[BrandonBe]

I just finished my SolidWorks model for my steel frame CNC. It will have a 1.5 Kw spindle motor, 15 mm linear rails, 24", 18", and 6" X, Y, Z travel respectively. The frame is 3" square tube with 3/16" thickness. The bearing block spacing is 12", 8", and 6" for X, Y, and Z respectively. I'd appreciate any design input as this is the final model before I begin purchasing materials.

[BrandonBe]

Here is the torque graph for the stepper motors I am planning to use. I feel like this should suffice as the pitch for the ball screws is only 5 mm, the max mass the X axis will need to move will be between 10-25 kg, and I do not plan on running the machine too quickly (>50 ipm). Any thoughts on these motors? Also what does the resistance mean for these motors and what is the difference between a motor with a higher current draw with a higher resistance and a motor with a lower current draw and lower resistance, performance wise?

The torque seems to drop off sharply but At 50 ipm it would drop to 1.25 Nm. But is that enough to mill aluminum, albeit slowly with shallow cuts?
Attachment 314362

[louieatienza]

I will say for wood you should be cutting at about 75-100ipm to start, spindle power depending, and maybe 40-100ipm for aluminum. This is because you have to meet the chipload requirements of the tool in order to get the best cut quality. You need to size the chips coming off the tool such that they actually draw heat away from the edge and the workpiece. Wood really machines better with higher speed and faster spindle speed. There'd be no sense in building a rigid machine, if you cannot take advantage of the rigidity.

To mill aluminum slowly would require a pretty heavy duty spindle with more torque available at low RPMs.

[dmalicky]

Nice modeling. Thanks for the many clear pics and transparent views -- very helpful!
Since you're using profile rail, I'm assuming you want it to be stiff. You're doing many things right along those lines, such as the moving table design and wide rail spacing. For higher stiffness, here are some comments:

Spindle mount and Z car:
See recent analysis on these parts to stiffen them a lot: http://www.cnczone.com/forums/diy-cn...ml#post1859452
To mount a cylindrical spindle, tie it into the Z car (C channel shape) with 2 mounting plates attached on 3 sides, similar to the pic in this post: http://www.cnczone.com/forums/diy-cn...ml#post1858294
except raise the upper spindle plate as high as possible on the spindle, for ~max separation. My FEA models have shown that type of design is super stiff -- the spindle and Z car are really locked together. Mounting the spindle only to the Z plate results in more flex.

Y car:
For consistent stiffness, it's good you have the Z blocks on the Y plate. But this plate will flex due to the loads from the upper Z blocks (see the latest Hardwoods post for a pic of that). Two fixes:
1) Raise the upper Z block to get them closer to the upper Y rail. They don't have to overlap, but closer. That will also reduce Z block forces and increase stiffness that way.
2) Add channel legs to the Y plate (that analysis isn't up yet, but you can probably imagine how it helps). A good strategy is nested C-channels for the Z and Y cars.

Y rails/blocks:
It's good you have large vertical spacing, to resist X-loads at the cutter. To be similarly effective at resisting Y-loads at the cutter, make the horizontal spacing similar to the vertical: i.e., a square pattern. Also lower the ballscrew as low as you can: that reduces the rotation of the Y car due to Y-loads at cutter, and stiffens that mode.

Gantry tubes:
A large single tube is always much stiffer than multiple smaller tubes. See: http://www.cnczone.com/forums/diy-cn...ml#post1413406
Some of the theory is here: http://www.cnczone.com/forums/diy-cn...ml#post1419700
Also see any large commercial router, like Onsrud and the pics below. A single large tube is also more likely to be and stay planar, for parallel rail mounting (critical for profile rail on a stiff machine, to not bind).

X table/rails:
Good square pattern for the X blocks.
The ideal rail spacing for a moving table is around 60% of the table width. When the rails are wide like shown, the center is more flexy than the edges. The goal is to get the center a bit stiffer than the edges, since machining usually occurs there. Examples:
http://westerncncinc.com/wp-content/...5-MT-Large.jpg
http://www.romanblack.com/CNC/5_CLOSlg.jpg
Also the table looks a little thin to me -- I'd go ~50% thicker and/or add bracing under it.

Spacers under all rails and blocks: avoid wherever possible.
- 20mm rails will give more room, and most prefer them because 15mm rail has tiny bolts.
- Try to mount the ballnut 'inset'/overlapping/at the edge of the plate for a tighter package.
- The narrow spacers under the Y rails will add a lot of flex. If spacers/levellers are needed, make them wide and attached very securely.

