[louieatienza]

Ah well, high end brushless AC motors - a far cry from this market - all our home machines and refitted BPs etc. Yes, point made that I ignored the high end.

But is my thinking antique? Once, only the rich could afford a car, and they had to have a resident chauffer. Then you had to be a serious mechanic yourself to drive one - adjusting the spark and compression etc. Now - teenagers and little old ladies hop in a compact and away they go. Will the servo industry follow in this path? I suspect so, but of course I am guessing.

While you maybe right, consider that as many consumer techmologies trickle down from industrial, aerospace, and military technologies. meaning, we eventually will have access down the road to what the big boys have now, but down the road teh big boys have something far more advanced! It wasn't long ago I dreamed of having a 16GB hard drive, now I have 16GB RAM on my laptop! The military had GPS 30 years or so before it made it to industry and moreso befoe it reached consumer goods, and even then we are only allowed an accuracy of 100 yards where the military can have an accuracy of 1/8".

But you can today build a system (or modify an old BP) to get 10 micron or less on an axis, fairly easily. How many people or companies really need better than that? For that matter, how many can hold 0.1 mm consistently on a manual, in production on simple machinery? Sure, an aerospace company might want 0.1 microns ... but there aren't many of those, while there's an awful lot of small engineering shops around. I don't think the low end market is dying.

I don't think anyone is saying that. The fact is, by the time consumers get "cutting edge" technology, it is no longer cutting edge technology by industry standards.

Maybe so, but then, 10 years ago, how many 'average hobby builders' even had a CNC? The 'low-end' market grows, in both volume and performance. History tells us this is always the way.

It's the level of sophistication in DIY machines that is growong. There were quite a few then, but now there are more types of DIY machines out there, from textile machines to 3D printers to pick-and-place machines. I don't think anyone is making CNCs anymore out of garage door track for example. But what was once cheap and commonplace on eBay is disappearing. I've seen the cost of NOS linear rail/block sets go up and transition from aution to "buy it now." Ground ballscrews still can be had relatively inexpensively, as well as precision gearheads though that may change soon as surplus high-end servo systems become more available. I used to see a lot of linear actuator assemblies go for very cheap but now they've gone up in price. Unless you are rich, you end up with the "dregs" of industry - and that is not always a bad thing!

That depends on the specs you want. In many cases a stepper does meet the user's requirements. If you want to go to the shopping mall, you don't need a chauffer-driven 'der grosse' Mercedes.

If someone wants to bequeath me several high-end servo systems (plus a 40krpm spindle drive), who am I to turn them down? But in the meantime...

They still use steppers in industry where the requirements warrant it. Most every printer, scanner likely still all use them for example. They're far easier and cheaper to implement.

[louieatienza]

How about the globoidal cam?!

Google Image Result for http://info.motionindexdrives.com/Portals/177433/Images/98.png

[RCaffin]

would I be right in saying that, for example, an encoder is a disc on the end of a final drive shaft that has many lines on it, and the servo motor, driving the shaft, does so until the required number of lines on the encoder disc have passed by the sensing head and then the servo motor stops and holds that position against all forces, and that it will attempt to keep on driving the shaft indefinitely until the line count is reached, whereas a stepper motor will receive pulses from the controller and attempt to move the output shaft for the pulse count and then stop even if the shaft did not move?

Sounds right to me.
The encoder does not HAVE to be optical, but the principle is correct.

Cheers

[RCaffin]

If someone gave you a high end servo system, you would not know what to do with when you got it, I have lots of them, mostly Yaskawa & Dmm

I'll call you on that one! Send me one and I will put it into action! A rotary one would be nice ...

There is a video a few posts back, with a Hardinge direct drive rotary table this has a accuracy of ( +/- 3 arc-sec ) but is only using a 16384 incremental encoder so are they telling stories about there accuracy??
Or do you know the answer as to how they achieve this, with a 16384 count encoder

Well, I can do the maths. Let's see. 360 degrees is 360 * 60 * 60 = 1,296,00 arc-seconds, according to my trusty HP32S calculator.
Divide that by 16,384 and you get 79 arc seconds. That's a bit more than 3 arc-sec.
Of course, this is possible if you have mechanical step-up gearing between the rotary table and the sensor. A step-up of about 26:1 would be required, assuming their 3 arc-sec corresponds to 1 step. There may be a factor of 2 somewhere, depending on how you interpret the +/- bit. Mind you, I think a mechanical step-up of that size is a bit high, and I would have reservations about using such a high value myself.

