Results 1 to 11 of 11
Thread: How to use a CMM
10-31-2011, 05:58 AM #1
How to use a CMM
This year I transferred from Electrical Engineering to a course in Engineering Systems Maintenance - this is more suited to my work. This is a big change as I am an Electrician and I don't have much Metrology experience. We have been asked to write a report on a CMM - it's structure, probe, etc. I am picking bits and pieces from the net but I'm confused over a few points:
1. What are the steps involved in measuring a part, is the first thing to calibrated the probe?
2. How and why do we align the axes - and what is the importance of this?
3. When our lecturer gave us a demonstration, he aligned the z axis. Then the machine was able to automatically probe the piece (gearbox) and give the results on the PC. How does this work?
4. What is the importance of the air bearings and do all 3 axes work on air bearings?
I know these are basic questions to you guys but it's nice to get the opinion of pro's. I would be grateful for any help with.
10-31-2011, 06:59 AM #2
3D Measuring with CMM
Welcome to the forum! There are many ways how you can use the CMM. I can tell you how I usually work in the below (simplified) description.
Question 1: which steps
Basically the steps are:
A) Evaluate which sizes from the drawing you want to measure (measuring plan) to know which probes you need AND who are the reference elements for your alignment(s)
B) Calibrate the desired probe(s)
C) Measure at least the reference elements for the main alignment
D) Align the part ("designate the Z-axis"; the secondary axis and a part origin).
E) Measure everything you need, as your measuring plan says.
F) Build a measuring report (enter nominals and tolerances)
Question 2: how and why aligning?
The need for aligning the part is to tell the CMM (-software) where the part is located on your CMM. When aligning, you determine the origin (the centre of a bore - in case of a gearbox) AND the direction of the part, since it can lie 'tilted'/rotated in 3 directions on your CMM. So you can put your part randomly on the CMM. By aligning it, you determine the main directions of the part. The CMM-axes don't matter anymore, the main direction of the part is the primary direction of your alignment. This can be a centre line between two bores, a mounting face; but this depends on the reference on your drawing.
See answer on question 2.
Question 4: air bearings
Most CMM's have air bearings on all axes. They run on highly accurate marble or ceramic guides. The combination of guides, airbearings and how square they are, determines the accuracy of your CMM. The more accurate ('straight') the guides were polished, the better (and pricier...). This is also how a CMM usually is more accurate than a CNC milling machine. And thus making 3D measurements a good way of inspecting machined parts.
Hoping this helps you in becoming more familiar in the world of CMM's. I do contract 3D measuring on a 5 axis CMM for advanced measurements (CAD, free form inspection, etcetera) in the Netherlands (so please don't mind my English...). Feel free to visit me online and/or send me an e-mail.
Kind regards and good luck!
10-31-2011, 08:32 AM #3
I am also in the newbie learning stages of running a CMM. (Trying to get OJT from a crabby nutbar who prefers to rant about everyone and everything is not working out too well.)
I would also appreciate any help anyone can offer. Are there any books I could get? Internet is baffling me.
We have an older Mitutoyo BHN715 with upgraded MCOSMOS 3 software, PH10M head, and TP200 probe.
I can perform most basic functions, both manual and auto, but am having trouble figuring out basic alignments. The best I can figure is to align the part as if I was going to make it (or the feature I'm trying to measure).
10-31-2011, 09:50 AM #4
Aligning parts in CMM Software
In my opinion, making a decent alignment requires skills and experience. It might be the hardest part of 3D measuring. CAD, CNC, free form and additional rotation axes are just not as essential as a good alignment of the part! Any book about CMM inspection would tell you all about how to make an alignment. And a bit of training could be usefull as well (by your CMM supplier).
Your PH10 & TP200 are great, but it doesn't matter for making a good alignment. That's up to you. Independent of the software you're using: either Mitutoyo's; Zeiss', Renishaws, PC-DMIS or mine: Metrolog XG.
You usually have a primary direction (which defines 2 axes of rotation); a secondary direction (which defines the third axis of rotation) and an origin which defines the 3 translation axes (X, Y and Z). This is fairly easy for box-shaped parts. And - like you said - the alignment is quite similair to the alignment you make on a CNC milling machine. However: a good alignment can be 'read' from a decent drawing. References (usually A, B and/or C with a square box around them) determine what elements will define your part's alignment.
Let me give you an example: plates for injection moulds usually have 4 holes on the corner. Most of the drawings have all dimensions starting from the centre lines. For your CMM aligment, you would want your origin in the middle of the 4 holes. CNC machinists usually align 2 holes on one side of the plate using a dial gauge. Then they call one of those centres X=0 and Y=0. Then they translate their origin to the middle of the plate.
