Slant-Bed CNC Lathe Build

[barnbwt]

You're making my head hurt. But a very interesting concept.

Yeah, mine too. It took three days for me to get the concept worked out.

It's like the spawn of a slant bed and a Swiss lathe. No guide bushing, but a tailstock

 

[barnbwt]

Hopefully this helps clarify things;

I did determine that the concept is very similar to an old-school horizontal knee mill. Except the knee moves the spindle up/down, as opposed to the X-Y table, almost more like a surface grinder in that sense. That makes me a bit hopeful the idea has legs (but maybe arthritic ones)

I think tool crashes could be avoided fairly simply/automatically; there's a 'safety plane' between the turning tools and the parting bit, so at the end of each turning op lift Y to the safe plane, at the end of each drill/bore retraction lift to the same safe plane as well. Tool changes would require sending the spindle to the midpoint between tool positions (the 'safety plane' in the X direction) before engaging them into the stock along X (or along Y, depending how the tools are ground and how 'smart' the Y axis is). I will have to do a LOT more research before I can determine whether/how those routines can be configured in the Galil so they are called up by the generic T(X) tool G codes.

 

[Karl_T]

You know, a taper 50 horizontal mill with power feed on all three axis can be had for scrap value. SUPER RIGID machine. Put a 2J head on one end of the table with a Z axis in the quill, change the spindle out to something like a D1-3, add ballscrews etc. Now you got essentially what you are after only 10X more rigid.

IMHO, building your own machine has one huge problem - making it rigid. This is such a huge challenge that I would never try to make my own machine frame. Just my two cents.

Karl

 

[barnbwt]

You mean essentially this?
Hardinge TM with a milling head on the horizontal bar
That does look attractive as far as the footprint and I love the idea, but for all the stuff that'd have to be replaced;
Spindle assembly (since these are all low-rpm units as best I can tell)
Ballscrews (so I'd have to ditch the ones I have & source new ones, like much larger than 16mm)
Linear guideways (maybe I get lucky & get a machine with good ways, but most of these machines look pretty rough <5000$)
Much more powerful servo motors (so I'd have to ditch my motors, amplifiers, and power supplies)

The mere fact the spindle would need to still be a scratch build makes dealing with that much iron less desirable. And it's not even that much more; 900lbs vs. 6-700lbs. Much of that is between the moving parts & the floor, too, and I could still make this machine a stand if I wanted (would be nice to keep a coolant tank & pump down there)

Maybe if I can get a broken down or incomplete machine for the price of just the castings it's worth looking into. I'll keep my eyes open.

Also, the entire world supply of Horizontal mills is apparently in Rhode Island, which presents additional issues.

 

[JimDawson]

I will have to do a LOT more research before I can determine whether/how those routines can be configured in the Galil so they are called up by the generic T(X) tool G codes.

You just set up a G53 (machine absolute coordinates) position and call it before a tool change. This would put all of the axes into a safe position for a tool change. In my software this is done with a mouse click, the G53 routine is already in there.

 

[barnbwt]

You just set up a G53 (machine absolute coordinates) position and call it before a tool change. This would put all of the axes into a safe position for a tool change. In my software this is done with a mouse click, the G53 routine is already in there.

Wonderful, I was hoping that it would be fairly straightforward

 

[JimDawson]

Just to expand on this a bit, there are two coordinate systems in my software, the Work Coordinate System (WCS) and Machine Coordinate System (MCS). The WCS is relative to the work and tool offsets are applied, the MCS is absolute relative to the axis home positions. The difference between the two systems is set when you set the tools. This is again is done with a mouse click for each tool and is stored in the machine parameters. The Z WCS is set when you load the G code file, the X (and Y in this case) is always set relative to the spindle centerline, this is stored in a file attached to the G code file. The only time you have to reset these values is when you change the tools.

 

[barnbwt]

So I'm still crawling down this Y-axis rat hole for the time being...

I'm trying to figure out my motor-mount layout, and have been looking at some industry implementations; between the 6" motor, 2" encoder, 3" coupling, and 4-6" reducer, the stack hanging off the ball screw mount is a long, heavy, awkward abomination. How is this dealt with in the 'real deal' machines, or do they just cantilever stuff way out there & roll with it?

