CMM or Mic Accuracy Question

[PeteM]

Just to add to all of the above good advice, measurement pressure can make a couple tenths difference in readings. That's why mics meant to measure tenths or better have some method of controlling and sometimes indicating pressure. Your customers probe may be gently touching, while your mic anvils are tightened down a bit on the part. If you're hand holding the mic, that bit of extra heat can also open up the frame a bit and give you lower readings.

Part of this pressure effect on readings can be surface finish, but another part can be flex in the mic body and a bit of give in the thread/nut.

Could be you and your customer will get closer agreement if you take this variable out of the mix, perhaps by using an indicating mic held in a stand and measuring with whatever light pressure (and room/part temperature) gives best agreement.

I'm assuming here that lobing has been ruled out. If not - a two point mic isn't the right tool.

 

[Micmac1]

So ive checked our mics on a master block and a master round, measure witin .00005 and thats by hand, ratchet stop. Obviously there is some room for error with it being done by hand, gaurenteed accurqacy on the mics is .00005. so that makes sense. They are nist traceable.
The parts are round, as said before they are ID honed & then OD ground on centers
Ill update on what we find out

 

[Larry Dickman]

check it with a height master and or gage blocks. That would be the bottom line.

 

[75sv1]

The OP did compare his results to CMM results. You have to understand the validity of the results. It was mentioned about Indicating mics, and that is what I used on fine tolerance parts. I bought my own after a while. Also, Starrett make use to make a 1 inch mic that has a tension sequence for tenths. I do have one. Microns is what I work in, so it is what I have the confidence in.
Also, if both mics are calibrated to the same standard, then they you probably read the same. Also, if I make parts to the same standard, then they would interface even if the standard is slightly off.
Also, on CMMs or even other measuring equipment, setting the info to ASME vs ISO can have an effect. So, lots of stuff to watch on CMMs.
Also, one instance I had, was I was supposedly measuring bad parts. They came in good on the CMM. A From Test was measuring them bad. Pictures were sent. I took one look and I could see the problem. They were clamping in a 3 jaw on one of the datums.

 

[StrawberryBoi]

That's a good point, that there are two ways of thinking about a tolerance. There is the universal interchangeability way, following ASME Y14.5-2009 or the ISO equivalent and doing everything you can to validate by verifiable standards that dramatically exceed your part tolerance requirements. There is also the single-supplier or single-Q/A approach, where the part is verified against literally the same actual calibration piece with the exact same method to verify each time (ie, always the same type and tolerance class of tool, not switching types of measurement tools).

Both achieve parts that fit together, but only the first is 'universal'. It isn't a more stringent standard excepting that it means you have to agree on all of the testing conditions. Temperature, pressure, type of calibration equipment, measurement equipment.

Sometimes you as the client need to provide the calibration standard to compare against, but you need to still own that standard. If you send the job to another shop, the standard goes with it. If you are selling the parts to your own client and they take the work somewhere else, then you need to be clear to them that the particular standard is the basis for their part fit and offer to sell it to them (or refuse if you are going to make that power play).

In the end, this should be a note on the print, ie:

X) All dimensions apply to a free part at 23°C, 1 atm, and in accordance with ASME Y14.5-2009 unless otherwise noted. Method of inspection, metrology tool part numbers, tool serial numbers and calibration certs at the time of measurement to be provided with each part along with the measured values and not simply pass/fail results.

You can even dictate something like:

Y) Dimension (3) to be checked vs a certified class ? granite slab while datum B is on the slab surface and part is free.

I've had situations where are part needed to be round in installation, but didn't need to be round when free. I used a note such as this:

Z) Diameter dimension (7) to be checked with part at known constant temperature through entire part volume while affixed to Pin B in the provided test fixture xxxx-yyy, with ID (6) installed on the pin until both Datum C and Face (9) are within the non-chamfered section of the pin length.


I mean, it's frustrating to get ridiculous, but when you start getting into tenths, you often have to begin to own the process. If your supplier owns the process, you are stuck with them until you can develop a matching process. The more detailed notes are you owning the process and owning any fixtures for executing that process, so that the supplier no longer owns the part business for that part and also so they know exactly what they are supposed to supply. They might be less happy to know you could take that process anywhere, but ultimately you'll both be happier because they won't provide parts they are confident are good and have you reject them.

 

[ttrager]

Our Procedures classify measurment tool use as follows. Context: We do work that has .0002" total tolerance.

I'm not going to list the coarser tools such as Calipers.

+/- .0005 Micrometers can be used.

+/- less than that, and certainly at the .0003 mark means a dial indicator and stand, or for OD's and such an INDICATING MICROMETER (.0001) calibrated to Class 0 gage blocks or Class X or better gage pins.

We also have a CCP Optical Comparator we use with auto-edge detection, but that is calibrated to a gage pin / plug prior to measurement to see if we need to comp the measurement manually. The comparator (auto edge detection) regularly reads - .0001 to -.0002 on larger diameters).

A good set of Micrometers are pretty good, digital or manual, but, in our experience at sub .0005 measurements, and certainly at .0002 to .0003 total tolerance, we've found that Indicating Mics / dial indicators are far more accurate and reliable.

 

[eKretz]

Bring one of your gage blocks over there and have them measure it while you're standing there...

I've run into this issue when getting cutters ground for milling v-ways on a press ram. The cutter came and was good on the 60° included angle but was tilted off-axis by a degree. I checked it before using it on a sine plate and found the error. Called the tool grinding shop owner and he told me "No way, I checked that with an optical comparator, it was not off." So I brought the whole apparatus over to show him. He insisted on putting it back on the comparator... and promptly found that it was 1° off-axis...

 

[gbent]

A good CMM operator can make it read what ever he wants it to read.
Don


Sent from my iPhone using Tapatalk Pro

But But But the printout says its good!

CMM results are more subject to programmer's or user's ability than physical gages. Any fussy dimensions off our CMM were also accompanied by form error on the printout. Without form error, there is no way to know if a bad point was used in the dimension calculation.

 

[TheBigLebowski]

Are the parts long enough to lay sideways on a surface plate?

When I have to grind a plug gage, I like to lay it sideways on a surface plate and comparatively check it vs gage blocks using a good indicator. I have found that regular thimble micrometers have a user induced "Fuzz" in the range of 0.0001" to 0.0002", the surface plate method gives a little bit more accurate/repeatable reading.

 

[75sv1]

But But But the printout says its good!

CMM results are more subject to programmer's or user's ability than physical gages. Any fussy dimensions off our CMM were also accompanied by form error on the printout. Without form error, there is no way to know if a bad point was used in the dimension calculation.

This is a point I meant to bring up. Form Error, Form Fault or Form Deviation can have various inputs. First is the Probe set calibration. That too can have various inputs. As stated a damaged probe. Also, a lose probe set or too long of shafts for the probe. Generally, I like 1 micron or less on calibration.
Another to watch is the Form Error in the measurements. If high, then is the probe shanking? Part surface rough?
And then there is the Super Machine attitude. Not just applicable to CMMs. What is the CMM rated for? We don't measure our ring gauges on our CMMs. We use a Form Tester.