[RCaffin]

Hi Ian

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

Like this?
Attachment 207020

From Belt Technologies, Inc. | Linear Drive
But welding the ends of the strip together - that's hard. Possible with electron beam welding...

Cheers

[louieatienza]

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.

If you're going to work on parts way smaller than the diamter of your rotary, use very small tooling, keep the tool in center of the rotary axis, and run the axis very slow setting acceleration very slow as well, then even with a belt you'd likely see zero backlash. My query to your proposed design is whether trading belt stretch for plate slippage is wothwhile.

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.

We can take any of the commonly used mechanisms for rotary tables today and make them completely zero-backlash under certain conditions. Since you've set parameters as to workpiece and cutter size, and speed, then that is compromise as well. I have a small C0 precision ground ballscrew in my parts stash which probably is the closest thing to absolute zero-backlash I'll ever hold; but even that can't have absolute zero backlash.

In fact I would postulate that any drive mechanism that utilizes friction cannot be absolutely zero backlash, since there will always be surfaces that slide, slip, or rub against each other. Luckily for most humans, we make parts to certain tolerances, no part is truly perfectly flat, nor round, but totally useful and functional.

[Eldon_Joh]

that steel belt idea is the way to do it in my opinion. either that or a #25 chain.

you could make that steel belt from galvanized pipe strap. spot weld it in exactly the right place then grind the metal down carefully, then anneal the entire belt in an oven. it will take a bit of practice to get the spot weld in exactly the right place, but if you make them too close, then you can stretch the metal across the joint to adjust it.

[handlewanker]

Hi, by a steel belt I mean actually a steel link belt, not a continuous band, but like a watch strap that has lots of small links with small pitches which make it a very flexible......the teeth, like the flexible toothed belt, would also be on the underside and pitched apart like the toothed belt.

If you have a continuous steel band you have a problem with the tooth drive geometry as the teeth of the pulley must contact and engage in holes in the band which is not a good interface.

So far I think the friction drive has the ability to drive without backlash, provided the load does not exceed the ability of the surfaces to remain driving and not slipping, it's just a matter of contact area and pressure.

It's horses for courses, if the drive is expected to perform hard work, use a worm/worm wheel or toothed belt and wear the compromise of belt stretch and backlash etc.

For a design I think you could use an old ballrace outer race with a diam about 100mm and attach this to a centre steel body, the outer race periphery forming the surface the drive roller will drive on.....both surfaces will be naturally hard so no extra machining, heat treating or grinding will be necessary.......and the drive roller will be the outer race of a 20mm ballrace also mounted on a solid steel drive axle with two more pinch rollers to apply the adhesion force.

That's just a simple explanation, but the actual design is different to this although the principle is the same.

With a drive like this you would only get a 1:5 reduction, and as the target is 1:20 two further drives coupled together in compound will do the work.......the final ratio doesn't really matter as we are not dealing in steps but sensitivity of the drive, it could be 1:26 or 1:24, it is only to get the servo motor turning at enough speed to allow it to give a fine enough drive for the encoder to interpret.

The encoder will have as many lines as the resolution calls for.

If the friction drive turns out to be too much of a fiddly task I might re-consider the gear drive, but a drive I must have, one way or the other.

With a gear drive you would use only a stepper motor as the drive is already exactly proportional to the input, with no slip to warrant an encoder at the end, and as the gears are close meshed, practically no backlash would be detected......my second choice and possible much simpler to do.
Ian.

[handlewanker]

BTW, I just remembered, anyone who has rolled 1" thick steel plates in a roller to form a cylinder will know that the drive roller and pinch roller exert enough drive force to roll the plate into a tube, and that is by friction drive alone.

All steel mills with billet to sheet forming processes do so with friction as the driving force.
Ian.

[Eldon_Joh]

I think you could use the outer race from a 180mm bearing as the driven roller, and drive it with a 10mm diameter roller directly mounted on the stepper shaft. let the stepper motor float and have dedicated bearings on both sides of the 10mm diameter roller to reduce friction and provide say, 100 pounds of force needed to get something on the order of 25 foot pounds of torque before slip. 18:1 reduction right there and that might be enough with a 425 oz-in stepper.

