Slant-Bed CNC Lathe Build

[Karl_T]

Z or index pulse is part of the encoder. Its a pulse once per revolution. I think, but not sure, that you could home the machine with encoders attached to steppers. Homing is a three step process - rapid to home switch - slew speed to Z index mark on encoder - creep back off index mark.

 

[barnbwt]

I've decided to ditch the D1-4 spindle flange feature, and do a 5C only. I was really tempted to do a 2J collet system since there's a lot of stuff that needs just a tiny bit more capacity than what fits through a 5C, but I have to draw the line somewhere. Plus this way I get to use much cheaper 45mm bearings and a far smaller spindle profile. Compare;

The spindle shaft goes from 4.5" diameter to 2.0" diameter, and 9" length to 7.5" length (all 1.5" shifting the collet taper closer to the bearings, and out of the work envelope). I also changed the square mounting flanges on the casing to round ones so I could add more mounting bolts and true them up on the manual lathe once welded.

I still not entirely happy with the tail-end labyrinth closeout plate, since I feel its thin profile and large diameter are a recipe for vibration. It's not meant to be a proper fly-wheel, but I am considering the benefits of making it one to help dampen cutting vibrations and function as a brake rotor. The added inertia seems to be both a blessing and a curse in certain ways.

I now plan to mount the future 4-5 position tool turret directly to the YZ table since there is more room in the cutting area. With a max 5" swing radius available, I'll mount a double ball nut on Y to reduce backlash/chatter in that direction & lose the extra inch of motion that I'd planned on using to dock with a fixed tool changer. Except for really long boring tools or large parts, there should be ample Z travel to move the carousel out of the way of the workpiece to spin.

The spindle bearings and 48VDC power supply have been acquired, so the only thing left is steel and cabling, and lots of work. There was a change of plans on the AC bearings; they are now a medium-preload NSK7209A5TYDUMP4 because of a somewhat misleading ad (the previous part was pictured as a complete matched pair, but was being sold in halves for some dumb reason). The NSK's seem just as nice if not moreso on the exterior, and were the same price as the 'half' Barden set. The medium pre-load is overkill on my machine, but even if I do pre-load them to spec the top speed is still within my limits (I'll probably under-load them simply to reduce friction load on the spindle motor)

So, lesson learned on bargain AC bearings; since there's lots of orphaned bearing halves out there that sellers will list with the original box, it's easy to get fooled unless you look/read closely. I got lucky & am only out return shipping, so no harm no foul.

TCB

 

[barnbwt]

NVM

 

[Karl_T]

Your headstock design is now virtually identical to a hardinge lathe headstock. These are built solid as a brick sh*t house with a brake and 5C collet closer. The headstock bolts right on to the lathe ways, so it could bolt right up to a slant bed. The ground spindle also takes hardinge lathe chucks - 3 jaw, 4 jaw, step 5c chuck adaptors. I have one - someplace - but scrapped hardinge chuckers are common.

The listing shows the headstock
http://www.ebay.com/itm/Hardinge-DV...ardinge-Turret-Tool-Cross-Slide-/132236584966

same headstock on a dozen models

 

[barnbwt]

Okay, back on this project after a long absence, new town, new career...

As I reexamine the progress I had made, it looks like one of the more obvious 'gaps' in the plan is the leap between the software I will be using to generate my G-code commands (Fusion 360) and the Galil motion controller (DMC 2183). Galil uses DMC language rather than G-Code, so something is needed to bridge that gap, unless I'm missing something.

A compiler & post processor native to Fusion that outputs DMC would be the most direct option, but I'm positive nothing like that exists. Which is a shame since older affordable Galil controllers are so plentiful, but still capable of impressive control. Seems right up the alley of the Fusion user base, which are basically cheap bastards willing to put in some elbow grease to get the job done for a lot less money

 

[JimDawson]

A compiler & post processor native to Fusion that outputs DMC would be the most direct option, but I'm positive nothing like that exists.

Nope, and it would be almost impossible. You need a translator and user interface between the G code and the controller. There is a lot of stuff going on behind the scenes.

