Looking to start a CNC project

Hello and Happy New Year,

This years project idea is to build a cnc workshop and funds are tight so I don't want to make expensive mistakes with purchase of controller boards and servos/ steppers etc that are not upto the job.

The end goal is machines capable of small production in a variety of materials. So maybe would convert more industrial sized machines such as a Harrison l5 lathe and Bridgeport sized mill.

I figure if the machine without cnc is capable of daily running and accurate work a conversion could do it or am I wrong?

Any thoughts and opinions are food for thought

Thanks,

Chris

Hello Chris,

Plenty of people have done CNC conversions. Some machines appear to be particularly amenable to CNC conversion and searching this site and the wider web should quickly identify some of the 'prime contenders'. One of the main criteria is usually the ability of the machine to be upgraded to have less backlash, usually, but not always, by adding ballscrews.

Neil

<Man on the Silver Mountain on the radio >

It would be helpful to know what type of work and materials you intend to use the equipment with, for a Bridgeport size machine I would definitely go with ball screws, and servos, I have limited experience of converting lathes to CNC I've done a 5" raglan but again also went with ball screws, however in this instance I used steppers, the next thing to consider is controllers and software Mach seems to be the most popular with model engineers but Linux controllers are also available, I think your best place to start is probably the forums for the above software, this subject had been covered many many times on these forums with some very helpful people ready to help.


Shaun

Something you might want to consider as an intermediate, cheaper and less drastic step is to cnc-convert an X-Y table which can then be mounted on the mill. You still have one mill-axis to do of course but at least the mill is available for making parts for most of the job.

I just finished a 2D conversion of a standard, Far-East X-Y table fitting it with ball-screws (tight) and steppers for use in engraving and cutting PCBs etc.

I'm part way into a BP 3-axis conversion. Currently in a state of hibernation due to moving back from Canada to Blighty but hoping to resurrect it asap. The photos and screenshots in my albums are not up to date, apart from the Z-axis which is essentially complete.

My original plan was to drive the X leadscrew directly and the Y leadscrew by timing belt, with closed loop steppers from Leadshine (as used on the Z-axis) but I'm now going with proper closed loop servo motors from DMM Tech with enclosed 2.5:1 toothed belt reduction. These motors claim to give flat torque all the way up to 3000rpm max speed unlike stepper drives and are competitively priced relative to open loop stepper motors and drives.

I could upload some more current views of the assembly and some pics of the compts which I've made (manually) on the machine if you are interested. I've created all the parts and assemblies in Solidworks - don't know if you have CAD? Once you get going, you can obtain most of the parts from the likes of Ali Express and ebay at reasonable cost.

If you are planning to convert large machinery with reasonably powerful motors (mine are rated at 400W ie 1/2 HP but are deceptively small), you need to be careful to incorporate limit switches, e-stops etc to ensure that the thing doesn't go ape and hurt you. Although you can get "breakout boards" from any number of sources, they generally have limited I/O and bandwidth (and questionable parentage). For not much more money, you can get more "proper" interface boards. I've gone for the Mesa 5i25 / 7i76 pairing which has a dedicated FPGA on board (rather than relying on the PC's processor and parallel port) and buckets of I/O. There are other options but as it's only a part of the expenditure, it seemed to me something I'd want to get right.

The other chestnut is the control software. I'm going with LinuxCNC which is open source and free but also pretty professional and well proven. It requires a PC running Linux of course but it's effectively got a proper RTOS, unlike the PC applications. It's also pretty complex ie certainly not quite plug and play. Many people use Mach3 (soon 4) but this is proprietary and runs on Windows but isn't so easy to customise. JAS who also posts here has considerably more experience than me in using LinuxCNC and has his own blog.

There are loads of conversion kits out there and many of them seem to be "hit and hope" or "it seems to work", with little evidence of any rational design process. It's worth looking at the Tormach "white papers" which are refreshingly open about their design process and give you a good starting point. Certainly, if you look at some of the specs for the stepper motors people have used and do some basic calculations, you'll realise many of them are massively overdesigned in terms of the headline stall torque rating. Even if you plan to do a lot of heavy drilling, it's hard to see why you'd need 11Nm motors and sure enough, the professionally implemented machines have more modestly rated servo motors. Many user manuals are available if you look for them...

Hope you find my thoughts interesting / helpful. There's no right answer but plenty of options!

Murray

Hi Neil,

I think ballscrews are a must from research without close to zero backlash the mills can easily stack up lots of mistakes and loose accuracy. I think play in the screw is less of an issue with a cnc lathe as it can be programmed to compensate but would still upgrade them.

Shaun I currently make replica parts for vintage bikes and can use a lot of different materials i'e brass, bronze, stainless steel and high tensile such as en19. I know you can't always tick every box with smaller machines but need to keep machining options open. One of the reasons I only use the model engineer forum is because of the mind set of anything is possible in your shed.

Bandersnatch it definitely is scary been without a machine I think I'll try do a machine I use the least first so I can get a feel for the conversion and programming. I'm not sure I could cope with everything cnc as there's so many times I like the trial and error approach of making things.