[louieatienza]

BTW, I just remembered, anyone who has rolled 1" thick steel plates in a roller to form a cylinder will know that the drive roller and pinch roller exert enough drive force to roll the plate into a tube, and that is by friction drive alone.

All steel mills with billet to sheet forming processes do so with friction as the driving force.
Ian.

I think that works by rolling, and tension and pressure keeps the metal on the rollers not friction, hence the name pinch roller. If there was friction tha would mean there is slip and if there was slip then you couldn't form a perfect cylinder. I thik any friction would come from the slight compression of the inner surface and expansion of the outer surface.

[handlewanker]

Hi, Louie, I don't think a roller could indent a steel plate over the distance that those pinch rollers operate on...it's pure surface pressure that grips the plate both sides, and that is how I intend to face the friction drive solution I'm working on now.

BTW, I have a pinch roller type steel rolls with 3 rolls, that I made, to roll 3mm copper sheet and that drives by pure friction alone.....it will roll 300mm wide X 3mm thick copper sheet to make a 125mm diam tube for a copper boiler.

Eldon JoH.....you must be on the same psychic wavelength as me.......that is exactly as I envisaged the drive.......a big ballrace with a smaller ballrace driving on the outside diam.....that is how I drew it at the start, and it just went on from there to develop and rationalise the layout.

As a development, I went a step further and had the big ballrace outer race only pressed onto a bigger steel spindle, and made the steel spindle with a recess and another smaller ballrace inner race pressed into the recess like a sleeve.

Now when the small ballrace is pressed against the outer rim there is another small ballrace pressing against the inner rim so that a pinch action takes place to resist any deflection caused by one way forces.

The outer small ballrace will have to be a solid type, because it is the driving force, and to do this I would make a spindle and just press the outer race of the small bearing onto it, so having the hard steel driving faces all working together.

I don't think you can go too small in diam with the driving wheel, but you can make it wider by having two small ballrace outer races driving against two big ballrace outer races so increasing the surface contact face......there would be two small ballraces on the inner diam recesses, one either side, as these are just pinch rollers and only provide the pinch force.

That is as far as I've got with the actual design, but I think with the diam of the big driven roller fixed at 100mm diam and the small drive roller at 20mm diam, that will give 1:5 reduction, and that is because the big wheel cannot be too big or it won't fit on the box I've drawn for the size of the 4th axis I want, and if the small drive roller is too small it will not drive......bigger would be better all round, an it might mean a 25mm diam drive to give only 1:4 reduction.......you can make a series of compounded drives in a very small envelope to give you I:20 which is my target figure.

With a further development, the drive roller will be spaced away from the face of the big driven roller as mall amount, and two more ballraces will act as wedge rollers either side of the drive roller and between the drive roller and the driven big roller, so increasing the drive force and area of contact.

Using stripped down ballrace components means that you get high quality steel material in a finished state with hardened and ground surfaces.

Once the prototype is up and running, the design can be rationalised, as it's quite simple in principle, by making the complete drive train from dedicated parts and having them all hardened and ground, but the drive needs to be proved for load capacity before any exotic work is anticipated.

The secret of the drive is in the pinch effect, without which the drive roller will just skid on the driven surface.
Ian.

[RCaffin]

Hi Ian

Intermediate wedge rollers on each side - about as good as it can get.

One problem though: if the table is to make more than one rotation, you will need to know the exact ratio between input and output. Measuring the diameters gets close, but may not be enough. You may need to include provision for a sensitive optical index marker: a flag interruptng a very small beam. These optical interruptors are a standard item costing only a few dollars.

Having an index pulse also allows you to check the system regularly.

Key point: absolutely scrupulously clean environment for the rollers to avoid any slip or hiccups.

Cheers
Roger

[handlewanker]

Hi Roger, The intermediate wedge rollers make direct contact with the outside of the wheel and will have to work with the internal pinch roller, much like holding a piece of paper between finger and thumb where the pressure exerted is direct as opposed to just pressing against the paper.