Alternately you can have the sensor output analog A & B signals, and do further digitisation on the analog signals. That is what Avago tried to do with the AEDA-3200 unit, and what DMM do with their sensors. Others may also do this as well. But there are speed limits to that technology, which is why they have to resort to rather expensive so-called 'intelligent servo systems'. The so-called 'intelligence' is an expensive way of getting around the limitations of the analog extension process.

You copied what other's had said in post 497 & agreed that this is correct (360*60/16384 = 1.3 arc minutes. Yes, I can imagine that would be very popular.) This is your words & what others said as well

No, I did the maths myself. There was ZERO copying. After all, the maths is pretty simple.

so how do they achieve (1 arc-sec with an Encoder with say a 13 Bit=2048 count encoder Its called Electronic gearing, with a simple bit of math you can have a rotary resolution of (1 arc-sec )

Sigh - yes, I know. See above.

I had better explain here that I am a consultant research scientist with ~40 years experience in designing measurement systems, with a PhD. Been successful too. I am very familiar with sensing and servo systems: a lot of my work has been deeply involved with them, and not only using them but also designing them.

I was already playing with servo systems when Galil was born. My PhD was in Control Systems, around 1972.
And I have a first edition copy of 'Motion Control by Microprocessors' by Jacob Tal of Galil. He sent it to me.
Oh yes, I am obviously way off-track. I cheerfully admit it.

Cheers

[RCaffin]

How about the globoidal cam?!

Oh, their stuff is just cute!
An offshoot of the quite venerable German Feinwerkmechanik school. Very nice.

Cheers

[louieatienza]

Here's a great video demonstrating Kollmorgen's direct drive servos... pretty cool stuff.

Kollmorgen AKD Direct Drive Servo Systems - YouTube

[RCaffin]

Here's a great video demonstrating Kollmorgen's direct drive servos... pretty cool stuff.

0.1 thou at 27" I think they quoted? Well, close enough anyhow.
0.0001/27 = 3.7 * 10^-6 = tan(theta)
theta = 0.00021 degrees = 0.0127 arc-minutes = 0.76 arc-seconds.
That's 1 part in 1.7 million, or between 20 bits and 21 bits resolution.

The demo is impressive, but that sort of angular resolution is what is expected at the high end, which is what I expect this represents.
OK, it's a bit better than what I can do at home!
I wonder what the motor with sensor and power driver costs?

Cheers

[AK47]

How about something like this?

[RCaffin]

How about something like this?

The problems with a worm and wheel config have been discussed a number of times. This diagram does not address them.
Basically, to get zero backlash you need both sides of each wheel tooth to be pressed against the worm. This leads to either huge friction and wear, or huge friction and lock up.
A secondary problem is that a reaction-force will push the worm away from the wheel if it can - if it is spring-loaded. That creates effective backlash. If you don't use spring loading, then the difference between spinning with backlash and locked up from friction starts to become very small. A bit of wear later, and you have a gap.

Yes, you can buy good rotary tables made with worm&wheel configs. They have backlash. It might be small, but it is there.

Hey - this is a problem the mech eng world has been trying to solve for 50 - 100 years so far!

Cheers

[handlewanker]

Hi, I fully love to browse the various exotic methods to get something to move from A to B, but at this moment in time all those methods are tantamount to building a space craft to get to the stars.......I just want to make it fly a few feet off the ground etc.

So far the rotary table worm/worm wheel drive is the simplest, and many people are using them with mixed result from the system limitations of backlash etc.

Next comes the belt drive, and that too has limitations not only in the reduction available but also in the ability to resist deflection from the cutter forces.....belt spring....... but as noted the belts are now available that allow higher tensile forces to resist the belt pitch elongation.

I previously mentioned friction drive and got a short answer......no joking suggested on the score.

