[Cheeseduck]

After around a month of design, I think that I might be ready to start putting together my second CNC. My first was all plywood and used allthread for the drive; as expected, it cut like a wet noodle and took forever to do it. For that reason, my new design's main goal is to retain as much stiffness as possible. I am hoping for a solid machine that can cut wood fast, engrave fast, and do a bit of light aluminum cutting. My budget is ~$250 (my old router will be glad to donate the electronics), but I can increase it if absolutely necessary.



My current design is, in a sentence, a mishmash of ideas from every commercial kit out there. The base is made out of 2x4's (Solsylva style) , and has 1/4" thick CRS on both sides (CNCRP style).



The gantry consists of a 2x4 beam of hot rolled 1/8" steel tubing with two 2x6x1/8" blocks of aluminum tubing acting as risers. The linear bearings are homemade clones of the CNCRP bearings (Momus style), and are bolted onto the bottom of the gantry risers. The 2x4 that rests on top is also the main rail that the z-carriage rides on (xyz-cnc style). A plywood plate attached to the side holds the bottom linear bearing in place and holds the motor/linear motion junk in place (don't ask why I made it blue). A small section of 1/4" steel angle also serves to reinforce the joint between the aluminum and steel.



The z-carriage is just a bit of 1/4" aluminum plate with a number of linear bearings bolted onto it. The portion that rides on the 2x4 steel beam uses 6082RS (skate) bearings, and the portion that guides the z axis uses 1/2 OD bearings with shields. The z axis itself is a solid block of 3/4x3 aluminum bolted to a 1/4" thick aluminum flat that acts as a rail.



#25 chain is what pushes the whole thing around (Buildyourcnc style). I have 425 oz.in motors and can use chain to reduce them by 2x, so I should have 1.5 inches per rotation. Since my CNC is mostly steel and aluminum, I should be able to put around 100 pounds of tension in it. The Z axis uses allthread.

My old CNC router will be making some of the wooden components for the new one (once the new one is done, I'll replace the wooden parts with aluminum versions). I have access to a small mill, so I should be able to drill most of the holes with a good amount of accuracy. The router is from harbor freight (probably not a good omen).

So, am I ready to build this? If you see anything that looks off, I'd love to know. I am determined to make this machine perfect, and I won't start until it gets there.

[pippin88]

You'll never get a perfect machine (period) and broadly speaking you get further from perfect the cheap you go.

The general consensus is that chain drive is generally pretty poor (cogging). Belts are better usually.

[askjerry]

I wonder about a single bar for the Y-gantry. With two you have more ability to adjust. I think if anything... I would evaluate 80/20 because it can be adjusted and trued in. I would also agree on the belt vs. Chain drive.

[kennethpaine]

Are you planning to place the X axis drive on one side only? You ought to drive the gantry on both ends to avoid parallelogramming.

Kenneth

[troyg]

What CAD system is it modeled in? I would be willing to pull it up and give it a thorough review if it is in SolidWorks or ProE. I have some decent experience with motion design - used to design medical imagers (laser printers using polyester media).

[Cheeseduck]

I suppose my statement about making a perfect machine was a bit frivolous . In better terms, I'd like to build a machine that gets as much performance out of every dollar as possible.

pippin88- I have seen quite a few machines use chain effectively- my main reason for using it comes from this machine (Home built CNC router cutting dovetails - YouTube) which also uses a geared down chain drive.
...then again, my biggest mistake with my first machine was not listening to any of the advice I got. If I were to use belts, then how could I do it? An alternative that I was considering was 0.37" steel reinforced XL belting with a 1/2 reduction provided by chain. Is this any better?

askjerry- I don't quite understand what you mean by "two bars". Could you elaborate?
I've also decided against 80/20 simply because I don't need any of its connection-related versatility. The y-bar's only connections are to the aluminum risers, and I can mill slots instead of holes to allow some ability to adjust its length. Thanks for the suggestion, though.

kennethpaine- Woops, I forgot to include that in the drawing. It is being driven by both sides by a shaft that goes through the center of the steel tube.

troyg- Wow, that's quite the favor! Unfortunately, I designed the entire thing in Autodesk Inventor 2012...