I'm not sure how you plan to join the 3x3 tubes. Welding will distort them (requiring post-machining) and shift over time unless stress-relieved.

[BrandonBe]

Wow, thank you so much for all of that advice David! I can see you really know what you're talking about.

I was planning on welding the bottom base and then the gantry portion of the CNC separately and then bolting the two together. I've heard a lot about flexing and it kind of concerns me. How would you recommend I create this frame? I initially figured if i'd only weld the edges of the base that are vertical or normal to the x/y plane, then I could restrict flexing to only the x/y direction keeping the base parallel and un-warped where the rails will attach. Would this even work? (I still need to get a lot of welding practice in before I even touch this frame.)

So what are my options here because creating the frame is the major thing keeping me from starting this project. I know I would be able to take care of all of the other aspects of the CNC with little trouble but I am seriously worried that I won't be able to create a suitable frame.

I am also already way over budget so having the frame post machined would probably be beyond my budget. Unless I can post machine it myself using a mill. How exactly does the post machining process work anyway?

[dmalicky]

It's true that only welding as you describe (e.g. for the base, only the vertical weld beads) will minimize out-of-plane distortion. Still, there will inevitably be some, and also parallelogram distortion. Maybe it would work, maybe it would not. For me, it wouldn't be worth the risk, since distortion could occur after months and ruin the machine.

It is possible to do stress relief DIY with a torch, but I don't know how effective it is for something like a CNC machine. Post machining means machining the rail and block mounting surfaces after welding and stress-relief, yes, on a mill or similar.

Welding without stress relief can work for the lower frame of larger machines, if the welding is only at the ends of the main rail-mounting tubes or other triangulation, and the machine is heavy enough to plant it to the floor. Then adjustable feet can be used to level the X rail mounting surfaces.

Other options are:
1. Bolt & pin together using lap plates (like your blue plates)
2. Rivet & epoxy together using lap plates
3. Maybe brazing. I don't know if it distorts over time. Probably much less than welding, if at all.
4. 8020 construction, for more $.
5. Baltic Birch plywood construction (for most parts) for less $. A strategic design can give a pretty stiff main structure, especially for a moving table config, and it's easy to build. The Y car, Z car, and a few other components need to be metal because the loads are too concentrated on them.
They can all work -- personal preference, how you like to work, and what tools you have available.

[wizard]

It's true that only welding as you describe (e.g. for the base, only the vertical weld beads) will minimize out-of-plane distortion. Still, there will inevitably be some, and also parallelogram distortion. Maybe it would work, maybe it would not. For me, it wouldn't be worth the risk, since distortion could occur after months and ruin the machine.

Distortion over time is a reality. One machine manufacture I knew of left iron castings to weather outside over the winter. That gave the castings time show any defects and to relax the stresses.

The problem with weldments is that how they will distort over time depends a great deal upon how they are welded in the fist place. That and the arraignment of the individual pieces of the weldment.

It is possible to do stress relief DIY with a torch, but I don't know how effective it is for something like a CNC machine. Post machining means machining the rail and block mounting surfaces after welding and stress-relief, yes, on a mill or similar.

A torch might work if you have a smaller object but for a larger machine components you really want to heat the entire structure evenly.

Welding without stress relief can work for the lower frame of larger machines, if the welding is only at the ends of the main rail-mounting tubes or other triangulation, and the machine is heavy enough to plant it to the floor. Then adjustable feet can be used to level the X rail mounting surfaces.

Other options are:
1. Bolt & pin together using lap plates (like your blue plates)
2. Rivet & epoxy together using lap plates
3. Maybe brazing. I don't know if it distorts over time. Probably much less than welding, if at all.

Heat will distort metal. You don't even need welding / brazing temperatures.

4. 8020 construction, for more $.
5. Baltic Birch plywood construction (for most parts) for less $. A strategic design can give a pretty stiff main structure, especially for a moving table config, and it's easy to build. The Y car, Z car, and a few other components need to be metal because the loads are too concentrated on them.

Baltic Birch is a good low cost construction material as are some of the other laminated sheet goods out there. Steel can be had pretty cheap though if you have a vendor that sells drops by the pound. At onetime I was picking up drops for a buck a pound. It is always a good idea to avoid retail for machine components if one can.

