Backlash free rotary table

[handlewanker]

Hi,.....plasma welding.......similar to oxy/acc but with a heat source that gets up to 8,000 deg C.....read up on MULTIPLAZ 3500 on UTUBE, there's a number of videos on the subject.

I've had a rethink on the bearings for the spindle on the 4th axis I'm working on, and I think that instead of having one deep groove radial bearing at each end of the spindle , I now intend to fit two bearings at each end, but the bearings will still be the deep groove type but will be half the width, hence at 7mm wide they will occupy the same space in the housing.

The advantage is that, as the balls are also smaller due to the smaller width of the race, having virtually the same OD gives a bigger bore size, which means I can have a bigger spindle and a bigger spindle bore to pass larger diam work through.

I already have a pair of 6007RS beaings which are 65x35x14, and if I go to the 6810ZZ deep groove radial I get 65x50x7, and they are rated at 10,000 rpm......LOL...... stand by for take off.

I feel that that having two bearings on the shaft at each end with a cross sectional aspect ratio that is rectangular as compared to a single bearing that is square.....by this I mean that if you look at the bearing in cross section across the ball area, the single bearing has a square section with the OD,ID and width forming a square, whereas the twin bearing set-up has a rectangular cross section even though it is the same width.......the advantage is more stability and a bigger spindle size without a bigger width and only a 3mm increase in OD.....the speed is irrelevant as the 4th axis won't be moving faster than 100 rpm at most.

It's academic as far as design is concerned and is not really a "must have" factor for a 4th axis function, but it does give a bigger spindle bore capability, and that is the main reason.

Having four rows of balls also increases the ability to handle some end loading forces, which again is a plus.
Ian.

Cost is about $25 per bearing, so the jury is contemplating on that one.
Ian.

[RCaffin]

having two bearings on the shaft at each end with a cross sectional aspect ratio that is rectangular as compared to a single bearing

Hum ..... Interesting thought.
Hell of a robust design then!

Cost - not high, compared to the hours of machining?

Cheers
Roger

[handlewanker]

Hi, it's probably over engineered for the job it's going to do, but I like to think a bit outside of the ordinary to get something that has more going for it.

The machining will be the same as for two bearings, just straight through boring.....the bonus I decided with the bigger spindle diam was worth the extra money for the bearings....they cost more than the two original bearings I had planned.

For the ultimate design I would be thinking of big needle bearings but as there is a very real thrust factor to counteract as well as sealing, that is a design I'll not worry about.......it would enable a very big spindle bore without increasing the box dimensions, but the limiting factor for spindle bore size is the hole through the 80mm diam chuck, so it is diminishing returns in the end.

The other side of the coin is that with an even bigger bore available with needle bearings, you can use a collet system and still have a fairly big material diam, but collets are very limiting in material standard sizes, and collets cost a bomb for complete sets which may never be used fully.....they are also size dedicated specifically for their one size, not like the ER series which has a 1mm close down overlap to size.

As this 4th axis is going to be specifically for the SVM-0, the size will be made around the table dimensions and slot pitching and taking into consideration that the SVM-0 mill is not all that big and the work I intend to do is also not going to be universally big.....you can go overboard attempting to cover all eventualities.

Thus far, I think I have covered the aspect of the backlash problem with simplicity that is easy to achieve, so now it only needs to be made.
Ian.

[RCaffin]

Given my druthers, I think I would prefer over-engineering to under-engineering. :-)

For the ultimate design I would be thinking of big needle bearings but as there is a very real thrust factor to counteract

I dunno. Needle bearings are great for sideways loads, but really poor at handling end-thrust - and a rotary table will get plenty of end-thrust.
I would be more inclined to suggest crossed roller bearings as the ultimate - as the pricing would indicate!

Collets ... hum, there's an idea. All you need is a suitably tapered bore behind the lathe face plate, and plenty of lathes have that. OK, they usually don't accept ER collets, but no matter. I fell in love with the ER32 series on my lathe.

Cheers
Roger

[handlewanker]

Hi, Cost is going to escalate to about $100 for 4 bearings, plus the cost of a worm wheel and matching worm, but I might use a worm I already have and get a worm wheel cut to match it.

The gains would be an increase in spindle size without increasing the box dimensions which for the SVM-0 I want to keep compact........the centre height of the 4th at the moment is 80mm which will allow full opening of the jaws without hitting the table.

The main machining will just be the boring of the bearing holes and the splitting of the box and final boring to bearing size.

The rest is just a bottom facing and drilling and tapping exercise.

