Interesting device but I still can't visualize the whole set-up. Does the indicator slide along that top rail? If so, I bet it sags or deforms several thou at least.
If your balls read at .002 difference, that isn't really acceptable error.
It may be a design flaw that you have to get balls lined up to less than a thou over a 30-inch length. Very hard to do
Of course, I may be simply not understanding the entire device - can you post a picture of the whole thing with a barrel on it? Also we need to see the end-blocks - are they one-piece or are they split?
How repeatable are your measurements? Have you tried to take a piece of tube of known wall thickness and tried to measure it?
I have made several wall thickness gauges in my shop and know that even if they appear simple they are hard to design so they work well.
