Thanks again. I've just done some rough beam bending calcs and, just as you say, it looks like the deflections will be OK if I ensure that the ends are adequately well constrained.

I've assumed, as a worst-case, that the beam length will be 800mm, with the cutting load applied at the 700mm point (this assumes that the head travels over the central 600mm part of the beams). I've also assumed that the load will be shared equally by both beams and that they can be considered to be simple point loaded cantilevers (not quite right due to the constraint on the "free" end, but a pessimistic simplification).

With a 50N force applied at 700mm from the constrained (driven) end, I get a total deflection of 0.01745mm, which is small enough not to worry about!

Torsional stiffness may be an issue, as I'm using two beams that are only constrained at their ends. I want to run the spindle between the beams, so can't box them in to increase the area M of I around the common long axis. I've done a rough calc of the amount of torsional deflection at the centre of a single beam, assuming that the cutter was offset by 250mm from the beam centre, with a 50N side load. This seems to give an angular deflection of 0.000056 degrees. This would equate to about 0.000243mm deflection at the tip of the cutter.

Assuming that the torsional deflection would be at least half this due to being shared between the two rails, and add in the fact that the true rail torsional and bending stiffness would be significantly greater due to the effect of the SBR16 rails bolted to them, and I don't think I need to worry too much at all!

Having also done some rough calculations for unsupported round rails, I am very glad that I opted to spend a bit more on the fully supported type. It would seem that total machine rigidity would be dominated by the rail deflection if using unsupported round rails, which would make building a hefty structure to mount them fairly pointless.

Thanks for pointing me in the right direction with the deflection calcs - I hand't thought to just apply simple beam theory to this problem!

Jeremy