Slant-Bed CNC Lathe Build

[barnbwt]

I have no idea if this is the right place for the build thread, but my home is above my shop, and this CNC project will be in it

I'll say up front this is a big, big project. Especially for something that's only the size of an oven. This first post will be a sort of rolling summary (so long as I remain able to edit it) of milestones passed & yet to be completed. I will try to mention all the questions I've had to find answers to in order to progress, so that someone embarking on a similar adventure might take a short cut. If it looks like I'm missing anything, or especially if anything needs clarification or is outright wrong, by all means speak up! My starting point is as a total ignoramus who only has a couple years' experience with a small manual lathe, and a degree in aerospace engineering. Having seen what a *well made* machine can do --even a small one-- and the near-limitless capabilities of even simple automation compared to my own abilities, I think it's worth reaching for the stars a bit.

I will have a lot of questions I'll ask here, every step of the way to be sure . I'm not posting on CNCZone since that place is lousy with abandoned machine-build threads, and it's depressing. It seems like plenty of ambitious projects happen here, and will way less pretense. Thanks in advance for tagging along, and of course for any help or advice.

So away we go...

Overall Concept:
-Benchtop size, trying to keep it inside a 3ft cube
-Workpiece size 2" x 7.5" (with boring capacity to that depth)
-CNC lathe profiling capability, including threading/rigid tapping
-Light milling capability with a live tool (engraving, slots, helices --real word)
-4000rpm spindle (well over 1000sfm for a 1.25" 'average size' profile)
-Capable of good results on all commonly machined metals (steels, aluminum, brasses, Delrin)
-5C Collet/Closer spindle, with D1-4 mount for chucks
-Design for future; tool changer, tailstock, flood coolant, full closed loop axes, chip auger


That last one (growth capability) is why the machine isn't about 1/3 shorter.

Design Constraints: All Good Designs Have Restrictions
-Residential 110VDC/60Hz 1-PH @ 15A, grounded GFCI outlet
-3ft cube external envelope (again)
-500lb weight limit (this is flexible within reason)
-"Easily" moved; it won't be on casters, but I plan for it to have lifting points for a hoist
-Bench top; heavy wooden workbench, but no 'proper' machine platform to stiffen it (also flexible)

Design Criteria: Brass Tacks by Category
Complete Parts List (with Purchase Price):
-Used Baldor BSM80C-375AF servo motor (135$ with extra 10:1 gearing) & pulley
-Used Baldor FMH2A09TR-EN43 (267$) & cables
-Nema 23, bipolar, 381 oz-in, 3.5A steppers (tentative) & mount @40$/ea
-Galil DMC-2813 motion controller, 8 axis, ethernet interface (350$)
--Galil SDM-20640 daughter board stepper drivers, 4 axis, 3A & 50V (300$)
-Used Cherokee International QT6A1D 24VDC @ 5A, 5VDC @ 20A Power Supply (24.99$)
-48VDC @ 8A stepper driver power supply (tentative)
-12/24VDC ducted cooling fans (tentative, for servo driver & power supplies)
-16mm OD, 4mm pitch SFU1604 ballscrew & nut, 300mm X axis (30.58$)
-16mm OD, 5mm pitch SFU1605 ballscrew & nut, 500mm Z axis (76.89$)
--BK/BF12 ballscrew bearing mounts qty(2) (0$ part of Z axis order)
--Stainless steel spring coupler 1/4"-10mm, qty(2) (0$ part of Z axis order)
-28mm steel ballnut mounts for 1604/1605 ballnuts qty(2) (20.52$)
-25mm x 400mm linear guideways qty(2) & HSR20CA carraiges qty(4) , X axis (238$)
--25mm x 600mm linear guideways qty(2) & HSR20CA carraiges qty(4) , X axis (0$ part of X axis order)

