turn off "deviation perpendicular to centerline" - will result in spherical, 3 axis true position.
I would recommend a 2nd look at your program, make sure you are using the most accurate feature to capture 3 axis (definately not measured features). The only reason I would see to use spherical true position would be a cone or a sphere; if a cone, I highly recommend 3 sample hits.
Strictly speaking, the standard calls for a cylindrical zone instead of diametric.
Technically correct, but misses the mark. All
diametrical True Positions are 3D, or 3 axis, and their zones are cylindrical.
I would guess that the OP is asking about TP of a hole on a compound angle to the FCF. It is still a cylinder, but unlike a hole square to an axis (at (0,0,1) you can locate the hole with only XY axes, at any given crosscut of the cylinder), you need all 3 axes to locate a compound feature.
OP, you can check "Deviation Perp to CL" to force the dimension to calculate as though it were looking straight through the hole. (This ignores which axes you have chosen to report out, whether XYZ or XY, etc).