Thanks,

Chris

Chris;

Firstly - thanks Murray for the blog reference. Sorry to loose you from Canada, although physically, you are probably closer to me now... (Canada's a big place, and Britain is not that far across the Atlantic...)

I did a little Unimat SL lathe conversion. If you can find a little lathe that works, making it into a CNC lathe is fairly trivial. If I said that you could do it (with care) for 50 quid, maybe less, I don't think I'd be far off. Look at my blog for info on my Unimat SL conversion, if you wish. I did use an expensive stepper motor controller, but I had it kicking around.

The thing with mills is that they really do cut better with climb milling, and while you can program in "backlash" compensation into them, you are better off with ball screws. Lathes are slightly more amenable to backlash issues using the as-supplied feed screws. So if you do a lathe, at least you'll have an easier time of it, and learn from it, but you may not have as useful a tool as a mill might be.

I *really* like LinuxCNC. It is incredibly stable, and people have made it do many interesting things, machine-wise, and, heck, the Internet basically runs on Linux, (as do your Android phones) so LinuxCNC is built on a great base. Why anyone would choose a windows-based controller (with all the kludges required to run CNC) is beyond my ken, but that's just my opinion, and is certainly not held by others.

For those experimentally-minded, machinekit.io (web site) is a fork of LinuxCNC for experimentation, and for differing platforms, like little single board computers. As an historical note, Art Fennerty "forked" LinuxCNC (known as EMC back then) and ported it to the Windows platform, and called it "Mach". The rest is history. Google it if you don't believe me.

AS to mill conversions, I'm basically finishing up my "G0704-style" mill conversion; the jury is out as to whether it is going to be great, or mediocre. We'll see as I gain experience with it. The kits and plans do have some things that I don't think are correct, but again, I could be wrong, and time will tell, and I may eat these words!

Happy New Year - off to a steaming party - warmer than last years' : (on youtube search for "conway steam canada" and see if you get a video of last year).

Edited By John Alexander Stewart on 01/01/2015 21:03:33

Hi Murray,

I've not yet got a Bridgeport Mill but I guess that's the most flexible in terms of work size and horsepower especially for the size of the machine. I've got an Elliot Jig borer and a Centec 2b horizontal and vertical mill I could cnc one of those but the the table sizes are only about 16" and 25".

Which model of the DMM Tech motor are you using? I think closed loop servos are good option if understand right better at avoiding loosing the position on the table mid job?

Are most interface boards compatible with either solution Mach or via linux cnc? I wouldn't mind buying the interface board and motors and hooking them up to the computer and just bench testing telling each motor what to do.

Chris

I also make parts for Vintage bikes and Cars, also custom bikes, and am currently making a Dry sump kit for an M3 BMW /F3 car conversion.
For good finish on any machining job you need sharp tools stiffness of setup and machine, smooth cutter paths, I would deffo be looking at a Bridgeport size machine servos and ball screws are a no brainer simple.

It a steep learning curve but well worth it, I have two friends who have with my help converted there millers to CNC, one an X3, the other an Elliot, both have had machinery for years, both tried to save money, both regretted it, both now have ball screws, both thought CNC was easier than manual (you just press the button marked sprocket) both soon discovered they where very wrong, neither can draw for toffee so I end up doing pretty much all the drawings, but they are me mates so I don't mind.......

Hi Chris

For my X and Y axes, I’m using the 60M-DHT-36 motor, which has 60mm flange (NEMA 23?), 1.3Nm continuous / 3.6Nm peak, overall length 127mm to flange face. The drive is the DYN-2H which is rated up to 750W and operates off 48Vdc. I’ll be running both drives from a 1.2kW PSU I have to hand, although I’m not expecting to be developing that kind of shaft power on the slides of a BP milling machine. The DMM drives can take either analogue or digital (stepper) inputs. As I’d already gone for the 7i76 I/O board which is designed for stepper drives, I’ll use the step/dir signals. Ideally I’d have got the 7i77 instead, which is suited to driving servo drives.

With 5mm pitch ballscrews and 2.5:1 reduction, I’m not expecting to stall the motors! I’m using double ballscrew nuts which you can preload to reduce (largely eliminate) backlash. They don’t cost much more. I even got the ballscrews machined to my exact design for something pitiful like $10 and they made a good job of it.

For the Z axis which I tackled first, I used a Leadshine “integrated easy servo” closed loop stepper which is also NEMA 23 and rated at 2Nm. The drive is built in to the motor and only requires step and direction signals. The technical data on these is a little sketchy but it looks as if they are using vector control which is pretty snazzy for a stepper drive. In contrast, some of the stepper drives out there seem to be little more than an off the shelf stepper driver IC in a box.

I reckon it’s got to be sensible to use a closed loop drive, whether it’s based around a stepper (eg Leadshine) or servo system (eg DMM). Apart from the drive taking care of following the required movement, it can also signal if the positional control is lost finally – in which case you can decide to abort with some level of control.

