Diameter occasionally measures wrong with RT

[[Er...]]

Anyone here that has a good explaination why a diameter measurement (RT scanning with rotation) gives the wrong size on some rare occations? Error range is usually around 0.03-0.05 mm smaller than actual size.

Everything else is near perfect, form (diameter feature, BA features, RT-Axis features on part, etc), position, RT-axis, part temperature, environment, cleanliness, there is no excessive speed, stylus are rigid and not worn, measuring strategies, and so on. Nothing wrong with the reference sphere dia or stylus qualification etc.

Remeasuring with lower speed etc does nothing. But measuring without RT rotation, same stylus etc and it measures correct. Not exclusive to a single part or program. Just rare random error. 

Had a discussion with local AE and didn't get much of an answer. Except acknowledgement that this as been spotted before. The machine just gives the wrong size.

Sort of annoying.

[[Ch...]]

This is 100% true.  We have several +-.004mm and +-.002mm diameter shafts.  Also with insane form tolerances.  

My rule for my programmers is:

Form of the shafts - use rotary table with rotation.

Size and locations - absolutely do not use rotation on rotary table.

We measure the same circle twice, one with rotation and use for form, then again without rotation of rotary table and use for size.   

What's even more weird, how does form report more accurate with using rotation.  If it was an RT axis issue, then I wouldn't think this would be the case. 

[[Ri...]]

Please sign in to view this username.

 If we are talking RT-AB then that would make a lot of sense. The rotational accuracy of the AB is insane, and I've seen plenty of applications where we used the rotary inside of moving the bridge to achieve much higher accuracies. This is also why Zeiss has the form package for the Prismo that will get you in the realm of a roundness tester. 

This is the first time I'm hearing about size/location though. I'll have to inquire some more about this one. 

[[Ch...]]

I would keep an eye on RT value : 'Angle to main axis' as my rule of thumb I prefer it under 0.050 I typically can get it under .025, but anything over .050 I'm not confident in results.

As a double check you could always try 'indicating the part in' to run true, you obviously don't want to do this everytime, but this may tell you more info about what is going on , if it is software related, etc.

You could also try measuring from a different side, for example if you are using a #5 probe, try measuring at +X while scanning ,or also try at +Y

I've solved these issues numerous times, so I think we can help find the problem eventually. Initial check is to see if part is not loaded as well as previous piece, etc. and checking angle to main axis. Are you clamping a round/cylindrical part in chuck jaws?

[[Ch...]]

Please sign in to view this username.

This is on RT-AB's.  I have a Prismo on the way and should be here early May, I'll have to check out that form package.  

I highly recommend you run the test yourself.  When measuring high quality gage pins or rings, you will see more accurate diameter measurements when not rotating the rotary table, and more accurate form measurements when using the rotary table.

[[Ch...]]

Agree, that measuring without R/T will always be a little bit better, because you have an additional axis and possible source of error, along with software corrections for the R/T compensation.

That being said I found (when having to measure master gears with sub micron tolerances), by indicating your part in prior to the R/T axis measurement on the part, this helps the software compensation. However if the part is clamped/located flat, and the R/T correction is done properly, you can have a very large amount of runout and it will compensate most things accurately, however I have seen the diameter size drift issue more than a few times.

If a chuck is being used, I find if you indicate that in to best condition, and parts are made precisely, and operators load properly, I've had many lots of parts run through without any issues.

 

Do I recall Zeiss having an 'auto-mechanically-leveling' rotary table? That might be cool - I might be something of someone else though.

 

 

 

[[Ke...]]

This thread has been an interesting read...

I've been using a RT-AB for over 5yrs now, and the only time that I've had an issue with the diameter reporting incorrectly was when I had a shaft that was chucked incorrectly (visually not perpendicular to the RT)

Please sign in to view this username.

 Our RT-AB is on an ACCURA II and the correlation of Roundness measurements to our TalySurf is already very good... Any idea if the Form package would be compatible? Does it just affect the accuracy, or are there Roundness reports that are unique to the Form package?

[[Ri...]]

Please sign in to view this username.

 It is unfortunately only available on the Prismo Ultra. :(, but here is some information on it. 

 

 

They guarantee 10% or less tolerance usage on 2-3µm tolerances (so essentially 0.2-0.3µm accuracy). 

 

Having said that, the accuracy on the standard RT-AB is already extremely impressive. 

 

I'll do some more testing on our RT-RB-100 because I'm interested in quantifying the measurement discrepancy between diameters measured with/without the rotary. 

[[Ch...]]

I think the R/T - AB select I think is the 'best produced' (e.g. most flat & round) of the all of the R/T's they purchase;   which I heard once upon a time that are actually  acquired from Klinglenberg. (which I'm now running Kling P40 and R300 equipment 😀 )

[[Ch...]]

Please sign in to view this quote.

Please sign in to view this username.

I started running master gears on my Micura with no rotary table for this very reason.  

I need to know who makes your sub micron master gears.  I'm not very happy with our master gears at the moment.  

[[Ch...]]

Please sign in to view this username.

  Not sure , bunch of different suppliers from InvoSpline to Broach Masters, I've been inspecting master gears roughly 20 years. Most in the midwest some from northern California - we had the capability of doing them ourselves although only have done them once or twice, because technically you  need a 'lab' to verify. 

I apologize for going off topic - but if most have a bore of 0.500" would be good idea  to make an arbor whose post protrudes and all diameters are concentric < .0001". Then you can mount part in chuck and indicate the arbor if needed. I've done most without an arbor in the chuck with measuring the bore and gear with R/T. I have lots of experience inspecting gear and spline gages on GMMs.

One note I will add is that most master gear and spline gage drawings lack a lot of information regarding proper datums, and tolerances per AGMA master gear chapter info, including bore and face runouts, etc. Spline gages - the teeth actually do not have to be true to anything on the outside of the gage, only to themselves, which can be a pain to establish the pitch diameter as datum, but I have also done many times. If I was making the drawings, I would just put a perp and parallel callout for inspection purposes only.

I'm impressed your able to inspect master gears WITHOUT R/T, star probe ? I think you have a Prismo Ultra too correct ?- would have loved to have one of those.

 

We ordered the master gear on he zeiss webshop last year - not a good experience, super late, and not the greatest quality Ive seen some profile error on one side over 2 micron , but within its limits.  Could have been a one off issue.

Edited 1 hour ago