S-Value vs. T-Value (Matt?)

[\v/inston]

a fixture shop gave us a pc-dmis cmm report with both T and S values reported, now everyone and their brother wants to know what the S-Value means vs. the T-Value. i'm having a hard time answering this since all the book says is this:

T = prints error along approach vector (for points on curved surfaces).
S = prints the deviation along the surface vector.

can anyone explain the difference in laymans terms and why you would use only T vs S. thnx...

[Jeffrey Lovely]

when taking an edge point t value is the blue arrow
s value is the green
t value is the edge
s is the surface
both are profiles

[Matthew D. Hoedeman]

T & S (t & s? ain't it T & A?) are only going to be used (together) for an EDGE point.

The T = Trim deviaiton
The S = Surface deviation

THe S only works if you are taking a surface sample hit.

If they are using the S axis for a vector point, change it. If they are using S for a surface point, change it. The S is ONLY for the surface deviation of a point (feature) that takes a surface sample hit. I just checked and it can also be used for a HOLE that has a surface sample hit.

SURFACE points should use T for (T)otal deviation along the vector
VECTOR points should use T for (T)otal deviation along the vector
EDGE points should use T for (T)otal deviation along the vector (edge vector) and S for the total deviation along the surface vector (only with sample hits)

You can also use RT or RS, but if you EVER see those in a report, THEY ARE CHEATING THE RESULTS! The help files are not much help, and they are WRONG, they say the default are RT and RS and they are NOT! RT and RS will report a deviaiton along the NOMINAL approach vector, regardless if you are using that as the TARGET vector. This is where teh CHEATING comes into play. They can change the target vector, which will change the probe compensation direction, but then report AS IF IT APPROACHED CORRECTLY!


VECTOR = T
SURFACE = T
EDGE = T (trim) and S (surface)

Accept no others!

[\v/inston]

thnx matt & jeff..

do you report S? i've always only reported T.

when you report an edge point that takes only one sample hit does the S work right or does it need 3 hits?

[Matthew D. Hoedeman]

If you only take a sinlge sample hit, then it uses the NOMINAL surface vector, 3 or more, it uses the actual surface vector, same as a surface point. I always use S for my edge points, it helps eliminate some point to measure by combining 2 readings with one point.

[\v/inston]

so you take 3 hits in the surface for every edge point you take? i normally only take one hit on the surface and one on the edge to speed things up.

[Matthew D. Hoedeman]

so you take 3 hits in the surface for every edge point you take? i normally only take one hit on the surface and one on the edge to speed things up.

NO, I only take one, I was just telling you what it does, single sample (MY WAY) uses the NOMINAL surface vector (for comp and depth calculation) while 3 or more uses the actual surface vector of the part. Unless you are WAY off angle, (like 10 or more degrees) you really won't see too much difference and we are WELL within that range, let me tell you!

[Matthew D. Hoedeman]

Worst case scenario:

Surface is 10 degrees off angle right at the programmed point of touch. Using a 4mm ball, the amount of error intruduced into the measurement is 0.030mm. This is how much error you might be able to remove by using 3 surface sample hits and letting Pcdmis use the actual part surface for the approach vector. For a 2mm ball, the error is 0.015mm. So, working automotive sheet metal with tolerances +/-0.7mm (or more, 99.5% of the time) that amount of error is not worth worring about, BUT REMEMBER, that is if the part is off 10 degrees. I would have to say we are within 1 or 2 degrees 99.999% of the time, the rare cases when it is more than 2 degrees would be heavy-metal-crash-form flanged parts. At 2 degrees with a 2mm ball the error is 0.000609mm, whooooooooo. I am really gonna worry about THAT!