Another One For Debate

[#2#]

Will an un-calibrated tip give good positional data?

I know that size data would be untrustworthy but are there cases where positional data would be unaffected?

Specifically, the operator here ran a program on a cylindrical part but did not calibrate the tip first. This program uses only one tip and one angle so correlation between tips/angles is not an issue. This program does not report any sizes (diameters) but only the location (TP RFS) of several different diameters back to the datum diameter.

Would the position data reported be accurate?

I'm trying to save some re-work and re-measure.

[Matthew D. Hoedeman]

I have done this myself, but ONLY if:

1) You are taking 4 hits, X+,X-,Y+,Y- so that the center point would be the mid-points of the 4 hits, thus giving a good center (no third axis vector!)
or
2) All the touches had the same vector.

Other than that, it will be wrong by whatever that calibration deviates from the perfect size.

[Jan d.]

I think you are fine. What is the ratio between accuracy required on the print and what your machine can do? If it is 10, you are sure to be OK. However, if you get under 4, you may need to run it again.

I think that next to ball diameter, you are also off in offset and probe length. That means the machine does not take the points at quite the correct location and you would introduce vector approach error. This error may get bigger the smaller the feature is.



Jan.

[Cumofo]

No,
Ever look at the theoreticals and the actuals? They maybe very different. If all probes were perp. to the CMM's axii then this wouldn't be an issue.
If this probe was using latent data from yesterday, it may be within the reported accuracy of the machine. If the prbe was just "built", then you have a great risk of uncertainty to consider.


G

[inspector212]

What if the probe is "off" in the X- vector, let's say .001mm. Wouldn't this shift the center of the cylinder the same amount in the same vector?

[Matthew D. Hoedeman]

It is ALL going to depend on WHAT he is doing:

If you touch off on the table and call that XYZ zero, then raise the probe up and stick a 0.5" jo-block on the table and take a touch on top of the jo-block, will the results be correct?

YEP!

If you measure a hole, that is in the Z plane (touch vectors ONLY in X & Y) and call it XY zero, then measure another hole, same thing (z-plane, XY-vector touches only), will the distance between them be correct?

YEP! IF, each of the 4 touches in each hole is 90 degrees to the previous (IE DCC holes, not manual)

IF you are measuring SINGLE HIT features, then all the features MUST have the exact same vector, then they will all relate to each other correctly.

IF measuring 2-D features (circles & slots), then ALL the touches have to be 2-D and all using the same 2 asic for touch.

NO 3-D features can be done correctly.

[craiger_ny]

Guys he makes shopping carts. Goodluck measure a 2X4. If the CMM says it is somewhere around 1.5 by 3.5 go for it, you are fine. You could have requalified it and remeasured your part by the time I posted you know. Or did you guys ship the parts? er I mean carts.

[bw_bob]

Guys he makes shopping carts. Goodluck measure a 2X4. If the CMM says it is somewhere around 1.5 by 3.5 go for it, you are fine. You could have requalified it and remeasured your part by the time I posted you know. Or did you guys ship the parts? er I mean carts.

ROTFLMAO.......thanks on a Monday

[#2#]

Guys he makes shopping carts. Goodluck measure a 2X4. If the CMM says it is somewhere around 1.5 by 3.5 go for it, you are fine. You could have requalified it and remeasured your part by the time I posted you know. Or did you guys ship the parts? er I mean carts.

Alright, sounds good. I've got a chunk of 2X4 under one leg of the cmm to level it. I'll just pull it out and measure that.

Thanks for all the input guys. I guess they are going to put the ummmm... 'carts' on hold for now and pull apart 5 and have us re-measure the part. If those 5 turn out OK then they are going to assume the other 295 are OK.

[craiger_ny]

Technically you should probably use a C=0 sampling plan or perhaps something from MIL 105D even though it is outdated it is still statistically sound.

[Wingman]

Will an un-calibrated tip give good positional data?

I know that size data would be untrustworthy but are there cases where positional data would be unaffected?

Specifically, the operator here ran a program on a cylindrical part but did not calibrate the tip first. This program uses only one tip and one angle so correlation between tips/angles is not an issue. This program does not report any sizes (diameters) but only the location (TP RFS) of several different diameters back to the datum diameter.

Would the position data reported be accurate?

I'm trying to save some re-work and re-measure.

If he used only one articulation of the probe and the diameter of the ball was defined correctly then you have no worries. Think about it. If the diameter is supposed to be 2mm and it was defined as 2mm it might measure +-10 microns or so. That is 10 microns MAX error due to size of ball being of from nominal. Features of size won't have this issue. Positionals will be very accurate as long as you equally spaced everything.

Worst case, remeasure one of them. If it is close to the previous data (I am sure it will be) then accept them.

[cmmguy]

The only problem(shopping cart tolerances not withstanding) is that when you measure single-sided features such as planes and lines and points. If these types of features are measured then the calibrated size of the tip will have an impact on the location of those features, which in turn will impact the location of diametrical features.

[#2#]

Technically you should probably use a C=0 sampling plan or perhaps something from MIL 105D even though it is outdated it is still statistically sound.

What one 'should probably' do rarely gets done around here. Anyway, it isn't my call so I'll just measure what they bring me. BTW, for this being such a hot issue one would think I would have seen parts by now Oh, well, I can't measure what I don't have and it will take just as long to measure if they bring it to me now or next week. They are the cause of the delay so I'm not going to hurry or worry.

[craiger_ny]

How did the "form, fit and function" check go? You did do that right? I know you were supposed to.

[John Kingston]

Technically you should probably use a C=0 sampling plan or perhaps something from MIL 105D even though it is outdated it is still statistically sound.

Craig,
Mil-Std-105 D was replaced by Mil-Std-105 E which was cancelled by the government body that oversaw it. Now there is ANSI Z 1.4 , which is identical to Mil-Std-105. With that in mind, what level and AQL should he use? If he uses C=0, are the characteristics Critical, Major or Minor. I feel that the shopping cart is should have all Major characteristics, as I may have beer in the cart. If it fails I would be ticked off having to get a straw and drink the beer off the floor.
John