They can all work -- personal preference, how you like to work, and what tools you have available.

One shouldn't underestimate how much easier these builds can be with the right tools. I do believe after being a member of CNCZone for a long time that many have an unnatural fear of working with steel. No material is perfect but if you have a drill press that you drill holes in wood or aluminum you also have a drill press to drill steel. Same thing goes for a handheld drill, you just need a bit more muscle and need to sharpen the bits more often.

The big problem with steel or aluminum for that matter, is cutting the materials square and precisely to length. It doesn't matter if you have a bolted together structure or welded one if you start out with your materials square and to length you have less work down the road to align everything. If your home shop doesn't have the ability to cut the materials properly then enlisting a shop to help makes sense. With Aluminum extrusions you do have the ability to order stuff cut to length, finding a machine shop for steel components cut to order takes a bit more effort.

[dmalicky]

Great advice by wizard, as usual.

>>Distortion over time is a reality. One machine manufacture I knew of left iron castings to weather outside over the winter. That gave the castings time show any defects and to relax the stresses.
Yeah, I've heard it called seasoning -- supposed to be done for any precision-machined casting (or a furnace relief). Of course, most of the low-end machines skip it, which is part of why they have a limited lifetime!

>>A torch might work if you have a smaller object but for a larger machine components you really want to heat the entire structure evenly.
Yes, a slow furnace relief is the only way to be sure.

>>Heat will distort metal. You don't even need welding / brazing temperatures.
Agreed on distortion. The harder question is about the residual stresses. With welding, steel melts around 2600F and the weld arc is extremely concentrated -- both of these create very high temp gradients. The short duration causes quick cooling. Both of those create the high residual stresses in welding.
A brazing torch by nature is much slower and more distributed heat, and much slower cooling. Some silver braze alloys melt around 900F. Both of those create much lower residual stresses. But yes, they will still be present, and so I agree it's riskier than room temperature options.

>>One shouldn't underestimate how much easier these builds can be with the right tools.
Completely agree. The tools have a huge influence on the methods and thus the design.

[BrandonBe]

So I've decided to go with the lap joint and bolt method for attaching the various beams. I also removed the rail mounts so the rails will now be attached directly to the frame. I also lowered the y axis ball screw.

I am going to leave the y axis bearing block spacing as-is as opposed to moving them into a square configuration. I understand this will be a bit less rigid but I feel like it would not have a very big effect on over all accuracy. Or would it? I am hesitant to changing the spacing because it would either eat up almost all of my y movement range or cost significantly more to extend the width of the machine.
I am also probably going to leave the plates as 0.5" thickness (with the exception of the plate on which the spindle motor is mounted to, which is 0.75" in thickness) as opposed to increasing them to 0.75". This is due to the fact that i simply cant afford a $400 aluminum plate for this project. I may upgrade to a thicker plate in the future but not as of yet. I was thinking about producing a plate myself though. I have cast aluminum before so I am fairly familiar with the process. Would a cast aluminum plate 0.75" thick made of 3000 series aluminum be adequate? I know cast aluminum can be brittle but 3000 series is also known for its ductility.

As usual, any advice is appreciated.

Attachment 315132Attachment 315134Attachment 315136Attachment 315138Attachment 315140Attachment 315142Attachment 315144Attachment 315146

[dmalicky]

Good work on the updates. I'm realizing I gave you incomplete advice earlier -- sorry. For your frame design, I see the X and Y rails have to cross over a tube joint at each end. This joint will not be very flat, and so the rails will have to negotiate a discontinuity at each joint. This will likely lead to binding in the axis travel. So we do need 'spacers' for those rails, but make them wide: I'd suggest making the spacer and lap plate as the same part: full length of the rail, and full width of the tube. Probably something like 3/8" x 3" cold-rolled steel bar stock -- that is quite flat.

For the small lap plates like on the base, each will need at least 4 bolts in a square pattern. Ideally the bolts are near the corners of the tube (stiffest); a bolt in the middle of the tube face does little. 3 bolts in a row allows parallelogram movement. I'd go wtih bigger lap plates, too.

To keep the lap joints from shifting over time, include at least 2 roll-pins on each tube (or rivets, or epoxy).

Good job getting the Y ball screw low.

The Y block pattern and Y rail spacing makes that axis imbalanced in its stiffness: it will be very stiff for X loads at the cutter, but pretty flexible for Y loads at the cutter.

