CNC daydreaming

[dkemppai]

It seems like you guys are putting a lot of effort into going the Linux route. If Mesa is slow at developing/deploying hardware, I would quickly move on. There are many other good choices in my feeble opinion.

This is just a discussion of possible controllers/concerns, etc. @rabler is mulling over a pretty adventurous VMC conversion (and he has Linux Experience). We're just musing over the options with him.

South Western Industries controllers are windows based. They have a GUI that runs on windows CE and talks to their own hardware. I don't know exactly how they're doing the interface, PCIe, USB, or other (most likely other). They almost certainly have to have a built in motion controller to offload the motion control work from windows. IIRC they charge to open up the Ethernet port on the box, so you have to pay extra to map a network drive for G code.

For me personally, I'm probably never going to run LinuxCNC (I did look into it for a build, but chose alternatives). But that doesn't mean I don't want to learn what's out there. Discussions of hardware and software limitations apply to many systems, as all CNCs have GUI and hardware based components. Having personal experience with Flashcut CNC, GRBL, Mach 3 (smooth stepper), South Western Industries, Heidenhain, and HAAS they all have their strong points. They all also have their limitations.

SWRI boxes are nice, but are a little harder to get to work with CAM.
HAAS is nice, but a lot of things are locked until you pay for the feature. Even high speed machining costs extra.
The Heidenhain controller doesn't like to play with canned drilling cycles (At least some cam post processors blow peck drilling into individual G code commands).
GRBL is a joke for anything but a small hobby mill. Not that it can't be useful, but it's quite limited.
TinyG address a lot of the issues of GRBL, and give 6 axis control). https://github.com/synthetos/TinyG/wiki/What-is-TinyG
Mach3/Mach4 supports separate motion control hardware which is good. But my experience with Mach3 keeps me from trying Mach4.
My personal feeling is that LinuxCNC is sort of that middle of the road option.
The Flashcut CNC controller also lands near LinuxCNC
Then there's the BuildBotics CNC hardware. That doesn't support backlash (or didn't when I looked it didnt), and has built in stepper drivers).
Many of the GRBL and Mach3 level machines are pretty limited, but great for beginners.
You mentioned Centroid, which I've heard good things about, but have not experience with.
And finally the one I want to try, the Kflop (8 axis) dedicated motion controller (With KMotionCNC).

In my own personal collection I've got four Flashcut controller boxes, and one Mach3 setup. I'm going to replace the Mach3 with a Flashcut as soon as I get a chance (just need the time). And then maybe use the Mach 3 boards for target practice afterwords!

The Kflop board has a lot of really advanced features, and I'm itching to try it in a build someday. Being able to close the Motor -> Encoder loop with fully tunable PID loops in HARDWARE! It also supports C, which is extremely powerful. It's the one out of all of the above I'd be willing to spend time to 'get to know'.

 

[rabler]

To expand on what I'm actually buying from Mesa:
1) A 7i92TM. This is and Ethernet interface, the FPGA, and two digital breakout interfaces.
2) A 7i85 (which connects to the 7i92TM). It has 4 channels of encoder inputs, plus 5 channels of RS-422 expansion. So it supports 5 serial expansion boards at 2.5Mbps.
3) A 7i84 card. Basic digital I/O, connects via RS-422. 5V-28V outputs. Several bells and whistles
4) A 7i83 card. Basically 6 channel analog servo interface. Connects via RS-422.

That's a start. I can add 3 more RS-422 breakouts beyond the above. Probably use one more in the front control panel since it is going to be 8-10 feet of wire from there to the back electrical panel. The front control panel break-out would be used to interface to an MPG and possibly a few "front panel" controls (e-stop, feed rate reduction). Front panel will have LinuxCNC monitor, keyboard and mouse. May also house the actual computer there, or in the large rear electrical cabinet.

If I run out of channels, the 7i92TM can host a second 7i85, adding another 5 channels of RS-422. So I can easily have up to 10 break out boards total.

