Slant-Bed CNC Lathe Build

You have a great start here but I think there is some room for improvement. Note that I am not writing from practical build experience but from having followed dozens of build threads and absorbing comments from people who know more about this than I ever will. So take what I say with a grain of thought but please understand that I hope my comments will help shorten your path to success.

I think you need to bring up the stepper voltage considerably. Performance at 24 volts will be poor. Anyone who, like me, bought a Taig mill with an Allegro based controller running high inductance steppers should be able to attest to that. According to Gecko Drives stepper voltage should be 32 * sqrt(inductance) for best performance. Thus the first motor you listed wants about 43 volts and the second one about 83 volts (in parallel, series is much higher). I don't know what drives you are planning to use but the first motor draws a fair chunk of current so something like the Gecko G540 would be out of the question. On microstepping: its main purpose is to smooth out the motor at low speeds. It isn't the greatest way to increase resolution since the distance between microsteps isn't always even. There is a point of diminishing returns with microstepping. That is around 10 microsteps according to Gecko Drives. See: https://www.geckodrive.com/gecko/images/cms_files/Step Motor Basics Guide.pdf

As for the relationship between the motor, support block and screw, one of the roles of the support block is to isolate the motor from the violence of the screw. Support blocks have robust preloaded angular contact bearings while motors have comparatively wimpy deep groove bearings. With the mounting arrangement you are proposing, the motor will be subjected to all kinds of stress as the screw expands as it gets warm or if it starts to whip. In any case for short screws like you will here, you don't need to support both ends providing the screws are a sufficient diameter. The 16mm screws you are specifying are more than adequate. Look for the Critical Speed calculator on this page: http://www.nookindustries.com/EngineeringTool/Index to figure out what will work. It is interesting to note in their fixity examples, none has the motor opposite a bearing block.

I don't know where Mach 4 is at but Mach 3 has well documented issues with threading and dealing with index pulses (it can only deal with 1 per rev so it doesn't track very well). You might want to consider Linux CNC. It is more work to get going but is more robust than Mach 3.

If you have access to FEA software, run an analysis of your design. I think you will find that very little of the material above the spindle/bed plane contributes to rigidity. Take that part of your weight budget and put it where it will do more good and build up the enclosure. For instance, extra mass around the spindle and the spindle/bed interface is rarely wasted.

Have you see this build? It is similar to yours: http://www.cnczone.com/forums/vertical-mill-lathe-project-log/281738-cnc-cad.html

bob
 
Absolutely right on the stepper voltage; I'm learning a lot and thought the drivers had a step up switch in them vs. just being a high-tech Morse code switch that passes current/voltage straight through when pulsing. I think my servo drive is similar (really hard to tell what all it does, what with the multiphase conversion, voltage/current regulation, and pwm; I just know it's recommended for this motor). I'm looking at some 60-80vdc supplies for servos, and possibly the 220 transformer I was hoping to avoid for the servo motor.

This really begs the question; why aren't there voltage step-up stepper drives? Seems like it'd be nice to have the same adaptability you have for your output (current & voltage selection for different motors) on the input side, so you only need one big nice/regulated 24vdc supply unit, for instance.

I know what you mean about the screw mounts always being fixed at the motor end. I suppose thermal growth is a very valid point, especially in big machines where they could get toasty. My X is only 10" long, though, plus there's a spring on the end of it at the motor. But by this same logic, there's no advantage to mounting the thrust bearing at one end or the other. I suspect the wider footprint of the fixed mount may be a bigger factor for rigidity in supporting a cantilevered motor than anything.

The upper frame is intentionally overbuilt, but is 3/16" vs 1/4" thick tube. I won't go thinner since I'm not a great MIG artist. I plan to mount a tool carousel in there, and at this point I may be putting some/all of the electronics in a cabinet/hutch up there, possibly on a thick aluminum slab for heat conduction (servo drive requires a fan *and* a cold plate). The lower frame will likely get heavier though; I am contemplating another tube along Z across the rail supports to better secure that critical plane against warp caused by the rest of the structure. I only want to mill the rail seats planar once ;)

TCB
 
I'm sure my beer budget will dictate some changes, but at this point I'm thinking;
-Gecko G214 high-res stepper drive for X and Z axes (plenty of growth room for stronger steppers & closed loop in the future)
-80VDC unregulated linear power supply, 4A capacity qty(2)
-23HS33-4008D: 400 oz-in NEMA23 stepper (lower current usage for peak holding torque; speed is not so important)

Is the higher voltage itself an advantage? I guess perhaps better second-order motor response due to electrons being pushed harder? Neither velocity nor acceleration are as important for turning jobs, at least compared to static holding torque and smoothness (which is why servos would be ideal, if only they were affordable)
 
"Have you see this build? It is similar to yours: http://www.cnczone.com/forums/vertical-mill-lathe-project-log/281738-cnc-cad.html"
That build & dumpsterCNC were my inspiration (that poor guy lost everything to the tornado near Emory, Texas just a few weeks ago as I was first diving into this, otherwise I'd probably be pestering him with these questions, lol). My hope is to scale that design down slightly, but not having that big box structure beneath the bed supports & above the bench top, and to make more efficient use of a small envelope with a stationary tool carousel (the tool holder will spin to index, and the carriage/slide will 'dock' with it to load & unload tools). The servo spindle is more of an added capability than an improvement over the original design, meant to make this thing way, way, waaaay more useful than my manual lathe is.

