[jevs]

I am building a Data - Cut router.

It calls for round rails, but I am going to use profile rails. I already bought THK SHS25 rails for the X axis. I just can't decide on the Z.

15mm, 20mm, or 25mm? It uses two rails and 4 bearings.
Right now I can find some 15mm stuff reasonable, but I want this thing to be strong and accurate. I will do some aluminum I am sure.

I just don't have enough experience with this rail. 15mm seems small even though the ratings are ok. Some of the twist ratings are only a couple hundred pounds I think though. 20mm and 25mm are a bit harder to find right now, but I am willing to wait or pay a little more if I should.

I am trying to buy new stuff off eBay or wherever I can find new surplus stuff etc. This is a big project for my little amount of free time, so I am trying to avoid headaches.

Anyway, I want the rail to be stronger than the machine can handle so I never have to kick myself later. If 15mm will not limit anything then that's great, otherwise I will go as high as 25mm if I have to.

Advice? Thanks.

[pippin88]

Linear profile rails are incredibly stiff and strong. 15mm would be enough for all axes. Preferably get ones with preload

[jevs]

I see a lot of the routers you can buy ready to go that say they have 25mm rails on all axis or 20mm. Why would they spend the extra money if 15mm was good enough? Of course I don't know if they are using two row or 4 row bearings or what rails they have though.

[ger21]

I see a lot of the routers you can buy ready to go that say they have 25mm rails on all axis or 20mm. Why would they spend the extra money if 15mm was good enough?

To make you think you're getting a better machine than you really are??
Probably because larger rails are easier to mount. 15mm rails use 4mm bolts, which can be a little more tedious to work with.

Typically, you see 20mm or 25mm rails on a gantry that carries a Z axis that might weigh 25-50 lbs?? Those rails would have no trouble carrying 1000 lbs.
15mm rails on a Z axis should have no trouble carrying 400-500 lbs.

[wizard]

I see a lot of the routers you can buy ready to go that say they have 25mm rails on all axis or 20mm. Why would they spend the extra money if 15mm was good enough? Of course I don't know if they are using two row or 4 row bearings or what rails they have though.

This is the thing, THK would likely not recommend there smaller rails for that load. Unfortunately I forget which size it is that has the greatly reduced load capacity in the up direction.

In any event I look at it this way, the Z is relatively short on most machines and also experiencing a mix of loading as the machine moves around. Going to a larger rail isn't a huge expense so one shouldn't get too worked up over the extra cost. The real question is does the extra size give you any other design advantages or disadvantages. Here I'm thinking a lead screw clearances. The physical height of the larger rail might help with lead screw clearance while the the larger bearings might tighten things up to much.

In other words one can't just randomly change rail sizes especially if a machine has been designed around one rail system. In this case I'm not sure if we have an existing design or if we are working with a ground up design. It really doesn't matter because you still need to fit the components to the design in a rational manner. So if we are talking about a rail that can handle all the loads applied in any direction then you have to see how it fits together. If it doesn't fit design wise then you need to change the rails or the design of the rest of the machine.

Somebody else mentioned screw sizes, frankly messing with a massive number of small diameter taped holes can be a good reason in and of itself to go to a bigger rail. A 6mm tap is far easier to work with when you have multiple dozens of holes to deal with. In other words there are practical concerns to address.

I hope this doesn't confuse the issue but when almost any rail will do, other factors come into play when choosing a rail.

[jevs]

Makes good sense. 20mm rail would put me closest to the original design as far as clearance goes I think. I believe the 25mm would push it out a little further than the original design, but not much. I might very well have to shim things if I went 15mm.

I will have to do some looking and measuring. You bring up a great point that may make the use of 15mm more work than it would be worth.

I might just draw the z axis in my own CAD program so I can see how things go together and how the drive screw fits etc. The drawings I have now are just 2D.

