Using a slight variation to Dinger's example mission, that allows the shell to follow it's default path. I get exactly the same results. Shells inparticular seem to be consistent. Bullets can vary a bit more, but they are always within the times posted above, give or take 0.002 seconds.
I think FWatch will cause the problems you have seen? If you want to grap the XYZ position of rounds to use in a graph, output the results to a dialog in OFP. Then you can just cut and paste them into the likes of Excel.