[RCaffin]

There's a long write-up of a high-end (I think so) DIY RT based on a Harmonic Drive at
http://www.cnczone.com/forums/linear...gineering.html
1000 steps per degree, no real backlash, 125 mm 3-jaw chuck.
(Disclosure: my RT.)

Cheers
Roger

[handlewanker]

Hi.....harmonic drives do have some backlash, but so little it doesn't really matter in the full scale of the compactness and reduction capability........for my yen, a completely backlash free drive, mainly to be able to create one and use it is my ambition.

I am working on the assumption that if you deep mesh involute gears with a resilient component the drive will run metal to metal without binding and that means no backlash ....as in 100% free.

Just making a 4th axis that is near the figure is not the aim....it has to ,be 100% or the quest for the Grail will go on.

So far I've come up with 3 different solutions, each having a different drawback, but all in the ball park.

The 1:18 or 1:36 reduction is the all important attribute it must have so that it will work with a 200 step motor and give a resolution of .01mm on the circumference on diams around 50 to 75mm.

The working principle has been established and now all it needs is the final design to actually make it.........one design has 4 gears, another 8 and another10, but all are practical and not hard to make.
Ian.

[louieatienza]

Most of the rotary tables I have been looking at either come with Nema 23 or 34 motor mounts, or the manual rotary dials can be replaced with motor couplers relatively easily. I need to do more research on motors for this but I will definitely read up on harmonic drive options. My intention is to buy a motor with a lot more torque than I think I might need. The message I have taken away from here is that the more holding torque the better.

Every part I work on is likely to start as a 2" x 6" rectangular slab that is between 12" and 30" in length. That should make it fairly easy to guesstimate the max weight it will need to hold.

I don't think you need to get too carried away. The great thing with harmonic drives is the fact that they achieve very high gear reduction in a single stage. With a planetary reduction it may require two or three stages to achieve the same reduction. So a harmonic drive would be more compact. But you'd probably find that surplus planetary gearheads are much cheaper than harmonic drives, generally. Also, unless they're branded "Harmonic Drive," I wouldn't trust the cheap knock-offs. Another thing, you can buy the Harmonic Drive gear assembly, of which you'll need to provide your own input and output bearings, or buy a pre-made unit that has both bearings and only requires mounting. Finally, the way the Harmonic Drive is mounted will determine its gear ratio - and they way you'd intuitively mount it and make the input and output would give you an oddball ratio like 51:1, 101:1, etc.

Probably the least expensive way that would still give you good accuracy would be to just find a planetary gearhead from one of the well-known brands - Parker, Wittenstein (Alpha), Stober, Thomson (Micron), Neugart, GAM, ABLE... even Harmonic Drive makes planetary gearheads which have a flex ring gear designed to take up lash. These will have large heavy duty output bearings and output shafts, so you could theoretically make a chuck adapter, and it should be fine for what you are doing. If you find something in the neighborhood of 20:1-30:1 you'd only need a 300-300W servo.

[Goemon]

Increasing the reduction seems to be fairly straightforward with most setups using any one of the ready made and inexpensive gear reductions available. As with most of these parts, compatibility is about finding shaft couplers to go from your motor to the gear reduction device and then to the spindle shaft that turns yours chuck or round clamp.

Most of the rotary tables I have seen also have their own gear reduction. Some look high (e.g. 80:1 or more).

I can't see any of this being an issue as long as I make sure that the motor and reduction combo has sufficient torque to index the heaviest blank stock I am likely to work with. In my case that is very predictable as I can google the weight of a 2" x 6" x 30" slab of aluminum.

[ger21]

many gearboxes have the coupling built in. They are designed to fit a specific size motor, so it's a simple bolt on procedure.

[handlewanker]

Hi, I get the impression that with a slab of hardwood or aluminium 2" thick X 6" wide and 30" long you would be working on something that looks like a large flat panel.......that is, working on each of the wider flat sides top and bottom etc.

It must be realised that as soon as you get away from the centre line of the work piece the torque reaction that the cutter applies to the 4th axis head, bearings and drive are increased.

No matter how firm you hold the work piece between the tailstock and the drive there will be a deflection on each side of the work piece as at 6" wide it is 3" off centre.....this is something like stepping into a boat and having it dip down on one side from your weight.

