Linear motors for hobbyists?

CNC machines normally use direct drive stepper motors driving ball screws.

find an ironless linear motor. Baldor's LMCF series would have been ideal, but it looks like they're no longer available.

nd anything suitable. All the linear motors are huge and expensive. Has any one else had any luck?

You might want to think about printed windings - lots of layers in a multil ayer board with very heavy copper on each layer - and neodynium iron magnet s to set up the magnetic fields. Neodynium iron magnets can be quite compac t, which might be important in your application.

You'd end up with a U-shaped magnetic structure, in soft iron or mild steel to hold the magnets on either side of the - moving - PCB, very like the Ba ldor motor (of which I've got an example, though it was too thick and too h eavy for my application, which needed a motor structure about 12m wide).

I was thinking about using ferro-fluid to centre the board between the magn ets - I've got no idea whether the centering forces (essentially magnetic p ressure) would be high enough for your application.

--
Bill Sloman, Sydney

Good sources of steppers are old printers and hard/floppy drives. The older the better. I remember 8" floppy drives with very good stepper motors. You might still be able to get them in garage sales, if you are lucky.

Linear motors make a lot of sense if the available force is enough for the task. Ball screws are expensive and they need good bearings to run in, which are expensive too, then you still need to buy a motor anyway and usually an encoder. I was surprised to find that a lot of large CNC mills (e.g. from DMG) now use linear motors. I guess the coils are probably liquid-cooled, in order to allow more current and therefore more machining force.

If this is a one-off project where the OP's time is not too expensive, I wonder whether it is worthwhile to build a motor rather than buying. If making a few of them then it may also be worth invesitgating whether a PCB place could make some multilayer boards with very heavy copper plating for the windings, though if large forces are required then it might be hard to get a high enough percentage of copper in the gap between the magnets. I guess with the right jigs they could also be wound from wire and cast in epoxy etc. Liquid cooling might be needed if the force and duty cycle are high.

If I were building a light-duty CNC machine I would also consider using a 2D planar motor (with a hydrostatic or aerostatic bearing under it, held in compression by the magnets or electromagnets of the planar motor). By using a 2D planar motor that can apply torque to the platform, (e.g. two simple planar motors stuck together side by side), the angle of the moving platform could be servo-controlled. This would avoid the need for linear bearings, which are also expensive, (though a second encoder would be needed on one axis to determine the angle of the platform). There are a few sorts of planar motors (induction, permanent magnets in checkerboard or Halbach arrays, etc.), and a few videos of these on Youtube.

Of course this all assumes that linear displacement transducers are available. I wonder about the practicability of interferometers using a single mode diode laser locked to an absorbtion line of iodine (red) (or an infrared one at ~1550nm laser locked to an absorbtion line of acetylene). A much cheaper, less accurate option might be a 2-d version of the PCB based capacitive transducer that is used on cheap digital vernier calipers.

BTW: In case any of you are interested in how the air bearings of PCB drilling spindles work, here is a book about aerostatic bearings:

I found it interesting, and the author did not object to me uploading it. I put his webpage address at the end of the pdf.

Chris

Our photo-plotter at Cambridge Instruments (for making printed circuit artw ork) used precisely such a 2-D linear motor running on a air-bearing.

With a 2-D linear motor you've got an array of magnets in the table over wh ich the moving coils run. Hall effect sensors can track the varying magneti c field.

You may need a bunch of them to average enough magnets to smooth out local variations.

Laser interferometers can track X and Y position with exquisite precision. At Cambridge Instruments we used a 20,000 UK pound bought-in package from H ewlett-Packard to control the mechanical stage in our electron beam micro-f abricators to a fairly small fraction of the He-Ne laser wavelength (632.8 nm) . Zygo introduced a rather better system in the late 1980s, but nobody wanted to switch.

Don't bother trying to roll your own. We got stuck with such a system devel oped at Thompson-EFCIS, and it was total rubbish.

There are lots of cheaper and less precise systems around designed for mach ine tools. Heidenhain is a good place to start looking.

--
Bill Sloman, Sydney

Or servo motors driving ball screws. And it's done for a reason -- you get a much more rigid assembly that way. A good linear motor doesn't exert any force at all as a function of position -- it only exerts force as a function of drive current. Any rigidity in the system would come from the control system, which means that in the short run the position would be at the mercy of the tool bit forces. A ball screw, on the other hand, may move when you push on the carriage, but in the short run it effectively has tons of inertia, which means that in the short run the system is rigid.

Rigidity is a Good Thing in a machine tool.

--
www.wescottdesign.com

There used to be linear motors in disk drives (the last time I saw one, was an Apple CD player), but they were generally low-force things, or very heavy (one for a 3-foot diameter hard disk had a 4" voice coil and about 30 lbs of magnets).

Long-throw woofers, of course, are still produced (but 'long' means a few millimeters).

