Stated Accuracy of Portable CMM Arm and Real World Use

[slimething]

My company is considering the purchase of a Romer 7520 arm (2m range) for one of our plants to use for machine setup/alignment (non-critical) and some inspection. In their literature the repeatability is quoted to be +/-.016mm and accuracy +/-.023mm and certified per B89.4.22.

Shouldn't the true reported repeatability be .032 and for accuracy .046, for example measuring the distance between two holes? Wouldn't the uncertainty be the sum of those? I find the +/- to be a bit misleading.

Our other plant had a demo performed and the plant manager is sold on the machine. My plant currently has a smaller manual CMM. I obtained a quote for a larger DCC CMM as we do different types of work than our sister plant. Rumor has it however, that instead of upgrading it to a newer DCC CMM, some are suggesting the Romer could replace the CMM for inspection.

I am having deep reservations about this because we routinely measure machined details with TP of .030, concentricity .013, perp .013, diameters <.025 tolerance and so on (metric).

In a meeting I brought up the accuracy issue. The quoted DCC CMM has a MPEe= 3.2 + L/300 accuracy and MPEp=3.2. which would suit most of our needs and provide greater throughput. The probe system is the PH20 with TP20.

The supporters of the Romer say the machine is more accurate than the spec when measured at closer ranges which is believable, however since Romer does not offer an accuracy statement for that, how can it possibly be known what the accuracy/repeatability is throughout the measurement range? Admittedly I've never used an articulating arm and do realize the advantages of them, however some tell me they are not well suited for close tolerance measurements, and since Romer only quotes the total volumetric error, how reliable can the measurement be? I don't think it should be any less than the stated accuracy in their documentation. If they are are much more accurate in the smaller envelope, than where is the cert to prove it? What do you do, measure the feature until it falls in tolerance and accept it as good sarc/off ?

Using the 4:1 TUR (test uncertainty ratio), I can't imagine ever trusting an arm to measuring with any confidence closer than +/-.13mm tolerance.

Anyone out there that can give me some insight into this would be greatly appreciated as I'm afraid by purchasing this arm my plant may be making a big mistake, but could be wrong.

 

[steve68steve]

I'm not super qualified to answer, but I lobbied to sell our Romer Arm.

The Romer Arm has twice axes of a bridge CMM, sensitivity to operator technique, often instability due to mounting... it's a lot of slop to compound, IME. Their stated accuracy was confined to a small measuring volume and dropped off quickly, IIRC (compare with the '+L/300' for your CMM - I'd be surprised if it wasn't worse).

For your application: "repeatability is quoted to be +/-.016mm and accuracy +/-.023mm...we routinely measure machined details with TP of .030(metric)".

What am I missing here? I assume the .016 accuracy is linear (and only over a certain length). TP is diametral, so your .030 TP is really a radial .015 dislocation. Let's say the TP is off 0 in X and .015 in Y - that's .030 TP. But you're going to measure that .015 Y location with repeatability of .016? ...and accuracy of .023? More than your entire tolerance is consumed by gage error.

I'm not sure I could even demonstrate TP of .030mm on our bridge CMM - accuracy is a function of part size (+L/300), and on a large part with datums far from features, to repeatably measure an X and Y accurate to tenths (sorry, .0001" - machine shop) is already pushing the machine's capabilities, IME.

Again, I'm by no means an authority, but I would NEVER want a many-axis, human-powered probing system to replace a 3 axis mechanically driven system... and I wouldn't consider a machine with stated accuracy and repeatability at or higher than my spec capable.

 

[slimething]

Thanks for your reply.

You are exactly correct. The gage error (quoted in their spec) is greater than the tolerance for a TP of .030! In fact, it is more than 2x greater.

The argument is the arm is much better than the quoted +/-.016 repeatability and +/- .023 accuracy when measuring smaller envelopes. My argument is I don't care what the salesman says, if they can't put it in writing and quote a standard used to validate it, they are just blowing smoke.

If they put it writing, they'd have to back it up with traceable data otherwise their lawyers would be very busy people.

