Servo Motor Spindle

[BobWarfield]

Had a crazy thought after coming across an article on converting a DC motor into a servo ( http://truetex.com/servomod.htm ). I have a Lathemaster 9x30 that I've converted to variable speed with a DC treadmill motor (now you see where I'm going). Seems like it would be straightforward to do this fellow's conversion of the treadmill motor into a servo and wind up with the ability to precisely index the spindle as it is now a servo motor.

Not sure where I'd go with such a capability other than that it would make it possible to do under CNC control a lot of things done manually once you can index and lock the spindle. One application I thought of would be to make a motorized drill chuck attachment for the tool post to enable CNC drilling of holes that are indexed precisely by spindle position. I would think a capability like this would also be nice for those that are thinking of CNC'ing a 3-in-1 machine. There you'd wind up with a 4-axis mill that would be pretty slick.

I'm not sure if Mach 3 is emotionally prepared to deal with a lathe spindle being a servo or not. I'll check with Art on his board.

Any thoughts? Probably someone has travelled this path before.

Best,

BW

[CJL5585]

You might have a problem using a regular DC motor if you plan on using it for indexing the spindle.

You would not have a problem with having a variable speed capability, but you might have to manually index the spindle into position.

The ability to index the spindle from software, would require a servo driver and an encoder on the motor shaft. The problem that arises here in attempting to use an-off-the-shelf DC motor is the placement of the poles inside the motor. Normally, DC motors do not allow the motor to stop in-between the placement of the poles, so the DC motor would move (or dither) between the two closest poles.

True servos have the poles of the motor laid out in a fashion that does allow the servo to hold a true position with analog or digital encoder feedback.

Jerry

[BobWarfield]

You may be right, Jerry, but that fella from the link I quoted claims to have done it with ordinary permanent magnet DC motors and on a Bridgeport CNC conversion to boot. I've seen posts from others who've done this sort of thing with treadmill motors like I have on my lathe, usually for B'port conversions.

Interestingly, the one drawback mentioned in the article is the inertia of these motors is much greater than "real" servos, so the acceleration is slower. I can't think that matters in my application.

It's also my understanding that the servo approach doesn't rely on the motor pole construction the way steppers do because they simply use the encoder to create a feedback loop to hold the position. In effect, I believe the motor dithers back and forth ever so slightly at the resolution of the encoder. With even a slight reduction ratio methinks this errant motion would be nearly invisible to the indexing application.

Of course I'm no expert and just talking through my hat here. Any others want to comment on this?

Best,

BW

[Al_The_Man]

There definately is a difference between a motor designed as a servo and a general purpose DC motor, the former usually will have more poles, plus 4 brushes, not 2, and a skewed rotor lamination design, this will all give superior performance at low rpm, the general purpose type is subject to cogging at low speed due to the lack of features, regardless of encoder resolution.
The motor in the link is a better quality Gen. Purp. motor compared to some that are made for Treadmill duty.
Personally, Considering the ability to purchase large surplus Servos at a decent price, I would only consider using a motor as shown for general spindle drive.
Al.

[wizard]

I'm with the majority here. The cheaper DC motors will not run well close to zero RPM. As mentioned the issue is cogging. However in the smaller sizes there is a huge array of DC motor designs, some are suitable for servo conversion some aren't. If you get the motors for free it might be worth looking into but if you are going to spend $$$$ better to buy motor suitable for servo duty.

Another issue is heat. Servos are designed or rated to put out a specific torque continously at zero RPMs. Many DC motors are not.

All that be given there might be some utility to servoing a cheap motor for doing somethng like indexing a spindle. Say for a tool change. Also note that for most spindle drive arrangements you would likely be better off with the encoder on the spindle. Still if you really want control over your spindle the only way to do so relaibly at zero RPM is though a motor designed for that.

Dave

[Stevie]

This brings up an interesting question

The big 1Kw Sanyo's

With high count encoders like they have; just how would you drive one to say 3000rpm with the limitations of 45,000 pulses avail in Mach3
seems to me like you'd be lucky to get over 500rpm

[NC Cams]

Keep in mind that gearing can deal with a lot of the issues endemic to a DC motor.

If you look at the RC servos, they have fairly inexpensive brush 3 or 5 pole PMDC motors. They do tend to cog. However, they also run tremendous gear reductions which is used to regain the torque.