On a CMM however, I use the surface of the plate for the primary direction. Secondary direction and the origin I set, using all 4 holes. For this I use an option called 'Best Fit'. This lies the origin in the CMM-software at the true centre of all 4 holes.
But maybe you can describe one of the parts you have, with a picture of the part and a copy of the drawing. Then everybody can give their opinion on how to align it...
10-31-2011, 10:53 AM #5
Here's a real-life problem from last week:
Had a shaft that was supposed to be bent. The two ends were the same diameter and the middle was necked down. Only the ends mattered. Overall length about 18 inches.
I laid the shaft flat in v-blocks and took two half circles of three points each, one at each far end, then used the center of these circles to draw a line to use as the basis of my alignment. Also had a plane taken on one end. All done manually.
Used this info to automatically take more half-circles, which were turned into full circles for diametral measurement. The two outermost circles were again used to create a line (presumably more accurate because more points were taken), which I then aligned to. In all, three circles were taken on each end. The result was surprisingly straight, barely over 0.001" when the centers of the circles were compared.
My rationale was that if I was dialing this in a lathe, I would dial just in front of the chuck jaws, and as far away as I could. Whatever fell in the middle was the runout.
My partner disagreed. He took the three circles on each end and made two cylinders. He then aligned to one cylinder and noted the difference between the two. His results were far greater than mine, about 0.009.
For the life of me, I don't understand why. I'm sure there wasn't that much room under the part when I rolled it on the CMM table, although I didn't try feeler gauges. Now it's gone and I can't.
Most things are simpler than that, but I'm still getting used to locking down all six degrees of freedom. I usually end up forgetting one and then not being able to run in auto.
All the books I've seen online seem to require prior knowledge of the basics, and I don't have the basics. Oddly enough, once it's set up I can usually do the fancier stuff.
(I learned to run CNC the same way many moons ago: one month OJT with a guy who was leaving and didn't care anymore, then I flew by the seat of my pants for almost a year until I was able to take night school courses. Made a world of difference.)
10-31-2011, 02:50 PM #6
Thanks for the speedy response guys. I'm going to try and digest all that info and I'm sure I'll have a few questions. As I said - I work in maintenance in a factory but I am not mechanically trained so all this metrology is new to me!
10-31-2011, 02:58 PM #7
Sean -- One important aspect of this sort of measurement is to understand how the CMM is actually going to interpret the instruction, or how the software essentially calculates what you THINK you are asking the machine to do. looking at an average centerline deviation among (6) circles is a different task than creating 2 cylinders and looking at the deviation between the two cylinders' center axes.
Check out the manual for the machine, and get some background on the calculations that are done for a given function. This is a perennial problem when using a CMM.
10-31-2011, 03:06 PM #8
Any recommendations on interesting reads on the subject?
10-31-2011, 03:22 PM #9
Thanks for all that - that's great. How do you calibrate a probe and if you are measuring a part - would you have the drawing of that part in your software?
10-31-2011, 08:33 PM #10
Always set your part alignment or datums over a long a distance as possible.
Every point taken by a cmm has a range of values returned just like any measuring instrument.
So lets say you have a 10 inch long shaft.
Now you take the alignment circles on one end with a total length spacing of 0.100.
Just to pick a number lets say that the machine has a repeatability of .0001.
Now if one of your circles is reporting .0001 off from actual your alignment is off by 0.057 degrees. (tan-1(.0001/0.100).
At the other end it would show you to be off by .010 inches if the shaft was perfectly straight.
If the part is not moved and rechecked often these errors will remain the same and the results will look very repeatable but wrong.
Slightly moving or reorienting the part sometimes "moves" the errors around resulting in wildly changing results. Although it does not always show up, if it does this is a sure sign of a problem with the technique used.
I have made many a person feel foolish by simply saying "Ok, lets move the part by a small amount and check it again".
I once had a customer check a master SNA-322 insert on his new cmm.
He was checking the angle between each side and no matter what he did the angles of the 4 corners would not add up to 360 degrees.
His inspector had come to the conclusion that the sides were "warped" in some strange way and that my master part was no good.
You need to assume that every point taken has some error to it and think about how that error effects your final output.
It is so easy to misuse a cmm if you think about the points as being absolutely true. Relying on a line projected outside of its actual probed endpoints is a recipe for disaster.
11-01-2011, 08:07 AM #11
You DON'T need to have a drawing (or model) of the part your measuring in order to take measurements; once again, it depends on the software you're using. Our shop has 5 CMM's: 2 manual, and 3 DCC (computer-driven) ones. The manual ones are exactly that: you move the probe head by hand, and manually take your "hits". (Probing technique is critical!) The DCC ones don't REQUIRE a model (or drawing), but are capable of using them.