Any deflection from the cutter forces will make any surface the cutter works on totally inaccurate.......especially in the Z axis area.

I think this would totally rule out a belt drive/timing pulley set-up mainly due to the low 1:6 reduction ratio, but also from spring in the belt drive, but the worm drive or harmonic drive types would be better on a broad faced work piece provided the backlash could be reduced to practically zero, and that was where the search for the Holy Grail was heading.

BTW.........if the work piece is a wide flat panel then I have to think that a 4th axis is not the ideal means to rotate it for surface machining as all you get is a curved surface......the Z depth will be constantly decreasing or increasing as the piece moves away from the flat plane......is that how it's meant to be?
Ian.

[louieatienza]

Hi, I get the impression that with a slab of hardwood or aluminium 2" thick X 6" wide and 30" long you would be working on something that looks like a large flat panel.......that is, working on each of the wider flat sides top and bottom etc.

It must be realised that as soon as you get away from the centre line of the work piece the torque reaction that the cutter applies to the 4th axis head, bearings and drive are increased.

No matter how firm you hold the work piece between the tailstock and the drive there will be a deflection on each side of the work piece as at 6" wide it is 3" off centre.....this is something like stepping into a boat and having it dip down on one side from your weight.

Any deflection from the cutter forces will make any surface the cutter works on totally inaccurate.......especially in the Z axis area.

I think this would totally rule out a belt drive/timing pulley set-up mainly due to the low 1:6 reduction ratio, but also from spring in the belt drive, but the worm drive or harmonic drive types would be better on a broad faced work piece provided the backlash could be reduced to practically zero, and that was where the search for the Holy Grail was heading.

BTW.........if the work piece is a wide flat panel then I have to think that a 4th axis is not the ideal means to rotate it for surface machining as all you get is a curved surface......the Z depth will be constantly decreasing or increasing as the piece moves away from the flat plane......is that how it's meant to be?
Ian.

The holding torque of the motor you're using will be multiplied by the gear reduction and the gear efficiency. It would take a massive amount of torque to back-drive a Harmonic Drive or planetary gearhead with only 30:1 reduction. And if a worm drive rotary table is used, they are non-back-drivable.

As to flat bar stock, the usual procedure would be to use it as an indexer first to rough each side, then use continuous rotary and/or parallel machining for the finish passes - unless you also have true simultaneous 4-axis CAM, but that can get very expensive.

[louieatienza]

many gearboxes have the coupling built in. They are designed to fit a specific size motor, so it's a simple bolt on procedure.

That's a good point. Also the type of gear box will be specific to the type of motor it's designed for. Some are NEMA sized and can fit steppers or some NEMA sized servos (some may need motor shaft bushings to work.) Others are sized 50mm, 60mm, 90mm, etc., and are designed specifically for servos (and will not fit steppers.)

[handlewanker]

Hi, I think we've answered the posters question as to the limitations of the cheap EBAY renditions for 4th axis drives, and the consensus of opinion is that totally another form of drive is the answer.

That drives the boat into more expensive waters where the solution for a drive to cater for the specific needs is just to throw more money at it.

You could possibly get close to the ideal if you used a rotary table on it's side and ignored any backlash that would be present.

I think I would pursue that path if I was only interested in an off the shelf solution without having to modify the drive in any way.

There was one solution mentioned on another thread some time back, and that was to wrap a wire around and attach it to the chuck and attach the other end it to a spring or counter weight to apply constant pressure against back driving from backlash.

That would be true if the drive only rotated a max of 360 deg and then returned back to 0 deg again......the spring pressure would keep the wormwheel teeth against the worm flanks at all times then........otherwise a lot depends on how much backlash you can tolerate.
Ian.

[RCaffin]

And you don't use cheap Chinese '5 for $2' cutters on hard steel either.

Cheers
Roger

[Goemon]

Hi, I think we've answered the posters question as to the limitations of the cheap EBAY renditions for 4th axis drives, and the consensus of opinion is that totally another form of drive is the answer.

That drives the boat into more expensive waters where the solution for a drive to cater for the specific needs is just to throw more money at it.

You could possibly get close to the ideal if you used a rotary table on it's side and ignored any backlash that would be present.

