Great advice from Gerry, as usual. Preloaded skate bearings are probably not terrible, though they can't compete with profile rail. For each skate roller, probably only ~2 of the balls are actually taking load. Also the typical 1/4" steel rail hangs out 1/2", which my FEA models show is enough to be a flexy spot in an otherwise stiff machine. With profile rail, there are many balls loaded in parallel, and they can also be preloaded.
The Momus design is ok, though the high placement of the x-rails is not best for stiffness. The best x-rail height is the same height as the cutter. Above or below that, cutter loads create a moment that increases load on and deflection of the x-bearings. A Momus height is probably optimal for high gantry acceleration (~same height as the gantry CG). Try doing some FBDs of each subsystem to see how to minimize moments--a lot can be optimized just from stick models. Slocum is a god of this kind of machine design and has a great pdf:
http://pergatory.mit.edu/kinematicco...Principles.pdf
For CNC, esp see his slides on Bearings, Abbe (sine errors), Structural Loops, Centers of Action, and Stick Figures.
Steel or aluminum are both fine. The key is getting a large cross-section and appropriate wall thickness. Typically the problem with larger tubes (e.g., 6"x6") is that the thinest steel walls available can be very heavy, so alum is often a better choice there. Steel is easier to weld, and welding makes a stiff joint, but welding creates distortion and slowly stress-relieves over time, so accuracy suffers. Alum is easy to power tap, so with lots of bolts and good joint design, it's not hard to produce a stiff structure with it, too. Bolting + epoxy should be equivalent to welding for stiffness.
Yes, 2 rails per axis, and 2 blocks per rail. Many machines use 1 long block per rail. These aren't very stiff; the blocks have to be spaced out. Profile rail has equal load rating in all directions, so they can face the same way.
For a stiffness target to cut aluminum efficiently, it looks like 20,000 lb/in is a minimum. Good links:
http://www.cnczone.com/forums/diy-cn...tml#post944411
http://www.cnczone.com/forums/diy-cn...tml#post911744
University of Utah - ME EN 7960 - Precision Machine Design"); if (currentModule == "index") document.write(" - Main"); if (currentModule == "courseInfo") document.write(" - Course Information"); else if (currentModule == "projects") document.write("
http://www.mech.utah.edu/~bamberg/re...e%20Design.pdf
It's great you're planning an FEA model. I've a lot of background in structures, but have still been surprised at where the flexy parts are in a given gantry machine. And it's challenging to get everything ~equally flexible (so not overbuilt or have an Achilles heel). Lots of complex load loops and the usual mechanics of materials equations don't predict some major flex areas like "collapse" of the cross-section for the main gantry tube (bulkheads or a diagonal sheet fix that very well). Be sure to model the bearings with contact conditions (so they can slide); this will show the importance of leadscrew placement (also inferred with a stick model, center of stiffness). It's pretty cool stuff!