4x8 Wood & Foam Router Design WIP

Greetings. I'm designing a machine for cutting wood and foam. I'd like to draw on the experience here so I can avoid making a nightmare machine that never works right. If the whole design needs to be shot down immediately, then by all means, fire away.

This design gives 4 feet x 8 feet x 18 inches of work volume. I'd like to have repeatability on the order of 1/64" if possible, but I can deal with worse. If I can't see errors without squinting, I'll be happy.

A big goal is low price. So far, it's looking like the mechanicals will cost less than $1000, and I want to keep it that way.

It's designed for light duty and future tinkering. I'm planning on carving foam for composite molds and routing wood mainly. In the future I'd like to experiment with a 4th axis, plasma cutting, tool changing, and maybe 3D printing a rough foam shell and carving it down rather than starting with big blocks.

Design features:

• 2" steel tubing for frame and gantry, 1/2" aluminum plate Z axis
• Bearing on frame for X and Y axis, SBR-12 rails for Z axis
• Belt drives for X and Y - I plan on using GT2, 12mm wide belts with 20 tooth pulleys. That should give me 0.2mm steps without the need for gearing.
• TR8 lead screw for Z axis
• Table has bolted longitudinal members for easier relocation
• Cheap 400W ER11 spindle

Some Concerns:

• How accurate can I honestly expect bearing on steel to be?
• Will warping from welds make and hope of accuracy vanish?
• Will the Z axis be stiff enough to cut wood?
• Is the table stiff enough in general?

So whaddaya'll think?
https://www.cnczone.com/forums/attac...d=442924&stc=1https://www.cnczone.com/forums/attac...d=442926&stc=1https://www.cnczone.com/forums/attac...d=442928&stc=1https://www.cnczone.com/forums/attac...d=442930&stc=1https://www.cnczone.com/forums/attac...d=442932&stc=1

Hi TE - You need a 600W spindle (prefer 800W 400W in timber will stall too easily) , you need to make the bearings captive to the rails in all directions. By the time you buy all those bearings and bolts you will spend more money then buying round rails and cars. Yes welding will distort your tubes way more than 1/64" ( 0.4mm) 18" of Z is 457mm which is huge. Need a very stiff machine to handle that. Commercial machines are about 6" for a good reason. I'd build a plywood machine before going down this path. Less then $1000USD is doable, 12mm wide belt OK but for 8ft machine I'd go to the next size. I use 16mm wide on mine could be a little stiffer. Your design will work but med term you will be disappointed with the tubing as rails and skate bearings for the carriage, too many bits to come out of whack which they do often. Rethink those to supported round rails.

If you use the high rail/low rail gantry you will have to connect them somehow to stop them wobbling. Good luck and keep designing.. Peter

It looks like quite a thorough modelling project.The advice in the previous post is very sound and I would throw a couple more comments into the mix;Why triangulate the components for the lower shelf and not the side posts of the gantry?Also of concern is the possibility of the top and bottom rails of the gantry deflecting sideways as the gantry traverses the table when cutting wood.For foam you don't need too much stiffness but rapid moves can still cause the rails to oscillate.

While you may be able to get halfway decent results in foam, you have no chance of this design working well at all with wood. Unless you get almost all of the materials for free, you're budget is far too low.
There are issues with almost every aspect of the design. And there's really no way to fix them and meet your budget goals.

Every bit of flex in the machine adds up, all the way to the tool tip.
Let's start there.
The the Z axis plate is going to flex a lot. A couple things that can help.
1) Bolt some angles along each side
2) Reverse the bearings, so the shafts are mounted to the moving plate, and the bearings are on the fixed portion. And space the bearings farther apart, with one set at the bottom ,and one at the top. This will require longer shafts, but will be much stronger.

Then you get to the gantry. As Peter said, it's going to flex a lot. The two beams need to be rigidly tied together. And even then, it's going to twist. What you really need is a single 6"x8" beam, at a minimum.

The belts are likely going to flex a lot. And I'd recommend against direct drive. Every large direct drive machine I've seen has issues with smoothness.

The bearings on tube is going to result in a fair amount of play. To get them really stiff, they need to be very tight against the tubes. But due to the tolerance of the tubes, getting them very tight will likely result in them binding. So, you'll likely need to keep them on the loose side, which will result in a "loose" machine.