Fitting the worm wheel is simply a key and screwed collar,

The more complicated part will be making the worm housing that pivots off the face of the back of the box enclosure.

The whole design is dependent on getting the worm to contact the worm wheel with a constant pressure, so that will be cut and try when I come to it.
Ian.

[handlewanker]

Given my druthers, I think I would prefer over-engineering to under-engineering. :-)


I dunno. Needle bearings are great for sideways loads, but really poor at handling end-thrust - and a rotary table will get plenty of end-thrust.
I would be more inclined to suggest crossed roller bearings as the ultimate - as the pricing would indicate!

Collets ... hum, there's an idea. All you need is a suitably tapered bore behind the lathe face plate, and plenty of lathes have that. OK, they usually don't accept ER collets, but no matter. I fell in love with the ER32 series on my lathe.

Cheers
Roger

Hi, needle bearings are OK for situations where a small envelope is required, but as I'm not stressed for that factor, and considering the aspect that you need to have a hardened shaft for them to run on, coupled with the fact that thrust means a pair of dedicated thrust races and the design can get quite complicated for no good reason.

It would be the way to go if the spindle just had to be much bigger and not increase the housing dimensions.

A rotary table is a different kettle of fish in the bearing department, as it's preferable to have a large diam angular contact type bearing to keep the table down on and another one much smaller on the bottom of the spindle with thrust capability, as the profile has to be quite shallow to reduce the height beneath the spindle with a with job on the table.....very different to a headstock or 4th axis horizontal box design.

I have a set of ER32 collets, but as they only go up to 19mm I don't think I'll complicate it further and have a taper that accepts them......You only really need a Morse taper in the spindle chuck end to have a centre as this is what dividing heads have for between centre work.

After doing some doodling on my graphics pad, I'm also going to bore a shallow hole, about 50mm diam and 6mm deep in either one side or the bottom of the 4th axis base.....this is to mount it on it's side on an angle plate so that it can be tilted up from horizontal to about 45 deg.

I think 45 degrees will be the limit as the height to the spindle might get reduced too much and there is also the worm drive mechanism on the back to worry about clearing the table.

There will be a time when you want to incline the spindle up a bit, so that is how I'll do it.....it could enable a 5th axis to happen, but not for my needs......... but who knows.
Ian.

[RCaffin]

.this is to mount it on it's side on an angle plate so that it can be tilted up from horizontal to about 45 deg.

um ... errr .... oh I see
Cute!
I have seen some 4th+5th axis units that look like that. Neat idea.

Cheers

[handlewanker]

Hi, a lot depends on the mounting, due to the stepper motor sticking out sideways at 90 deg to the main spindle.

I think an angle plate made up without webs and with the box mounted on the inside face with a bug on the side face instead of the bottom and mounted to allow at least a 45 deg inclination upwards would cover most needs.

As the box inclines to the full 45 deg, the worm housing at the back moves downwards and towards the angle plate and can foul it if the design is not carefully worked out.

Having the worm and stepper motor in the verticle position with the spindle horizontal would place the stepper motor up off the table when the box is inclined, so this might be the way to go.

The box by itself is just big enough to house the bearings, and adding the worm wheel and worm drive housing to the back makes them overhang the sides of the box due to the worm being offset to do the driving, so this would mean that the worm would benefit from being mounted in the vertical position on the far side of the box and in that position would not hit the angle plate as the box inclined upwards.........it could also mean that a full 90 degree rotation to the full vertical position is possible, provided the table to spindle height clearance was sufficient, which I doubt on the SVM-0.

The design now takes on a different approach.

Being able to incline upwards, and attached to an angle plate for that purpose, means that it no longer needs to be mounted on a base plate to the table and can be just a square box attached to the angle plate which will now become the base plate for bolting to the mill table.

Making the box is now even simpler as it's just a square object and could be made from a piece of heavy wall square tubing with ends welded on and a round bung welded on one side for attaching to the angle plate.

I am having the box split as originally designed, so this will now occur in the vertical plane instead of a horizontal one.

Time to get the graphics tablet out and put some sketches together for the "new design".
Ian.

[RCaffin]

and a round bung welded on one side for attaching to the angle plate.

In another context ... I used a shoulder bolt for this. Worked well.

Cheers
Roger

[lixingwen1987]

I AM GLAD TO TAKE THIS OPPORTUNITY TO INTRODUCE OURSELVES AS ONE OF THE LEADING SUPPLIERS OF Precision Bearings use for CNC and Machine tool ,AND Radial-Axial bearings WHICH ARE MAKE OF Wafangdian Tianjiu Bearings Technology Co.,Ltd.

