Slant-Bed CNC Lathe Build

You mention LinuxCNC..... I have been rather impressed with the operation of LinuxCNC on the Tormach mill over the past couple of years. I would certainly think it would be worth a look. It takes some messing around to get a non-supported machine up and running but their forum has some great followers that would get you pointed in the the right direction.

Robert
 
My two cents. The work you are considering is duck soup for a fourth axis on a mill. Any of your four routes would work. For hardware just add an indexer with a drive belt to a servo or stepper motor. Pic of mine attached, its plenty robust to run a 1/2 endmill in steel.

If you drop lthis from the lathe project, now it becomes simple also, any of your choices would work.

Lots of people have done both these projects, you won't be breaking any new ground.

index head.jpg

milling.jpg
 
I already have a manual lathe, but no mill in which to cut keyways/splines, or even do engraving. It'd be nice to make candle holders & chamber reamer profiles, but I can do that already more or less on the manual.

Call me greedy, but given the difficulty, uncertainty, and expense of a homebuild, I'd like a significant addition to my capabilities. Even if the servo isn't strong enough now due to power restrictions, the capacity to swap in a bigger unit & have it work within the system at a future date is desirable.

Checking into it, it sounds like any worm arrangement I am likely to make will have a lot of backlash, unless there's some trick to that. I'm not seeing how it would be superior to a servo spindle. Swapping in a bigger/faster drive & gearing it down seems like a better move (even though the gearing complicates some things)

TCB
 
Alrighty, next series of major purchases has been made;

Galil DMC-2183 8-axis motion controller
Galil SDM-20640 4-axis stepper motor driver daughter board

So while the wallet charges back up, I am now looking at the next link in the chain; stepper motors!

My design constraints are at least well-defined since I have a driver circuit to work within, which is kind of backwards from how you're supposed to approach this. The Galil driver is certain to be 'good enough' for my rather modest means, so it's down to milking the most I can out of it for the money --same as everything else so far.

1) Driver current is 3A maximum:
This effectively puts a cap on the amount of torque I can get from my motors. Every motor winding design has a different torque:current ratio, but peak torque is always attained at peak current (and at slow speeds). According to materials on Gecko's site about sizing servo power supplies, the bipolar-parallel winding configuration my steppers will use can only draw at most 2/3 of the rated current in practice (I think the rated value is for a *stalled* motor condition, which a good machine design won't operate them in). Unless I'm mistaken --please tell me if I'm mistaken-- that means my 3A driver circuits can run 4A steppers at most, and I'd need a +8A power supply to feed them.

2) Driver voltage is 50V maximum:
Voltage limits put a cap on how fast you can go, and also what types of motor windings can be used. Current is held to a fixed level during the peak/constant torque portion of the motor's performance so the windings don't fry, and since torque & speed correspond to power, a higher operating voltage capability means higher speed before torque falls off (i.e. current drops due to impedance effects at high speed). To get the most from my drivers, I need power supply that can operate at around 48V (50V isn't as common, is all)

48V isn't super energetic, but it is high enough that it can damage certain stepper motors (it can drive current too hard into a coil that isn't build with enough inductance to resist it) or at least cause them to run poorly, and hot (more likely with only 3A to play with). Going by the voltage-rating formula out there in many places (Vmax = 32*sqrt(inductance)), it appears that if I wish to use the full 50V capacity of my driver for power output, I must use motors with impedance near 2.44mH, or slightly higher. If much lower, the motors will cook at 50V, if impedance is much higher the stepper driver's voltage won't drive the motor aggressively enough to operate at higher speeds (torque falls off badly)

3) Motor wiring is bipolar-parallel:
There's three ways the two sets of coils inside the motor can be hooked up to power and driven; bipolar series, unipolar, and bipolar parallel (in rough order of increasing torque rating). From what I gather, unipolar is an archaic configuration that older drivers were designed around, before more modern switching circuits became available; they can still be wired as series, or as parallel (but only using half their coil turns), but we generally can't drive stuff in unipolar wiring anymore. That leaves series and parallel, and is basically a function of how much speed you need; series can be driven at low current, but has very high inductance so lots of voltage is needed to spin the motor quickly (more than any driver can supply). The trade off is not equal, so unless you don't have much current *and* don't need speed, bi-polar is more efficient all around. For simplicity, this is the only motor configuration I'm looking to use.