[jevs]

I cannot find any bearings with extra preload on them for my X axis 25mm rails. Do you think the normal radial clearance would be fine for this router considering its 25mm rail? Otherwise I am going to have to spend twice as much on bearings.
The ratings are:
No designation -8 to 0 µm considered normal preload and the only thing on eBay or elsewhere that I can find without buying from THK
C1 -14 to -8 µm considered light preload
C0 -20 to -14 µm considered medium preload
CS would be custom ordered preload

[dmalicky]

Good questions and advice above. Unless you'll be doing low rpm heavy cuts in steel, any of those profile bearings will be plenty stiff, assuming they will be placed with reasonably good leverage. Normal preload is also fine (clearance would be another story--avoid that). And actually, the more preload, the more likely the bearings would bind up if the mounting surfaces are not super flat. For a point of reference, Haas uses 25mm rail on their Toolroom mills and lathes. I think 20mm is the best choice for us--as mentioned above the 15mm rail bolts are small.

The much harder part will be getting the rest of components in the same stiffness ballpark as the rails/bearings. Or to put it another way, on a DIY machine, the tiny extra deflection-at-the-tool of 15mm bearings (vs 25mm) will pale in comparison to the massive deflections due to the gantry, Y-car, Z-car...

[jevs]

The good thing is the gantry on this thing should be strong. 1" thick uprights and 1" thick bearing plates. Attached across underneath with 1/2". A 1/2" Y backer plate with a Thomson 2DA-12-JOBL36 bolted to it.

It should be pretty strong for a router. That is why I want to make sure the rails are good enough (just started learning about their specs this week).

I will probably choose 20 or 25mm. I need to get it drawn to see which works out best, then further base that on what I can find for a decent price.

[dmalicky]

Just saw the Data-Cut webpage CNC Router
Yep, that gantry cross-member will be very flexy. It looks like 1/2"x6" alum, plus rails and mounts. Generously assuming 3/4"x6", my FEA shows the deflection-at-the-cutter due to a 100 lb cutter load is 0.031" (30" Y travel, 6.5" Z). By comparison, the deflection-at-the-cutter due to the flex of 15mm no-preload Hiwin bearings would be about 0.00033". Almost 100x less! Many other parts of that design will have similar flex as the gantry. If you only need to cut wood, it will likely be fine. For cutting aluminum, I'd suggest a redesign.


EDIT: Oh, just saw your post. To clarify, the issue isn't "strength" (how much load it can take before it breaks). The problem is stiffness (or the inverse, flexibility or compliance). It's often surprising that "solid metal" isn't all that stiff.

[jevs]

Did you add in the fact that the Thomson 2DA-12-JOBL36 is bolted to it? 2.62" wide mounting surface and .752" thick, of course this does not include the rails and the rest of the extrusion that would also make it stronger.....

I don't have any FEA, can't afford to spend for that option at this time (and not worth it for my limited use).

That part is considered part of the strength of the Y backer. It would have to be thickened up surely if just bolting rails to it. But wouldn't this make it quite a bit stronger?

I am open to suggestions though as I am just beginning this really. I attached the data sheet.

The only thing is I am using a 2DB-12-JUBQL36 because I found a new one reasonable on eBay. It has a lead screw and motor mounting built in. My plan was to dismantle it and make it into a 2DA-12-JOBL36. The only difference is that two inches of the extrusion will be gone from each end (where motor and end plates were). However, I might be able to leave the end blocks. I have to get this stuff drawn to know. Leaving those in would beef it up even more because they completely capture 2" of the rails and they could be bolted into from the uprights.

I could still get rid of this and beef up the backer and use profile rails I suppose, but that would be more cost...

[jevs]

EDIT: Oh, just saw your post. To clarify, the issue isn't "strength" (how much load it can take before it breaks). The problem is stiffness (or the inverse, flexibility or compliance). It's often surprising that "solid metal" isn't all that stiff.

I would think that would stiffen it up a bunch too (less flex)? I just added some drawings you can look at. Surely the 1" cast aluminum uprights and bearing plates won't flex a lot? Will they?

I will have to get this all drawn and then get some feedback I suppose.

[dmalicky]

It's really great that you're asking questions early in the process -- so much easier to revise.

Yes, that extrusion will stiffen a 1/2" plate substantially, but it will still be flexy compared to a "stiff gantry". I've found the ~only practical way to get a stiff gantry is to use a large square (or rectangular) tube. That member is in torsion, and only large tubes are effective at resisting torsion (or large solids, but those are very heavy and $$).

