Design Question: About Lovejoy (Spider) Couplings

[Odin]

Hi all, forgive me if if this post is in the wrong forum- I am both hung over right now and and sheer awe of the number of forums on here, so here goes.

I've been mulling over the notion of building my own 3 or 4 axis CNC mill like alot of people have already done on this fantastic site, and in the course of agressively scouring designs all over the net', and here as well, I've realized that it's not incredibly complex, it just takes some time and careful planning.

Which brings me to my question: in looking over the designs others have built, I've noticed a lot of people choose to use "lovejoy couplings" also called "spider couplings" to mate their servos to the various ballscrews/leadscrews/other linear motion transferers. What I can't seem to figure out is WHY.

I know that no matter what you do, the servos you choose will never EXACTLY repeat positioning perfectly, but can still be fairly accurate within various ranges. The nature of these couplings, however, means meshing two metal joints by placing a rubber or plastic intermediate hub jammed between the metal joints to transfer rotational motion. This should theoretically make it even more prone to positional errors, since some flex is allowed, so why not just DIRECTLY mate two metal couplings to each other, with no rubber/plastic middlepiece to allow flex, or even weld the servo shaft to the ballscrews? Wouldn't that be more repeatedly accurate? Maybe I'm missing a point everyone else sees- I'm not an engineer, more like MacGuyver.

And for that matter, assuming an infinite budget, what do the big boys for the professionally made CNC machines use to couple their drives to their X/Y tables? Do they also use these, or something more exotic?

I am one of those insane people who will take a week to build something that should take 10 minutes, if it means being closer in tolerances or higher accuracy (I suppose it comes from doing gemcutting as a hobby, where you can spend days polishing a single gem facet on a good stone), so I want my machine to be absolutely as accurate and repeatable as possible.

Can someone explain these matters of power transfer couplings to me?

[RICHARD ZASTROW]

Odin, Personally, for the reasons you mentioned, I would not use those couplings. That is because: 1.) I over-engineer everything. 2.) The machines I generally am involved with require a high degree of accuracy. 3.) I'm as much a zealot as you are, maybe worse.

Having said all that, there are reasons for less extremes. Trading a bit of accuracy for a bit of "cushion", easy aligning, less cost etc.

Building an accurate mill requires you to design stiffness into and compliance out of the machine.

Instead of rubber/plastic membered couplings, you might consider metal bellows torsion resistant type coupling, direct mounting or a belt drive arrangement. All have pros & cons of their own.

Infinite budget? hollow spindle AC servo mounted directly on the ballscrew. If the application permits, round or flat linear motor(s).
Dream on,

[Al_The_Man]

Personally I would not use these either, there are many others to pick from, Lovejoy themselves make a variety, the Uniflex is one I have used for zero backlash.
http://www.lovejoy-inc.com/content.aspx?id=352
http://www.lovejoy-inc.com/content.aspx?id=286
The true Oldham coupling is also better.
Al.

[awerby]

If there's a little misalignment between the motor and leadscrew, something that's pretty hard to avoid even if you're careful, then the motor will tear itself apart as it revolves, since every revolution will torque it sideways a little. Either the shaft will eventually break or the front bearing will. That's why the Lovejoy and other flexible couplers are used. If you're worried about backlash or positional error, there are backlash-free solutions, like rubber-cored couplers, bellows, helical and slit types. Which one to use depends on how much power you're transmitting and the degree of misalignment you're dealing with. A good source for these things is Stock Drive Products (sdp-si.com), which also happens, coincidentally, to be a sponsor of this site.

Andrew Werby
www.computersculpture.com

[ger21]

What I can't seem to figure out is WHY.

Because they only cost about $2 per axis. Most guys don't want to spend $30 per axis for zero backlash couplings that will tolerate misalignment. Keep in mind that ANY misalignment with a rigid coupling will cause binding and other problems, as Andrew said.

[Odin]

After only 3 replies, and being less hung over, I checked out some of the links you guys attached.

I had no IDEA there were that many types of couplings! Now I see Lovejoy is a company, not a coupling (I am apparently a moron), and the types of couplings I was referring to look like the Oldham couplers- two metal sides joined by a rubber/plastic variant middlepiece.

As such, I am completely flabbergasted now as to WHICH is the best. With so many technical terms thrown at a layman, I have no idea how to tell any apart. I've only managed to figure out that Zero Backlash is a good thing- meaning it stops and won't jerk backwards on position rest. As far as accomadating servo missalignments from mounting to a precision ground ballscrew, I guess I would have to somehow measure that to choose properly... So go ahead and start laughing at this inept fool...

I am now thus even more lost. I feel like my attempts at over-engineering are catching up to me now, for I have no idea what to make of all these specs! Suddenly I'm wishing I were an engineer instead of a Japanese specialist.

[Al_The_Man]

The Oldham are similar to the spider coupling, but I think you will find that they are much closer to zero backlash than the spider.
Al.

[paulC]

Its also looking at how much backlash figures into your machine at this point. While backlash is critical when you transfer rotational to linear motion, a 1 degree backlash in rotation may be insignificant in some configurations. Its worth doing the math before spending money in the wrong area of your machine. I had concerns about some couplers I was using and on working it out the error translated to .005mm. This was a woodworking machine so it was nothing to worry about. If this had been using ballscrews with a small TPI it would have been a major issue.
Paul

[blksmith]

there is also a problem with slide accuracy with a rigid shaft mount, if there is any misalignment it will tend to make the screw revolve off center and push the slide from side to side a little. so if you have .0005 play in the slide you could end up with .001 movement

[Odin]

Judging from the various types of couplings available that offer the critical thing, zero backlash, for various given torques and speeds, I assume no matter what you do, you can never PERFECTLY align the output shaft of either your servo or stepper to the input shaft of your ground ballscrew, leadscrew, what have you.

In that case, since perfect alignment is nigh impossible for a home builder, WHICH coupling is best to transmit torque to your screw input, with zero backlash, that allows only as much flex as needed? How do you know which one has enough flex for what we do?

I guess now I'm asking, taking the utmost care in design, what do the majority of all of you who are the sticklers for ACCURACY in building your machines, above all else, use for couplers? If I pick a coupling that has too much flex but still zero backlash, won't that throw off positioning by my servo since fractions of it's motion is lost to eccentric flex between shaft couplings, which should be inevitable?

Forgive me if I seem like I'm going overboard, but if I'm going to put out the money to build this thing properly at some point with ground ballscews and good servos, I want to do it right.

It seems like Oldham couplings are used a lot... Are they what I should be using, and for those of you who are the sticklers, what did YOU choose, and why?

Thanks for all the help guys.