I anticipate that the complete assembly will be inside a case on the end of the main housing box which is absolutely necessary to keep the friction surface clean.

I was under the impression that with a servo and an encoder the ratio is not a concern, as the servo just keeps driving to get the line count......the lowest ratio would give the most accurate resolution.

Driving by friction makes any ratio a fairly random affair as you will get slip along the way, but the slip is a continuous progressive thing...... backlash is another story.

The main object is to eliminate the backlash at the point of reversal........distance travelled is when you get there...stop, and as far as I know that is the job of the encoder and the servo just drives blindly until told when to stop.

As long as the read head can read the lines of the encoder disc, you can have any number of turns before the stop point.

I think also there has to be a home point that you can go to in order to zero the table when you want to make some specific indentations along the axis as opposed to all the way around the diam.

The backlash part is at the point of reversal, not along the way.

If the drive hesitates at the point of reversal you have backlash, and that is where I'm at hopefully in the design to eliminate it.
Ian.

[louieatienza]

Hi Roger, The intermediate wedge rollers make direct contact with the outside of the wheel and will have to work with the internal pinch roller, much like holding a piece of paper between finger and thumb where the pressure exerted is direct as opposed to just pressing against the paper.

I anticipate that the complete assembly will be inside a case on the end of the main housing box which is absolutely necessary to keep the friction surface clean.

I was under the impression that with a servo and an encoder the ratio is not a concern, as the servo just keeps driving to get the line count......the lowest ratio would give the most accurate resolution.

Driving by friction makes any ratio a fairly random affair as you will get slip along the way, but the slip is a continuous progressive thing...... backlash is another story.

The main object is to eliminate the backlash at the point of reversal........distance travelled is when you get there...stop, and as far as I know that is the job of the encoder and the servo just drives blindly until told when to stop.

As long as the read head can read the lines of the encoder disc, you can have any number of turns before the stop point.

I think also there has to be a home point that you can go to in order to zero the table when you want to make some specific indentations along the axis as opposed to all the way around the diam.

The backlash part is at the point of reversal, not along the way.

If the drive hesitates at the point of reversal you have backlash, and that is where I'm at hopefully in the design to eliminate it.
Ian.

Ian,

The way I understand it is that the pure definition of "pure rolling" exludes any form of friction. The pressure applied to the mating surfaces caused deformation, even if macroscopically or microsopically, and that is where "friction" comes into play. And wherever there is friction, there is sliding, and thus the potential for backlash. I am not sophisticated enough to estimate what amount of pressure is needed to prevent slippage for a given torque applied, but I'm pretty sure this is a variable that would change with temperature, speed, and "cutting forces." For a "pure" system to have "zero" backlash would require both mating surfaces to have perfect fitment with no contaminants, and just enough "pressure" so as to not produce even the slightest elastic deformation. Unfortunately that probably would not be enough to overcome cutting forces or even the inertia of the load. In the worst case scenario, the pressureneeded to sufficiently couple the drive surfaces would cause the surfaces to harden to the point where they start to deteriorate and flake, which is not an uncommon occurence with overloaded bearing surfaces.

Like you mentioned it may be best to use the widest needle roller bearing outer sleeve that you can find. I may be apt to use non-hardened tool steel for the "pinion", to possibly maximize any elastic deformation and prevent slip.

There are quite a few linear motion systems that use friction for movement, such as some differential roller screws, "threadless leadscrews," and freewheeling ballscrews. They are not usually used for positioning without feedback because the effective "lead" can dependat on the load. While I am sure you can get accurate "psotioning" with encoder feedback, I am unsure how this would affect machining.

edit: As to the encoder disk, they are availabe with an index track to indicate zero or "home"....

[RCaffin]

Hi Louie

The way I understand it is that the pure definition of "pure rolling" exludes any form of friction. The pressure applied to the mating surfaces caused deformation, even if macroscopically or microsopically, and that is where "friction" comes into play. And wherever there is friction, there is sliding, and thus the potential for backlash.

(My bolding) Fortunately, the primary assumption here is wrong. What we might generally call 'friction' has two components, which we may term 'dynamic friction' and 'static friction'. The latter is more commonly known as 'stiction'.

