TOS FN20 from manual to EMC to Mach3 to LinuxCNC

If the subject looks cryptic, let me explain.
It was originally a (semi?)manual machine with X,Y and Z(knee) driven by a separate motor through a gearbox and the direction chosen by a "gear lever". So not spindle syncronized. In other words, it had constant feeds in one selected direction at any time. The construction is like the german Maho machines of the time. More info here: www.lathes.co.uk/tosfn40/page2.html



Then I converted it to EMC as it was called then. Not EMC2. All axes driven by step motors. Those of the kind where you do not want to drop it on your foot, or poke around the wiring while powered up. They are driven by separate step drivers from Phytron at 70V up to 10A. The interface was a Pico Systems USC. At that time the EMC was very far from what LinuxCNC is today. Especially in the number of installations and followers.

Then I converted to Mach3. Not that I can recall today why I did, it is many years ago now. It may have been just out of curiosity. There is now no card between the PC parport and the Phytron drives. They don't need it, as they have optocoupler inputs and can be driven directly by the 5V TTL signals.

After having converted my lathe from Boxford CNC to LinuxCNC I now have found Mach3 to be far behind LinuxCNC in pretty much all respects, and still moving forward, which Mach3 have not done for many years. The lathe also use Pico Systems USC, which I am satisfied with. It is capable of doing the job, and I have not even used all the discrete IOs. But I wanted to try Mesa, mostly because I'm curious and want to try new things. This may prove to be not so wise, but I can get back to firm ground by ordering another Pico board if need be. LinuxCNC will be what I see in use anyway. The HW is 5i25 and 7i78. Strangely enough the kit with cable was not in stock when I ordered it, but the boards were. At least that's what the shop pages said.

That is pretty much where it stands now. I just downloaded the latest DVD image and burnt it to a disk. Then I will swap in the hard disk that contained EMC and use that, so I can go back to Mach3 if I need to use the machine before it is up and running LinuxCNC. Come to think of it maybe I can run it using the parport, as the Mesa stuff was for some reason not shipped very promptly and it seems it will not arrive before the holidays.

Most of the configuration regarding steps/mm etc. I think can be pulled from Mach3 before I pull the disk. That should give me some of the parameters I need to set. Best even to just copy the disk to my NAS.

There also is a mechanical problem: The Z movement is by raising and lowering the knee (actually not a knee on this machine, but it works the same.) This is a slow movement and not free from backlash. The backlash is taken out by gravity so not really a problem in itself. But the knee is heavy so even if I put in a 170V/10A drive I have "in stock" it cannot be driven fast as when contouring in Z. Then it would start shaking the house. But the spindle can also be raised and lowered if I make some brackets and add a motor. But not by much, so the knee still have to be driven. My thought is that the knee can do large movements for setting up the workpiece to a suitable level. Then the spindle can do the contouring. But then LinuxCNC will have to deal with two Z's. Is this something that can be handled by LCNC, and how?

My thought is that the knee can do large movements for setting up the workpiece to a suitable level. Then the spindle can do the contouring. But then LinuxCNC will have to deal with two Z's. Is this something that can be handled by LCNC, and how?

Just set the knee up as your "W" axis.
If you have a working mach configuration it can be converted to linuxcnc.


Rick G

I think I have seen somehwhere that there is a utility that will extract mach3 info to help build a LCNC configuration.

If set up as a W axis, will it be controlled by using W in place of Z in the G-code?
If that is the case it will be easy to use without knitting together some complicated G-code.

I think I have seen somehwhere that there is a utility that will extract mach3 info to help build a LCNC configuration.

Take a look here...
www.forum.linuxcnc.org/forum/16-stepconf...-wizard?limitstart=0

If set up as a W axis, will it be controlled by using W in place of Z in the G-code?

Yes, this is how I have my knee mill set up. Typically I just use the knee to position the work and then the Z to make the final moves.
But W could do all the work although the moves are slow.

Rick G

My thought is that the knee can do large movements for setting up the workpiece to a suitable level. Then the spindle can do the contouring.