Some key questions: what do you plan to cut and how aggressively? If only plywood, softwoods, plastic, etc, then some moderate changes to the Data-Cut design would be likely be fine. If you want heavy cuts in alum, major changes will be needed and I'd probably start fresh.

Yes, FEA is usually $$ and complex (although if you have time to learn it, a student version of Solidworks is pretty reasonable.) I reran the model with a 1/2" plate and a rough outline of that Thomson section (added on a 2.75 x 3/4 bar, with two 3/4x3/4 'wings')--this assumes a perfect bond between the Thomson section and the 1/2" plate, which is optimistic. Deflection is down to 0.015", about 50% of the solo 3/4" plate, but a long way from "stiff".

1" thick uprights will show some flex, although probably not as much as the gantry, Y-car, and Z-car: those 3 are where most of the action usually is.

If you want a "stiff" machine, I'd suggest starting with a lot of research on:
1. How much stiffness is needed for heavy and accurate cuts in wood, alum, steel.
2. How to design the machine to achieve that.

Recent threads and the archives have a lot of good info. For 1), see the middle of this post:
http://www.cnczone.com/forums/diy_cn...ml#post1408740
Of course, many machines cut those materials with a lot less stiffness; they use small depths-of-cut and feedrate to avoid chatter.

For 2), here's an overview post with links to some of my prior posts:
http://www.cnczone.com/forums/diy_cn...ml#post1430502

[jevs]

Can you send me a link to a router that is stiffer than this one in the 3'x5' size range? Maybe if I look at their design it would help me decide if I want to make it stiffer.

The heaviest thing I might do is light aluminum cuts. It goes down in stress from there. Its really a wood machine. I do have a decent size CNC mill for much more than that (what I will be milling the CNC router parts with).

[dmalicky]

That's a fairly unusual size range, but you could scale and borrow the important features of the good 4x8 designs like MultiCam, Techno, Onsrud, etc. E.g.:
http://technocnc.com/cnc-router-pres...UTER-LARGE.jpg
http://multicam.ca/wp-content/uploads/2012/10/1000R.png

The important features are:
- Profile rails and bearings for at least the Y and Z axes
- A large tube for the gantry cross-member. For wood cutting and 30" Y travel, the easiest is a Misumi GFS8_100200. A good choice for alum cutting would be a 4x8x0.5 alum tube. Or 3x6x.5 for wood cutting.
- Y rails at the top and bottom of that cross-member, with vertical spacing similar to the Z clearance
- Y and Z bearings arranged in roughly a square pattern: both widely spaced and tall.
- Stout uprights for the gantry, with good bending stiffness both laterally and fore/aft
- X rails near the table height or up to ~4" higher, depending on stock loading needs vs upright simplicity.
- Dual X drives. At 5' instead of 8', dual 2525 ballscrews would be better than R&P. Dual drive allows the table to be thick and braced from below.
- Long 'feet' for the X bearings, to prevent tipping. Extra long if using a single X drive.
One thing they don't do, should do, and is easy to do: Y and Z cars with high bending stiffness.
Of course, the industrial CNCs have some design features (like the fancy welded uprights) that are hard to do DIY, but those aren't essential to copy exactly, just functionally (stiffness).

For the hobby designs, CNCRP and ShopBot do a good job with the basic layout and proportions, although ballscrews and profile rail will be far better, of course.

If you're ok with light cuts in aluminum, then some moderate changes to the Data-Cut design would be fine:
- Change the gantry bar to a Misumi GFS8_100200, 4x8x1/2, or a 3x6x1/2" alum tube.
- Thicken the table. Preferably with dual X drives, 2525 ballscrews. 30" is a wide gantry for a single drive; if single, the gantry feet need to be pretty long.
- Raise the X rails about 3-4" above the table so the uprights are shorter. Or make the uprights stiffer in the lateral direction.
- Make the Y and Z plates from at least 3/4" thick alum.
- Space the Y and Z bearings in about a 6"x6" square pattern, or wider.
- Tie the router into the Z plate and Z bearings with beefy mounts.
That should be very good for wood and ok for light aluminum.