The model train guys do the same thing with skewed arm, 5 pole, 2 brush DC motors - they use worm gear drives and you'd be amaxed at the control/power/speed they can generate with an N scale locomotive using such a drive.

Using a tread mill motor as a servo is an interesting idea. I bought one to play with and found it to have a lot of promise, as long as you only ran it in one direction.

However, to hold a part stationary might pose a problem. Here's where a gearbox with a brake might be in order.

Yes, a treadmill motor can be turned into a servo. All you do is blah blah blah encoder blah blah blah hook it to a Mach blah blah cut parts.

A bit of an oversimplification but doable IF you replicate the technology glossed over by the blah's. That's the key - replicating the not so readily apparent linking technologies.

[BobWarfield]

Originally Posted by Stevie This brings up an interesting question The big 1Kw Sanyo's With high count encoders like they have; just how would you drive one to say 3000rpm with the limitations of 45,000 pulses avail in Mach3 seems to me like you'd be lucky to get over 500rpm

Yep, this is true. However, there is always Mach IV and a GRex for just such an occasion! I got my GRex yesterday and haven't had a chance to try it yet. What a sexy little beast! Tiny little thing for what all it does. Chock full of I/O, 6 axes, and it runs off the LAN. It will be a while before Mach IV is mature enough to experiment with this kind of DC motor treadmill stuff, but I'm in no great hurry. Encoders can be had cheap, and I'll need to get some when I convert my steppers to "unstallable mode" anyway. I'll probably pick up an extra at that time and revisit this issue.

I've got an application where it sure would be nice to drill holes off-axis in a freshly faced surface. If I can fabricate a little drill motor with chuck that drops into my QCTP and slap an encoder on the spindle motor to get there, life would be good. If not, I'll just have to continue moving the part off to the mill for the holes.

Best,

BW

[WilliamD]

I too am working on a project which turns a PMDC motor into a servo. I bought a good quality DC motor, like the one referenced in the article above. I also have the famous 2.5 HP motors off SurplusCenter. I'll have encoders for both at the beginning of next week, and should have results by the end of next week. I'm pretty anxious, seeing how others have made it work. But we'll see.

[Stevie]

Originally Posted by BobWarfield Yep, this is true. However, there is always Mach IV and a GRex for just such an occasion! I got my GRex yesterday and haven't had a chance to try it yet. What a sexy little beast! Tiny little thing for what all it does. Chock full of I/O, 6 axes, and it runs off the LAN. It will be a while before Mach IV is mature enough to experiment with this kind of DC motor treadmill stuff, but I'm in no great hurry. Encoders can be had cheap, and I'll need to get some when I convert my steppers to "unstallable mode" anyway. I'll probably pick up an extra at that time and revisit this issue. I've got an application where it sure would be nice to drill holes off-axis in a freshly faced surface. If I can fabricate a little drill motor with chuck that drops into my QCTP and slap an encoder on the spindle motor to get there, life would be good. If not, I'll just have to continue moving the part off to the mill for the holes. Best, BW

I think I'll wait on Mach4

[wizard]

Originally Posted by Stevie This brings up an interesting question The big 1Kw Sanyo's With high count encoders like they have; just how would you drive one to say 3000rpm with the limitations of 45,000 pulses avail in Mach3 seems to me like you'd be lucky to get over 500rpm

There is and always will be trade offs in motion system design. It is all in the math. Not that I'm going to try to remember all that but there are a few things to consider.

Yep high resolution encoders can slow things down. Either due to the data rate capabilities of the hardware being feed or to signal degradation.

That is one issue. Another thing to watch out for is that high resolution encoders increase the gain of the system. You could concievably have issues tuning the servo loop and other issues that high gain systems exhibit.

On some of the very high resolution systems I've worked on the solution was to switch in midstream to a low resolution encoder for fast moves. This gave the machine speed when needed and resolution when making the parts. Of course now adays the electronics can be faster to keep up with the high data rates from the encoders. That is fi the encoders can keep the signal quality up.

Dave

[lerman]

Use a large 3 phase motor for normal turning, facing, etc. Have a second, lower power, servo motor for indexing. Have both on the same belt driving the spindle.

Ken