By observation I conclude that.......if you have a hardened roller, rolling on a hardened flat surface and the roller is subjected to a load to ensure it remains on the flat surface, then by turning the roller you will get a forward movement without the roller slipping on the surface.

This is the simplest form of friction drive, but is not relevant to the drive I theorise on as there are design factors that dictate simple surface contact as being totally impractical.

Simply explained, means if you have a cylinder of steel with a hardened and ground surface of 100mm diam and a hardened and ground roller of 20mm diam driving against the surface you will have a 1:5 reduction drive.

Not much force can be applied to the roller in that design, but if the roller is set a distance of 10mm away from the surface of the 100mm roller and two separated rollers are used to bridge the gap, one either side of the gap, you can force the two rollers into the gap and so get a great surface grip without undue slip, provided the drive and driven rollers are in a fixed position and can withstand the wedge effect loading of the two side rollers without deflection.

This is similar to the toothed belt drive where you have a fixed drive and driven toothed rollers and two jockey rollers mounted to deflect the belt to get tension.

My contention is that with the belt drive you will get belt stretch and so variable drive, whereas with friction drive you get a continuous motion without hesitation at the point of reversal when the load is applied, all within the co-efficient of the rollers friction grip.

I agree that you will get slip (minimal) during the drive, but that is the encoder's job to convince the servo motor to keep turning until the destination is reached.

My main concern is for the backlash characteristic that is the main point of the exercise.

PROVIDED.......the drive is not over powered by the work load, the friction forces will drive the rollers continuously and have an exact response to drive reversal without any hesitation at the point of change, which interpreted means no backlash.

It cannot have backlash if the drive is within it's loading capacity........that it can have positional migration is indisputable, but that is not the object of the exercise.

I am attempting to cure one ill at a time, and the conquest of the backlash battle in my opinion is solved.....for me at least.

Now I must get the information for the design to apply an encoder to the output spindle so that the servo motor can keep going until the encoder orders it to stop.

The forces of milling under CNC drive as applied to a 4th axis job will be small compared to a regular mill job in a vice etc, as it will be mostly with the high speed spindle and pointy cutters digging out small amounts of material.

However, the drive will also be called on to remove material by milling on a rotating piece of work, and this will place a strain on the drive mechanism where slip can take place, but as the drive is encoder dependent, the servo just keeps driving etc etc.....the heavy work will not be a positional problem and when the load reduces to the finishing cut the load will not be significant anyway.

The advantage of the friction drive 4th axis will be the ability to have rapid forward and reverse movements in a small area without losing ground at each reversal, in the same way that a ballscrew allows rapid reversals in the X & Y axis travels with regular milling where backlash is not tolerated at all.......driving an X axis ballscrew with a stepper motor mounted off the table end via a toothed belt is a compromise, whereas a direct drive is preferred.

I've looked at the various drives so far that are practical and within the scope of getting anywhere near being made, and bought off the shelf, and buy in custom made items are very far down the list due to the cost.
Ian.

[handlewanker]

Hi, BTW, I just wanted to add, one of the biggest useages of friction dive for the last 300 years has to be the locomotive.

The engine driver does not care if the wheels slip a small amount from London To Glasgow, he just keeps the throttle open until the station is reached.

Don't even suggest to me that friction drive is not a viable possibility.....without rail transport, the World would come to a standstill.

It's time to get the thinking caps on and apply the science of applied mechanics in the friction department.
Ian.

[RCaffin]

Hi Ian

two separated rollers are used to bridge the gap, one either side of the gap,

That design should work fine. Yes, a friction drive can slip a bit under shock loading, but with feedback that does not matter.
Ah - fully enclose the drive to keep all dirt out!

Cheers

[AK47]

Well, if you want theoritical zero backlash with no wear and no huge forces - it is nice problem to think about You are rigth about it.

I was only try to give practical view - does system really need to last xxxx hours continuous use under maximum load?

In my point normal industrial barands like nikken, tsudakomi and so on have working quality nc-rotary tables "as is" - one worm screw and brake.

I understand that making it yourself is what can give people lot of satisfaction.

I can manage with 0.01-0.025mm backlash (0.0005-0.001 inch).

