Home built CNC router resources

[David King]

Hugh will be my guide here and anyone else that wants to tag along is welcome.

I have 4 different Cad programs that I'm trying to evaluate and make up my mind on.
First up is Rhino 3d which that have ported to mac os as a free beta. I've made no headway with this one, totally unintuitive and I'm too lazy to sit through videos when my screen looks nothing like what's on the PC version

Next is Dassault Systemes DraftSight, a free 2D version, ported to all OSs. These guys make SolidWorks and clearly they want you to get hooked and step up to the varsity version for $3-4K. Their only web-based tutorials make it look so simple but then I can't seem to ever make what they want me to make.

AutoCAD 2013. ported to mac, seems like the real deal, a bit more intuitive than Rhino but doesn't seem to have Rhino's facility with surfaces in 3D. I always kind of feel like I'm lost in the bowels on the DeathStar. It's all a bit sombre.

Lastly Autodesk Inventor 2013. This one seems like a winner albeit still from the dark side. Very intuitive and more importantly fully parametric. Unfortunately it's PC only and totally worthless with Apple's magic mouse on a windoze seven driver. Lots on nice interactive video tutorials to get you started. It's tempting to buy a $300 mouse to go "native".

Everyone who has a seat of SolidWorks is telling me to just forget about the others but I'm not about to commit fiscal suicide to discover that I'm too brain dead to learn any new tricks. That said I can handle G-code and do all sorts of useful tasks with a cnc machine with mach-3 for starters. Once I have it working for me i'll be far more motivated to jump into 3d head first and not forget what I learned last week if I'm using it everyday.

[ReWind]

Subscribed.

I see lots of small-scall CNC plans and kits out there for milling, but most of the large scale stuff I see is for woodwork. Not to say it isn't out there or that some of the more modular smaller scale systems couldn't be adapted with more robust materials to be somewhat more industrious.

I've also put a bit of consideration to taking something like a used Southbend or Bridgeport or even a used larger Grizzly or Jet and converting it to CNC. There's a mean J-head with an auto feed and quill autofeed near me for $1,800. …including a Bobcat to lift it. It would be an impulse buy for me in a way though, as I really have much more to consider before taking that sort of plunge, deal though it may be.

The same fella from CNCDudez in England sells a couple versions of his CNC router frame / kit: CNC Design Limited - CNC Desktop Router

It's not really up to the task of machining steel as a regular task though. That's a "must" for me, but may not matter to many.

[David King]

Steel isn't hard to machine provided you get the speeds and feeds down to an appropriate chip loading so the cutter doesn't burn up or lose all it's teeth. You can't expect to run a 6" dia face mill over case hardened blocks if you don't have several tons of stationary iron and a 20HP spindle in play. If you have the room for it and enough amps at your service panel you can pickup a used, mid 80s-90s bed mill for just a few thousand $. Any of these will eat a Bridgeport for breakfast.

[Bruce Johnson]

I know that this is a big subject, but in general, the hardest part of building a metal-cutting CNC machine is making up the frame and sliding parts which are rigid enough to take the cutting loads without significant deflection. It's less difficult with a wood-cutting machine, but it's still an issue. With any CNC machine, if the frame flexes under load, you've lost any accuracy that the scales and software might have.

If you want to make a metal-cutting CNC machine, the simplest path by far is to buy a used manual (old school!) milling machine, and convert it to CNC operation. Old manual milling machines are being sold off for scrap iron price these days. You can pick one up for a few hundred bucks or less. That will give you the basic frame, moving XYZ axes, and powered spindle, all strong enough to handle metal cutting. You convert it to CNC by essentially adding servo or stepper motors to drive the leadscrews of the three axes, attaching the Scales (the digital measuring devices), and wiring it up to a controller. The really nice conversions replace the old leadscrews with ballscrews, but many homebuilt conversions keep the old leadscrews. I'm not an expert on this, but there is a ton of information out there. Converting old mills to CNC is a cottage industry and hobby.

For someone like David, who already owns an old Bridgeport, converting it to CNC is easy. I believe that there are commercially available bolt-on kits for almost any model of Bridgeport mill. And a CNC Bridgeport can do both metalworking and woodworking. It isn't ideal for making bass bodies, but it could handle all the tricky parts.

[David King]

Bruce as much as it would seem to make sense I have yet to see a CNC design that makes use of scales for position feedback. All the designs seem to think step counts with no feedback or rotary encoders with tight feedback are all that's needed in spite of inconsistencies of rolled thread etc.

[David King]

The questions I want answers to are: What brand, size and torque stepper? Geckos? Ground or rolled threads? And what diameter? How do I estimate the loading on a linear rail?
I'm definitely thinking cheap Chinese water-cooled spindle but will one of those hold up in a crash? What HP?
Are there designs where the gantry and table move in opposite directions to minimize inertia?