The Z block pattern is more square, but very short -- about 1/3 the height of the Y blocks. So all that prior X stiffness (from the Y rails separation) will not have much benefit.

If you can't widen the Y blocks, it would actually be best to drop the upper Y rail, so the Y block pattern is more square. Also enlarge the Z block pattern (at least the height, ideally the width, too). The Z block pattern should be at least as tall as it is wide. For no obvious flexy mode, the Y block pattern and Z block pattern should be of similar dimensions (Z can be a little smaller): e.g., if the Y pattern is 6x6, a logical Z pattern would be 5x5 (but not much less).

But probably more flex is going to come from the twin 3" tubing as the gantry (see the FEA in the Hardwoods thread, page 2). It would be simpler and stiffer to use a single large tube for your gantry cross member. E.g., a 6" x 3" steel tube. Then the Y rails don't have to negotiate a tube joint, and it would reduce the number of lap plates. A single tube face is likely to be more planar (than 4 tubes bolted together), for rail parallelism. To support the 6x3 tube, use a lapped 3x3 tube behind it, and small lap plates at the front.

Something we haven't addressed yet is how to produce a flat enough surface for the X and Y rails to mount to, so the axis doesn't bind. Leveling/mounting pads that are later machined flat is 1 good way. Alternately, go to SBR rail which is forgiving (but much more flexy).

The most flex in the current design is probably in the spindle mount and Z plate. Again, see the end of the Hardwoods thread. I would raise the spindle (it looks about 2" too low; bottom of spindle body should be at the same height as the bottom of Z plate), use 2 of those spindle mounts, and make the Z car channel section.

I know it's lots of text to digest; hope it helps.

[louieatienza]

So I've decided to go with the lap joint and bolt method for attaching the various beams. I also removed the rail mounts so the rails will now be attached directly to the frame. I also lowered the y axis ball screw.

I am going to leave the y axis bearing block spacing as-is as opposed to moving them into a square configuration. I understand this will be a bit less rigid but I feel like it would not have a very big effect on over all accuracy. Or would it? I am hesitant to changing the spacing because it would either eat up almost all of my y movement range or cost significantly more to extend the width of the machine.
I am also probably going to leave the plates as 0.5" thickness (with the exception of the plate on which the spindle motor is mounted to, which is 0.75" in thickness) as opposed to increasing them to 0.75". This is due to the fact that i simply cant afford a $400 aluminum plate for this project. I may upgrade to a thicker plate in the future but not as of yet. I was thinking about producing a plate myself though. I have cast aluminum before so I am fairly familiar with the process. Would a cast aluminum plate 0.75" thick made of 3000 series aluminum be adequate? I know cast aluminum can be brittle but 3000 series is also known for its ductility.

As usual, any advice is appreciated.

For maximum clearance and usability, I would situate the spindle clamp on the Z plate such that you can align the bottom of the collet with the bottom of the Z plate. That would give you maximum travel. In your design, you've reduced the Z travel by how much the spindle is sticking out below. You can always slide the spindle up and down afterwards, as fit for the job at hand.

You should also be able to find Mic-6 cast aluminum plate for a decent price on eBay. And it's already ground flat, saving you the work.

[BrandonBe]

I've added a 1/2" steel plate (most likely hot rolled due to cost, or is cold rolled required for this application?) for the gantry cross member. Will I need to mill the mounting surfaces flat for such a plate or are they most likely going to be fine as is? I feel like it will increase the stiffness of the gantry because i know that was an issue that was brought up last time. Will I need the beam I have placed behind this plate for support? It is a half inch of steel but I don't want it warping over time from use.

I also removed the middle beam on the base since extending the ball screw was cheaper than having the shorter screw with the beam underneath it to mount to. Will this affect the rigidity of the frame at all?

I added 1/4" steel mounting surfaced on the x axis for the rails. Again i'd prefer hot rolled steel for the price unless I need to use cold rolled. I would also be able to mill the surface flat if necessary.

The lap plates have been enlarged on the frame hopefully to increase rigidity. I will be using 1/4" x 3 3/4" bolts.

Any input for now?

[wizard]

The lap plates have been enlarged on the frame hopefully to increase rigidity. I will be using 1/4" x 3 3/4" bolts.

Any input for now?