The Mesa cards don’t seem to be suitable for Mach – all of the info I could find about them was exclusively focused on LinuxCNC. On the other hand, it sounds as if most of the parallel port-based breakout boards will work with Mach. That’s probably not surprising, as Mach seems to be a commercial application aimed mainly at hobby users. I’m not an expert on the history but AFAIK LinuxCNC was originally developed with US government finance and it’s a professionally respectable application. Doesn’t surprise me to hear that Mach branched from LinuxCNC / EMC some years ago.

You can install Ubuntu Linux with a current version of LinuxCNC onto a spare PC and try it out (it’s all free, being open source) if you follow the instructions on the LinuxCNC website. If you have Windows on the machine already, you can make it dual boot. There are “simulation” installations that allow you to run a virtual CNC system without any real hardware. You can install interface boards, drives and motors to them and easily get shafts spinning, then build your system up from there.

Murray

Chris, if you don't already have a mill to convert could I suggest looking around for an old Denford cnc mill? A lot were used by colleges and schools, often seemed to get little use in fact, but the build quality is great, they have ballscrews fitted, and with modern controllers work very well. I have a Novamill which is much smaller than you probably want, but they did much larger machines too, also cnc lathes.

Sorry daft question what is a ballscrew ?

A better machine would be a Bridgeport BOSS 1 to 4.

They have steppers already fitted and ball screw, pressure oiling and set up for CNC.

These mechanically are a very good machine and it's rare to find one worn as the original electrics of the day were that crap they all broke down.

Many are still stood in corners of workshops acting as benches and the going price for these is from free to £800. Anyone asking more than 800 is never going to sell it.

Because of their size they are not attractive to home shop guys like a Denford or Boxford and you get more for your money. They take to bits very easily with a one tonne engine crane and everything can be transported in a transit van so that's another hurdle out of the way.

.

Jim,

See here

MichaelG.

If Linux is used for cnc in industry then I'm up for trying it, in a previous life I worked with Linux. The simulation idea is a cheap way to find out how it all works first. The spec Murray has mentioned has at least given me a bit of a bench mark to work off when purchasing.

John if you know of a Bridgeport Boss in that kind of price range let me know it would certainly save some money and time. What kind of places do people find machines? Anything on eBay seems to go for much higher and ends up been worn out.

The denfords seem to turn up a bit but don't the specs not mention cutting steel?

Shaun I don't expect it to be easy but hope my computer experience may help, I did once try do some 3d drawing in A program called Creo or something that's meant to be like solid works. The bit I'd like to see is how that drawing is then converted into tool paths as I guess that you need the orientation of the part the way it's clamped up and parameters setting according to the depth of cuts and cutter size on the machine? Is it more complex than that?

Thanks,

Chris

I certainly wouldn't argue with JS's suggestion. If you could get a Boss for "retrofitting", it would save a lot of your time. I bought my manual BP clone without any thoughts of CNC conversion and only started on that route last year. Having said that, I'm more of a machine designer / builder than a scale engine builder so that's what I like doing. Perhaps if I can acquire a decent sized workshop that has room for more tools, I may look into acquiring and retrofitting a proper CNC machine. Obviously I'd be wanting to keep the BP clone!

Personally I'd struggle with anything bigger than my manual BP clone which weighs (allegedly) 950kg, as that's right at the limit of my Weber engine crane. Anything bigger and I wouldn't be able to move it about myself - something I've done quite a bit of over the years.Looks as if a Boss is closer to 1500kg ie 1.5 tonnes.

Converting the CAD models into g-code for the machine is indeed one of the tricky bits. It's a lot easier if you are essentially just operating in 2D but even then, the clever part is getting the tool loads and cutting parameters right. As I'm still at the machine building stage I can't recommend anything with any authority but I'm sure others will be able to. As I'm using Solidworks, I plan to use HSMXpress, which integrates with SW and is free for 2.5D use. Real workshop warriers are able to generate g-code manually for some jobs but I don't expect that's something I'll be able to manage!

Murray

The Denford Novatmill has no problems cutting steel. When working for a living I bought a couple of these for use in our R&D department. These machines were aimed at schools so their advertising tended to emphasise the sort of materials schools would use for demonstration purposes. However, the Novamill is a robust, though small compared to a Bridgeport, mill with a 1kW motor running to 4,000 rpm (if I remember correctly). If you can get a machine with a running copy of VR milling then this will convert DXF files into 2 1/2 D G code. In this software you can specify the material to be milled and the machine will calculate the appropriate spindle and cutting speeds - steel is one of the options but it is possible to define your own materials as well. I wish I had room for one of my own. I guess that there are quite a few on the 2nd hand market because schools seem to have moved on to 3D printing and routing.

HTH

Rod

I have found its ok to have some space for a machine but getting the machines in require egyption techniques my fault as i built a garage over the vehicle access. The last machine I had to resort to my garage door as a ramp I had half an hour to get the mill out and the van back to hire company on my own.

Rod what about the denford micro mill and micro turn lathe?