Slant-Bed CNC Lathe Build

[JimDawson]

All of the newer boards can have isolation option for the drives, and in reading through the documentation for the AMP-20440 documentation it looks like the motor power is isolated from the board power by default. The board requires separate power from what I can see.

 

[JimDawson]

I think your better options might be to: 1) save your nickels & dimes until you can afford at least 400W AC servos & drives (about 300/set), or 2) to get running for now, buy some cheap NEMA 34 steppers & drives (about 90/set). You would need 60V power supplies in either case.

I read back through the thread and maybe I missed what you are going to use for a spindle motor.

 

[barnbwt]

The spindle is a just over 1kW @ 4000RPM Baldor, driven off a 10A Parker amplifier. Is 400W per axis really needed for a 1" swing lathe slide? A friend has 570oz-in NEMA23 steppers moving his 100lb mill table around against a 3hp spindle; I figure this shouldn't be anywhere near as demanding as that (40lb table, 1.3HP spindle). My goal is also keeping the power consumption well below 20A @ 110VAC, which I'm aware is rather limiting. The machine needs to be 'oven' sized, not refrigerator sized, too.

I would like to stick with NEMA23 form factor for steppers if possible, and 381oz-in appears to be about as good as I can get through my amp board for that size. It appears NEMA34 gets me to 465oz-in, but with faster fall-off. Part of the issue, which has been an issue since the start, is knowing how much holding force my tools actually need for this task, since that is what actually drives the motor sizing. I believe I ended up with 400lbs minimum motive force being my goal through some questionable calculations last year. Feel free to weigh in on whether that seems realistic or not.

May I request a quick sanity check from someone here? Is 3A at 48V yielding 381oz-in correct, or are the spec sheets giving parallel-wired torque figures and series-wired current draw? There's a lot of misleading or outright incorrect information in many listings

Second sanity check would be servos delivering just over double the torque of a same voltage/amperage stepper at their operating speed (a little over half way to max rated in both cases). As I see it, servos have 1/3 the torque, but run over 6 times faster.
My references are;
http://www.animatics.com/products/smartmotor/sm23305d (being run at a 1/3 lower-than-rated amperage of 3.3A, 40Oz-in @ 4500rpm)
Stepper Motor (being run at a 1/7 lower-than-rated amperage of 3A, 120Oz-in @ 700rpm)

Those servos appear to gin up about as much torque as my buddy's 5A 570oz-in steppers, even at a lower-than-rated amperage (assuming linear torque response). But with much less juice, less heat, and presumably much smoother operation. Once again interested in making servos work (darnit)

 

[JimDawson]

I'm running my mill 750W servos on a 120V, 20A circuit and have never had a problem. Actually that circuit is running the lighting in that area, the mill controls (two 750W servos, one 1200 oz/in stepper, NEMA 23 stepper for the 4th axis, and the computer), my various grinders, the lathe computer, and anything else I may plug into that system. No problems so far. Most of the time the various motors are using a small fraction of their rated power, so the power system does not have to be sized as if everything was running at 100% power at all times.

Given the small mass that you need to move, NEMA 23 steppers will probably work fine. Worst case using the NEMA 23 motors is that you lose some rapid speed and you have to keep the acceleration down to a reasonable level, they should have plenty of power at normal cutting speeds. Unless you are trying to maximize production, who cares if it takes an extra second or so to move the next tool into position. If power does become a problem, then gearing down 1.5:1 might be an option. The problem is that when using steppers that may be a bit underpowered it is possible to lose steps without encoder feedback. The good news is that encoder feedback can be added at any time, and is not required to get you up and running. It needs to be noted that without encoder feedback, the DRO display will only be able to read where the tool should be rather than where it actually is. So if steps are lost, the DRO will not reflect that.

My theory on steppers is to run them in the lower RPM range to keep them in the higher torque band. This means high lead leadscrews, 0.250 or even 0.500 lead. The leads on my router screws were 25 and 40 mm for the X & Y axes using 1200 oz/in NEMA 34 steppers, moving loads in the 700 lb range and capable of 600 IPM rapids.