Well about those bolts, You don't always want to bolt through the entire thickness of square tubbing. The reason is you can risk crushing the tubing and you may also have the bolt losses up over time. Most of the time you will want to run the bolts through just the walls that are directly joined. This way you can leverage proper torquing of the bolts and not have to worry about crushing stuff. Now it should be obvious that you can't always do this due to accessibility, at this point you just have to be careful and make sure torquing down the bolts isn't having a negative impact on the tubing.

The other option is to avoid through bolts where you can and instead drill and tap for cap screws. This is often a better solution than trying to bolt through a large square tube. The major problem here is that you need square tubing thick enough to support the screw thread reliably. So imagine 6mm screws for tubing with a wall thickness close to 6 mm. Basically you want a tubing wall thickness equal to or better than the screw diameter or vice versa. to get a good structural bonding you may want to use more screws than you may fist imagine if they are in the 6 mm range.

Also back to the through bolts. Where you can run through bolts, don't be whimpy use 3/8" or 10 mm bolts, maybe larger. The idea is to clamp parts in place permanently, with structural rigidity, so where you can install bolts that are larger than 1/4" and torque those bolts down good.

When you get this all assembled and aligned, consider getting some dowel pins and reamers. Install the dowel pins to keep your machine from ever going out of alignment even if the screw loosen up.

Now you may think that reaching inside the tubing is difficult, it can be but buying the right tools can make a big difference here. Ideally you will minimize the number of drilled and tapped holes or places where the bolts go all the way through both sides of the tubing. A key thing to realize is that if it can be distorted by clamping pressure on the tubing it will be distorted.

[BrandonBe]

How would I go about using the dowel pins? I haven't had experience with them before and am not quite sure how they are implemented.

And I thought that the force exerted on the tube by the bolts would be distributed enough by the lap plates such that it wouldn't be at risk of damaging the tubes?

The frame tubing will need to stay 3/16" thick which will probably be too thin for tapping so how about if I use shorter bolts and bolt through the plate and one wall of the tube, then screw the nut onto the bolt from the inside of the tubes. Should I use thread lock so the bolts wont come loose over time?

[dmalicky]

Good work on the updates.

Cold rolled may be flat enough to use as is, so long as you aren't looking for high accuracy. But it cannot be milled flat: it will warp more as you mill.
Hot rolled is definitely not flat enough as is. But it would be needed if you mill it flat yourself, and will stay flat after milling.
I.e., you have to decide before-hand if you will mill it flat. Then buy.
Another option is cast Mic-6 aluminum tooling plate -- it is ground very flat, not too expensive, and would be my choice.

The only step backwards is the structure behind the Y rails: the rails need tubing directly behind them. Imagine pushing in the middle of either y rail, perpendicular to the plate -- the plate will easily bend rearward. But again, I'd suggest 1 big 6x3 tube rather than 1 or 2 3x3 tubes: a 6x3 is flatter and much stiffer (then mount a 6" tall flat-plate to the front face of that 6x3, and mount the rails to the plate with ~5" vertical separation (max) ). Again, see the Hardwoods thread, page 2, or any commercial router, for why 1 big tube is best. When not an expert, it's best to copy what is known to work.

The lap plates are looking beefier, and great job on those bolt patterns. You got the bolts in a big rectangle and right next to the edge of the tube -- perfect. With those bolt locations, the clamping force will be taken by the tube edges and it won't collapse the tube.

It's not ideal, but I have found 3/16" wall steel tube is thick enough to tap 1/4-20, where necessary. But you can't torque too much. As wizard says, use 3/8" for thru-bolts. As long as the bolts are right next to the side wall of the tube, the tube wall thickness is 3/16", and the lap plates are good thickness (1/4"+), the tubes should not collapse. That strategy wouldn't work with 16 ga tubing, of course.

The X rail plates look great.

The middle base beam loss is no problem at all. Actually I like it better without -- cleaner.

Yes, pins will be needed to maintain alignment. If you haven't done much machining, I'd suggest roll-pins or rivets over dowel-pins for apps like this -- you can just drill instead of ream.

In the bottom view, I think it would be better to replace those 8 lap plates with 2 big and long lap plates, one on each end. Simpler, stiffer, easier to make, better alignment.

Since the 1/2" thick table is staying 1/2" thick, I'd add some reinforcement underneath it. I'd change the 4 spacer blocks (on top of the bearing blocks) to long bars or tubes. Maybe 1x2 steel tube, flat wise. Those tubes need not run 100% of the plate length: 80% would be fine. Then cross-bars connecting them at each end, if those will clear the ballscrew support bearings.