WE ARE IN THIS BUSINESS SINCE LAST FIFTEEN YEARS; IN THIS SPAN OF TIME WE HAVE SUPPLIED PRECISION IN ALMOST ALL THE CONTINENTS OF THE WORLD.

BESIDES, WE DEVELOP HIGH-PRECISION YRT ROTATED BEARINGS, MACHINE TOOL MAINSHAFT BEARINGS AND MOTOR PRECISION BEARINGS. THE PRECISIONS OF OUR BEARINGS HAVE REACHED LEVELS OF P5, SP, P4 AND UP.

FEATURED BY HIGH PRECISION, LOW NOISE, HIGH ROTARY SPEED, HIGH QUALITY AND LONG SERVICE LIFE, OUR PRODUCTS SAVE USERS’ COST AND SHORTEN PURCHASE TERM. WIDELY APPLIED TO COMMON MACHINE TOOLS, CNC MACHINE TOOLS, LARGE / MIDDLE MOTORS, PRECISION ROLLING MILLS, ETC., OUR PRODUCTS FILLED THE DOMESTIC BLANKS AND CAN SUBSTITUTE IMPORTED PRODUCTS.

THIS IS TO INFORM YOU THAT WE HAVE PRECISION BEARINGS AND OTHER GOOD QUALITY BEARINGS AVAILABLE, CAN SUPPLY FOR YOU WITH BEST QUALITY.

WE HOPE CAN CORPORATION WITH YOUR COMPANY.

WE CAN SUPPLY OUR BEST PRODUCTS FOR YOU.

THANK YOU !

REGARDS?

--------------------------
Lixingwen?????
Wafangdian Tianjiu Bearings Technology Co.,Ltd.
(Wafangdian Heavy Bearings Researching & Manufacturing Co.,Ltd.)
Brand?WZFB
Web: ??????????????? ????????????????
Tel: 0411-85366785
Fax: 0411-85366785
Mobile:+86-15942617721
E-mail:lixingwen@wzfb.com
Q Q : 178702120
Skype: lixingwen1987

Do our best to give customer best service.

[handlewanker]

Hi, one man's meat is another man's poison.....I get a bit pedantic with design and in this instance the bung (about 50mm diam) will be clamped in the upright face of the angle plate, which will be split across the centre line to allow it to clamp the bung and the box at the desired angle.

If the bung was replaced by a deep groove ball race, (probably two) then the box would be free to move and if the mounting spigot for the bearing(s) was extended to allow another worm wheel to be attached and another pivoted worm also, you have a 5th axis in the making, but that is a lot farther down the track, but still retrofittable if ever the occasion arises, but it won't cost a fortune to go to a 5th axis capability.

Being very cold at the moment.......5 deg C this morning at 9AM, the welding of the box, which is the next step, has not yet occurred, which is a blessing in disguise now that the design has evolved to an inclinable design.

With the box being able to incline, a tailstock is now totally unusable in that mode, except for horizontal applications and it would need to be able to be adjusted to rise above centre for shallow angle or long tapered work pieces, but as necessity is the mother of invention, anything can happen.....just another bridge to cross as I haven't even go the mill yet to use it on.
Ian.

[RCaffin]

Hi Ian

> 5 deg C this morning at 9AM
Ah, quit complaining. -2 or -3 C some mornings here, with a cold wind.

Yes, you could convert this unit into a 5th axis, but think of the fun of starting a whole new design in a couple of years time!

Cheers
Roger
PS: off the air for 2 months henceforth.

[ftkalcevic]

Great thread. I've just spent the last couple of days reading it.

I'm planning (still in the dreaming stage) of building a 5-axis machine capable of machining steel.

I want a B-axis (rotate around y-axis) to rotate the spindle 0-110 degrees, and a rotary table as the C axis (rotate around z).

I was planning on using harmonic drives, but a comment in post #17

Harmonic drives can be purchased in zero mechanical backlash configurations but the main drive flexspline is made of spring steel and will deflect under pressure (although not as much flex as a split worm gear). Grab the end of a robot arm and move it back and forth. Older robots are almost all harmonic drives.

made me wonder if they are suitable for high cutting loads (I have no idea what high is. Some of the mid size harmonic drive gearboxes I've been watching on ebay have a dynamic torque of 600Nm).

Is the springy-ness an issue?