-1.5" box tubing, 1/4" and 3/16" thicknesses (tentative)
-3/8" or 1/2" steel plate (tentative)
-1" steel plate for carriage & slide table surfaces (tentative)
-1" square cold roll steel bar for spacers (tentative)
-2" square aluminum bar for stepper motor mounts (tentative)
-Spindle assembly (ball bearings, spindle shaft, spindle casing, pulley)
-45x85x19mm Barden 7209/209HCDUL paired P4/ABEC7 angular contact bearings (130$)
-45x85x19mm Fafnir 6209/MM209K P4/ABEC7 deep groove bearing (80$)

Machining Envelope:
-Intended part size is 2"OD x 7.5" stickout (min cutter travel 1")
-Spindle bore pass through 1.25" diameter
-No tail stock at this time, but room to add
-Ability to move to tool changer docking position along X-axis; total travel must be 6"
-Boring depth capacity of 7.5", means Z-travel of carriage must be 15" total
Machining Capability (Forces/Power):
-Capable of machining steels, aluminums, plastics, brasses, and irons
-Carbide cutter speeds (>1000sfm, for a ~1.25" OD part is 3000rpm, so spindle = 4000rpm)
-Shallower/faster cuts acceptable so long as surface finish is not compromised (if possible)
Optimizing Rigidity:
-Beefy, welded truss frame, with several hundred pounds minimum weight
-Optional attachment of mass/dampening material later (iron plate or epoxy granite)
-Heavy gauge plate steel motor & spindle mounting plates at headstock
-Thinner gauge sheet closeout bolted to exterior
-Direct-driven stepper motors
-25mm timing belt drives spindle
-16mm ballscrews, 25mm linear guideways both have far higher load capacity than needed
Optimizing Accuracy:
-Use of linear guideways to constrain axis motion with very little off-axis backlash
-Use of ballscrew drive for axes to reduce backlash (also friction)
-Use of servo motor on spindle to precisely track its position & speed for threading & milling
--May ultimately add a secondary encoder to the spindle to even more precisely monitor position
-May eventually add a second ball nut to either/both axis if the lost cut envelope is acceptable
-Addition of friction brake to spindle to hold it steady during stopped moves (slotting/side drilling)
Minimizing Size:
-Work envelope and machine travel kept to minimum practical size
-Simple single-belt spindle drivetrain
-Z-axis carriage and X-axis cross slide made from 1" steel plate with inset mounting surfaces
-45deg bed slope is the best compromise between height & depth, and a short load path down