I think you will be OK to start, and you can upgrade the system later if needed.

 

[barnbwt]

The plan was to add some el cheapo magnetic scales fairly early once the steppers were running open-loop, to at least mitigate issues with any step skipping. I already have ballscrews (5mm Z pitch and 4mm X) but peak rapid speed isn't really a critical factor here. The X moves only 6" and the Z 12". It would be 'nice' for the machine to run to limits in only five seconds, but so long as I can cut fast thread pitches at proper surface speeds without losing steps I'm a happy man.

Those Animatics servos are a pretty cool architecture, kinda wish I'd known about them before diving in since it's an interesting alternative to a centralized motion controller. From what I gather, a computer generating motion commands instructs the servos, which each contain their own controller, driver, motor, and encoder, so the first control loop never leaves the motor housing. At that point, the only additional feedback needed are limit/home switches to the relevant motors, and scale outputs to go back to the controlling machine. All done via RS232 which is apparently scalable to over 100 coordinated axes, lol. The onboard brains are apparently smart enough to run their own stored programs, even while being sent external data (macros called up by the controller program data, or pre-planned moves in response to I/O conditions). I believe the RS232 is also what keeps them all synced up with each other in real time? Very decentralized approach to automation, doesn't even need a central 'brain' really.

I'm not sure they're useful for my particular needs, since I've already got the driver & controller parts handled by the Galil & it'd just be redundant and potentially incompatible. If I could find a standalone servo (motor/encoder only) with the same motor specs I think that might work really nicely.

 

[JimDawson]

I think you still have to have a motion controller connected to the Animatics motors. Normally the communication protocol would be DeviceNET or CANOpen, and there are also Ethernet motion controllers. Something like this http://www.galilmc.com/motion-controllers/multi-axis/dmc-52xx0

Rockwell Automation, Rexroth, and many others make controllers that use the CAN protocols
https://en.wikipedia.org/wiki/CAN_bus
https://en.wikipedia.org/wiki/DeviceNet

It might be possible to write your own motion control software using the CAN protocols and maybe even get it to run on a Windows computer, but it really requires a real time dedicated operating system to do this properly, and would require a lot of development time to really produce anything useful. Not something I would want to tackle I'll stick to the high level languages. The thought of coordinating 8 axes makes my head hurt. That's why they invented motion controllers, just tell it what you want it to do and let the controller figure out how to do it.

One system that perhaps you haven't looked at is the ClearPath motors. They are a self contained servo and are relatively inexpensive. But still require a motion controller. https://www.teknic.com/products/cle...ervo-motors/clearpath-sd-stepper-replacement/

 

[Karl_T]

My two cents here. You got a Ferrari control and you are planning to use it to drive a Hyundai. You should either go steppers with Mach, or get quality servos.

A bit of history. When my Bandit control died in '95, I went to the only PC control available at that time (AHHA). had to go from servos to steppers. Started ruining parts due to lost steps. Went to huge steppers at high cost, still had trouble. Then I re-upgraded back to the servos just like what I took off - problems solved.

IMHO, the only reason for steppers is to build a low cost machine. Steppers with mach cost maybe 10% of my servos with camsoft control.


<EDIT>

Just one option here. Like I said I like DC servos => lots of bang/buck

30 seconds of shopping found this servo: (you can do better - get all servos the same)
https://www.ebay.com/itm/Dynetic-Sy...m=112955618697&_trksid=p2047675.c100005.m1851

An AMC drive will cost less than $50.

You'll need an encoder - US digital has them for about $75 each. Or go real top end with linear strips. many servo offers include the encoder.

You need a DC power supply - one for the whole system.

If you decide to go this route, I can hold your hand. I don't think its terribly difficult.