As an alternative, for the B axis which doesn't have to rotate 360, I was think of using a hydraulic actuator - a ballscrew mechanism pushes and pulls a hydraulic piston, which turns the hydraulic actuator. Actual position can be feed back with a magnetic strip encoder wrapped the B axis (like this Wholesale Product Snapshot Product name is DMB-A series magnetic stripe(Pole pitch:1+1mm)), so there will be 2 feed back loops to manage (hopefully doable with Linuxcnc). What are the issues with hydraulics? (I know zero about hydraulics). Is it responsive? Is there any springiness when loaded?

Thanks,
Frank

[RCaffin]

Hi Frank

I was planning on using harmonic drives, but a comment in post #17 ... made me wonder if they are suitable for high cutting loads

Nothing has zero springyness. Nothing. Small HDs move more than the larger ones. I have been measuring the compliance of an HD I have and it is small. (ie stiff) Much less backlash than many other things. (Note: measuring micron deflections under load in a calibrated manner can be fun.)

Your hydraulic solution is workable (in theory) but requires some precision in the manufacture to avoid the problem of tiny leaks causing a slow drift. Or you have to factor in a periodic reset of the zero - possibly while in use.

Magnetic bands - well, yes, but... That unit is designed to stick to a flat surface. How it would go on a curved surface - I don't know. Would it stay there 'for ever', or would the curvature cause it to slowly peel off? That said, some mfrs have used the idea. In addition, you would need to experiment to get the butt joint exactly right, or there could be a small discontinuity in the calibration. Yes, it can be done, but you might need to buy a fair bit of tape to run the experiments. Where you might come to grief is in the read head: the one shown is designed to be flat as well. Could it handle a curved surface? Dunno. Maybe you can get a curved read head?

No, I am not trying to be discouraging. It's just that improving precision can be difficult and expensive. Sigh.

Cheers
Roger

[ftkalcevic]

Hi Frank
No, I am not trying to be discouraging. It's just that improving precision can be difficult and expensive. Sigh.

Not discouraged at all. I welcome the feedback.

The magnetic strip indicator was only for a B-axis that had 120 degrees rotation. I wouldn't try to join it - poles every 1mm, 10um accuracy. The plan was to wrap it around the axis, or a flange - I'd make it a large diameter as not to bend the strip too. I bought some of the strip a while back and it came rolled up. It is a thin steel strip, so I thought I stretch it tight. When mounted, I'd just calibrate it. I was going build my own sensor using the AS3111 chip.

Of course, that's plan B. I'd prefer Harmonic Drive gearboxes and direct drive, but I'm struggling to determine the forces involved and how that effects rigidity (and my requirements are a bit airy-fairy at the moment).

[RCaffin]

Not discouraged at all.The magnetic strip indicator was only for a B-axis that had 120 degrees rotation.

Only 120 degrees? And you have some of the strip already? I'd say go for it! Try a quick lash-up first maybe?

I was going build my own sensor using the AS3111 chip.

I'd like to hear how that goes. Interesting.
Um - do you have a URL for the AS3111 chip? I can't find it.

I'd prefer Harmonic Drive gearboxes and direct drive, but I'm struggling to determine the forces involved and how that effects rigidity (and my requirements are a bit airy-fairy at the moment).

There's a guy on eBay in Korea who sells second hand HDs at a fair price. I bought one from him: good condition.

Cheers
Roger

[ftkalcevic]

Um - do you have a URL for the AS3111 chip? I can't find it.

Sorry, typo - AS5311 - AS5311 - High Resolution Linear Position Sensor ? Hall IC - ams [edited link]

[RCaffin]

Interesting.
And they also sell RINGs of encoded magnetic material, with the proper joints as it were.
Hum ...
128 magnetic poles around the ring, and 4k pulses per pole pair ... beats the hell out of an optical encoder, doesn't it?
Hum

Cheers
Roger

[ftkalcevic]

Unfortunately, I only have the chip - no breakout board to test it. I keep forgetting to build one each time I visit itead studio.

My 120degree axis is based on a fixed gantry design. I got a Sony Robokid off ebay and liked the way it had an X axis wider than the table. I thought that could be useful for auto tool changing, or a 4th and 5th axis.

The attached drawings are of the Robokid, but it is too flimsy to cut anything significant. I'm looking at putting together a similar style machine using thick wall RHS (rectangular hollow section) and fill it out with epoxy granite - so I can knock a couple of things off my bucket list.

[ftkalcevic]

There is another post on cnczone about the AS5311. The guy didn't have great success with it - http://www.cnczone.com/forums/genera...ml#post1140288