I think I would pursue that path if I was only interested in an off the shelf solution without having to modify the drive in any way.

There was one solution mentioned on another thread some time back, and that was to wrap a wire around and attach it to the chuck and attach the other end it to a spring or counter weight to apply constant pressure against back driving from backlash.

That would be true if the drive only rotated a max of 360 deg and then returned back to 0 deg again......the spring pressure would keep the wormwheel teeth against the worm flanks at all times then........otherwise a lot depends on how much backlash you can tolerate.
Ian.

I am assuming that the amount of backlash will be somewhat dependent on the quality of the rotary table. Like anything else with CNC, there is a broad range that starts at $60 and goes up to many thousands.

My assumption (in the absence of any direct experience) is that a functional rotary table has to have accounted for issues like that - given what their primary purposes is. I.e. Even a manual rotary table main use is holding and rotating metal parts while they are being cut by a metal milling machine. In theory, the requirement for strength and precision should be even greater than what is needed for a CNC router.

I don't think they aren't designed to hold work still while it's being milled but that would be the job of the motor in this case. I am assuming that this is the importance of having enough holding torque.

I'm not sure I understand how adding additional spring tension would help if you have enough holding torque and the shaft couplers are suitably robust for the purpose. If they aren't robust or there isn't enough holding torque, it's hard to imagine that a spring would work as the last line of defense.

Please can you explain where your see the spring adding value here? Are we talking about correcting for any additional play in the motor shaft and coupling components while the motor is in a holding position? How does this work on a quality ready made 4th axis?

[handlewanker]

Hi, the spring I mentioned (specifically for a 4th axis) is to drive the worm deeper into the worm wheel teeth to make metal to metal contact but with some resilience to prevent binding in that mode.

This is purely an idea I proposed to remove the backlash that ALL worm/worm wheel derived 4th axes have.

It's a modification that pivots the worm with spring tension in a pivoted housing that allows metal to metal contact and so removes the backlash.

The backlash is caused by the need from manufactured tolerances to have some very small clearance in the worm wheel tooth contact with the worm as it rotates.

Normally, for indexing and positioning purposes in dividing heads and rotary tables, the table or chuck is in a locked state so no movement will occur while the tool is in contact with the work piece.

In a 4th axis design deriving it's rotary motion from worm /worm wheels .....some backlash can occur, and it was my proposal to "find the Holy grail" and have a 4th axis that responds to movement forwards and backwards without lost motion.

All dividing heads and rotary table have worm/worm wheel contact adjustment for backlash, but they are of a cam type adjustment and so are rigid in that manner.......too tight and the worm binds in the worm wheel and too loose for fast rotation and you have backlash.

A lot depends on the quality of the build as all designs follow the same principle but not in the internal build quality of the components........quality = price tag on a rising scale......don't ask me how I know.

For simplicity and cost effectiveness the belt driven 4th axis is probably the cheapest bang for the buck you can get and the simplest no frills design, even if the reduction is rather low, but I prefer an alternative design for the ideal solution to achieving 100% backlash free.
Ian.

[louieatienza]

I am assuming that the amount of backlash will be somewhat dependent on the quality of the rotary table. Like anything else with CNC, there is a broad range that starts at $60 and goes up to many thousands.

My assumption (in the absence of any direct experience) is that a functional rotary table has to have accounted for issues like that - given what their primary purposes is. I.e. Even a manual rotary table main use is holding and rotating metal parts while they are being cut by a metal milling machine. In theory, the requirement for strength and precision should be even greater than what is needed for a CNC router.

I don't think they aren't designed to hold work still while it's being milled but that would be the job of the motor in this case. I am assuming that this is the importance of having enough holding torque.

I'm not sure I understand how adding additional spring tension would help if you have enough holding torque and the shaft couplers are suitably robust for the purpose. If they aren't robust or there isn't enough holding torque, it's hard to imagine that a spring would work as the last line of defense.

Please can you explain where your see the spring adding value here? Are we talking about correcting for any additional play in the motor shaft and coupling components while the motor is in a holding position? How does this work on a quality ready made 4th axis?

