Probe tolerance needed to pass calibration

[bigtallanddopey]

Hi guys, I am hoping you can give me some advice regarding the annual calibration.

Usually our CMM is calibrated using an SP25M probe, however this has become faulty and so far management will not replace it. So when calibration comes round I will have to have on the machine the TP20 probes we have. However these are also faulty but they at least pass probe calibration on the sphere.

I have been trying to get the SP25M and the TP20's replaced (as they are both over 12 years old) but my Boss has very deep pockets and extremely short arms.

The TP20's we have (I have 4 at my disposal) all have the same fault of sticking sometimes during use. So when they take a hit the red light goes off on the probe head and sometimes it stays off as I am guessing the connectors inside the module are stuck open. I then have to manually touch the probe and reset.

My question really revolves around will this problem effect the calibration that Hexagon are going to be doing very soon. I keep saying to my Boss that the replacement cost of the probes is less than the calibration and if they don't pass you will end up buying both anyway, but he is deaf to it all as the CMM still runs.

Is there any kind of measurement routine I could run that could give me an indication as to whether the machine will pass the calibration with the probe modules that are being used?

Would be great to have some advice for this situation.

[brian10m]

As a quick check on my probe I measure the sphere using a0b0, origin SPHERE then measure sphere using tips A0/B0, A90/B90, A90/B-90, A90B180. All results (XYZD) should come in within ± 0.010mm, any bigger could indicate issues with probing system. Hope this helps

[DAN_M]

I'm sure that you'll do what Brian says. You'll report back to your boss in a professional, common sense sort of way. And he'll stare at you like you're speaking a pig-latin version of korean.

I've worked for a guy like this before and thinking back to that, if I were you I'd have one of two choices...

1) Suck it up and deal with it.

2) If the CMM being operable is tied in with customer requirements/the machine is something you DRASTICALLY need, just let it fail, Your boss will be forced to pay for it and maybe will learn a little bit about not listening to you in the future.

My old boss started listening to me when he realized that every time he DIDN'T it would cost him more than if he did.

[grandpaal56]

If it were me I would do exactly like Brian suggested but I would repeat the process several times to show repeatability. If it doesn't pass, it's "GARBAGE IN, GARBAGE OUT". you can't get reliable results from faulty equipment.

[DJAMS]

ISO 10360-4 describes a probing test where you measure a sphere, set origin, take 25 additional hits in a described pattern and get 3d distance from sphere center to each of the 25 points. Max distance - min distance is the error.

To evaluate the error value - use the first value from the accuracy spec for your machine. So if your machine spec is 2.5 + L/500 the error should be <= 2.5 micron

It's at least a way to evaluate just the probing error (mostly). May still fall on deaf ears.

[bigtallanddopey]

ISO 10360-4 describes a probing test where you measure a sphere, set origin, take 25 additional hits in a described pattern and get 3d distance from sphere center to each of the 25 points. Max distance - min distance is the error.

To evaluate the error value - use the first value from the accuracy spec for your machine. So if your machine spec is 2.5 + L/500 the error should be <= 2.5 micron

It's at least a way to evaluate just the probing error (mostly). May still fall on deaf ears.

Sounds like an interesting test, mite look into it.

I do run a quick probe check program like Brian10m describes, A0B0 is often fine, Other angles are often extremely close or out of tolerance. I do however have the tolerance set to 0.005mm

I just ran the probe check program straight after calibration this morning. doesn't look too good really.

Capture2.PNG

[NinjaBadger]

I do this (except just A0B0 and one other angle per probe) also with a 0.005 limit - Usually they're all in but occasionally one or two probes might have errors, so I re-calibrate those.

What probing system are you using dopey? Is it a TP20 straight on the PH10 or are you using the TM25 adaptor on an SP25?


Also how many hits do you use for the sphere test - is it the same as you qualify with?

Also what's your method? Do you never move the ref sphere? and go straight to it? Do you recall the sphere position as defined?

[bigtallanddopey]

I do this (except just A0B0 and one other angle per probe) also with a 0.005 limit - Usually they're all in but occasionally one or two probes might have errors, so I re-calibrate those.

Mine have all been in easily until recently, now calibration doesn't sort it out.

What probing system are you using dopey? Is it a TP20 straight on the PH10 or are you using the TM25 adaptor on an SP25?

Yeh I am using the TM25 adaptor. Will this affect the results?

Also how many hits do you use for the sphere test - is it the same as you qualify with?

yes they are the same amount of hits.

Also what's your method? Do you never move the ref sphere? and go straight to it? Do you recall the sphere position as defined?

The sphere is never (or very rarely) moved. to calibrate I always answer yes with DCC hits as the CMM is switched off every night. The sphere position is recalled from the calibration data.

I now cant even get probes to calibrate, pretty sure this is unrelated but when I went to calibrate a different probe it drove straight into the qual sphere. I then restarted and selected DCC with manual hit to locate. I take the single point, the CMM then takes 3 DCC points. I then get an error message, error in best fit 3d. No idea what ive done for this.

EDIT: scrap the blue stuff, sorted it. Just reset everything.

[NinjaBadger]

Do you use a master probe to define sphere position (assuming you have multiple probes etc)?

[JEFMAN]

ISO 10360-4 describes a probing test where you measure a sphere, set origin, take 25 additional hits in a described pattern and get 3d distance from sphere center to each of the 25 points. Max distance - min distance is the error.

To evaluate the error value - use the first value from the accuracy spec for your machine. So if your machine spec is 2.5 + L/500 the error should be <= 2.5 micron

It's at least a way to evaluate just the probing error (mostly). May still fall on deaf ears.

+1 !
You just have to use another sphere than the calibration one, in another place.
Instead of measuring the sphere and origin on it, you can also
ASSIGN/V1=QUALTOOLDATA("xyz","sphere_2")
ASSIGN/V2=QUALTOOLDATA("diam","sphere_2")
Then create a generic sphere located at V1.X,V1.Y,V1.Z with a diameter of V2.

This test with A0B allows to see how the sphere "moves" from homing, or from other parameters (temp comp...)

[brian10m]

Dopey, The reason I use the probe positions I listed above is because it will measure maximum distance between positions for my setup, the positions you have used won't give you the full picture also the sphere is normally not exactly 25.0000mm there is usually a small deviation which should be reflected when you setup your sphere for calibration (something like 24.99285) usually stamped on the base or label. It may only be a few microns but it's important to be as exact as possible, The nominal for the sphere should also reflect the exact size in the report.

So many things can influence the output.