[rncbme]

You might want to consider Acme Threaded rod for the z-drive, etc. You can drill the nut(s)
for a grease fitting.

[troyg]

I would echo their concerns with chain drive - it is typically avoided for precision control. XL belts would be a significant improvement, GT would even be better.

[troyg]

Why would you use the chain to provide the reduction instead of the belt pullies? You would just be adding extra parts (cost and reliability) using the extra chain/sprockets.

[Cheeseduck]

troyg- Will I still have the ability to make aggressive cuts with steel reinforced XL belting? My whole CNC is designed for that sole purpose, so losing that ability because of my belts would be silly at this point. I'll change out the sprocket-based reduction, though, and replace it with belt(no need to half-ass it, I suppose).

rncbme- Sure thing, I think I already have a cheap place to get that. On a separate note, would rack and pinion work for the Z axis too? I know that it seems a bit strange, but I think that my motors can actually handle it.

[Trotline]

Chain has its place. It's by far the least expensive option for long travels. Chain drives are a bit temperamental and quirky. For instance, most people intuitively want to have the chain banjo-string tight, which is usually not the optimum setup. However, chain can be used effectively. When properly adjusted, (about 1" of sag over 6'6" of travel, on my machine) the cogging is extremely minimal, and you have to know where to look to spot it on a carved piece. I can do high resolution lithophanes, as long as I keep the stepover on the chain driven axis, and the rastering passes on the other one.

For travels of 48" or less, though, I don't think chain is the best option. Acme screws would be a better choice.

Luke

[Cheeseduck]

Huh...well, this is confusing. My goal is, put simply, to have a machine that can cut fast without sacrificing accuracy. A belt drive seems very reasonable at this point, but I don't know how XL belts compare to metal chain. Is there an appreciable difference between the "stretchiness" of chain and steel reinforced XL belting?

[Trotline]

How fast do you want to go? Neither chains nor belts are especially good for performance in any arena except speed. Both have issues that screws don't have. Acme leadscrews aren't particularly slow. Here's a machine profiling a guitar body. All ACME driven.



Luke

[dmalicky]

In the simple and low $ segment, I'd say the design is very good, and a great start. It's smart to copy good ideas, and that you want to get the details right before building. Performance per $ is a great quest. Best features as I see it: CRS/CNCRP style linear motion, a tube for the gantry, smart rail heights, simple overall. Suggestions...
- More spacing for the x bearings (longer 'feet'), to prevent tipping deflection.
- Make sure the 2x4 tube you get is very uniform in width and height. Any deviation beyond a few thousandths will cause slop. To minimize the effect at the cutter, increase the width between the y-bearings. The 2" dimension is less important since your design elastically loads those bearings.
- Cap 1 face of each riser tube (or both, but 1 does plenty of stiffening)
- More vertical spacing for the 1/2" z-car bearings, and make the pairs that oppose each other ~at the same height. The vertical spacing is ideally at least as much as the z-clearance.

The best drive system depends on what's important to you. Chain is speedy, but gives up stiffness and accuracy (#25 is a good choice). Belt is somewhat more accurate than chain, but less stiff. ACME is good for stiffness and accuracy, but probably more $. Steel reinforced belt vs chain is a tough call. Chain should be stiffer, but that extra steel creates sag and that affects accuracy and stiffness just after changing directions--softof like backlash but it ramps up in stiffness, as the sag is fed from one side to the other.

For a 2x4 machine, I'd agree ACME is easily the best of those options, especially for cutting metal. Fast travel precision ACME is about $10/ft, but nuts can be DIY, motor couplings are cheap, and no pulley reduction needed. But ACME would require either a single screw under the table (whole new design and racking stiffness is much lower), or dual screws with 1 motor and a belt/chain transfer (complex, low stiffness), or dual screws with 2 motors and drives (best option but more $).

