Snap points

[AuRules]

Hit the F1 button and read the info on snap points kinda have a idea, but someone give a example or a better definition of, does it work best with sheetmetal or what? Help a poor brother out....

[Matthew D. Hoedeman]

Hit the F1 button and read the info on snap points kinda have a idea, but someone give a example or a better definition of, does it work best with sheetmetal or what? Help a poor brother out....

SNAP POINTS:

1) Not a great idea to use for ITERATIVE alignment points
2) Great idea for all other VECTOR points in the program
3) Can be used in SOME instances for constructing other features, sometimes not, however, all features can be constructed with NON-SNAP points.

Snap Points simulate a perfect machine measuring a vector point, the perfect machine being able to STAY exactly on the approach vector, not deviating by as much as a micron. Snap points are a short cut for something YOU could do yourself with about 6 more steps.

a) Vector point nominal IJK used to create a 3-D line going through the nominals of the vector point
b) Plane perpendicular to the 3-D line is constructed THROUGH the actual measured point
c) 3-D line and PERP. Plane are intersected giving the reported XYZ location of the point. This puts the point EXACTLY on the nominal vector line eminating from the nominal XYZ location of the point.

By using SNAP point, the formula SQRT(Xdev^2+Ydev^2+Zdev^2) will match exactly the "T" axis as reported by Pcdmis

By NOT using SNAP, you must use Xdev*I+Ydev*J+Zdev*K to get the matching "T" axis value.

Both formulas are correct for the RADIUS of a SPHERE, which is really what the "T" value will represent, the distance from the center (XYZ NOMS) to a point on the surface (XYZ ACTUALS). HOWEVER, if NOT using SNAP, the ACTUALS will not match due to the 'drift' of the CMM, thus requiring you to use the VECTORS (IJK) of the point in the calculations to determine the "T" value.

[Glenn.Russell]

Here's another basic way to look at it... Say you were interested in measuring table top (some plane parallel with your granite)... Often, all you really care about is the Z location. But, usually, you wind up with some small deviation in X & Y. This can be caused by tunneling error in the machine, etc. Snap allows you to look at just the deviation in the primary axis, and deviations in your secondary and tertiary axes will be 0. -Glenn

[AuRules]

I think I'm catching on ole' "GREAT ONES OF WISDOM" so since I'm measuring a spar which is 100" long and 10" wide and .080" thick has ship cordinates once I use a interate alignment to align the part then all of the other points should be snapped due to having no blueprint and having only the cad model and T values to go by. Could this be correct?

[Matthew D. Hoedeman]

I think I'm catching on ole' "GREAT ONES OF WISDOM" so since I'm measuring a spar which is 100" long and 10" wide and .080" thick has ship cordinates once I use a interate alignment to align the part then all of the other points should be snapped due to having no blueprint and having only the cad model and T values to go by. Could this be correct?

That is the way I would do it. BUT, snap or no snap, the T value WILL be the deviation along the nominal vector from the nominal XYZ, or a 3-D distance. However, by using the snap point, you will not have some yutz asking you "Why, if the point is straight up and down in Z is there an X and Y deviation and why does the T value ignore them?"

[emingeri]

Slight Correction...

You indicated that the formula "SQRT(Xdev^2+Ydev^2+Zdev^2) will match exactly the "T" axis as reported by Pcdmis" this isn't exactly correct, the resultant value will always be positive. As you are aware a positive vector deviation indicates that the measured point exists outside material. This formula does not allow for a negative vector deviation resulting in a report where all measured deviations lie outside part material. That said, you provide the correct formula for T (T=Xdev*I+Ydev*J+Zdev*K), this formula is correct whether you use snap or not, it will always report the deviation along the normal vector. As for what the T-Value represents, it would be better to state that the T-Value is the 1 dimensional distance from the target to the measured along the normal vector.

SNAP POINTS:

1) Not a great idea to use for ITERATIVE alignment points
2) Great idea for all other VECTOR points in the program
3) Can be used in SOME instances for constructing other features, sometimes not, however, all features can be constructed with NON-SNAP points.

Snap Points simulate a perfect machine measuring a vector point, the perfect machine being able to STAY exactly on the approach vector, not deviating by as much as a micron. Snap points are a short cut for something YOU could do yourself with about 6 more steps.

a) Vector point nominal IJK used to create a 3-D line going through the nominals of the vector point
b) Plane perpendicular to the 3-D line is constructed THROUGH the actual measured point
c) 3-D line and PERP. Plane are intersected giving the reported XYZ location of the point. This puts the point EXACTLY on the nominal vector line eminating from the nominal XYZ location of the point.

By using SNAP point, the formula SQRT(Xdev^2+Ydev^2+Zdev^2) will match exactly the "T" axis as reported by Pcdmis

By NOT using SNAP, you must use Xdev*I+Ydev*J+Zdev*K to get the matching "T" axis value.