Quote:

---

I'd like to have repeatability on the order of 1/64" if possible

---

Repeatability will likely be a function of the stiffness of your belts, and the amount of friction in the machine. Basically, if your belts are strong/rigid enough to get the machine to follow them.

If I had to guess, I'd say that your spindle will oscillate much more than 1/64" just when the machine stops and starts, and changing direction. You could probably see 1/4" of flex at the tool with pressure from 1 finger.

peteeng, routalot, ger21 - This is excellent feedback; exactly what I was looking for. I'm willing to restart from scratch if it makes sense to. I want a machine that looks clean and intentional. Let me break my response down by system:

Belts: I was thinking even the 12mm wide belts would be less than ideal. I designed it so it could be easily upgraded to a system similar to ServoBelt later on. That's the one with a static belt glued to the frame that the drive belt meshes with.

I didn't know about direct drive causing smoothness issues. It makes sense that another belt reduction would absorb a lot of shock. Would I still expect smoothness issues if I got a good controller with fine microstepping?

Bearings on Steel: Bearings on steel did seem like a much cheaper option going in, but I will look into the difference in pricing with round rails. Could someone suggest a source? RobotDigg has crazy low prices, but they're kind of unbelievable. I've also heard of people getting real garbage from some vendors on Amazon.

Spindle: I have the 400w from a previous project. I doubt it'll last long with how much I spent on it, but I'm sure it'll get through some foam. When it goes out, I'd probably go ahead and get something over 1kW. It's easier to justify splurging when the machine itself is already working, and not just a pile of parts and a dream.

Z Axis: I see what you mean about reversing the Z bearings. That would add a lot of stiffness.

Gantry: I had been thinking about making the gantry a 3 chord truss, perhaps with lighter steel than the 2" tube. I worry that it'll end up too heavy, and I'll have to go up to NEMA 34s and way thicker belts.

Base/Table (I'd like some feedback here specifically): It sounds like I generally underestimated the required stiffness, so I better go back to the base first.

I've never been very trusting about the dimensional stability of wood, and I'm in a very humid climate. The CNC will go in a detached garage, and the AC won't be going all the time. Strong bias towards steel here.

Assuming I just modify the design shown, what would ya'll suggest? I'm thinking of going with heavier, maybe 3/16", tube for all static structure. I might use bolts rather than welds on all the gussets to reduce warping (and wherever else possible). Cross braces connecting the legs might be a good idea too.

Again, thank you for your responses. I'll post pics as I make significant progress to keep this thread interesting.

Some Concerns:
How accurate can I honestly expect bearing on steel to be?

[Not very. If you're lucky, they may make it all the way down the axis; more likely they'll get stuck partway there. You'd be better off getting slides and attaching them to the steel members rather than trying to slide directly on them.]

Will warping from welds make and hope of accuracy vanish?

[Yes. It makes it less likely to work at all, let alone accurately.]

Will the Z axis be stiff enough to cut wood?

[The Z axis, with the profile rails, looks like the stiffest part of this design. It's the rest of the structure that needs beefing up.]

Is the table stiff enough in general?

[The table - underneath the place where the workpiece sits - is okay; that's not what needs to be especially stiff. It's all the rest of it that needs to be much more rigid. The box tubes need to be bigger in section, and the structure needs to be braced with triangles. The gantry should be a single unit, not a couple of tubes loosely connected. And reducing the height of the gantry would make it less likely to wobble. ]

Its a positive thing to see the concerns raised being given some consideration.One of the items nobody seems to have mentioned is constraining the parallel rails at the top of the table.There is nothing in place to prevent them moving apart or squeezing together and further there is nothing to stop them moving longitudinally when the stepper torque is applied to the tool at either end of the gantry.How about three cross braces and at least one diagonal to prevent lozenging?I do recognise that a spoilboard will perform part of this function but given the comments about wood based materials and stability that have been expressed,there will be only a limited amount of accuracy from that source.

One other consideration that may be worth keeping in mind is that if you use bolts,rather than welding,it can be a lot easier to add improvements is the need becomes apparent.

Hi TE - I think the servobelt idea has issues and a good std design will be better. I can discuss the servobelt if you like, its mainly to do with meshing. A belt is not designed to be flat, its designed to wrap around a pulley so when used flat the belts do not mesh correctly.... Direct drive on a belt is fine depends on the application.