Dynamic friction is basically the shearing forces between the two surfaces at the microscopic level. There are two components to this. The first is the straight mechanical deformation which happens during shear - a bit like shearing gear teeth off with too much torque. The other part is due to what we call 'Van der Walls' forces. These happen at the atomic level: surface atoms on one side actually making chemical bonds with surface atoms on the other side. The whole area constitutes a majort part of a rather hairy subject known as 'Tribophysics'. Anyhow, it takes energy to break these bonds once they have been made.

Stiction is a special case of the above, when the macro forces are not sufficient to break the Van der Waals forces. Under these conditions there is no movement between the two surfaces. This is what a friction drive aims for.

So what's with the need for high contact pressure? The higher the pressure, the more of the tiny lumps and bumps on each surface are flattened down, so that the contact area rises. The more contact area, the more Van der Waals forces available.

Incidentally, these same Van der Waals forces are what lets a gecko walk across a ceiling upside down. Great stuff! And, I suspect, part of the reason soft aluminium sticks to some cutters. The right lubrication aims to keep the surfaces apart - one molcule diameter is enough.

Bottom line: if the torque levels are not too high, there may be no slip at all in a well-designed friction drive.

Cheers
Roger

[louieatienza]

Hi Louie


(My bolding) Fortunately, the primary assumption here is wrong. What we might generally call 'friction' has two components, which we may term 'dynamic friction' and 'static friction'. The latter is more commonly known as 'stiction'.

Dynamic friction is basically the shearing forces between the two surfaces at the microscopic level. There are two components to this. The first is the straight mechanical deformation which happens during shear - a bit like shearing gear teeth off with too much torque. The other part is due to what we call 'Van der Walls' forces. These happen at the atomic level: surface atoms on one side actually making chemical bonds with surface atoms on the other side. The whole area constitutes a majort part of a rather hairy subject known as 'Tribophysics'. Anyhow, it takes energy to break these bonds once they have been made.

Stiction is a special case of the above, when the macro forces are not sufficient to break the Van der Waals forces. Under these conditions there is no movement between the two surfaces. This is what a friction drive aims for.

So what's with the need for high contact pressure? The higher the pressure, the more of the tiny lumps and bumps on each surface are flattened down, so that the contact area rises. The more contact area, the more Van der Waals forces available.

Incidentally, these same Van der Waals forces are what lets a gecko walk across a ceiling upside down. Great stuff! And, I suspect, part of the reason soft aluminium sticks to some cutters. The right lubrication aims to keep the surfaces apart - one molcule diameter is enough.

Bottom line: if the torque levels are not too high, there may be no slip at all in a well-designed friction drive.

Cheers
Roger

I know of Van der Waals forces in terms of atomic atttraction and but never knew it applied to friction? (Go easy on me, my physics classes occurred over 20 years ago.) I don't think we need to overcomplicate this. I'm sure the coefficient of friction of a hardened and ground steel surface can be measured, if it's not already known. And thus one could calculate how much force is needed to overcome this friction, or how much friction is needed to overcome a given force. Since the coefficent of friction for a hardened and ground steel surface is likely very low, if both surfaces are such then it would take a relatively large amount of pressure to create enough "friction" to overcome a relatively small force. This is especially true since there is not a whole lot of contact area (or footprint) to begin with, that is, unless a larger "pinion" is used. I don't think we need to calculate Van der Walls forces here to figure this out!

Bottom line: if the torque levels are not too high, there may be no slip at all in a well-designed friction drive.

If the torque is not too high, it's possible to have no slip in a rubber-band drive as well. I bet most any drive under the right condition can measure at or near zero backlash. But it's a bit hilarious to say that under the right condition, a well designed friction drive can exhibit zero backlash then say that a belt drive is inadequate since under certain conditions it can exhibit backlash. Or say in one sentence that a frition drive is fine since very light cuts will be used with the tool aligned with the rotary axis, and in another a belt drive is inadequate since there is a risk of belt stretch when a larger cutter is used in conjunction with climb milling?!