You took a break from this project at about the same time I picked up the Hardinge CNC. That was actually good timing. I now have my software running on the Hardinge.

 

[barnbwt]

(continuing the conversation from CNCzone) I think I've read some of the thread(s) on that build, but somehow I'd missed the more recent updates; I'll peruse those next in the coming days.

So I guess for Fusion to talk directly to the Galil, you'd need a ridiculously sophisticated post-processor to handle all the setting & configuration options defined locally in the controller? Maybe I'm still a little confused about exactly what data is being sent to the controller and how it differs from the G & M code commands put out by the post-processor.

Oh, another question I just had; do these controllers (the 21X3 series anyway) run commands near-real time as they are fed to them via ethernet & a small local buffer, or are they preloading large chunks or entire programs to local memory and only running standalone? Does the computer with your CNC interface hooked up to the controller "do" anything while the controller is executing DMC code?

 

[JimDawson]

(continuing the conversation from CNCzone) I think I've read some of the thread(s) on that build, but somehow I'd missed the more recent updates; I'll peruse those next in the coming days.

That was an interesting project.

So I guess for Fusion to talk directly to the Galil, you'd need a ridiculously sophisticated post-processor to handle all the setting & configuration options defined locally in the controller? Maybe I'm still a little confused about exactly what data is being sent to the controller and how it differs from the G & M code commands put out by the post-processor.

There is a lot of configuration and other functions going on in the CNC program. Assuming it was possible to create a post processor that would output DMC code directly, it still would not run the machine. You still have to deal with the tool offset tables, cutter comp, and a number of other things that the Galil does not do.

Each of the G and M functions have a corresponding sub-program in the Galil non-volatile memory. To do a tool change for instance requires about 50 lines of DMC code. But it is called by sending single line command, giving it the new tool number and telling it to execute that sub-program. ''NEWTOOL=5;XQ#M6,3''

Oh, another question I just had; do these controllers (the 21X3 series anyway) run commands near-real time as they are fed to them via ethernet & a small local buffer, or are they preloading large chunks or entire programs to local memory and only running standalone?

There is a 512 line x 80 character buffer that streams the linear interpolation commands to the Galil. It will keep going as long as there is data in the buffer. It's the CNC program job to keep the buffer full for smooth operation of the machine.

Does the computer with your CNC interface hooked up to the controller "do" anything while the controller is executing DMC code?

It monitors a lot of housekeeping activities, keeps the DRO updated, provides needed messages to the operator, and of course keeps the command buffer full, and sends commands as needed.

EDIT: I might add that unlike any other CNC program that I'm aware of, you can also surf the net, or work on drawings while the machine is making parts.

 

[Karl_T]

There is one GREAT, but expensive option. Camsoft is designed from the ground up to use CAM generated gcode and run Galil cards. It is the only thing I use on two CNC lathes and two CNC mills.

 

[barnbwt]

That was an interesting project.



There is a lot of configuration and other functions going on in the CNC program. Assuming it was possible to create a post processor that would output DMC code directly, it still would not run the machine. You still have to deal with the tool offset tables, cutter comp, and a number of other things that the Galil does not do.

Each of the G and M functions have a corresponding sub-program in the Galil non-volatile memory. To do a tool change for instance requires about 50 lines of DMC code. But it is called by sending single line command, giving it the new tool number and telling it to execute that sub-program. ''NEWTOOL=5;XQ#M6,3''



There is a 512 line x 80 character buffer that streams the linear interpolation commands to the Galil. It will keep going as long as there is data in the buffer. It's the CNC program job to keep the buffer full for smooth operation of the machine.



It monitors a lot of housekeeping activities, keeps the DRO updated, provides needed messages to the operator, and of course keeps the command buffer full, and sends commands as needed.

EDIT: I might add that unlike any other CNC program that I'm aware of, you can also surf the net, or work on drawings while the machine is making parts.

Okay, I think I get it now; I was a little soft-headed there, I think. Most of my experience is in CAD and CAM, not machine operation. Obviously a post can't perform the control, monitoring, and display functions that the operator panel does. Can't indicate a tool via CAM software output