A direct drive usually likes running fast so for cutting foam its ideal (15-20m/min on my machine). My routers usually run direct drive belts and I'm happy....But they don't like running slow say <400mm/min. Most of my timber cutting I do at 1000-1500mm/min and the steppers and belts are fine...

Smoothness comes down to the quality of your motion controller and driver. usteps don't directly solve that issue.

Design is always a compromise on what you need to get done. You will hit the same issues with screw drives (large pitch is fast won't like slow,,, small pitch heaps of grunt but won't run fast) so you have to decide on its operating parameters. Plywood is quite stable if you saturate it with epoxy or a suitable resin. Again you need to figure out the operational & cost parameters that the machine has to meet. Plywood or mdf make great machines for the purpose if done right. There are several very good ones in the forum. The consequences with welds warping and supply tolerances is just as real with metals. You just have to live and learn about your preferred medium...Peter

Oh yes Nema 23/24 will move mountains don't worry about weight yet. Nema34 can have speed issues so don't go there yet either.

Ok, so I did a pretty extensive redesign. It's definitely going to raise the cost.

The gantry is now a single 8x4x3/16". Not quite 8x6, but I feel like that's a bit much for what I'm trying to do. There's room to change it, pending analysis.

I dropped the Z travel from 18" to 12". Still pretty long, but I also fastened the rails to the Z plate for added stiffness.

Most of the table is still 2" steel tube, but I use 3" for the long axis. The 3" tube is bolted directly on top of the legs. The way I had it before had a lot of welds on one side, and would've resulted in some gnarly warpage. The Table also has a few new angle irons.

So... what do ya think?


https://www.cnczone.com/forums/attac...d=443108&stc=1https://www.cnczone.com/forums/attac...d=443110&stc=1https://www.cnczone.com/forums/attac...d=443112&stc=1

That looks a lot better. I'm wondering about those rollers dangling down off the gantry, though. I don't think they'll add any stability, and most likely will introduce disruption as that tray fills up with sawdust and the rollers have to deal with it. Instead of that, put in more triangular bracing between the gantry beam and the box section it's mounted to. Realize too, that the hot-rolled steel those tubes are made of is never very flat, so figure on something like a piece of extruded aluminum bar to mount the rails to, with shims between it and the tube.

The rollers are for his belt drive.
Put the belts on the outside.

Looks a lot better, but I think the belt drive will be an issue.

Better-but still no triangulation of the gantry beam to the carriages.Do remember the leverage of that enormously long Z axis backplate.I doubt it will be an issue in foam.On the topic of the Z axis I suggest you model a spindle and cutter even if it is a fairly generic shape and not 100% accurate.It will allow you to see what kind of angle will be the limiting factor for cutting in any particular direction.It also reveals how close to vertical walls you will be able to machine if a feature is more than about 70mm below the next plateau.Sometimes the revelation comes as a surprise.

See pictures below for updates.

@awerby - Do I need the aluminum bar? I was just going to put shims between the tube and SBR-20 extruded base. The flange of the extrusion is 5mm thick.

@ger21 - Redesigned with a belt reduction. The design accommodates direct drive or reduced. I'd probably start direct drive to lower the upfront cost for a functioning machine.

@routalot - It took me a second, but I see what you mean about limiting angles. Parallel with the gantry, it looks like it's a little more than 50 degrees from the horizontal at best (just eyeballing). That might be just fine. I'll have to think about it.

It feels like this design is getting a bit pricey. If there are no more major design concerns, I'll probably start on a BOM next week and come up with a price estimate. It might be a week or two before I can get to it. Would you guys mind scanning through a material takeoff list and telling me if any prices are questionable?

https://www.cnczone.com/forums/attac...d=443170&stc=1https://www.cnczone.com/forums/attac...d=443172&stc=1

Hi TE - You could remove the square tube that the bearings are on and use a thicker triangular brace you have. This will be stiffer then the current arrangement and less cost. Its better to leave the belt in free air, it will help as the "track" will not fill up with dust as it won't be there.One way to reduce cost is to remove un-necessary bits. The connection of the belt to the structure is very important so needs to be as direct and stiff as possible. Maybe integrate the brace to be the drive structure? If the belt is vertical it does not fill with stuff. So maybe the "brace " could become the motor mount as well....cheers Peter