I did find some nice +4000rpm right-angle Nema 23 planetary reducer gearboxes out there (and not insanely expensive, surprisingly) but that doesn't seem to solve the problem much. Actually kind of makes the cantilevered loading worse, but at least keeps the machine envelope small. I was really surprised to see +4000rpm worm drives out there too (didn't know they could do that) but only in comically large sizes for this application (5" cube)

The closest thing to an 'elegant' solution I have now involves a wider/taller tool plate, hollowed out underneath for the ball screw stuff which lays right on top of the servo, linked via 8:1 25mm belt. Most concerning is the loss of material, which leaves the tool plate looking like a U-channel with a .5" thick floor. I'm just not sure how much it would impact rigidity for the load path to take the longer route up & over the hollow portion.

 

[JimDawson]

Maybe I'm missing something here but why do you need to gear down the servo so much? I'm not sure any reduction is needed. Both my X and Z are direct coupled to the ball screws, granted my servos are more powerful, but they are moving 2500 (or more) lbs of carriage. They are both cantilevered off of the frame.

If gearing is needed, then why can't you use a belt drive and tuck the motor back under the frame on a mounting plate, or maybe off to the tailstock side in the case of the X axis. You could do the same thing in the Y axis, put the motor at the bottom rather than the top of the assembly. That way it would be under the Z axis drive system. The Z axis servo could go at either end of the machine, where ever there is the most empty space. It might make sense to put it on the tailstock end to get it out of the way of the spindle hardware.

Just some things to think about

 

[barnbwt]

My servos only generate about 45 oz-in of torque, but spin most happily at around 3000-4000rpm. I only need about 500rpm on the ballscrew to get the rapids I'm looking for, so an 8:1 reduction seems like the obvious course of action. That also gets me about 1000lbs of holding force on each axis. That's just a really fast input RPM, though, so I'm not sure the best way to access it; a rated gearbox is the most obvious answer, since I don't know what the speed limit on 25mm timing belts is.

The cantilever of a reducer, servo, and encoder would be 10-11" off my mount; it just seems goofy when the X-axis rails are only 16" long themselves. Also makes the machine much, much larger. The right angle boxes work, but you still have a 12" long lever arm on the 6" rail spacing which also seems goofy. I'm well aware the rails can take it; that little moment is nothing, it just looks silly & 'wrong.'

Belt drive seems to be the best option at this point, even if it does mean a lot of the Z-axis carriage structure gets blown away. Assuming it can go fast enough. There's still enough space to bridge the two X rails with some pretty beefy braces, so maybe that's good enough.

Please let me know if hanging the headstock off the rails is stupid for some reason; it seems like a nice way of making the machine shallower (but taller) and potentially giving better access for swarf cleaning etc. Oh, and God bless Hiwin & Electrocraft for having auto-generating CAD filer tools on their websites. The big black cube is the Baldor BLDC spindle motor.



The spindle is also coming along nicely, closely based upon the previous design (the main purpose of all this CADdery is to move the design into Fusion vs. NX while incorporating changes & lessons learned). The casing is 4.5" OD, 45x85x19mm bearings, 5C bore. Spindle is extra long since I'm not certain how to implement the pulley, brake, and possible collet-closer just yet. I think I've got a handle on how to go about the assembly/disassembly though;
1) Press AC bearings into nose of casing against an internal abutment (per Hiwin dimensions)
2) Press spindle shaft & nose cap (for labyrinth) through AC bearings while supporting their inner races. Tail of spindle is a close slip-fit
3) Bolt on fixed nose cap (mating portion of labyrinth) to casing
4) Slip spacer sleeve over spindle tail against inner AC races, slip rear DG bearing onto spindle & into slip-fitted casing*
5) Torque a jam nut against the DG inner race to clamp everything along the spindle axis, then stack the pulley/brake/closer on that

Disassembly is done by removing the fixed nose cap & DG bearing/spacer, bracing the spinning nose cap, and pressing the spindle out the front. The AC bearings are driven from the casing by driving a tube against their outer races from the rear.

*Please let me know if an inner & outer slip fit for the deep-groove pulleys handling the pulley load is a bad idea; it seems like the best way to allow both inside & outside AC races to be press-fitted.