Screw-and-stepper-motor is more common nowadays. There are hydraulic solutions, too, if you can afford metering pumps (or rely heavily on positional feedback). Heavy equipment can use acme screws and servomotors to good effect.

+1

For a cheaper option you can use trapezoidal screws with a mating nuts, but at the cost of increased backlash.

I have built my own CNC router rig using ballscrews and NEMA 23 steppers. For 900mm X and Y, 150mm Z axis the ballscrews, steppers and beam couplings came to around 350UKP new. This gives a theoretical resolution of 12.5um or about half a thou. for you USA types (20mm * 5mm pitch ballscrew and 400 step/rev motors)

To that you need to add all the running rails/bearings, support structure and bed - oh and you will also need to put together a controller for your drive system.

I think for the basic XYZ chassis I spent around 1700UKP - plus you need to add router and cutters of course.

Andy

One thing a CNC machine drive needs is stiffness. That's why good milling machines are massive, forklift transportable. Lead screw dives are stiff; a linear motor won't be. Even with precise position feedback, it will be gooey at high frequencies.

There are hobbyist machining newsgroups, where people do this stuff.

How much precision do you need? How big might the work be? Metals? Plastic? If you intend to machine metals, a good way to go is to buy a few hundred pounds of Chinese drill/mill machine and add steppers to it. Software out the backlash. It's good to leave the manual controls in place if you can, because sometimes you want to just hog something out by hand, not program the computer, get a feel for the metal.

--

John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

"High frequency" is defined by the inductance of the drive coils and the voltage you can use to change the current through them. Lead screws are limited by by the speed of sound - mechanical displacements - in the screw and it's mountings.

At those kinds of frequencies, everything is "gooey".

--
Bill Sloman, Sydney 
>  
>  
>  
> There are hobbyist machining newsgroups, where people do this stuff. 
>  
>  
>  
> How much precision do you need? How big might the work be? Metals? 
>  
> Plastic? If you intend to machine metals, a good way to go is to buy a 
>  
> few hundred pounds of Chinese drill/mill machine and add steppers to 
>  
> it. Software out the backlash. It's good to leave the manual controls 
>  
> in place if you can, because sometimes you want to just hog something 
>  
> out by hand, not program the computer, get a feel for the metal. 
>  
>  
>  
>  
>  
>  
>  
>  
>  
> --  
>  
>  
>  
> John Larkin         Highland Technology, Inc 
>  
>  
>  
> jlarkin att highlandtechnology dott com 
>  
> http://www.highlandtechnology.com

It doesn't seem to bother these guys, who seem to use ball screw and linear servos pretty much interchangeably on many of their machines:

Whatever the drive system, the mass and elasticity modulus of the table would be the only thing determining the rigidity in the really short term, bearing in mind the speed of sound in steel.

Ball screws are quite stretchy compared to some other parts of a machine tool, being sometimes over a metre long between the bearing and the nut, and only having a fairly small cross section. Therefore they stretch or shorten when you put them in tension or compression, and so the rigidity of the screw itself is far from perfect.

On the other hand, with a linear motor, if you move the table a fraction of a micron, the servo electronics and linear motor can apply a large force within tens or low hundreds of microseconds (or less if you try hard), to counteract the disturbance. With a ball screw, even if the motor were fast enough to respond, the part of the screw within the nut would not even start turning until enough time has elapsed for the rotation to propagate from the motor along the screw to the nut at the speed of sound in steel for torsional waves (perhaps around 3000m/s?). The mechanical advantage that is provided by a fine pitch ball screw also works to make the moment of inertia of the servomotor and ballscrew more problematic. Therefore it is likely that the control loop would have to be made quite slow with the ball screw in the loop, in order to stay stable. I believe a lot of ball screw based machine tools measure the rotation of the motor and don't measure the linear position, and just accept whatever stretching and pitch errors result. This would allow a fairly wideband servo loop. Obviously you then have to spend a lot of money on the ball screw to minimise these errors, or have some sort of dual loop system and pay for a linear encoder as well as the rotary encoder on the motor.

Of course the servo loop would have to be fast to make a system with linear motors be rigid, but why not do that? An FPGA could easily handle the speed requirements. High voltage transistors may be needed, but they are affordable these days. The only reason why I would consider a linear servomotor to be inferior is if it were unable to provide sufficient force for the task. Until this year I had assumed that was the case, but then I found out that a lot of modern machine tools already use linear servos. I guess the motor windings are liquid cooled. Some effort may be needed to keep swarf away from the permanent magnets, though I imagine that a linear induction motor could also be used, which would make swarf removal easier and may be cheaper in large sizes, though with a rather more complicated control algorithm.

Chris

Ball screw driven machines for machining parts normally do not use the rotation count of the screw, that would not be such a smart idea...

Most machines have a encoder on them of some sort, attached to the base and the other side to the table, not the screw..

Jamie

(snip)

Did you miss the part in the thread subject that stated "for hobbyists?"?

[snip]

Some hobbyists take their hobbies seriously.