How much difference there migth be in forces if you use two worm screw and zero gap in every moving/contacting surface, compared to one worm screw and spring load to get gap zeroed as i drawn?

What size machining we are talking about (tools, materials)?

And sorry about my english, it has "gap" too, aswell rotary table does

I have my own little rotary table project . Fitting manual "news -yamatokoki" 250mm (10 inch) rotary table to fanuc red head servo as 4th axis on small fanuc tape center... Like haas was doing several years ago fitting servo to manual rotary table. But not planing to do it as good as what is level of this topic.

Keep thinking, someone find out better and better solutions.

[RCaffin]

Hi AK47

if you want theoritical zero backlash with no wear and no huge forces

That's what this thread is about -how could it be done? We dream on.

Cheers

[louieatienza]

Hi, BTW, I just wanted to add, one of the biggest useages of friction dive for the last 300 years has to be the locomotive.

The engine driver does not care if the wheels slip a small amount from London To Glasgow, he just keeps the throttle open until the station is reached.

Don't even suggest to me that friction drive is not a viable possibility.....without rail transport, the World would come to a standstill.

It's time to get the thinking caps on and apply the science of applied mechanics in the friction department.
Ian.

You wouldn't care about a little slip if your locomotive weighed 200 tons! If it weighed 2 tons it wouldn't be able to haul a darn thing, no matter how much horsepower were applied to its wheels.

I think to do something as you envision would require expensive components and means of manufacturing anyways which pretty much defeats the purpose of this "mental" exercise. Plus there will be much overhead needed to overcome the excessive preload needed for this to work let alone correct for slippage, meaning larger, more expensive servo. If you're using an encoder to correct positioning based on slippage than it can be used to compensate for whatever perceptible stretch is with timing belts.

Have you seen the NuVinci CVT in action? It uses spheres between two disks to transmit rotation and the gear reduction is done by "tilting": the axes of the spheres, so that the input and output disk tough different radii on the spheres. A fluid that reacts like a "solid" is used to prevent slipping.

[handlewanker]

Hi, on the topic of the locomotive of 2 tons not moving anything, you must realise I come from a varied background of engineering application so when I mentioned a locomotive it was from actual experience with such a machine.

When I worked on the diamond mines in Namibia in the 50's we had small diesel powered locomotives that hauled 30 or more cocopans or mine cars loaded with gravel etc, and the hauler was a small loco half the size of a Mini or VW and weighed about 1 1/2 tons ........the driving wheels (4 of them) were approx. 400mm in diam.

I am examining the aspect of friction drive because it will give an instant response on reversal, which is zero backlash, my prime objective, and the load is dependent on the power of the mechanism to sustain it.

Friction is the microscopic hills and valleys of two surfaces interacting with one another like a rack and pinion, and if rack and pinion can drive a load so can an apparently dead smooth surface, or locomotives and trams would not exist.
Ian.

[handlewanker]

Hi Louie, I am not concerned with a variable drive at the moment, but the drive you mention has similar aspects to a single fixed reduction I mentioned as it is also friction dependent for it's motive force.

Anything that moves against anything will be retarded by the coefficient of friction of each object.

If variable speed drives can be obtained with enough power to drive a load then the principle is sound.
Ian.

[RCaffin]

I think to do something as you envision would require expensive components and means of manufacturing

Not quite true. The point of the intermediate roller is that it can move ever so slightly (I am talking maybe 1 micron) so that it wedges between the driving and driven rollers. Start with a bit of preload and the rotation of the driver does the rest. To handle reversing you have one on each side, both preloaded. This may mean that there is still one or two microns of deadband.

I saw a picture of a commercial unit which does this recently - or maybe a patent?

Cheers

[louieatienza]

Hi, on the topic of the locomotive of 2 tons not moving anything, you must realise I come from a varied background of engineering application so when I mentioned a locomotive it was from actual experience with such a machine.

When I worked on the diamond mines in Namibia in the 50's we had small diesel powered locomotives that hauled 30 or more cocopans or mine cars loaded with gravel etc, and the hauler was a small loco half the size of a Mini or VW and weighed about 1 1/2 tons ........the driving wheels (4 of them) were approx. 400mm in diam.