Electrical Components:
-Baldor BSM80C-375AF Servo motor is ~1kW, with max RPM of 4000rpm, 2500 pulse/rev encoder, generating 3.6N-m continuous force, and operating at rated 6.29A & 300VAC. The power figures triple at maximum, but for a lathe such violent power moves are generally unnecessary and indicate more serious problems (crash). That, and the 1kW resultant power draw make me confident this motor won't blow my breaker in operation (inrush, perhaps). The way these AC servos operate, is a fancy VFD called a servo driver feeds them a continuous current across the whole speed range, but increases voltage to drive it harder, while increasing frequency to keep time with its spinning (greatly simplified). This constant torque quality manifests as a nearly flat torque curve across the operating range of speeds. While the motor is not hugely different from any other induction unit apart from its low inertia, the real magic of servos stems from their precise control; an encoder tracks the exact angular position & feeds that data to the servo driver in real time, where high performance control systems adjust power delivery to quickly react to applied loads on the motor or commands from the controller.
-Servo driver is a 9A unit set up for analog command signals, but digital on the inside so it is also capable of accepting step/direction inputs just like a stepper motor (which are then parsed by a computer & sent to the motor, then adjusted as the encoder reports the response). The driver actually replicates & outputs the encoder signal, so it can be sent back to the motion controller itself for a complete closed-loop of feedback for all components. The controlling input pulse is a low level signal, and the computer must be isolated from the enormous & variable motor current, so a separate 24VDC supply is necessary
Component Block Diagram:
http://hobby-machinist.com/attachments/block-diagram-png.239021/
Controls Block Diagram:
-Laptop running Mach3 CAM software -> ethernet -> motion controller
-Motion controller -> servo & stepper drivers -> servo & steppers
-24VDC power supply->motion controller, servo drive
-48VDC power supply->stepper drive
Power Consumption:
-15A 120VAC available, 16A total peak demand (dang, no tunes!)
-9A continuous draw by servo motor drive (at max power)
-2A/120VAC continuous draw by 5A-24VDC/10A-5VDC power supply
--motion controller requires 1A 5VDC
--stepper driver fans require .5A 5VDC
--servo driver requires 1A 24VDC
--servo controller cooling fan requires .25A 24VDC
-5A/120VAC max draw, I think, from 8A 48VDC power supply
--3.5A steppers (3.5*2*.67) comes to 4.7A 48VDC
Mechanical Components:
-25mm linear rails; these carriages are large enough to have zerk lube fittings, and also have external flanges that broaden their mounting footprint. They also seem to be made to a slightly higher standard than budget smaller rail sizes.
-16mm ballscrew; the 5mm pitch is plenty for my needs and the screw itself is far, far, far, far stronger than any load I'll be applying to it via the stepper motors or cutter. I wanted more control & resolution in the radial/X direction so I found a rolled 4mm screw. The balls are smaller so the load capacity is lower, but the greater number of points of contact is ultimately more rigid.
-Ballnut mounts are an off the shelf steel variety paired with my ballnuts
-Stepper motors will be mounted directly to the ballscrews on machined standoffs, the shafts connected by helical couplers
Spindle Construction:
-The spindle assy is a 'cartridge' type that is mounted into the frame in alignment with the axis rails
-The casing will be a simple hollow cylinder bored for the bearing races, with a large square mounting flange welded at each end that will bolt to the frame for mounting
-The spindle will be a 5C collet bore with space for a tubular closer, and D1-4 faceplate for chucks
-Two opposed & preloaded roller or angular contact bearings support the business end, one set of deep groove bearings support the side load of the pulley on the tail end. All are P4/ABEC7 super precision grade, and 45IDx85ODx19mmW
-All bearings will be labyrinth sealed against debris and will have lube points nearby in the casing
Tooling:
-At least initially, there will be an AXA quick change post mounted on the slide that holds the cutters
-I plan to add a quick-change capability to the surface of the slide itself, of a wedge type similar to the Aloris system, that will be operate as part of the automatic tool change system

Frame Construction:
Maintaining Alignment:
Mechanical Assembly:
Electrical Assembly:
Function Testing & Adjustment:

 

[Boswell]

Barnbwt, This looks like it is going to be an interesting build. Looking forward to following your build.
what makes a Slant Bed lathe special? In other words, what can you do on a Slant bed lathe that you can't do on a flat bet? Just curious

 

[barnbwt]

The slant bed is cool for a number of reasons, but only one really has to do with the slant itself; the spindle rotates 'backerds' what we manual guys are used to, so the chips fall down off the upside-down cutter on the back side of the spindle, and fall straight down, below the ways. Chip clearing at the cut is improved, and machine cleanliness in general is improved.

The slant also makes for a slightly more compact machine volume, front to back, because of the incline and because the carriage is on the back side where the motor is already taking up space, and not sticking out the front where your hands can reach the handwheels on an apron

The other advantages, which have nothing to do with the slant layout, involve the use of linear guideways as opposed to V-ways. This is just more a modern high tech development, basically extending the rolling vs. sliding concept behind ballscrews to the surfaces that constrain the linear motion of the axes. The result is incredibly low friction (basically none compared to motor forces) and extremely low backlash/play in non-axis directions (which you only get from V-ways on larger/heavier machines with tuned up gibs). With the unpredictable friction out of the way, it's now far easier to control the machine, especially at really fast travel speeds, which is how the modern fancy tooling centers can run hundreds of ipm cutting moves while dynamically playing with ballscrew forces to compensate for chatter or other cutting forces and achieve a perfect finish.

For the home builder (or commercial builder, I suppose) the advantage of these rails is that it's much easier & more straightforward to build a machine that is square & in alignment. All I need to do is provide two roughly-correct sturdy surfaces to bolt the rails two in my frame, then face those contact areas in a larger mill or grinding setup, and then scrape or shim the rails to final tram upon installation. No more pain in the butt V-ways that require lots of material removal in gigantic grinders followed by induction hardening, that also can't be replaced when they wear out much faster than the hardened ballbearing races.