Both formulas are correct for the RADIUS of a SPHERE, which is really what the "T" value will represent, the distance from the center (XYZ NOMS) to a point on the surface (XYZ ACTUALS). HOWEVER, if NOT using SNAP, the ACTUALS will not match due to the 'drift' of the CMM, thus requiring you to use the VECTORS (IJK) of the point in the calculations to determine the "T" value.

[oddam]

If you have a check fixture that uses a go / nogo feeler check and you measure the part on the gage, (especially if the measured area is curved) and report the T value, that deviation should tell you if the go/ng check will be acceptable or not. Eg. the gage uses a 6mm feeler with a tol. of ± .5mm. You measure the part in the area of the feeler check. Now look at your t value deviation, is it a + or a -, now look at the deviation, is it less then .5mm? If it is, your go feeler should go between the part and gage where you measured, the nogo will not fit between the part and gage. If the dev. is more than .5, depending on the reading being a + or -, the go should not fit, or the nogo will fit. Turn on the graph to see the arrow dev. direction to confirm if you have cad. Hope I said this correctly, it's quitting time, I am out of here.

Odda

[JEFMAN]

You indicated that the formula "SQRT(Xdev^2+Ydev^2+Zdev^2) will match exactly the "T" axis as reported by Pcdmis" this isn't exactly correct, the resultant value will always be positive. As you are aware a positive vector deviation indicates that the measured point exists outside material. This formula does not allow for a negative vector deviation resulting in a report where all measured deviations lie outside part material. That said, you provide the correct formula for T (T=Xdev*I+Ydev*J+Zdev*K), this formula is correct whether you use snap or not, it will always report the deviation along the normal vector. As for what the T-Value represents, it would be better to state that the T-Value is the 1 dimensional distance from the target to the measured along the normal vector.

You're right !
"T=Xdev*I+Ydev*J+Zdev*K" is exactly the dot product between the vector of deviation (PT1.XYZ-PT.TXYZ) and the normal vector (PT.IJK or PT.TIJK if it's a vector point. For a surface point, TIJK and IJK should be different, and it should be better to use only IJK !)

[WolfMan]

I am a visual person, if its not a burden, could some one point me to some graphics explanation of the snap

If I have a vector point with Snap ON, and the THEO vector is .707,0,.707, so the Actual vector will be exactly .707, 0, .707 without any deviation?


If that is the case, why would you want not to use Snap ? Dont you always want the THEO vector match the Actual on all the features?

[pwhitsa]

I am a visual person, if its not a burden, could some one point me to some graphics explanation of the snap

If I have a vector point with Snap ON, and the THEO vector is .707,0,.707, so the Actual vector will be exactly .707, 0, .707 without any deviation?


If that is the case, why would you want not to use Snap ? Dont you always want the THEO vector match the Actual on all the features?

You cant use them on iterative alignments as you want the cmm to align to the actual vectors and not the theo vectors.
I have snap turned on for all surface points on machined surfaces after the alignment is done.

[WolfMan]

You cant use them on iterative alignments as you want the cmm to align to the actual vectors and not the theo vectors.
I have snap turned on for all surface points on machined surfaces after the alignment is done.

what about on other features like circles, planes etc.

[pwhitsa]

what about on other features like circles, planes etc.

Doesn't apply.
Only on surface and vector points.

[WolfMan]

Doesn't apply.
Only on surface and vector points.

Plane is a surface, so you would use SNAP on all of your planes?
never mind, there is no option for the AutoPlane

[JEFMAN]

Snap point is usefull (really ?) only when you want to skip deviations on axis perp to the vector. For example, if you want to measure a point along Z axis at theo 0,0,0, the cmm will measure (for example) 0.003,-0.002,0.004 without snap on and 0,0,0.004 with snap on. On a plane, or on a low curved surface, it's not important, and in this case, the z value won't change a lot for a few microns defect. If you do it on the vertex of a cone, the error along x and y will create a defect on Z, and with snap on, you won't see them.
Maybe snap point is only available to have a nice report...?

[WolfMan]

Snap point is usefull (really ?) only when you want to skip deviations on axis perp to the vector. For example, if you want to measure a point along Z axis at theo 0,0,0, the cmm will measure (for example) 0.003,-0.002,0.004 without snap on and 0,0,0.004 with snap on. On a plane, or on a low curved surface, it's not important, and in this case, the z value won't change a lot for a few microns defect. If you do it on the vertex of a cone, the error along x and y will create a defect on Z, and with snap on, you won't see them.
Maybe snap point is only available to have a nice report...?

Well, personaly I can see how it could be useful, for example if you have a Gage point at a specific XY, or a contour that is based on a bunch of XZ points. And where you tol is+/-.001 or less. every minor deviation does counts.

In fact I had a contour where the profile was .0005, so yeah, and the XZ came from a gage point. So it was very important that the gage point was in the right spot.