I've already found a reasonably-priced 48VDC power supply that can deliver up to 9A. So, I'm looking for bipolar-parallel motors in the 3-4A range with impedance around 2.5mH. *Not* looking at torque figures except for comparison (like I said, I'm having to do this a little backwards ;) ). If any of you know of some vendors that have a good selection, I'd like to hear about them; checking out the OMC/StepperOnline catalog, there are a plethora of NEMA23-size stepper motors in all manner of power levels, but I'm not seeing that 'perfect' combination of my design requirements that will let me get the most out of my machine. If I have to compromise on anything, I think it should be high impedance since that mostly affects high speed operation, and as I've mentioned, lathes don't gain as much from rapid traverse moves as do mills and routers. My target rapid speed of 420rpm is less than half the 1000rpm that the inductance figures are measured at, but I don't know how much lower they would be.

Here's my list of hopefuls;
(I'm also listing inertia to get an idea of the 'agility' of these motors)
24HS34-4004D; 35$, 4A, 3.5mH, 325.7oz-in, 840g/cm^2 **NEMA24** my understanding is these still roughly fit the NEMA23 form-factor, which is fine)
23HS45-3004D; 75$, 3A, 9.0mH, 354.0oz-in, 810g/cm^2 (inductance is rather high, by my 'rapid speed' rpm is only 420rpm, so maybe OK...price is silly)
23HS45-4208D; 75$, 4.2A, 2.3mH, 276.1oz-in, 810g/cm^2 (would have to run at slightly-lower torque at all speeds, but curve is flatter...price is silly)
23HS33-4008D; 25$, 4A, 1.8mH, 283.2oz-in, 530g/cm^2 (I'm limited to around 42V and less power, but torque/inertia ratio is much higher)
*this last one is the front-runner, unless my assumptions are bad or there's a better option

There's lots of others, but this 4A/3mH region seems to be a bit sparse. Tons of options one amp higher, or if I had 80V to drive a higher-inductance motor with, but these are the only options I have unless I want to step down to <250oz-in torque ouput (probably still okay, but why not get more if it's available?)

TCB
 
I found a fifth motor, I think it may be 'the one'
23H2100-35-4B; 40$, 3.5A, 2.8mH, 381oz-in. Seems just about perfect; near full-rated current, very close impedance match to 50V, and not insanely expensive. These also appear to be a favorite of the G540 Gecko users (which makes sense because that also has a 3A limit, but at 80V)

I've also started putting my electrical block diagram together, in advance of firing up some of these components for testing (see attached). One slight hiccup that I decided to take on is that the Galil controller & daughter board do not hook up as they are currently configured. The secondary board is meant to snap onto the top of the main one via a 96pin DIN connector as well as some secondary pin blocks, but the DIN on my 2183 unit is the right-angle variety; no va. So now I'm looking into connecting the two via ribbon cables (which seems noise prone) or going all out & removing/replacing the right angle connector with a straight variety (which costs like 3$ for five pieces). Seems like both routes will cost me ~60$ in either wires or soldering tools

Block Diagram.png

While that copper-based mess is percolating, I'm also finalizing the spindle cartridge design. I had been planning to use 50mm ID bearings, but I've decided to go with smaller 45's since they are apparently half the price for a marginal reduction in spindle thickness (still about .19" thick walls). This has enabled me to acquire proper P4/ABEC7 super precision spindle bearings for both the nose and tail sets, unfortunately neither are sealed so I am working out how best to protect them from swarf & coolant. My plan is to have a labyrinth seal at each end of the cartridge (more complex at the nose-end), and two internal grease baffles/slingers near the bearings to help keep the grease stay put in operation. Also one Zerk fitting next to each bearing set for the occasional squirt.

Spindle Concept2.png

The cam-lock spindle complicates the seal arrangement since the mounting cam holes are so big, but at least with the narrower bearings there is now room for a small series of labyrinth ribs. In practice those holes will be filled, but they will still allow debris to ingress to some extent during chuck changes/etc. To be honest, I'm having second thoughts about the D1-4 capability altogether; it makes the spindle large & expensive, eats some precious bed length, and most worrisome, I'm having doubts about its utility. Being inside the machine, manipulating the cam-locks chuck jaws will be highly annoying, and I'm not sure that the ability to mount 3 or 4 jaw chucks is all that useful in a CNC slant bed like this, where the vast majority of jobs will be single-setup operations held in a collet/closer. You can even mount ER40 collets in a 5C with an adapter, so my need for that ability is satisfied. Still, the ability to use all my current D1-4 accessories (3-jaw, 4-jaw, ER40 collet chuck, face plate [which may not even fit through the door, lol]) is somehow tempting.