The disadvantage, is that between the carriage and tool position, you really couldn't run one manually if you wanted to. At best, it'd be a fly-by-wire system with no feedback, where you couldn't see what you are doing very well. So it's really a dedicated/optimized CNC only kind of deal. I already have a manual Chinese lathe for those jobs, and I learned a lot about what I can do with it, what I can't do, what I'd like to do going forward, and what capabilities I'll need to do that. I want to be able to make Glock barrels from blanks & AR15 bolts from raw stock (not those exact parts necessarily, but those kinds of mostly-round widgets)

TCB

 

[Eddyde]

Wow, an awesome, ambitious project! I look forward to following this one, Thanks!

 

[barnbwt]

The current candidates for the stepper motors. I could actually use some advice, or at least confirmation of my assumptions here;
-23HS33-4008D: 400 oz-in NEMA23 stepper, 5.66A bipolar parallel (2.83A serial), and 7.2mH/1.8mH inductance
-KL23H286-20-88-ALY: same thing, but rated at 425 oz-in, 2.8A bipolar parallel (1.4A serial), and 6.8mH/27.2mH inductance

The vastly different inductance figures for each in parallel/series configuration suggests they are wound quite differently inside, despite the identical external profile. From what I understand, the higher-current draw & much lower serial inductance figures of the first one suggest it's meant to run stronger at higher speeds than the second. Now, seeing as I have limited current/power to work with, and the moving carriage is only ~50lb (20lb for slide), and the axis travel is not very large, I am looking at the second KL stepper in series configuration. If I understand correctly, it may lack in power at speed (during rapids at hundreds of rpms), but would be just as powerful when resisting cutting forces & directing the cutter during turning moves, which require far lower speeds (tens of rpms). I think this may be one of those cases where the greater serves for the lesser.

And the 24VDC power supply (ya'll be sure to let me know if I should run independent PS's for the stepper drivers & control boards);
-used Siemens 20A 24VDC power supply.

By my calcs I should never be running more than 13-14A 24VDC, and that's for the worst-case bipolar parallel stepper configuration. This power supply, in addition to being overbuilt for the task as well as German in origin, is also much more efficient in how it uses its source power than the claimed figures I've seen from the only alternative; the Meanwell clones that are visibly 'lesser' in every way. I figure the extra saved .5A is worth an extra 20$ or so for the used unit (even if I end up having to replace some ancient capacitors upon first powerup). I'm told the greater always serves for the greater when it comes to power supplies

Servo drive showed up today. It also looks like I'll need to add a cooling fan to the servo driver for sure, and probably also the stepper drives for good measure. I figure if I'm gonna bother adding a 12VDC step-down line for the fans (there aren't a lot of cheap options for 24VDC) I might as well work up something for the motors themselves while I'm at it. Maybe attach or integrate a heat sink & fan into the aluminum standoff the steppers mount to? Maybe make that giant 3/8"-1/2" end-plate the servo motor bolts to out of aluminum? The stationary items are easier to cool, just put a draw-fan nearby to suck the heat away.

TCB

 

[barnbwt]

I have decided to split the stepper drivers off into their own independent power supply for isolation reasons, since even steppers can dump current into the control devices when slowing down. This also gives me the ability to run higher voltages for the drivers, for better high speed performance & torque control.

I have a proposal for my stepper driver setup for each axis I'd like to get some thoughts on;
For the Z-axis, use a nice Gecko-type driver that has both a high microstep count (my carriage travel is .001/full step) with smoothing, and high speed morphing to single-step so the longer axis has full rapid speeds. Unless it breaks the bank, this drive would feature closed-loop encoder input capability for a later upgrade (not sure how important this is since the motion controller board will also have this feature for sure)