[Joe Gwinn]

The questions I want answers to are: What brand, size and torque stepper? Geckos? Ground or rolled threads? And what diameter? How do I estimate the loading on a linear rail?
I'm definitely thinking cheap Chinese water-cooled spindle but will one of those hold up in a crash? What HP?
Are there designs where the gantry and table move in opposite directions to minimize inertia?

First, list the desired uses and required work volume, accuracies, and time to complete a sample project.

By the way, depending on production volume, it's often better to use the CNC to make router jigs that are then used to cut out the guitar bodies and the like.

[Bruce Johnson]

Bruce as much as it would seem to make sense I have yet to see a CNC design that makes use of scales for position feedback. All the designs seem to think step counts with no feedback or rotary encoders with tight feedback are all that's needed in spite of inconsistencies of rolled thread etc.

Really?? I'm surprised by that. I'm no expert on CNC machines, but my understanding is that all "real" CNC machines (Haas, Fanuc, etc.) use scales as their position feedback. And I believe that all the factory Bridgeport CNC Turret Mills do too. So, there must be retrofit kits that work the same way. Even those cheapo import CNC Mill/Drills use scales, don't they?

Using step counts and rotary encoders on the motor is fine for little tabletop non-load things like a CNC pickup winder, but I wouldn't use that method for anything with heavy cutting loads.

[Bruce Johnson]

By the way, depending on production volume, it's often better to use the CNC to make router jigs that are then used to cut out the guitar bodies and the like.

That's so true. For most 2-D work, a guy with a pin router and a good fixture will turn out parts faster than a guy wearing a lab coat running a CNC router. A CNC machine is capable of doing some very intricate work that would be difficult otherwise, but for simpler work, it's slower. Often any gains in speed during the cutting time are lost in the part-change time. It takes a pretty good CNC setup to match or beat the guy with the pin router.

By the way, I've heard that while running a CNC machine, you must wear a lab coat and safety glasses, hold a clipboard, and have a serious expression on your face. And you have to be spotlessly clean. No oily rags in your pockets, no smear of grease on your chin. The floor is freshly waxed. That's what it looks like in all the CNC ads I've seen. It doesn't look like any fun.

And, if all the data is in the machine in digital form, what is the clipboard for? Is there a box to be checked that says "Green Light On, Making Humming Sound"?

[Joe Gwinn]

And, if all the data is in the machine in digital form, what is the clipboard for? Is there a box to be checked that says "Green Light On, Making Humming Sound"?

It's to smite the unbelievers.

[Hugh Evans]

One of the most common mistakes at the entry level, which is extreme detrimental to machine precision is using data aside from known specifications of the linear drive components. If you're using lead screws/ball screws and stepper motors, decent quality screws will have better tolerances than your machine often by a full order of magnitude. Planning your speeds and feeds will prevent missed steps. Providing the CAM software with good data on your drive system is the first step to realizing its full potential. Generally, closed loop controls are not implemented on entry level machines. By the time you need it, you're ready to move on to professional equipment.

If you really want to go down that road, I say build the machine first. Then, if necessary, rotary encoders will be the best starting point.

[Hugh Evans]

This is going to be fun. I have some advice and feedback about CAD software, but before I post them I want to refresh myself on the latest details. Among other things, Solid Edge may well be the best entry level 3D software with almost no learning curve associated with moving to SolidWorks.

Setting up a blog might also be a good idea to make this project easier to follow.

[David King]

My envelope probably needs to be 16 x 32" with 6" of z (but maybe 3" is plenty, I just don't know..)
Cutting speeds of 200ips and jog speeds of 600ips seem pretty standard or is that way fast for a home brew?
I figure almost everything will be done with a 1/2" ballnose at 1/4" depth and 1/4" or more lap on roughing in hardwood.
I'd expect expect 1/2 those numbers in aluminum.
A good spindle should do 30K all day I imagine. I figured a 2.2KVA model (3HP or so)

I've been told that the table should be 3/4" thick 6061, that the gantry should have dual leadscrews with either a timing chain or dual motors and drives.

Cam? The folks I know with Mastercam seem happy. There's a way to snag a free cam option with Solidworks. I don't quite remember how.

Rhino cam looks like it works for all the guitar makers i know.

