Randy - You're are absolutely correct - the rail can only be supported on the ends because the slides (saddles) are fully enclosed. I'm doing this on purpose actually, because my original design was held together by gravity, and I wasn't happy with the outcome. The original design started off as a homemade pattern copier, actually:
Old design still:
Old design moving:
http://frederic.midimonkey.com/cnc/IM001188.AVI
Anyway, this machine as you can see isn't going to be that large, (30x21" work area, machine about 3" larger in both directions) so I'm hoping the rails won't flex because of the small size and reduced weight of the thing not being so large. If your concern turns out to be a problem, know that I can fix it with this design
The design of my end supports for the rail and the leadscrews just below, has enough space in the design that I can install up to a 1.25" diameter rail, therefore I can gain more stiffness in the rails if necessary. The saddles as you can see in the pictures I posted already are adjustable - longer bolts and wider spacers - means I can ride the saddles on larger diameter material. The "rail" in the picture is a short section of 1/2" thin wall tubing I had laying around and used to illustrate how the saddles will mount, but the "real" rails are 1/2" diameter hardened round stock, which is by far stiffer than the thin wall tubing in the picture. How much stiffer? I don't know. I can bend the tubing against the back of my neck, whereas the solid, hardened round stock I cannot. I can barely flex it. But like I said I can increase the diameter of the rails up to 1.25" without having to machine new supports, make new leadscrews, and so on.
While I do intend to make a dremel head eventually, the main purpose of this machine is to perform plasma cutting, which is an advantage - plasma cutting heads don't weight that much, and if operated correctly "floats" about 1/16" above the material being cut, so there is no contact that will cause resistance in the movement of the machine. So at least for plasma cutting, the rails/saddles only have to support it's own weight, plus the weight of the other axis' installed on top.
One of the intermediate designs I came up with would have allowed for a supported rail, and still force the saddle to only move in the direction of it's assign axis - using three bearings per end, in a "Y" type of arrangement, looking down the length of the saddle. I made about six of these types of saddles before I finally accepted *I* was unable to make an accurate enough "Y" mounting arrangement so I went the easy route with a square, four bearing arrangement. Though I was making the "Y" style saddles by welding steel flats in a wooden jig I made, trying to get the 135 degree angles as accurate as possible. I just couldn't do it. If things do flex really bad, and upgrading the diameter of the rail for stiffness doesn't help enough, I may have to revisit the "Y" arrangement, and support the rail underneath with say, a 1/4" wide flat of appropriate height.
If I had a milling machine at my disposal, making the "Y" arrangement would be easy - I'd just make them out of solid aluminum stock and machine the shape, and machine slots for the bearings, then drill through to tie it together.
One of the things that was important to me, is to have the bearings supported by bolts that are in double sheer.
For about a year I was looking for a used "all in one" lathe/mill combo, however being unemployed for the past 18 months didn't help me afford such a unit, so I ended up with a used Clausing lathe instead:
Anyway, good eye Randy. Also, just in case it's not obvious, I'm not arguing, merely sharing my thinking at the time I started on this adventure. I realizing my writing style doesn't always facilitate that. One too many corporate memos I guess. Anyway, I very much appreciate the comments.
Rod - having a flat surface to attach the other axis too is key, and I put a fair amount of thought into it. Randy's concerns above concern me as well, however as I stated above if there is too much flexing I have an upgrade path outlined to reduce or eliminate that.