[BrandonBe]

Thanks David!

I added 1" by 2" 3/16" steel tubes as spacers for the X axis and Z axis. I was avoiding these because i wasn't sure if they were strait enough. Will I need to mill the surface of these tubes or will they be flat enough?

I'm going to use hot rolled steel for the gantry, X, Y, Z axis plates, and the mounting plates for the rails on the X axis. I'm planning on milling the surfaces for the rails on the gantry plate, X axis (is 1/4" hot rolled steel thick enough to have the surface milled?), and the z axis plate.

Will I need to mill the plates where I mount the 1" x 2" spacer block tubes to as well?

And I'm still a bit confused about the pins for alignment. Like where exactly will I need to insert the pins?

[volvox311]

Thanks for sharing your designs. The approach of using outside gussets to join thick steel tubing makes me think even someone without metal fabrication experience like me could attempt a steel build.

Question: Is it necessary to have so many bits and pieces to achieve rigidity? The more parts the more difficult the task. Would it help to center the rails, ballscrews, and bearing blocks on tubing and never have them cross a gusseted joint or any joint for that matter?

If you check out this design DIY CNC Router Build; Fixed Gantry, Steel - Wade'O Design he manages to get the proper elevation for the rails and ballscrews without a whole lot of funky gymnastics although he did have to compromise when sizing components. Good luck. I'll go back to lurking and monitor your progress.

[wizard]

Thanks for sharing your designs. The approach of using outside gussets to join thick steel tubing makes me think even someone without metal fabrication experience like me could attempt a steel build.

In the old days perfectly good bridges where either bolted or riveted together. Rivets, real ones not pop rivets, can be used to produce very solid structures. The approach is labor intensive though if rivets are used and you can get into special equipment. Bolted together structures can certainly be strong, while not used much in industry when we do the assembly is often supplemented with dowel pins.

Question: Is it necessary to have so many bits and pieces to achieve rigidity?

I might even add more. For example at all 90º inside corners a 90º brackets might make sense. Part of the reason for the seemingly high number of brackets is that you want to keep the machine balanced. Well maybe better stated the joints balanced. You put a lap plate on the top and bottom to prevent loads on the joint flexing a single plate like a hinge.

There are other approaches that could be taken and still keep a bolt together machine, for at least part of the machine. Imagine the tubing sitting on top of each other, crossing at 90º. You would still need gusset plates and possibly other enhancements to make the joints robust.

The more parts the more difficult the task.

Not really. The task is simple but repetitive.

Would it help to center the rails, ballscrews, and bearing blocks on tubing and never have them cross a gusseted joint or any joint for that matter?

I'm not sure what you are getting at here. In generally you need to look at things by how much they impact rigidity.

If you check out this design DIY CNC Router Build; Fixed Gantry, Steel - Wade'O Design he manages to get the proper elevation for the rails and ballscrews without a whole lot of funky gymnastics although he did have to compromise when sizing components. Good luck. I'll go back to lurking and monitor your progress.

There is no one right way to build a machine. There are however good practices when it comes to machine tool building. The wadeodesign is a good one buy you need to understand that it isn' a cheap solution. For one machine tool bases like he is using aren't cheap. The other problem is lots of welding and welments and if your goal is to avoid that then the design isn't ideal.

Personally I like weldments. However to do weldments right and to get the precision a machine tool requires, you have to post weld stress relieve and then machine the parts. That is difficult for most people whom aren't familiar with the metal working industry. Mind you we are talking about steel here many would suggest cast-iron as a better alternative. One can pick apart a design in an endless manner if they wanted to.

The machine that started this thread has evolved into something easy to produce. It should work pretty good really, with possibly a few tweaks after first assembly. Is the machine perfect - nope, no machine is, but it will be very usable if built to plans.

[volvox311]

I'm not sure what you are getting at here. In generally you need to look at things by how much they impact rigidity.

I appreciate that there are multiple design choices that all result in similar answers. Without, hopefully, being too intrusive I'd like to take some opportunity to learn by asking questions.

What I was getting at was using a single gusset to cross the entire front of the machine to brace 5 pieces of tubing instead of using many smaller gussets for the same purpose. I can see that adding 90º inside corners would help too.

I noticed that the X linear rails crossed a 90º joint with a perforated plate between them. Would it be useful to instead mount the linear rails on a single tube each and not have the rails cross a joint? Or conversely is it necessary that the rails cross that joint?

