I too have fought with similar things and am still learning. If the ways are not made correctly, parallel, then the gib gets loose at one end and possibly over tight at the other. Your tramming errors may imply that the ways are not square or that they are just not parallel and so the table becomes loose at the ends.
So the hazard of a gib which is too loose is that the table can torque about the z-axis. If this happens while it is translating then you could indeed see shapes like an hour-glass or others. At the least the table motion will not be square (Y axis is not 90 degrees to the X axis). On the other hand, if you over tighten a gib then you find that there is excessive drag and so the backlash increases as the lead screw and ball nut (1/2 nut) tend to jam before allowing the table movement. That is any slack has to be over come with a harder force being exerted by the lead screw, bearing, etc.
So I have run experiments on my CNC machine which has ball nuts. I fined that indeed by measuring the backlash as a function of gib tightness you can minimize it. But you want to check the backlash at several points along the axis of motion to make sure it is the same everywhere. If it is not then there is a good change that the ways are not parallel or that they are worn or that they .... have some other problem.
So one experiment that I think works is to just go a head and loosen the gib a lot. So much so that you can feel the table torque a bit. Then check the backlash. This backlash will then not be due to drag at the gib, but due to other parts of this axis of the machine: bearings, couplings, etc. Anyway, it this is a baseline value you can use when you are trying to check on the effects of the gib tightness.
A gib has a defined tapper to match the ways. You can take this out and measure it by measuring its thickness as a function of length. The tightening screw has a known thread pitch. So when you make 1 turn of the screw you push the gib 1/TPI of the screw and this represents, via the tapper number, how much space between the ways you are taking up. On my mill, as new from the factory, had a gib that was cut off improperly. Hence, it was too narrow and would go too deep into the tightening hole. By doing this sort of measurement I figured if I shimmed it by about 0.005-0.007" I could bring the end of it back up to where the tightening screw had not bottomed out! Someday I will make a new gib, but for now the shim works. I tried a couple of materials before I found a shim to use. Keeping the shim in place as the stage travels, and the gib slips a bit, is a bit of a problem, so I found some metal tape (Al, duck tap, which was about the right thickness) to use. ( It goes on the clamping side not the side that rubs the ways.)
I have deduced that my y-axis ways are not parallel. So compromises must be made until they get fixed. A big job. Not really removable like the x-axis as this is the base of the Mill!
Dave L.