I would even postulate that for a system to be 100% non-backlash, it would have to be 100% efficient. There are always compromises. We can sacrifice some efficiency to gain a tigher system, but it comes at a cost of wear and periodic adjustment, as well as the need to larger motors. We cam overcome positional error due to stiction and slip, but we need expensive encoders and more powerful servos. We can get excellent resolution with steppers with larger gear reduction, but at a cost of slower speeds and the expense of the larger gear reduction.

I don't see any shame in using surplus "parts." Many of the stuff on eBay are from semiconductor, instrument, and other companies where these parts are "lifed out" for their use but perfectly useable (and many still within original manufacturer specs.) Some are from bankrupt companies (the onslaught of machines dismantled from the debaucle of solar companies like Solyndra)... It's a great time for individuals to take advantage.

I don't say this to discourage any experiments, ideas, or such. I just question the wisdom of preferring "friction" drive, which is pretty inefficient at power transmission, over belt drive, which is likely much higher eficient at power transmission, when the measured stiffness and backlash of each are likely similar for all pracical purposes? Just to say it can be done?

I remember early on someone mentioning using two worms and electronically preload them using servos. Seems pretty reasonable to me, as they already use this principle for high-end rack and pinion systems. I think for most of us here, timing pulleys and belts gets us most of the way there for a relatively small investment, and for light loads, probably as anti-backlash as one could get for the money.

[691175002]

I remember early on someone mentioning using two worms and electronically preload them using servos. Seems pretty reasonable to me, as they already use this principle for high-end rack and pinion systems. I think for most of us here, timing pulleys and belts gets us most of the way there for a relatively small investment, and for light loads, probably as anti-backlash as one could get for the money.

Preloading with a second servo will only work when the mechanism is backdrivable (and the secondary servo is used to apply a set torque against the driving servo). This is not possible with two worms.

You could, however, use one worm drive for the primary servo, and some form of reduction (belt/gear/direct drive) for the preload servo. Assuming the cutting forces never exceed the torque of the preload servo it will have zero backlash.

Even if the cutting forces overpower the preload servo it will simply revert to a standard worm-driven rotary table which is unlikely to be a problem. Operations that apply huge forces (drilling off-center, etc...) are unlikely to require zero backlash.

You should be able to hit reasonable speeds (for a worm, anyway) because the mesh between the worm and wheel doesn't need to be horrifically tight, and you can remove the preload when doing rapids.

[RCaffin]

I know of Van der Waals forces in terms of atomic atttraction and but never knew it applied to friction?

Oh yes indeed! Arcane subject, Tribophysics, but rather fundamental to our world. What do you think keeps a nut done up?

Since the coefficent of friction for a hardened and ground steel surface is likely very low

It is actually quite high. For clean hard steel on clean hard steel you would be looking at a figure of about 0.8. Add grease and it might be 0.1. This is why train wheels work, until British Rail gets wet leaves on the track ...

But it's a bit hilarious to say that under the right condition, a well designed friction drive can exhibit zero backlash then say that a belt drive is inadequate since under certain conditions it can exhibit backlash.

A belt drive will have backlash if the teeth are of the wrong shape - like on the older design (1st gen) timing belts. Just the same as backlash in a (any) gear train: the teeth on the belt have lots of clearance between the teeth on the pulley. And I am the one who was claiming that a toothed belt of the right design would have negligeable backlash.

I would even postulate that for a system to be 100% non-backlash, it would have to be 100% efficient.

Good point, and I believe it is close to true. There is some energy loss due to the very slight distortion of the steel wheels under load, but otherise, efficient!

but we need expensive encoders and more powerful servos.

That is one method. If you are willling to trade off speed, you can reduce 'costs' in other areas. Ingenuity!

I don't see any shame in using surplus "parts."

Seen my barn? I KNOW I will never use 2/3rds of it, but which 1/3rd will I need???

Cheers

[handlewanker]

Hi, entering the forum sword in hand.....LOL.......on the gritty topic of the belt drive, there is no final tightening of the belt that ensures it will remain at that condition, the tight belt will increase the pitch of the teeth variably and any further tightening by the need to overcome the resistance of the cutter forces will make the pitch so variable it will not give a true stepper count.