[dmalicky]

Nice job on the updates and improvements. The gantry beam and dual spindle mount will be much more rigid. I think the main flexy spot now is the Z car. When raised up, it will be stiff; but when down, the plate will hinge just above the upper spindle mount (see the Hardwoods thread, last page).

The Z axis bearing-block spacers can usually be avoided by overlapping the nut with the Z plate, and the BK bearing with the Y plate (or an FK mount can help). I'd try to avoid spacers here, as any Z axis has inherent rigidity issues, and for drilling, the extra overhang increases torque on the gantry.

Yeah, good point on the 1x2 tubes -- they'll probably need machining, and that may distort them since they are cold finished. So probably not tubes for the rails or bearing blocks, but something flat or flattenable.

That's also a good question about milling 1/4" hot rolled plate flat. Success will depend on how big the plate is, the machine tools, and the skill of the machinist. At some size, a plate isn't stiff enough to manage its own weight. E.g., when put on the mill table, its weight will tend to flatten it to the table, but under stress. After milling and standing upright, it springs back and distorts. 1/2" plate is 8 times stiffer than 1/4", and seems about right to me for plates this size. Wizard may weigh in on this.

So, for X rail strips/mounts, I think the best options are either:
1/4"+ cold rolled steel, unmachined. Maybe make 1 axis with this to see if it's flat enough.
1/2"+ hot rolled steel, plus flattening of both sides. Probably machine 1 plate to see how it goes, before committing to all.
1/2"+ Mic-6 alum pre-machined plate.

For X bearing mounts / table reinforcements, and Y and Z plates, I'd probably go with those same options but 1/2"+ thick for all.

Note that a hot rolled plate needs to be machined flat on both sides, and parallel to each other. This may take 3+ operations if the first attempt is unsuccessful. I prefer to buy it flat, even if more $, but of course this is not always practical.

Basically everything that a bearing block or rail mounts to needs to be flat. The hard question is "how flat"? With a very stiff machine, a slightly curved mounting surface would make profile bearings bind. But the same or worse surface can be ok in a flexible machine, because the frame flexes to let the bearings move how they need. Your machine is somewhere in between, and so I don't know a spec.

Pins would go on any plate where alignment needs to be maintained, which is pretty much all of them. At least 2 pins on each side of the joint, separated far apart for best alignment control. 3 pins per side is better, 4 is plenty. Dowel pins are most rigid; roll pins are easiest and I think rigid enough for this machine. As wizard said, solid rivets are much better than blind rivets, but they are generally not practical DIY. All alum blind rivets are the worst; all-steel blind rivets are pretty good.

As Volvox suggested, bigger and fewer lap plates is better; it's mainly a question of material cost. For the lower frame parts, a good but expensive approach is a single plate over the entire top surface. For example, 3/8"+ Mic-6. Alternately, a single plate just big enough for both X rails would help those X rails be ~automatically in the same plane (like your current design for the Y rails). Otherwise, we'll need an assembly step to get those 2 X rail strips in the same plane, and keeping them that way might be difficult.

And it's true that where the rail crosses a joint, there's a risk the rail will bend there, and motion will bind. In the current design, the rails could be stopped short of the joint, but then we lose travel for a given machine size. Some alternate approaches:

1. For X, put the main cross-brace tubes underneath, so the rail can be full-length on 1 tube, with no joint to cross. Very popular, especially for moving table designs:
http://www.cncmotion.com/standard.gif
http://static1.squarespace.com/stati...1394877648838/
http://www.techspex.com/cdn/cms/00f6...srud_super.jpg
http://westerncncinc.com/wp-content/...5x12-Large.jpg
This approach is well suited for a weldment; maybe less clean if bolted.

2. For Y, make the gantry a single 6x3 tube, going full width, with one upright tube attached underneath (bolted with angle brackets), and a long "lap tube" behind it to stiffen the joint. That would be my choice, as it is also much stiffer.

3. For either, in the current design where the rail necessarily crosses a joint, get the tube surfaces well matched, then carefully file off the high spots, then overlay the flat plate. Maybe a sandwich layer of epoxy to fill in low spots. After it's built and tuned, we'll either get full travel, or have to stop short because of bind. So it is indeed an added risk, but it can probably be shimmed to fix it.

Wizard explains the pros/cons of the different approaches very well. The current design is meant to be easier and less $ than Wade'O.