Alignments using Cylinders

[[Co...]]

I have a part that uses a plane as Datum A, a cylinder as Datum B, and a plane as Datum C. B and C are perpendicular to A.  There are other cylindrical features(ie OD threads) that are concentric to datum B.  I believe that there is projection error being introduced using a cylinder as the secondary datum that makes calypso report that the features are out of tolerance for position.  I check the part using other measurement techniques and it says its good. 

Does anyone have any tips or suggestions to dealing with potential projection error that gets introduced using 3D features(Cylinder) as a datum? 

What is the recommended way for building these types of datum structures in Calypso to insure that the machine reports accurate results?

Thank You!

[[Ma...]]

To make it clear:

Using that setup as base alignment -> avoid using cylinders - use more circles to get 3d line. This will avoid flipping direction axis.
Next - using it in base alignment is not following correct calculation for alignment.

Correct way of using this setup is after base alignment by defining gemoterical alignment.

Problem is, that your secondary datums have to be perpendicular ( in this case ) to primary datum, which is not in base alignment.
Great example is shown in Zeiss Inspect using ISO/ASME calculation and Geometrical calculation.

[[Pa...]]

Cylinders can be a lot of fun.  In the primary position, they constrain more DOF than planes.  Where a plane is primary and a cylinder is secondary, there's always a problem if you adhere very strictly to the specs (ASME or ISO).  The problem gets worse the longer the cylinder.  Even worse, it's relatively safe to assume your results ARE accurate, and that your drawing is the problem.  For a quick demo, just swap primary and secondary datums just long enough for you to see the values drop precipitously.

Remember that the primary datum is going to (among other things) orient the coordinate system.  The coordinate system (alignment) is <first> perpendicular to the primary datum.  Further, your Datum Feature Simulators are all mutually perpendicular.  No correction for orientation is available.  Martin points to the graphic in Extras / Settings / Measurement / Form Datum.

So now we must assume that your primary plane and secondary cylinder are not perpendicular in real life, because they aren't.  They're close but not perfect.  In your case B in secondary locates the previously oriented DFS, which we know is perpendicular to A.  Now it should be easy to see how the DFS for your secondary datum will skew by some amount.  Assuming an OD cylinder, the DFS is the smallest cylinder which is <first> perpendicular to A and which will contain the entirety of the secondary datum. This smallest cylinder will always be bigger than the actual cylinder.  Worse again the centerline of this smallest cylinder will always be offset from the nominal cylinder, and  is now the CL of your DFS.  Because we are locked for orientation, the true CL of your DFS is never going to be your nominal CL for the cylindrical feature.  It's not really projection.  It's what's drawn and it will be different in every workpiece.

Depending on the situation I have used 2D circles at one end of a short cylinder to good effect.  I've also used constructions like the Perpendicular construction (this most often in the planar rotation) to resolve some of this.

Here at my shop in Maine we don't make any squares for customers.  We make big cylindrical shafts for jet engines, power generators, and oil & gas exploration.  In just about every case where a drawing shows up with plane as primary and cylinder as secondary, we have found it worth a call to the customer to describe the problem.  I've seen a 10" plane as the primary datum for a shaft 30 feet long.  By definition .0005" perpendicularity of the primary to the secondary means I should be putting the holes designed into the far end of this shaft in a different room.  Nobody ever wanted this, but it's what they drew. 

There's no way a machinist on the floor can locate this center so they'll always be shooting in the dark.  Every piece has a slightly different condition so every piece gets a slightly different result.

Apart from the government guys we usually get good results asking for the datum swap.  It's so clean it's always worth a shot.

Good luck!