The complication I faced is that the older board format was to have more integrated functionality in one board, with the option for one or two break-outs. So the orignal board I was looking at was basically the equivalent of the 7i92TM, the 7i84, the 7i83, and two RS-422 breakouts. The documentation and forums haven't completely caught up to the newer hardware, although it is basically equivalent from the linuxcnc perspective since it sees a "register" interface provided by the FPGA. (Some configuration required).

 

[7milesup]

Does southwest need too much $ to upgrade???

A friend has a recent mill of theirs and it is one cool machine.

I asked them about a color screen for the control pendant, and it was $950 just for the screen! The servo drivers are $900-ish each, and the motors are $1500 or so. The SWI Prototrak machines are very nice, but they are for industry, not the hobbyist guy like me. That being said, the ballscrews and other hardware on the machine are top-notch. Mine is old enough that it takes floppy drives. I did try to convert it to USB, but for some reason I could not get mine to work. There are a lot of people that use SWI machines and are very happy. Since mine is only a 2-axis, trying to go to a 3-axis and stay within the SWI ecosystem would be very expensive. I could probably remove the pendant and sell it on eBay to help defray the costs of an upgrade. I am hoping to utilize the current servo motors on the X and Y with a new control system, but I may have ask Jim Dawson for help on that (if he is willing).

EDIT: A recent machine of theirs would be awesome. I believe it would be top notch in all respects. Mine is a 1995 with a 3.5" floppy and DOS or basic Windows system.

 

[7milesup]

@dkemppai and @rabler

In no way am I meaning to take away from your adventure with diving down the CNC rabbit hole. It is a very interesting discussion that I too am enjoying.

 

[rabler]

@dkemppai and @rabler

In no way am I meaning to take away from your adventure with diving down the CNC rabbit hole. It is a very interesting discussion that I too am enjoying.

I appreciate the ideas and alternatives. Much rather hear about other options and don’t mind sidetracks and questions.

 

[dkemppai]

So, with the VMC is the tool changer treated as another axis? Or does it have it's own controller of sorts?

 

[rabler]

So, with the VMC is the tool changer treated as another axis? Or does it have it's own controller of sorts?

The short answer is that I'm not sure.

The first real step is going to be to reverse engineer the existing system. I have some documentation but mostly owner/operator stuff not repair/schematic level. I want to use the bridge crane to help pull all of the shrouding off so I can start tracing wiring BEFORE I disconnect anything.

I'll be surprised if I don't end up doing some fairly extensive hardware level interface work.

 

[rabler]

Mine is a 1995 with a 3.5" floppy and DOS or basic Windows system.

Yeah, this VMC as built is Win NT 3.5 based. USB is not supported. 10Mbps networking is, but primitive (SMB version 1.0, a security hole). Leaves 3.5" floppies as the only viable way of moving G-code.

Previous owner messed up the CMOS/NVRAM settings when the battery died, couldn't get it to find the boot drive. I've imaged the boot drive (using linux dd) so I could try to restore it. I don't really want to spend any time on the old controller, but I suppose if I could get it to boot it might actually be worth selling as spare parts on e-bay.

 

[Karl_T]

So, with the VMC is the tool changer treated as another axis? Or does it have it's own controller of sorts?

Generally just use LOTS of I/O. can do it with a separate PLC or within the control if its made to handle I/O.

On a turret lathe, I seen one done as an axis with an encoder. I did the same lathe with four inputs arranged so that 16 locations can be determined. So, it can be an option to do an axis.

 

[dkemppai]

If you get the old controller running, it might be a chance to see how they operated everything.
Then again, you might just be tempted to use it as is!

I wouldn't be afraid of using NT networking if it's on it's own private network. Just drop an extra card in the main PC, and run a dedicated cable to that controller. Drop files to it only from that PC. No need to let the VMC NT beast see the outside world. (It's how I run the old WinXP box that still runs Mach3 for the router...)

Dan