A Question Regarding T-Value...

[Mike Johnston]

PC-Dmis v.3.7mr2 CAD++
Windows XP
DEA Mistral 7-7-5
Programed off of a STEP file

Can someone give me a detailed explanation of exactly what is reported when asking for the "T" value? I'm pretty sure it is a vector deviation of some type but to what extent? I'm curious about some values that I've been seeing in one of my inspections. My overall width comes in perfect but the individual T values are showing the part to be undersized by .2mm. It just doesn't make sense. I've never used this tool in PC-DMIS so it is new to me.

Mike

Thanks in advance for any response.

[kbotta]

Normal surface deviation.

[Paul Sarrach]

It is basically the profile of a point. It is the distance 90 degrees from the surface or edge. If you use with on an edge point and use the "S" it will give you the suface dimension.

[Bill **********]

Since you're using CAD your T value is deviation normal to the CAD NOMINAL surface vector. Effectively a -T value is saying that your surface is shy from nominal and a +T value is saying that your surface is full. The fact that your WIDTH is spot on makes sense because WIDTH is effectively a distance measurement and where it's at isn't considered in the calculation.

Bill

[Matthew D. Hoedeman]

sqrt(xdev^2+ydev^2+zdev^2)

This is how it works for vector and surface points. It will also work that way for edge points, IF you have GAP ONLY turned on. Otherwise, it will not quite appear to work that way for edge points, even though it does, the values will not quite work out because they will also include the surface deviation IF you are taking surface sample hits. If you are NOT taking surface sample hits, then the deviations will be the same as if you had GAP ONLY turned on for the edge point.

It is what you really SHOULD be reporting for the vector, surface and edge points as it will give you the real, true, 3-D vector deviation from nominal of the point.

SNAP points really will have an effect on this as well. The T will report the 3-D vector deviation, but if you are NOT using snap points, the XYZ deviations may not match the T (by the formula) because your machine might have drift, in fact, it probably does have drift. This will show up very well if you have a point with a 'perfect' vector, 0,0,1 or 0,1,0 or 1,0,0, etc. The 2 axis with NO vector SHOULD always report zero deviation. If the machine has any drift, they will not. I have never seen a machine without SOME drift. So, for a point at X12, Y24, Z36 with a vector of 0,0,1, you might see an actual of X24.0035, Y23.9955, Z36.0123 which should report a T value of 0.0123, but the axis will show deviations of X+0.0035, Y-0.0045 and Z+0.0123. Using the formula, this would give a T of 0.0136, which would be incorrect for the actual deviation. This is where SNAP points lord it over non-snap points. They will give you actual values ALONG the vector line and ON the vector line and the calculations will come out perfect, thus eliminating the ******** or customer who just doesn't 'get it' and never will. I use SNAP points for every vector point I take.

[Mike Johnston]

You guys are good... Thanks for the great explanation Matt. That's just what I was looking for.

Mike

[jmgreen]

I went thru the same thing with BnS when I noticed the T value did not match the sqrt(xdev^2+ydev^2+zdev^2) as Matt pointed out. Attached is the official BnS explanation of T value. They also put it in their last news letter. Well I can't get the dang thing to attach as a document file so I zipped it.