Against a load with a belt already tensioned, you have the condition that when the drive reverses the increased load generated by the cutter ceases, and you lose steps by the nature of the belt creep when it relaxes and then re tensions in the opposite direction.

At the point of reversal the forces change sides and the stepper has to add or subtract steps to compensate.

It is quite possible for a stepper motor in this condition to oscillate about a point a few steps with the tensioning and retensioning of the belt without actually moving the table forward or backward.......this is backlash......a light load will be hardly noticeable, if at all.

Within it's torque handling capacity, I think the friction drive has the capability to not lose directional status at the point of reversal, due to there being no flexible parts in the drive train......it's hardened steel on hardened steel, and when the drive reverses there being intimate contact with the two surfaces, there will be an instant response.....if it slips at the point of reversal it will slip all along the drive path, but as we are talking about "within it's torque handling capacity" that is a none event.

The higher the torque drain on the drive train the larger and heavier the drive components need to be, that is a fundamental law for any load carrying requirement, be it for the family shopping in the family car or a ton of sand in the same conveyance.

In the friction drive set-up, the need to apply the force by a pinch grip is of the utmost importance.

I cite the disc brake with twin opposing brake pads that if they did not apply a pinch grip to the brake disc would not be able to apply any significant braking force whatsoever.

In the friction drive even the best prepared surface will fail if the rollers and drivers are not in a pinch grip mode.

It's purely speculative at the moment as the drive, although pretty clear on paper, needs to be applied in the very best state of engineering expertise, IE good workmanship and surface preparation for the components.
Ian.

[RCaffin]

Hi Ian and others

For what it is worth, I will spell out what I am currently thinking would work.

I agree with Ian that a friction drive would likely work, up to a certain loading. My problem is that I do not know what that loading might be, and the possibility of a slip during a machining run worries me. On the other hand, a belt drive with the right belts would have very little backlash - small enough that a servo LOOP approach would be fine.

* I would use a mechanical box arrangement similar to what Ian and Katran have described, as far as the bearings go. Steel or aluminium is probably not critical at this stage. Line boring the holes for the bearings would be critical however.

* I am not sure whether I would bother with a tubular shaft: I have seen relatively few cases where a thru-chuck holding arrangement is really needed. That decision affects the bearing diameter of course.

* For front end bearings I would go for either twin opposing deep-groove bearings or twin opposing angular bearings, with a 3rd bearing at the rear to take the drive load.

* I would use a 25 mm wide 5 mm pitch GT3 belt running on steel/iron pulleys for the first reduction of about 5:1. These use a fibreglass tensile member in a Neoprene matrix, and wrap around the pulley in a non-chordal manner. Very little stretch! That means the rotation does not have a ripple superimposed on it like with a chain drive. Yes, a jockey wheel for tensioning would be needed.

* I would use a 15 (or 20) mm wide 5 mm pitch GT3 belt for the second reduction of again about 5:1. Yes, I suspect that I would need a two-stage reduction to match the motor holding torque to the table torque requirements.

* I would be happy to use a DC servo motor for the drive. They can have more grunt. (Happens I have a stack of Baldors and better on the shelf. That helps. But a large stepper is viable too.)

* I would use feedback from an encoder on the table shaft. This avoids all the hassles about slippage and backlash. Two possibilities exist:

# ~5:1 reduction using a 2 mm pitch GT3 toothed belt to a HEDs-style 500 or 1000 line encoder

# direct coupling from the main shaft to a 10,000 line optical or magnetic encoder

* Standard feedback servo drive in the Mach3 style.

I have been vague about the reduction ratios for a very good reason: the exact values do not matter! On the drive side, a DC motor won't even be aware of the ratio. On the output side the lines/rev in case 2 does not matter either because I would calibrate the whole system after it is built. Mach itself contains software for this calibration, but I could do it myself very easily.

The bearings, belts and pulleys are not all that expensive. A 500 line HEDs encoder is not expensive. A 'genuine Western machine-quality' 10k line encoder would cost under $400 - and would probably be worth it. After all, the quality of the sensor sets the quality of the result in a feedback system. But it is optional.