So long as your spindle has enough travel to cut the depth you require, do just that.

That is exactly what I do on my turret knee mill.

The other issue you have to consider is the accuracy of the knee rise mechanism compared to the spindle.

On my mill you would have to completely refit with a different mechanism to use a ballscrew.
The current mechanism uses a very coarse acme thread within a rotating threaded geared collar arrangement, driven at 90 deg via a manual winding handle.
It works well lifting a heavy table and even has graduated collars by the handle, but quite honestly winding it up to roughly where it is required, locking it off and then touching off using the Z axis is much more accurate.

regards

@Rick: Thanks for the pointer to the conversion thread.

@ArcEye: My mill have the same setup. And I exchanged the acme screw with a ballscrew. It positions the Z with enough accuracy for my needs even though there is a bit of backlash in the angled drive. But it is taken out by the weight of the knee+table. Possibly an acme screw with lube will not work quite as well in that respect. But it is slow because of the mass. And even if putting on a bigger motor could make it faster it would make it shake a lot. And apart from making the house shake it would also affect the accuracy of the work by rattling every joint in the machine.

I wish I could devise a simple Z mechanism with a 3Hp high RPM spindle and separate Z movement that could sit on the existing head without it and the existing spindle being in conflict. The idea is that the high speed spindle would be useful for fine work in aluminum and plastics. While the existing spindle (max 2500RPM) will do heavy work and roughing with shell mills and other large cutters where fast Z movements are not required. Then the manual spindle Z would be locked to assist a rigid setup.

A separate Z and spindle could be attached to the side of the spindle head. It would not have its centre over the centre of the table. However the workspace on the table is 60cm while X movement is just 30cm. So a workpiece offset on the table will still be well within the small work envelope of this machine even with a spindle somewhat offset in X. Just a thought that have been on my mind often, and will still be as you tell me there is little or no problem with LCNC.

I tried to load the LCNC CD in the machine, but it will not boot from CD for some reason. It did before. So I will probably take another PC and just swap them. My impression is that the latency figures are not as critical with a Mesa driver board. Is this correct, and what would be the number I should not exceed? I have several PC's to try it out on but will first try one where I have two with the same motherboard. Having a backup PC could mean peace of mind.

It all depends on what you want to do.
If you are making a heavy cut / slot in steel the speeds are slow so the knee might get you where you want to be.
Go down to depth then cut away.

But if you are thinking about 3d profiling at a fast rate and light cuts that is a different story.

If you look up linear slide actuators you will find plenty of pre built router mount Z axis assemblies that you may be able to adapt to your set up.
You could create more than one configuration.
You could then set up an XYZW machine using the knee as W and the router assemble as Z.
Also a XYZ configuration just using the knee as Z and the original spindle locked.
And a XYZ configuration using the router assembly as Z.

Rick G

The boards were shipped. But just late enough they were delivered after christmas.
Had they been sent one day earlier, I would have had them in time.
Not a very big deal, but the original plan had to be scrapped.
I will get stuff from Pico Systems or Eusurplus next time.

Now the boards are here, but I now have some tools to make on the mill to put a power steering back in one piece.
Mach3 can do it.

I installed LCNC on a PC that I plan to use and had an issue with the mouse. (Other post) But solved now.
The latency jitter is not great. Mostly 15us/24us when browsing, listening to Youtube music and such.
When starting GLXgears it shoots up to 50us/50us. But I don't imagine I will run anything heavy while milling.
Probably an editor, some webpages (LCNC site) and not fiddling about with them while running the machine.
And although this is somewhat unclear to me, I believe the step generation is in the Mesa card, so I should be fine?

Also I now run without limit switches and home switches. So I may as well make the brackets and stuff to have them mounted before doing the switchover to LCNC. And put in some connectors so the mill and cabinet can be easily separated. Maybe I'll just pick some motors and stepperdrives from the drawers and make an "air mill" in the computer room to emulate the real machine. Figuring out the correct wiring and config before connecting up the machine may actually be a good thing. So I don't have a crippled machine longer than necessary.