[Cheeseduck]

Thanks for the advice, dmalicky. I've updated my design with most of the changes you recommended. It already looks better!
I think that belt may be the best option for me at this point- chain looks too finicky and sounds like a purchase I'll regret later. ACME is out too due to price- the dual leadscrew approach is around twice the cost of all the others, and may be doomed from the start by my motors. They have bad resonance problems (even dampers don't help much), so my speeds are limited heavily by that. (Fun fact- the modified Solsylva posted as an example above was yet another inspiration for my router!)

The system I have in mind right now uses XL pulleys to provide a 1/2 reduction. The actual drive will use 16T5 steel reinforced belt on a 14 tooth pulley. Is this overkill, underkill, or just right? If that checks out, I'll be ready to start my build.

[harryn]

The 2x4s on their thin side still won't be stiff enough for that size table. Consider instead to put them on their 3 1/2 deep side and make a box with 3/4 ply on the top and bottom. (sort of like a house wall frame) A 2 x 4 every 12 - 16 inches boxed with 3/4 ply is pretty stiff. If you can get kiln dried, that is a lot better, and 2 x 6s are ideal if kiln dried.

Skip the 1/8th wall and go for 1/4 in wall. It is so much better to work with and can accept a tapped thread if needed.

For a heavy duty machine, a belt length : width ratio goal is 10:1 or maybe 15:1. A light duty machine can go double that, so a good target is 1 - 2 inch wide belt for 48 inch long. Brecoflex makes some AT5 and AT10 that are about right for your machine.

[dmalicky]

Glad to hear it's helpful. Resonance is usually most related to the stepper drivers and settings.

Agreed with harryn on the Brecoflex AT5 belts. This pdf is great: http://www.brecoflex.com/download/B212.PDF
The ATL series is designed for higher accuracy (page 15).
Pages 136-147 shows each belt's Specific Stiffness. Those stiffness numbers are approximate/linearized assuming some initial pretension, as belt's spring rate is non-linear.

Wider is certainly better for stiffness, but keep in mind:
- Wider belts need more pretension to get into the higher stiffness regime.
- The motor (or jackshaft) bearings need to be able to take the overhung load (wider belt --> more overhang)
- Wide belts can get pretty pricey. At some point, ACME makes more sense.

For a timing belt drive, the gantry needs to be kept light or accels would need to be very slow -- I'd switch to an aluminum tube.

Also, the 3rd pic shows the y-car with the backside portion connected via 4 long bolts. Those will flex a lot when transmitting lateral loads. I'd increase the bolt diameter a lot, or extend the solid portion so the unsupported bolt length is short, or both.

Good luck!

[Cheeseduck]

AT5 it is, then- it's only $6 more and looks to be twice as strong. Not bad!
I've also added two plates to link the front and back halves of the carriage to help out those four bolts. My drive should have a theoretical force of around 40 pounds at speed, so I think that acceleration shouldn't be too big of an issue.
I am starting to get a bit nervous about the gantry's drive system, though. The rod that links both sides and supports the drive pulley is only 5/16. It is supported well on both sides of the drive pulley, but I'm not sure if that 50 mm region will flex under tension. Here's a pic to clarify.


(The outer pulley is the drive pulley- the others are for reduction)

[dmalicky]

In double-shear like you show, I'd guess 5/16" quality steel will be ok. It can be calculated, but it's a little involved--the failure mode is combined bending and torsion with the motor at stall torque, and we'd need to know the belt pretensions, too. If 5/16" is the biggest stock pulley ID, I'd probably use 1045 steel (http://www.mcmaster.com/#grade-1045-steel/=lx6qic). Or hardened alloy steel dowel pin is very strong (http://www.mcmaster.com/#catalog/119/3304) if you don't mind grinding flats for the setscrews, and that it might be a slight press-fit in the pulleys/bearings.

[Cheeseduck]

Well. I've made the plunge. My final machine has a 2x6 steel beam in place of the 2x4 for the gantry, and uses 16 mm AT5 belting for the drive. I'll probably start a new build thread once all the parts ship. Thanks for the help!