That's why I want real world users to give some feedback as to what they actually experience, and hopefully did correlation studies of an arm compared to a standard CMM. I suspect if such studies exist, the arm would fail miserably.

Again, I know there is a place for arms, and would definitely be beneficial in performing machine setups and the like.

 

[steve68steve]

"The argument is the arm is much better than the quoted +/-.016 repeatability and +/- .023 accuracy when measuring smaller envelopes."

That makes no sense at all...
Our CMM is quoted at 2.4micron + L/250mm. The "L/250" means (if I understand correctly) that for every 250mm long the thing I'm measuring is, I lose another micron. So, if something is 500mm long, Zeiss claims 2.4 micron + 500mm/250 = 2.4 + 2 = 4.4 micron. Applying the 4.4 micron at 500mm long on my CMM is about right - back of the envelope says I can hold about +/-.00017". Experience backs that up - I can measure a part about that size and watch the result bounce around about .0002".

The Arm numbers you quoted - .0XXmm - are already an order of magnitude worse than our CMM.

Whoever is pushing for the Arm is either being seduced by it's coolness, or is a victim of "truthiness" or confirmation bias or whatever: just because they want it to be true don't make it so. The numbers are staring them in the face.

I took a 2 day class at Hexagon on operating the Romer Arm. The guys in my class used the Arm for verifying WELDING setups. Another guy was fabbing SHEET METAL. They needed something with a lot of travel, portability/ flexibility, which can measure to specs which are +/- 1/16" or 1/32". The Arm is IDEAL for this. It's strength is it's portability, flexibility, and ability to run a PC DMIS program - NOT it's accuracy. Use the right tool for the job.

 

[slimething]

I hope this study is not typical of an AACMM. I don't have link posting privileges, so Google this document:
Application of a Force Sensor to Improve the Reliability of Measurement with Articulated Arm Coordinate Measuring Machines


Are there users out there that have considerable experience measuring bore diameters with an arm? What kind of form/size errors do you typically see when the true diameter is known such as a ring gage?

Thanks.

 

[Kingsld1]

The other issue is DCC vs Manual. A DCC machine is designed to load a part, press GO and walk away to do other things. Any type of manual machine will absolutely require someone there to push it around.

Romer does offer rentals and leases. It might be easier to convince the powers that be (or get sold on it yourself) to try before you buy. One good realworld trial easily beats a thousand 'expert' opinions.

 

[Kingsld1]

Are there users out there that have considerable experience measuring bore diameters with an arm? What kind of form/size errors do you typically see when the true diameter is known such as a ring gage?

Thanks.

I would be very careful of using either a DCC or Manual CMM to measure bore diameters. They will not give you the same results as a plug or ring gage unless your parts have very low form error. Unless you are scanning there is a lot of unmeasured space between the points taken to measure the circle/ cylinder.

 

[hogheavenfarm]

I use a Faro, 8ft arm, accuracy given as +/- 0.002". Thats on a good day. As stated, mounting, extension, and human probing all add to those tolerances. Using the results in constructing lines and points (using their software) can lead to some completely bizarre results. It is frequent that I need to make multiple measurements and multiple calculations to get a dimension I am comfortable with. It has its uses, but absolute precision is not one of them.

 

[True Position]

You'll never get the tolerances you need with a manual arm type machine.

I'd be suspect of anything less than a scanning CMM being able to do everything you want. At the tolerances you've described (~0.013mm) form becomes a big issue and taking large amounts of points per feature on a touch trigger machine kills your throughput.

 

[j.s.kushwah321]

I use a Faro, 8ft arm, accuracy given as +/- 0.002". Thats on a good day. As stated, mounting, extension, and human probing all add to those tolerances. Using the results in constructing lines and points (using their software) can lead to some completely bizarre results. It is frequent that I need to make multiple measurements and multiple calculations to get a dimension I am comfortable with. It has its uses, but absolute precision is not one of them.

Hello everyone
i am using portable faro arm with cam 2 measure 10.6 software
Facing a problem to in calculating egocentricity of to internal cylinders......
in which appropriate mandrel is going threw out but faro in calculating 2.2 mm...
please give me some feedback