works on a arch shaped part.
Bob

CarbideBob said:

I've designed and built many air gauges but admit to being confused how this works on a arch shaped part.
Bob

Click to expand...

You got me Bob. I have no idea either, but figured you'd be able come up with something.