View Full Version : Noob looking for little advice...

12-03-2018, 12:59 AM
Hi folks!

So I have no experience making CNC machines, alas that wasnt a part of mechanical engineering when I went to uni, and I am looking to build myself a CNC wood lathe. But because I have lots of ideas for more complex products, and have a bad habit of biting off more than I can chew, I am intending to make it a 4 axis lathe / carving machine. Preferably with laser engraving to boot. The 4th axis can be indexed - I shouldnt need continuous, and I am semi-confident I can figure out the mechanical aspects of the design, plus I have a lot of experience in wood turning. When I comes to the electronics and programming though, I am... more shaky. I'll be doing mostly thin spindles up to 50cm long but want to be able to come in and do bowls and the like as well. Am intending to use Fusion 360 to spit out the code, and CAD is pretty much second nature to me, but I really am clueless when it comes to the electronics and putting it all together into a working package.

- So, this is a big project, and question is - in my shoes, where would you direct your time for research? I am thinking of doing a small, almost toy sized version with 3d printed parts etc just to test out the concepts and get used to it all first, before creating the final build, does that sound reasonable, and how big a project would you say this really is? Like, big or huge?? Any other advice much appreciated!

As a PS, on the mechanical side, when it comes to hitting 3500rpm, being able to turn bowls and goblets, but also being able to do precise c axis movements, I figure I'm going to need a beefy motor and a strong stepper, with only one connected at a time. So each on its own individual bed, moving in and out to connect up with couplers etc. But that seems... very inelegant. Do you know a better way?

Thanks for your time!

12-03-2018, 11:59 AM
Something which sprang to mind for your PS is engaging the stepper with a clutch with interlocks to cut power to the spindle motor before the clutch can be engaged, especially easy if your stepper is driving the head through a belt as the driven pulley can be on bearings and run freely with the clutch disengaged.

12-03-2018, 03:45 PM
Thanks for the reply - I did look into whether there were any off the shelf CNC clutch mechanisms that I could buy but couldnt find anything. I mean I could probably design something, or use e.g. lawn mower clutches but it is still essentially using steppers to mechanically engage each drive, one at a time. Which I can do, I just thought maybe there was another way! The only alternative that I saw elsewhere, was a suggestion to use a variable reluctance stepper, which is happy to freewheel, but my inclination would still be to disengage the main motor when the stepper is active to reduce mass being driven and backlash. At which point I might as well put both drives on a clutch. Perhaps I am being over the top with it though, and over-thinking the problem? Anyway, the interlock is an interesting idea, I hadnt thought of that!

12-03-2018, 06:51 PM
In that case a DC Servo with encoder might make more sense, then you can run it as a positional axis or at a set speed.

12-03-2018, 08:47 PM
I looked into servos before, but the combination of speed (3500rpm), power (500W+ ideally) and high accuracy together in one package seemed like a stretch, my concern was it would be always be inferior in one area - or least without paying an arm and a leg, and then adding the control complexity on top... But perhaps I just gave up too soon, I'll have another look into it, thanks.

EDIT yeah definately gave up too soon, can get for 400 or so, much better option.

13-03-2018, 11:46 AM
There's a reason why lathes with live tooling have huge horse power motors. It's not that you're ever likely to use 30+hp while turning, it's so they still have enough torque to maintain a stationary position.

Given you're working with wood, the forces involved shouldn't be as high as metal, but you can still calculate the potential forces. If you know how big your live tooling motor is, and how much torque it can produce, you can then calculate the potential maximum force on the cutter edge. You can then take that force and calculate the torque the spindle needs to produce to hold steady (plus add a safety margin).

For ideas, search out the In-turn (there's a huge build/development thread on the mach support forum), which is a dual speed 4th axis add-on.
However, having a suitably geared motor only connect for positional work is a valid option. At low speed, having the high speed motor still connected shouldn't be a problem. It will add a bit more inertia, but I'd be looking at keeping the low speed at under 60rpm, so that additional inertia is going to be minimal, and reduces the level of complexity you need for changing speed.
The big problem is how you handle backlash in your positional motor drive system.

13-03-2018, 12:31 PM
Nice thanks - I have found the thread, or another related thread, lots to dig through there.

You mentioned you didnt think that extra inertia will be a problem, but that backlash in the positional motor will be - so, where is this backlash coming from?

The servos I was looking at last night were in the 1.5-3kW range, not exactly 30+hp! Damn that would be a chunky servo :). I'll see if I can figure out my torque needs then.

13-03-2018, 12:48 PM
Looking at figures, say you have a live tooling spindle capable of 1Nm (pretty high for anything wood cutting).
Now you have a 10mm diameter tool, which means the force is acting at a radius of 5mm, or 0.005m. So 1Nm/0.005m gives us potentially a 200N force at the cutter edge.