For the X-axis, use a much cheaper Chinese two-phase driver with minimum 16 division microstepping, but no fancy smoothing or high speed morphing functions, or even idle-current cutoff. My thinking is the X-axis is always going to be nearly stationary for any lathe or milling operation, so the fancy motion control features won't be that useful, and the lack of 'sleep mode' will keep the stationary stepper fully energized to resist X-axis cutter forces at full holding torque (since the X won't be sent pulses at all during a simple profile move, for example). I also found out they make cheaper rolled 4mm ballscrews, so I may go with the slower speed for this shorter axis vs. a more precise motor/control setup (though I'm torn since it is also the higher-stressed axis, and a 4mm screw is 1/4 weaker, 9.2kN vs 13kN, but still over 2000lbs capacity)

TCB

 

[barnbwt]

Another question that has occurred to me;
Besides the convenience of using ready-made ballscrews with their journal profiles already completed, is there any advantage to placing the preloaded/fixed end of the ballscrew at the opposite end from the motor-coupling? It seems to me, that by placing this end at the lower end of the table, the loaded length of the screw is greatly shortened, resulting in a more rigid system (albeit only slightly compared to other factors). The more common approach with the motor right at the fixed end seems tailored for CNC mill/lathe conversions, where the motor uses the space previously occupied by the handwheels, where there is also the structure to solidly mount the fixed-end screw support. The slant bed is kind of backwards, in this regard.

TCB

 

[cmantunes]

If your budget permits, consider replacing the steppers with integrated servo motors, such as ClearPath. Not only will you get 6400 steps/rev if desired (w/ enhanced version) but the quietness, the smoothness of movement and lack of vibration will considerably enhance the build.

 

[barnbwt]

Of course a full servo-controlled closed-loop system on every axis would be nice, but that is far out of the budget for this project (hoping to keep it under $1500 for starters). For now, I mainly want to make sure I am not boxing myself out of such an upgrade at some future date. If I have problems with threading or engraving, I'd consider it. A friend is running steppers & getting good results with engraving on his mill, but they are geared down 1:3 or so; it is possible the direct drive's reduced resolution & holding torque could be a limiting factor, in which case a precise/sensitive servo is a good solution. I do think that simply going with a tighter ballscrew would get me most of what I'm looking for, for the time being, for not much money, though.

I spent today re-tweeking the frame a bit more; now the slide is 1.5" wider (basically 8" square), but I was able to tuck the ballscrew supports between the linear guide carriages, reducing the length of the carriage about 1.5" up where the stepper mounts, and bringing the lower edge of the carriage in by an inch. Also added a horizontal stiffener bar to the upper end of the carriage (a section of 1.5" box tube) since the trench I had to mill for the X ball nut clearance left it U-shaped in a structural sense. End result, the ballscrews are a bit shorter and now hidden, the linear rails didn't change much at all.



This means the depth of the machine can be reduced about 1.5", and now that I've also updated the models of the servo and stepper motors to be more accurate, I have a much better route to stiffening the headstock area to support the spindle. Two X-braces three inches away from each other & filled by two thick shear plates should be crazy stout.

I'm pretty happy with the structural layout at this point, with only one possible, but unlikely modification; I think there is just enough room to put flip the stepper motors around and place them beneath the ball screws. Instead of a fat block mounted to the ballscrew supports that mounts the motor, there would be a thick plate that hangs below for the motor face to bolt onto. The idea of fully hidden motors is very appealing, and this would allow for some degree of gearing to a different ratio (namely on the X-axis). However, I also believe it wouldn't be quite as rigid, both due to the belts and the cantilevered motor mount. I'll have to think on it some more, and decide if it's enough to matter to the ballscrew.

TCB

 

[barnbwt]

Next round of stuff bought, pretty much everything but the bulk steel & electronics;
2X 400mm linear rails (X)
2X 600mm linear rails (Z)
300mm 1604 ballscrew & nut (unfinished) (X)
500mm 1605 ballscrew & nut (Z)
24VDC & 5VDC power supply
2X BK/BF12 ballscrew mounts
2X stainless steel 1/4"-10mm couplers

The cost this time was 408$. Provided I don't go crazy with the motion controller, I'm on pace to squeak by under 1500$ in materials. The first post will be updated with the particulars of these latest items.

TCB