ok, first attempt using sketchup 8

[Jonathan]

There you go again Jonathan completely missing the point.!! . . . . YES AGREE there are stronger and lighter ways to build a Gantry BUT and here's the point.!! THIS WILL EASILY HANDLE THE JOB OF CUTTING WOOD AND BE MORE THAN STRONG ENOUGH

Of course it will handle cutting wood well. A machine built from wood is capable of cutting wood but that doesn't mean you should make one.

My aim is to make as good a machine as I can, not one that's just meets my current needs.

From what I under stand Dean intends mainly wood use so the Z extension will be minimal, or should be for best results, so given that it's only 1500mm long and twin screw driven being securely tied into both the X axis bearing plates and the gantry end plates and with a good steel thickness on RSJ then the twisting(torsion) will be minimal.

A substantial joint at the ends, i.e. welded plates, will help but it's not the whole story. Good end fixity makes a big difference to beam bending but not torsion.

For instance, using the spreadsheet to calculate torsional stiffness relative to 180x100 RHS:
180x100 RHS section, torsional stiffness, 1
180x100 box section, 3mm wall, torsional stiffness, 103
180x100 box section, 5mm wall (same mass as RSJ above), 164
100x50 box section, 3mm wall (biggest from metals4u), torsional stiffness, 13.9
152x12.7 plate, torsional stiffness, 1.1

So you're talking 10-100x difference. The mass of the RSJ is greater than all of the above and the bending parallel to Y on all the above sections, except the plate, is negligible (much less than the torsional deflection).

So I suggest deannos should use the design Jazz suggested earlier with two 80x40 box sections, with the ballscrew between and a plate on the back. Easy to make with hand tools, no worry about parallelism of the rail mount surface and bending is less than all of the cross sections discussed above according to the spreadsheet. Unfortunately can't calculate torsional stiffness without FEA but it's clearly less.