Design assistance needed please!

Standard lathes have a single leadscrew on X (the cross slide). Not sure why you'd need two motors. Maybe if you post a diagram we could figure it out?

Also, assuming it's a normal horizontal lathe, Z is left to right and Y is up and down. Z is always concentric with the spindle bore (except with machines like grinders). A standard manual machine lathe is an X-Z machine. Multi-axis CNC lathes with Y axes also typically have the Y-axis non-perpendicular to the X-axis to provide more clearance over the (slant) bed. Y then becomes an interpolated "virtual" axis. I don't know if you'd need to do this for your build, I doubt it, but worth knowing just in case.

In either case, you'll need something more advanced than grbl to run 4 axes plus any additional IO like spindle brakes.

 

Gotcha. I think I can follow that easily enough. To include a Y axis, it may even make more sense for an extrusion build to run as a gantry machine. You're not gonna see much racking or other typical dual-drive issues with it being so narrow. The gantry will be X as normal, and Y and Z are flipped around (mostly important when it comes to CAM).

Well, the Z motors would need a driver each due to variations in inductance, etc, but you'd split the STP/DIR/ENA signals from the controller to both, no problems there.

The spindle is gonna have to run as a C axis for the pocketing since it has to have positioning, which might cause issues going back and forth between being a 3-axis dual-spindle XZY live-tool lathe and a 4-axis XYZA (in setup, not actuality for CAM purposes...Maybe) machine. Some kind of switchable spindle-speed-to-step-rate/step pulse passthrough breakout might be an option, so you have less to do in software... Because the software side of creating a true mill-turn appears to look like this: 4th axis as lathe spindle: question about index - it's complex, though skimming through the semi-working code on the second page it does look nice to just switch HAL pins around in software and not have to worry about anything.

@phil from seattle, does grblHAL have sufficient abstraction to do something like this? Seems like it should if it's a true HAL on top of grbl. And it would be much easier/cheaper to do the other basic controller parts, in theory.

 

Step/Direction/Enable

 

The BlackBox is basically an Arduino Uno and four stepper drivers in a box. It doesn't gain you an axis, because grbl is inherently limited to 3 axes. Two of the drivers are wired together internally as I described above.

If you can figure out how to do an X-Z-C only build, you could use a grbl controller, but if you have to have a Y axis to put flats in, it's not a viable option. I think you can get away without treating the "lathe spindle" as a spindle, just run the whole thing continuously as a standard XYZA 4-axis mill. The lathe part of this is confusing things more than it needs to, and with only live tooling and no static lathe tooling, there's really no advantage to thinking of it as a lathe. Think of it as a mill with a rotary axis.

Not sure what you mean by "a breakout board". Depending on what you need, a really basic CNC control system is a g-code interpreter/GUI, a planner/stepgen, and the drivers that power the motors. For grbl, the first part happens on the PC, the second part on the Arduino/ATMega chip, and the third in the drivers- whether they're standalone boxes or all together on a board like inside the BlackBox.

You need a more advanced controller, which is going to be more complex to install, configure and run. I use LinuxCNC, which can do up to 9 axes, but it's not easy.

Alternatively, consider swapping between your Y and A axis motors in a single driver on different programs, and only run three axes at a time. I'm slowly convincing @Batcrave to do this with his half-a-LEAD, it's definitely a viable option and allows for the use of grbl.

To do it as you originally envisioned is a much bigger project than you realise, I think. The mechanics is easy. The electronics definitely isn't!

 

I was referring to this:

Don't worry about the spinning, just keep it in "indexing mode" (which can also do continuous motion, of course). Let the program do the work.

grbl is a controller that runs on an Arduino. It's 8 bit, so not super accurate, but good enough. Simple, easy to work with, quick to get up and running, well-documented. You just send it g-code with a computer. It's limited to 3 axes though, and doesn't deal with rotary motion very well. Doesn't accept very many g-codes either, because it's almost at the limit of the hardware.