Well, there you have it. Fire away.

Cheers

[RCaffin]

And if you want to see what can be done, have a look at Boris.
Machining Delcam's 'Boris the Spider' on a Hermle C50U with PowerMILL - YouTube
Aluminium spider milled from a 700 mm x 145 mm disk by a 5-axis machine.


Cheers

[louieatienza]

Hi, entering the forum sword in hand.....LOL.......on the gritty topic of the belt drive, there is no final tightening of the belt that ensures it will remain at that condition, the tight belt will increase the pitch of the teeth variably and any further tightening by the need to overcome the resistance of the cutter forces will make the pitch so variable it will not give a true stepper count.

You live by the sword, you die by the sword. You can make this argument against ANY drive system. There's no known quantity as to the pressure needed to keep your rollers from slipping under working conditions. And any variation in pressure, or variation in surface, resistance from cutting forces will just as likely cause a variable drive ratio.

Against a load with a belt already tensioned, you have the condition that when the drive reverses the increased load generated by the cutter ceases, and you lose steps by the nature of the belt creep when it relaxes and then re tensions in the opposite direction. At the point of reversal the forces change sides and the stepper has to add or subtract steps to compensate.

Again, You can say the same thing about most any drive. And again, if these forces are lower than the rated specs of the components any creep/stretch is mitigated. If you order your belts from Misumi, their engineers will calculate for you the size and profile timing belt needed given your application, expected loads, etc. The right sized steper will NEVER lose steps. In the "friction" drive, as in any friction drive, encoder feedback is absolutely necessary for positioning. Consider a version of your drive proposal - the "threadless leadscrew" system from Uhing and Zero-Max (Roh-Lix). In their systems bearings are mounted on a bias in the leadnut, and make contact with a hardened ground shaft using only pressure. This produces very smooth movement, but will slide when their load capacity is exceeded and their effective lead is dependant on the pre-load of the system. Of course there are applications where slip is beneficial as it negates the need for a clutch or won't cause an overload situation which can damage equipment. Of course both companies recommend using encoder feedback when their systems are used for positioning.

Within it's torque handling capacity, I think the friction drive has the capability to not lose directional status at the point of reversal, due to there being no flexible parts in the drive train......it's hardened steel on hardened steel, and when the drive reverses there being intimate contact with the two surfaces, there will be an instant response.....if it slips at the point of reversal it will slip all along the drive path, but as we are talking about "within it's torque handling capacity" that is a none event.

Again, one could make the case for rpactically every well-designed drive system under the right conditionss. In my view, the fact that friction alone is relied upon for movement in my view precludes instant movement. It's the main reason why the dovetailed ways on a CNC mill are replaced with linear bearings. Even in that instance the linear bearings are always lubricated, because the bearing balls are not purely rolling on the tracks (and bump against each other in the case of non-caged linear bearings.) Without this lubrication the races would deteriorate.

It's purely speculative at the moment as the drive, although pretty clear on paper, needs to be applied in the very best state of engineering expertise, IE good workmanship and surface preparation for the components.

Once again, the same can be said for any drive mechanism. You can use better engineered timing belts (kevlar belts, steel reinforced) and pulleys (custom machined for zero clearance). You can use a high-precision planetary gearhead, which by the way, no matte how perfectly made, requires lubrication. Maybe even build a "hydrostatic" worm drive. The end game to me is what is done with your rotary once it's built. Does the backlash show up as out-of-tolerance features in a part? Does it cause you to run your machine at a ridiculously low speed in an effort to stay below the threshold of the system, especially considering the more robust linear axes (balance of performance between all components in a system)?

[louieatienza]

And if you want to see what can be done, have a look at Boris.
Machining Delcam's 'Boris the Spider' on a Hermle C50U with PowerMILL - YouTube
Aluminium spider milled from a 700 mm x 145 mm disk by a 5-axis machine.


Cheers

Nice... I wonder how much DelCam 5-axis costs. More than all my entire machines I suppose.