 

[barnbwt]

I think you still have to have a motion controller connected to the Animatics motors. Normally the communication protocol would be DeviceNET or CANOpen, and there are also Ethernet motion controllers. Something like this http://www.galilmc.com/motion-controllers/multi-axis/dmc-52xx0

Rockwell Automation, Rexroth, and many others make controllers that use the CAN protocols
https://en.wikipedia.org/wiki/CAN_bus
https://en.wikipedia.org/wiki/DeviceNet

It might be possible to write your own motion control software using the CAN protocols and maybe even get it to run on a Windows computer, but it really requires a real time dedicated operating system to do this properly, and would require a lot of development time to really produce anything useful. Not something I would want to tackle I'll stick to the high level languages. The thought of coordinating 8 axes makes my head hurt. That's why they invented motion controllers, just tell it what you want it to do and let the controller figure out how to do it.

One system that perhaps you haven't looked at is the ClearPath motors. They are a self contained servo and are relatively inexpensive. But still require a motion controller. https://www.teknic.com/products/cle...ervo-motors/clearpath-sd-stepper-replacement/

I'll make one last attempt & ask the Aminatics folks if the brains can be bypassed, and run like a motor/encoder pair. There's a pair on Ebay that look really good for this application. Those Clearpaths look nice as well. In either case I'd have to bypass the servo (or stepper) amp boards and run them directly off my controller & an external amplifier since they are brushless.

My two cents here. You got a Ferrari control and you are planning to use it to drive a Hyundai. You should either go steppers with Mach, or get quality servos.

A bit of history. When my Bandit control died in '95, I went to the only PC control available at that time (AHHA). had to go from servos to steppers. Started ruining parts due to lost steps. Went to huge steppers at high cost, still had trouble. Then I re-upgraded back to the servos just like what I took off - problems solved.

IMHO, the only reason for steppers is to build a low cost machine. Steppers with mach cost maybe 10% of my servos with camsoft control.


<EDIT>

Just one option here. Like I said I like DC servos => lots of bang/buck

30 seconds of shopping found this servo: (you can do better - get all servos the same)
https://www.ebay.com/itm/Dynetic-Systems-D-C-Servo-Motor-230091/112955618697?_trkparms=aid=222007&algo=SIM.MBE&ao=2&asc=53210&meid=1293ba83eed64c169a584bc9970a9deb&pid=100005&rk=1&rkt=3&sd=141869370410&itm=112955618697&_trksid=p2047675.c100005.m1851

An AMC drive will cost less than $50.

You'll need an encoder - US digital has them for about $75 each. Or go real top end with linear strips. many servo offers include the encoder.

You need a DC power supply - one for the whole system.

If you decide to go this route, I can hold your hand. I don't think its terribly difficult.

I was looking at those motors, actually. My board controller only does brush motors, which are much rarer for servos as I can tell. So I'd just need motors & encoders (and a 60V power supply to get the most power from them).

 

[JimDawson]

Those Clearpaths look nice as well. In either case I'd have to bypass the servo (or stepper) amp boards and run them directly off my controller & an external amplifier since they are brushless.

The ClearPath motors have the amp built in to the motor. You only need to supply power and a step & direction signal. The downside is that there is no provision for encoder feedback directly from the motor, for feedback an external encoder would need to be used.

The Animatics also have the amps built in, so you would need to supply power and the control seems to be some derivative of CAN or EtherCat

 

[barnbwt]

Man, Clearpath has some cool videos of their stuff. I always thought servos were annoyingly whiny; those things are dead silent, especially with their acceleration damping algorithm. I think it'd be like $700 for a two-axis system using those things, but that actually seems like a good deal for what you get. I still need to look into their service life/reliability to convince myself. Those things along with a linear-encoder on the axis giving feeedback to the controller should be a pretty well-regulated system, right?

In the coming days I'm also going to work on migrating my NX-based CAD models of the machine over to Fusion. We'll see just how limited that program is when assemblies start getting a little complex. Hopefully I'll have some ideas for improvements along the way as well.