Mach 3 and LinuxCNC are both PC-based controllers. Mach 3 is usually used with a SmoothStepper for high-quality motion, where Mach 3 just does the planning and the SmoothStepper actually does the stepgen. It's an expensive way to go and doesn't seem to be all that effective, to me? I've heard not-so-good things about its support and general capabilities, tbh. LinuxCNC does everything and can output signals to stepper drivers through a parallel port (to a C25 5-axis breakout, say). Again, you get higher performance with more expensive hardware, but it can be inexpensive.

No idea what microstep is.

 

Yes, grblHAL has the ability to control up to 6 Axes. It is also easy to gang axes together.

I know a basic lathe isn't what's wanted but there is at least one lathe (Emco conversion) being run by grblHAL. Here's the Hackaday posting of it. Doesn't say grblHAL but it is most definitely running it. There is a video of it cutting threads - I've been drooling over that. Pretty cool stuff. I am working on a showcase of grblHAL machines and the lathe will be in it.

If I understand correctly one of the axes will rotationally position the workpiece for cutting. This is pretty much standard "4th axis" stuff. Any 4 axis machine should be able to handle it. I think this is essentially a moving gantry machine (ganged motors for that) and a motor for rotational axis. n'est ce pas? 5 motors are thus needed. You will want to get a motor with fairly high torque for the 4th axis. This Old Tony did a similar sort of thing and had some interesting observations.

My Teensy 4.1 breakout board is set up for 5 Axes. You can read about it here. You would gang Y and B together and run A as the rotational axis. Pretty easy to do.

And, when is a grblHAL version of the BlackBox coming?

[edit] one other thing, if your gantry is moving (vs a moving bed), you definitely want two motors to drive the gantry. The bearings on the rails for a moving gantry simply can't be tight enough to prevent racking. You will be MUCH happier with a motor on each side.

 

I assume you mean using steppers to control the table height? I don't see why not. You would use the sender or a pendant to control Z.

This is what I think you are asking - X and Y motors for mirror positioning, Z (maybe ganged to B for 2 motors) to control bed height and A to drive the rotary axis. Totally doable with grblHAL. Though, I don't know how you generate the GCode - would like to learn about about that. Cost of the controller is about $60 - BoB, Teensy 4.1 and connectors. grblHAL is free. The new sender (currently Grbl GCode Sender but soon to be renamed to IoSender) is really well done. If you liked Grbl Panel, you will love IoSender.

 

Would probably have to do it the way I used to do it before I started using LaserWeb more- use Fusion 360's 4th/3+1 milling features, then manually modify/custom-post-process them to laser code (mostly modifying Z-5s to M4s, etc). It's a bit of work with Notepad++, but once you've done it a couple of times it'd be pretty easy to algorithmize the process.

My laser already has a Z axis and that's never been an issue. A axis might be another matter.

 

Then I would use A as a bed height motor and Z for the rotary axis. It doesn't matter which axis for bed height since it is manual.

 

For what it's worth, here is a showcase of several grblHAL based machines. First one is an OB CBeam machine.

 

Perhaps a bit OT but since VCarve and K40 lasers has been mentioned above:

Laser PPI mode and tube coolant handling plugins now under development for grblHAL.

ioSender has support for Vectric Output direct to machine for those who do not want to fiddle with saving and loading of .nc-files.

 

You must be

Welcome!

 

Nothing slips by you Rob! Yes, Terje is da man. You can see several of his projects in the showcase. Amazing work.

 

Haha, I was like hmmm, I recognise that name from somewhere....

And fwiw, @terjeio, I'd like to see this little discussion thread that Phil started updated and added to by someone who knows the system inside out: CBeam and a Workbee using grblHAL - while it may not be ready to replace my LinuxCNC controls or run any random machine I throw it at, I'm thoroughly intrigued by the idea of a 5-axis, 32-bit control running on hardware the size of a quarter (man, living in the future can be cool)!