TCB
 
Last edited:
Another paycheck, another step toward completion...

Today I ordered the spindle bearings;
-Barden 209HCDUL matched P4/ABEC7 angular contact bearings
-Fafnir MM209K CR P4/ABEC7 deep groove ball bearings

Both are wildy in excess of what I need as far as load capacity & especially rated speed, but only cost about 50$ than the less-precise P5/ABEC5 equivalent on the new-old-stock market (eBay). From what I can tell, class 5 is about as low as you ever want to go for a spindle, since the rated speed and service life fall drastically; unless you are running at car-tire speeds of several hundred rpms, they simply generate too much heat & introduce too much variation at the cutter.

I had what I thought was an interesting idea regarding Karl's comment about weak holding torque for milling operations. Obviously a brake of some kind is the logical solution, but I was concerned its creation would add significant complexity to this already ambitious project. Then I remembered those little disc-brakes that have taken over the bicycle industry the last several years; they come in sizes from 140mm to 203mm and it's only a couple dozen dollars for a rotor/caliper setup that is ready to bolt-on as you wish. More expensive hydraulic versions are out there, but the simple pivoting-clamp type pulled by a cable would be sufficient for my needs, and would be very simple to control with a solenoid (that, or mount a bicycle hand lever on the machine, lol). So I may end up adding a brake disc rotor next to the drive pulley for this purpose. I would think this direct-brake approach would be more rigid than a brake mounted at the motor shaft, too.

And thinking about it some more, since these pads/rotors are designed to actually work while in motion (unlike most servo brakes) and the replacement wear parts are so cheap, this could potentially work as a generic spindle brake so the inertia doesn't have to be dissipated through a braking resistor (I'm guessing this requires some more development on the programming side)

The next major acquisition is likely the 48VDC power supply and stepper motors; after that it's all steel (finally). In the mean time I will be designing & building the wire harnesses I need to begin testing the electrical components.

TCB
 
You won't need this for a while, but I'll offer it now. The attached program would work real slick to jog your machine. I mention it now cause it uses inputs on the Galil card. You would need seven to do three axis. I'd suggest keeping two or three more inputs and a spare axis to do your MPG handwheel entirely inside Galil. FWIW, a top quality JOG and handwheel really improves setup quality and time.

I'm a fairly accomplished galil programmer, if you ever need help.
 

Attachments

  • JOGnHOME.txt
    2.8 KB · Views: 6
Talk about putting the cart before the horse (or is it pearls before swine, lol)! Thank you regardless; at least as far as the logic side for the jog script, it's comforting to more or less understand what I'm looking at. I need to study up on Galil machine code a lot more before the homing routine will mean anything to me. At least this language doesn't require a labyrinth of GOTO commands to do anything (a crusty old prof made us learn FORTRAN for his class --in 2010!)

A buddy has his CNC mill outfitted with an XBOX controller for simple positioning/indicating during setup, and I agree it makes things a lot easier than driving the table strictly by numbers in Mach.

TCB
 
I considered Fortran an advanced language in 1972. the prof made us learn machine code and assembly language first. I was an early programmer of automation robots, where GOTO meant move the robot to a certain position or point in 3D space.

The homing function does not work with steppers, you need a "Z" index pulse. Just one of the reasons i never use steppers any more. Main reason is no feedback, but Mach won't accept feedback anyway. And that's why I don't use Mach.
 
By index pulse do you mean an actual encoder feedback, or just a once-per-rev deal? I plan to use dual-shaft steppers so that encoders may eventually be added, or maybe even NEMA23 servos if I ever get enough electric capacity and the funds to buy them & the necessary gearboxes. Z-axis could actually run a NEMA34, but it'd be a bit much to hang on of those off the back of the poor saddle, lol (maybe with a right-angle worm drive I could pull that off)
 
Back
Top