[Hugh Evans]

There's not a huge price jump to increase the cutting area a bit, so unless that's the biggest size you expect to need consider going up to 24"x36". The IPS figures are definitely conservative. There are no universal figures for feeds and speeds: everything from the size of the cutter, number of flutes, maximum RPM rating, machine rigidity, spindle limits, stepper motor torque, etc., play a role. 300 IPM is not at all uncommon in home brewed machines cutting wood. Spindle vs router is a major decision point. Spindles are much more expensive, quieter, generally more powerful, and have wider speed ranges available. Most require 3-phase power, and converters only go so far... Topping out at sufficient levels for a 5 to 8 HP spindle, but if you have 3-phase service the only limitation is current. My router can run as low as 10,000 RPM and 3-phase service is not available in our current shop space, so it was a simple decision. If you go with a spindle, buy as nice of one as your budget allows. Cutting bits are really an afterthought, factoring them in as a consideration right now would be similar to designing a car based on the intended tires.

For this size of machine, one leadscrew per axis, each with a single stepper motor, bearings, and anti-backlash nut is more than enough. Adding more will increase cost without increasing resolution and even when operating perfectly, interference is a very real possibility. Investing more funds into superior linear motion systems will easily outperform multiple drive motors.

3/4" extruded aluminum with T-slots makes a great working surface, thinner pieces can work as well. Using simple beam bending calculations can give you an idea of how much of a load they can carry and for the working area you have proposed the number will be huge when the rest of the frame is considered. The entire surface does not necessarily need to be covered. Aluminum is a very expensive material to risk damaging during normal machine operation. It's primary purpose should be providing a stable platform on which fixtures can be affixed. In my shop we reposition and sometimes remove sections of aluminum t-slot for various applications. They are very convenient when used in conjunction with carriage bolts, vacuum lines, quick release levers, etc.

CAM often has a couple of components: creating cutting paths, and translating code into signals that the power supply/motor controllers will use to drive the machine. It's commonplace to use two pieces of software to handle these tasks. This is the portion with the worst learning curve, and many people go into it expecting a fully automated process. It most certainly is not. Cutting path optimization is payback for the ease of CNC operation. MasterCAM, RhinoCAM, and every piece of high quality CAM software is not cheap. It can easily surpass the price tag of CAD licenses.

I understand you have a lot of questions, and that's a good thing. Bear with me, the depth of a complete CAD/CAM system can be daunting and I want to give you the best answers possible... Not just to help with your project, but to make this something of a useful guide to others who might have other needs.

[defaced]

Hugh will be my guide here and anyone else that wants to tag along is welcome.

I have 4 different Cad programs that I'm trying to evaluate and make up my mind on.
First up is Rhino 3d which that have ported to mac os as a free beta. I've made no headway with this one, totally unintuitive and I'm too lazy to sit through videos when my screen looks nothing like what's on the PC version

Next is Dassault Systemes DraftSight, a free 2D version, ported to all OSs. These guys make SolidWorks and clearly they want you to get hooked and step up to the varsity version for $3-4K. Their only web-based tutorials make it look so simple but then I can't seem to ever make what they want me to make.

AutoCAD 2013. ported to mac, seems like the real deal, a bit more intuitive than Rhino but doesn't seem to have Rhino's facility with surfaces in 3D. I always kind of feel like I'm lost in the bowels on the DeathStar. It's all a bit sombre.

Lastly Autodesk Inventor 2013. This one seems like a winner albeit still from the dark side. Very intuitive and more importantly fully parametric. Unfortunately it's PC only and totally worthless with Apple's magic mouse on a windoze seven driver. Lots on nice interactive video tutorials to get you started. It's tempting to buy a $300 mouse to go "native".

Everyone who has a seat of SolidWorks is telling me to just forget about the others but I'm not about to commit fiscal suicide to discover that I'm too brain dead to learn any new tricks. That said I can handle G-code and do all sorts of useful tasks with a cnc machine with mach-3 for starters. Once I have it working for me i'll be far more motivated to jump into 3d head first and not forget what I learned last week if I'm using it everyday.

SolidWorks guys tend to be very vocal, but it's far from the only 3D modeling game in town. ProE, Catia, Unigraphics, SolidWorks and Inventor are the programs I run into the most when talking to customers. Depending on what you're interested in doing, and how deep your wallet is, one may be a better fit than the other. I use Inventor and for what I do (which is pretty basic modeling work) is quite capable. One word of warning though, one of the big strengths of 3D modeling software is the ability to do assemblies. Inventor LT does not have this capability which bumps you to the full version of Inventor - at $3500 a seat. I have no knowledge if any of these do CAM work and spit out machine code. Most of the stuff I have CNCed is simple enough to be programmed by hand which the machine shop does when they make my parts.

In addition to the big guys, there are smaller CAD houses out there. I'm not too familiar with them as a group, but IronCAD was a company I stumbled on while doing some research and if I remember right, their software is a bit more affordable. Each modeling software has it's own work flow and since I know Inventor, I found the way IronCAD did things to be very unintuitive, so I didn't consider it.