TCB
 
A few more odds & ends, in anticipation of the servo motor's arrival & testing;
-High voltage connections, 8-pin & qty(2) 2-pin Molex connector terminal blocks; $5.78, $10.16
-Low voltage connections, 14-pin mini-Molex connector terminal block,
-Encoder feedback connector, 15-pin D-sub; $18.87
-Encoder feedback connector, Fancy-pants mil spec 15-pin round bayonet connector; $46
-Computer interface, RS485 cable; $6
-Servo power cable; $52.98

The power cable is Keb brand, but the 8-pin round connector & insert arrangement appears identical to the Baldor. I will probably end up buying a short length of shielded cable with 8 twisted pairs for the encoder line; while their prices are insane, Baldor was nice enough to clearly explain what is needed as far as the cable requirements in their installation manual. Once I have the RS485 cable in hand I should be able to put 24VDC into the servo drive to turn it on, and talk to a computer via the (apparently crummy & poorly supported) Baldor Mint Workbench software, which is where I can play with step/dir gearing settings and so on.

One lame thing about this servo driver is the input pulse speed of only 1 MHz; with the 2500 line encoder on the motor, that's like 400rpm tops at 1:1. Cost of doing business in 1998. Through this software, I believe I can increase the encoder pulse/drive pulse ratio by a factor of ten (250 steps/turn) or greater and reach the top speeds desired. I need to learn some more to find out if this is a setting that can be modified dynamically via M-code during a program; if that is the case, step/dir would definitely let me have my slow & precise cake and eat the high speed stuff, too.

Should I wish to run the motor in velocity (or even torque) mode via the analog signal, I may have some problems. This drive responds to -10VDC to +10VDC signals, but all the motion controllers out there seem to be 0-+10VDC outputs, with relays that reverse the output poles for the opposite rotation. Chapter 5.2 (no page numbers, Baldor?) of the manual linked below shows that the inputs can be set up as differential or single-ended. I think that even if I reversed the polarity of the motion controller board output, I would have problems with the single-ended configuration, since the +10V lead would be hooked straight to a grounded terminal. What I don't know is if putting +10V on that same ungrounded terminal for the differential configuration would actually work, or if its circuitry is still looking for a negative-to-ground voltage, albeit an unreferenced one. There's an "internal reference" resistor on the circuit diagram in this area that could still be going to ground for all I can tell.

http://www.baldor.com/Shared/manuals/1919-803.pdf

I know enough about circuits to know some simple MOSFETery can pass-thru a 0-10VDC signal, or turn it into a -10-0VDC equivalent depending on a binary direction signal voltage & power supply, but I didn't see such a ready-made device amongst the other spindle-control doohickeys at the usual vendors. If anyone has any suggestions I'm all ears, since I *suspect* interacting with driver settings via Mint Workbench as little as possible vs. external solutions is a good approach.

TCB
 
When I'm not torturing my mind with electrical/programming stuff, I'm working out the spindle for the next parts order. Originally I'd planned to use a pair of opposed angular contact bearings at the business end, but I'm leaning more toward a single-race dual-row angular contact setup for a couple reasons;
-takes up less space & is simpler
-is more available in a shielded configuration
-50mm seems to be a very common size for HVAC uses (so much cheaper)
-plenty strong for any cutting loads the ballscrews can resist without damage themselves
-No PITA preload adjustment necessary once assembled

That last one is where I'm not so sure. Not needing to squeeze the two halves of the support is nice, but I also worry that the loss of this ability is a bad idea for a lathe spindle. It seems to work fine for A/C compressors, but who knows whether that has anything to do with the loads I'm contemplating? Would the standard light pre-load bearing be rigid enough to prevent obnoxious acoustic patterns from fouling all parts turned at speed? I suppose another solution would be to use two of these two-row bearings, and lightly tension them against each other similar to how we normally do the single-row jobs (possibly one at each end of the spindle)

https://www.vxb.com/3210-2RS-Bearing-Angular-Contact-Sealed-50x90x30-2-p/kit16655.htm

TCB
 
This drive responds to -10VDC to +10VDC signals, but all the motion controllers out there seem to be 0-+10VDC outputs, with relays that reverse the output poles for the opposite rotation. Chapter 5.2 (no page numbers, Baldor?) of the manual linked below shows that the inputs can be set up as differential or single-ended. I think that even if I reversed the polarity of the motion controller board output, I would have problems with the single-ended configuration, since the +10V lead would be hooked straight to a grounded terminal. What I don't know is if putting +10V on that same ungrounded terminal for the differential configuration would actually work, or if its circuitry is still looking for a negative-to-ground voltage, albeit an unreferenced one. There's an "internal reference" resistor on the circuit diagram in this area that could still be going to ground for all I can tell.

Depends on the motion controller. When you have a motion controller that will output a +/- 10 V signal, it is normally setup as single ended output, with the command signal going to the drive command input and then the controller ground going to the drive signal ground. I have never done it, but setting up as a differential input might be the way to go in your case.

I would not think a relay switching the command signal polarity would be very fast system, but maybe it works. I guess for the spindle, it wouldn't really matter.
 
Well, it is for velocity/torque control & not position; I can't imagine needing to rapidly vary the speed instruction at speeds beyond the step/dir regime. I also think the negative values are for reversing the spindle (not simply slowing it) which would be an even odder command for an already-turning machine.
 
Thanks for the link, jbolt, lotta material to go over there
 
Back
Top