Now if you apply that 200N at a radius of say 50mm, or 0.05, it translates to 200Nx0.05m or 10Nm.
And that's before you consider what can happen if the tool grabs the work and takes a far larger cut, at which point you have the additional inertia of the live tooling spindle to drag the workpiece around, so you need to allow a reasonable safety margin. And any backlash will increase the likelihood of grabbing, unless you stick to conventional milling.

Off course, those figures are dependant on the tooling being able to transmit that much torque without snapping. It's worth spending some time running some figures through a speeds and feed calculator (FSWizard online I think gives torque figures, if not HSM Advisor has a free 30day trial which definitely does).

13-03-2018, 01:46 PM
OK, I understand, ratio of forces from the diameter of the cutting head to the diameter of the workpiece - good to know. I'll check out the resources.

Think I gave my specifications wrong to begin with too - I see that people are classifying the slow rpm rotary as another axis to the high speed. So, I need continuous on the 4th axis and I would very much like to be able to change the angle of my tool, but that doesnt need to be continuous. And perhaps I could add this in later. 5 axis then.

... just to get a "finger in the air" idea though, how big a project do you think this is? Can I do it on a budget of under 2k??

13-03-2018, 04:25 PM
If it's your primary spindle on a lathe it's your Z Axis, not your Fourth Axis.
If it's not then add a diagram as I'm lost! ;-)

13-03-2018, 05:29 PM
All these axis! OK so, two axis for positioning the tool (X and Y), then a rotary table axis on the tool (no idea what thats called... doesnt need continuous, maybe just change angle a couple of times per part). Then the workpiece spindle axis which should be able to both go at speed, and precision angle controlled (which I always thought was the C??).

13-03-2018, 05:57 PM
(which I always thought was the C??).

There are VMC, HMC and Lathe Conventions, Mill/Turn adds complexity and a degree of the manufacturer making it up as they go along.
On a lathe the main spindle is on the Z Axis.

13-03-2018, 06:02 PM
A conventional lathe has no Y axis ;-)

14-03-2018, 01:42 AM
Lathe axis are confusing.

As magicniner says, inline with the spindle is the Z axis.
Standard across the spindle line is the X.
Y axis is typically if the entire X/Z axis can be moved (typically it won't be at 90 deg to the X)
C is typically if the spindle can be positional controlled.

Off course, by the time you get into that level of complexity, you're verging on mill turn, in which case it really does vary.
I just tried searching for a suitable diagram, and have just learnt nobody really uses an A-axis on lathe (should be rotary around the X axis), and B-axis which should be rotary around the Y axis, is more likely to be used for the tailstock/second spindle housing linear Z movement.

However I'm sure I've seen diagrams where the tool holders are mounted on rotational axis labelled A and B.

14-03-2018, 09:54 AM
Back to the main question though -
If you want continuous rotational positional control of your lathe spindle then you will need a sensor on the spindle and a motor/controller capable of the speed for conventional turning and also the hold for stationary and torque for slow mill-turn operations with live tooling, (that eliminates steppers) a controller capable of operating a brake on the spindle may be required.
Given the power in a motor required for slow moves and stationary hold for live tooling, even with a brake you will need a higher power servo motor than you think (multiple rather than fractional horse), I think your proposed budget may require expansion, you'll only really find out when you start asking for quotes.
Before you buy anything you should start with your chosen CAM software and proposed controller as this will have to deal with running the hardware and if your project ends up needing a 4/5 axis CAM and a genuine Fanuc controller you may find your budget not going far.

14-03-2018, 06:59 PM
The circuit board side is my biggest worry... I'll focus on researching that for now then, really I am so clueless in that area it's embaressing. Hopefully I'll find some sort of bodge workaround instead of paying an arm and a leg. I mean I could pay more, but I kinda dont want to!

Interesting with the power of the servo - looking on alibaba at similar 4 axis lathes with live tooling, they seem to be good with 3kW, which is reassuring.

Re axis - OK I am using X, Z, C and maybe B. Got it!

15-03-2018, 01:45 AM
Dynomotion KFlop could handle what you're trying to do, provided you understand the kinematics involved, and aren't averse to a bit C programming.

Mach could be used, but I think that would involve CAM being responsible for handling tool offsets/angles, as I don't think it can handle the tool offsests when you get into angling tools around.

I'm sure LinuxCNC might be able to handle it, but I've got no idea how much it can do without getting deep into custom programming.

15-03-2018, 01:32 PM
Mach could be used, but I think that would involve CAM being responsible for handling tool offsets/angles, as I don't think it can handle the tool offsests when you get into angling tools around.

Once you move into 3D parts with 4 axes and above it's best to have your CAM handle all that, that way provided your machine model is accurate your simulation will be accurate, as will any machined parts saved out of your simulation as solids.