I am examining the aspect of friction drive because it will give an instant response on reversal, which is zero backlash, my prime objective, and the load is dependent on the power of the mechanism to sustain it.

Friction is the microscopic hills and valleys of two surfaces interacting with one another like a rack and pinion, and if rack and pinion can drive a load so can an apparently dead smooth surface, or locomotives and trams would not exist.
Ian.

Great story about the diamond mines. But I don't think there was ever a need to precisely control where that train started and stopped. Though I'm sure you COULD have controlled the train to start and stop exactly at the same points, I doubt its location and speed could be as accurately controlled. You really don't care if a train's wheels slide in a section and hook up in another as long as it gets you there. But that causes variations in speed. And then would such friction in such a drive you desribe be able to overcome cutting forces? You may say that any slip would be conpensated for by a servo/encoder, but my thinking is that this inconsistent movement of stick/slip can be problematic in terms of surface finish and accuracy.

Thanks for teaching me about friction, but I think I have somewhat of a grasp of friction. Once you get a little slippage in your friction drive, yo'll start to wear areas of your drive disk, and then you'll end up wth variable preload. I also am not sure at what speed would a reversal cause slippage, or what the upper limit of accereation would be for such a drive.

[handlewanker]

Hi, for a 4th axis.....speed does not come into it, relatively moderate fast moves back to the beginning are OK, and excessive drive forces are not on the cards either.

The main function of the 4th would be to rotate the work piece against a cutter of small dimensions accurately, and this does not enter into the realm of serious milling forces, so slippage on hardened drive surfaces will not be a problem, especially as a lubricant is present.

Citing the locomotive tractive force derived by tons of pressure acting on a very small area, usually with 4 or six or more driving wheels means that friction drive is practical and a very much used method of moving a body.

Bringing it down to the nano level of a 4th axis, would just mean tailoring the forces generated by the cutter to not overcome the forces holding the roller in contact with the drive wheel.

There is a limit of adhesion whereby the roller will skid on the driven wheel, but as that is beyond calculation for the exercise, it must just be judged by comparison to what could work and what has already worked, even if it is on a totally different scale.

In the case of the locomotive, each drive wheel contacts the rail on a very small footprint, and if you know your wheel tread profiles it is not across the full face of the rail surface.

So, if tons of load can be transmitted on a small footprint, then at the 4th axis level it is pro rata the equivalent.

The main gain is that it will transmit motion without backlash and at the same time exert a force for the drive that is not by any means insignificant.

You can overcome the force but why break a tool just because you want to overload it.

My contention is that, having analysed somewhat the effects of toothed belt drive and found some areas of need, the alternative friction drive is now for me a reasonable solution to a backlash free drive, but totally and wholly dependent on the ability of the servo motor and encoder to do the necessary positioning.

I would have liked to use the gear method, as it can give a large amount of reduction for little design problems, and chain drive falls into the same category.

If all else proves to be impractical for one reason or another, then belt drive can still be used without too much mind wrenching pain.

The thread is after all devoted to achieving a backlash free rotary table without compromise, and the 4th is in the same bed, so near enough is not good enough.....it's all the way or keep seeking.

As far as I'm concerned the friction drive does cover the need for a backlash free drive, and the capacity of the drive is just another goal to be scored, but the prime object has been achieved.....in my humble opinion of course.

I got quite excited when I rolled a ball race on a plate of steel and realised that the solution was staring me in the face all the time.

Now the hard part comes with the application of the principle to the design, it's all been done before even though the application was for a different purpose.

I think I could safely say that if the toothed belt were made in a metallic form, that is.... totally metallic with no elasticity or stretch, but like a chain drive that it emulates and having a constant pitch, that would be the ultimate, but I'm thinking of the same toothed profile on the drive face as the toothed resilient belt not as a chain with the common roller design driving a sprocket.

You can get an idea of what I'm thinking about if you have a jointed metal watch strap and you wrap it round your finger....it's so flexible.

That would mean more links but thinner ones and a smaller pitch with the teeth spaced at 3 or 4 chain pitches apart so making it very flexible and able to wrap around a pulley like the toothed belt, and the teeth would not bottom in the pulley tooth gap.
Ian.