If I now understand you correctly, in step 4 you're basically verifying that the reference square is truly square. You're flipping the square in that step, not the part being tested (this is what caused my confusion).
My (possibly faulty) understanding is that the bearing point on the DTI simply establishes a straight line (two points) between the base (your faceplate) and the object-being-tested (the whole point of swinging a curved surface at the bottom to find maximum deflection is just to ensure you use the same point on the base surface each time you measure).
You want to know whether this imaginary line is truly perpendicular to the surface plate or not. One way to verify is to compare to the line produced to one known to be perpendicular (like a cylinder square). Another way, if the object has a flat top and bottom that are verified to be parallel (easily verified with the same DTI and surface gauge) is to flip the object upside down and re-measure. Any variation will be double the angle off of perpendicular.
Unless I'm missing something, the best way to verify an
angle-plate-looking-object (possibly one that can't be flipped top-for-bottom) is truly square is to compare it to something known to be square.
My very slightly modified sequence (for clarity) would be something like the following. (DTI: dial test indicator, OUT: object under test):
- Find or borrow a known true surface plate (doesn't need to be huge, but big enough to hold the OUT and gauge at the same time) and a DTI.
- Create a squareness gauge with a dial test indicator on some form of curved base (either a mag-base on a faceplate, or a surface gauge with a curved base).
- Find or make a precision square (a 1-2-3 block, for example, or a cylinder square).
- Verify your precision square is truly square as follows:
- Verify with the DTI and surface gauge that the top surface is the same distance from the surface plate at all points.
- Place the point of the DTI on a vertical/side surface of the OUT, with the curved base of the gauge touching the OUT.
- Swing the curved base to find the point of maximum deflection and zero the indicator.
- Flip the square top-for-bottom, re-measure, and verify the dial is still zero (if not, it isn't a square)
- Replace the square with the OUT and re-measure. If the dial doesn't read zero it isn't square.
I suppose if you were really anal, you would re-measure at multiple horizontally spaced points along the vertical surface of the OUT and 1-2-3 block (to test for wind in the vertical faces). The advantage of a cylinder square over a rectilinear square is that there is only one vertical line at each point around the cylinder, not a complete surface (potentially in wind).
And if you really, really, really, want to start from first principles and only "stuff on hand" you'll need to make three surface plates using the "
principle of symmetrical distribution of errors" before step 1. You'll need to fabricate a scraper and find some marking medium (red-lead or hi-spot blue).
Wait.
First you'll need to make a blast furnace and to read up on refining iron ore....
The more I learn about metrology (and this entire hobby) the more I enjoy it.
--
Rex