Made of aluminum, Made for aluminum: Would love design critique!

[Nr1madman]

Wait.. don't you have any bearings for Z ballscrew? Only direct connection with motor?

[routercnc]

Q1: For Z you need a ballscrew bearing block with a pair of angular contact bearings to take the axial thrust and support the weight of the spindle. The bearings in the motor are not designed for that amount of load. This is usually at the top of the ballscrew.
At the lower end of the ballscrew it is better to fit a bearing but I think you can run without one.

Q2: I would not use 2 bearings on the Z axis, it needs 4. Also I would swap them around so the bearing blocks are on the Y axis and the rails are on the Z axis. Then add a Z axis plate to mount the rails to on one side and the spindle bracket on the other.

Q3: 15mm side plates are OK, 20mm is preferable.

Q4: For what you have drawn a pair of Nema 23s with the right power supply (~70V) will move that no problem.

Q5: Can't quite see the detail around the interface of the gantry side plates, but make sure there is a good connection down to the bearing blocks.
Do you already have the 15mm rails for X and Y? If not, go to 20mm.

[Zeeflyboy]

As above, while you can get away without a bearing on the floating end (you just end up with a lower critical speed which isn't generally a big deal on a Z-axis) you absolutely do not want to go with the way it looks like you've designed it there with the screw attached to the motor. Ideally you want a set of angular contact bearings on the fixed end.

Given the design you already have, it wouldn't be too much work to raise up the sides and then sit the gantry flat on the rails - that will give you much more rigidity than using the longer gantry sides, then just move the ball nuts to the side. Unfortunately I'm very limited in width and I couldn't get away with that design on my own machine given the work area I wanted, but it's definitely the way to go with a machine like this one.

what is the span on the gantry? how thick is the extrusion? looks to me like you'll get some fairly significant deflection fore-aft which may limit your surface finish and speeds/feeds in alu. I'd consider beefing that part up personally given the fairly decent looking span.

[Neale]

.
Do you already have the 15mm rails for X and Y? If not, go to 20mm.

Consider 20mm rails for Z as well. Not needed for strength, but if you go with the two-plate Z structure as recommended, 20mm rails make it much easier to fit the ballscrew and nut between the plates.

[gamingan]

Thank you all for the incredibly useful suggestions; learned a lot by trying to think about why each of those suggestions made sense.

I've refined the design now (along with some significant changes in overall geometry) -- let's see if you guys like it.

Here is the 2nd version of the design...

Front view: http://i.imgur.com/Ik9s7uV.jpg


Back view: http://i.imgur.com/iKuChuA.jpg


Below are the changes I made:

1. Most important of all, altered the geometry so that the rails and ballscrews are protected. This also simplified the Gantry side plate design a lot!
2. Gantry side plates now are 20 mm thick (instead of 15 mm).
3. Strengthened the X-axis (i.e., the horizontal "gantry axis") by using 80 x 160 mm aluminum profile instead of 40 x 160.
4. Replaced ballscrews with 20 size (instead of 16), and linear bearings + rails with HGH20 size (instead of HGH15).
5. Added a fixed bearing block (with AC bearings) for Z-axis ballscrew (previously it was fully floating beneath the shaft coupler).
6. Added two more linear bearings (HGH15) for Z-axis.

Feedback?

I hope this one is OK to start assembling!

[Zeeflyboy]

Significantly better in most regards, nice job.

Consider going one step further and putting the rails on the top of the side extrusions, you can quite literally get rid of all unsupported gantry side arms and it is certainly easier for alignment purposes than side mount.

larger extrusion on X is a good decision, I would still go for heavy profile though if that's an option in that size.

Also consider flipping the Z-axis around so the router mounts to a plate with the rails and the carriages are fixed to the gantry side... 'tis what all the cool kids are doing these days. Gives better rigidity. I'd also say just upgrade those to 20mm rail too for simplicity's sake.

[routercnc]

... you can quite literally get rid of all unsupported gantry side arms and it is certainly easier for alignment purposes than side mount.

In case it is not obvious you have trapped the long axis rails between the gantry sides, which means you cannot easily control the pre-load on them when it is all assembled.
As drawn the side extrusions which sit on the bed - one of them needs to be adjustable laterally so that the pre-load on the rails can be set. Otherwise if you bolt this together rails may wear prematurely or bind.

[gamingan]

In case it is not obvious you have trapped the long axis rails between the gantry sides, which means you cannot easily control the pre-load on them when it is all assembled.
As drawn the side extrusions which sit on the bed - one of them needs to be adjustable laterally so that the pre-load on the rails can be set. Otherwise if you bolt this together rails may wear prematurely or bind.

Consider going one step further and putting the rails on the top of the side extrusions, you can quite literally get rid of all unsupported gantry side arms and it is certainly easier for alignment purposes than side mount.

Hmm... I will try to go this route but I really prefer the rails mounted on the sides because it also does a bit of dust shielding. Is there any reliable strategy you can think of (screw-based adjustment perhaps?) to adjust/align things if I *HAD* to go with rails on the side?

And to be honest, I don't see how top-mounting the rails allows adjustment either... routercnc mentioned the need to do lateral adjustment, which I agree with, but how exactly does this become possible in practice for the case of top-mounted rails?

Zeeflybloy: Regarding flipping things (rails <-> bearings) on the Z-axis: Will check to make sure it doesn't conflict with any dimensional constraints I have, and give that a try.