A worm drive is non-backdriveable. On a manual mill, you can do precise work even if the rotary is not of ideal strength or precision because you can "tweak" its position. On a CNC you actually need more precision, because unless you have an encoder directly on the table itself, you'd never be 100% positive of its absolute position. Luckily for most the work done on DIY machines, the surplus gearheads and Harmonic Drives on eBay will give us more precision than we'd actually see.

The one thing I'm not too keen on with most worm drive rotaries is that they usually have a relatively high reduction ratio, like 60:1 or 90:1, so unless you were just purely indexing, it's going to turn pretty slow if you're using a stepper. And the ones rated for servos cost significantly more. As to holding, as the gearhead reduces the speed by the gear ratio it also increases the torque by the same amount. On some rotaries a brake is employed to hold the spindle for indexing, but I think for your needs that's probably an unnecessary complication.

[handlewanker]

Hi....there are a number of reasons why a certain type of drive would be preferred.....first, the simple toothed or timing belt drive can be made by even the less gifted in the ability class of DIY mechanics and still provide a true 4th axis drive.

If you want to get more sophisticated then the mechanical drives that need lots of mechanical skill to build and set up will also do the trick with better results but will take a workshop skill level and decent machinery to make them accurately.......you only get out of it what you're capable of putting into it.

I think if low reduction and cost is on the cards then an off the shelf belt drive model is the simple solution.

That means comparing complete EBAY priced items to what you can cobble up at home with off the shelf or DIY manufactured components......your labour input will work out at 20 or 30 hours of work for the same results......a $400 EBAY model is a hard act to follow.
Ian.

[johncarpenter]

Another cheap harmonic drive 4th axis going together. Uses a "Harmonic Transmission" again. I like this one better than my previous integrated inline stepper one. Less work to fabricate. 200 to 1 reduction times 2 with a 425 in-oz stepper. 8 inch lathe plate in place but not mounted. Less than $200 and this one can be built with an accurate bandsaw.

[handlewanker]

OH dear I think the plot has been lost........the reason to use a harmonic drive is because it has high reduction and stiffness in the holding power of it's design........adding a belt drive makes all this a bit......what can I say without offending someone's design preference?..........well, it's like having a horse to pull the family car.

My opinion would be to hard couple the stepper directly to the drive, and thaty would make it more compact too.

I wouldn't personally use this set-up as the 1;200 is way too much reduction.....my opinion only......even 1:60 is a lot.
Ian.

[johncarpenter]

Well, remember that that belt is on the input side of the gear reduction so any sloppyness gets reduced. But I don't think belts are all that bad. My full size Excello CNC knee mill uses belts between the steppers and the ballscrews and I'm very happy how it works. And yeah that is a lot of reduction. But I use what I can find.

[ger21]

A lot of big machines use belts to turn the ballscrews, with zero backlash. There's very little load at all on the input of a gearbox, so driving it with a belt should be an issue at all.

[Goemon]

A worm drive is non-backdriveable. On a manual mill, you can do precise work even if the rotary is not of ideal strength or precision because you can "tweak" its position. On a CNC you actually need more precision, because unless you have an encoder directly on the table itself, you'd never be 100% positive of its absolute position. Luckily for most the work done on DIY machines, the surplus gearheads and Harmonic Drives on eBay will give us more precision than we'd actually see.

The one thing I'm not too keen on with most worm drive rotaries is that they usually have a relatively high reduction ratio, like 60:1 or 90:1, so unless you were just purely indexing, it's going to turn pretty slow if you're using a stepper. And the ones rated for servos cost significantly more. As to holding, as the gearhead reduces the speed by the gear ratio it also increases the torque by the same amount. On some rotaries a brake is employed to hold the spindle for indexing, but I think for your needs that's probably an unnecessary complication.

In my case it is only for indexing and nothing else. It is just about turning the work piece over so the other side can be machined without any indexing issues. If I can't do that with whatever kit I buy then it has no value to me. Similarly anything other than indexing also has no value to me. I won't be doing any lathe type work on it so speed is not particularly important to me. My main concern is being able to hold 2" x 6" x 30" aluminum work pieces with enough stability for the machine to do it's thing.

[RCaffin]

My main concern is being able to hold 2" x 6" x 30" aluminum work pieces with enough stability for the machine to do it's thing.
In that case flex in the workpiece itself may be your limiting factor.

Cheers
Roger