Backlash free rotary table

[RCaffin]

Hi Mactec

Yes, I have done my research on moly grease over the years. The problems I quoted can be found on the web, coming from vendors and other experts.

I mentioned the problems because it was suggested that Eros might be using moly on his gearbox. Fine on sliding surfaces, but contra-indicated for ball races. EP or lithium greases would be better there.

Yes, I have moly grease in the drawer, and graphite., and silicones, and ... Actually, I think the moly I have is from Rocol.

cheers
Roger

[handlewanker]

Wow......been off the thread for a while as the site did not email me on current posts as they occurred.

One thing I will say.......and that is purely my personal and biased opinion, and that is....if you delve into the world of complication and exotic materials with even more exotic methods of applying them for a less than predictable outcome, you are living in your own world of exoticism and for the ordinary person seeking an ordinary and achievable method for whatever purpose, which in this case is a viable and achievable method to have a backlashless rotary table....whatever....... the plot is definitely getting lost in translation.

The DIY harmonic drive manufacture is tantamount to going to the Moon in a garbage bin propelled by the power of a bundle of fire crackers when placed in the hands of the average or above average machinist to make with the machinery generally available at that level, and by that I mean a lathe and a mill without any grinding equipment you could shake a stick at.....the outcome is definitely not predictable.

If the outcome is not predictable, then the method is hit and miss at best.

I think I will stay with the spring loaded worm and worm wheel method as that to my mind seems to be the ideal and achievable simple solution that will produce results that are predictable.

How can that be so categorically stated.........well a worm wheel can be very accurately produced and the worn too and applying one against the other produces very accurate results as all dividing heads are witness to, despite the fact that all dividing heads and rotary table have backlash to some degree, and 4th axes or rotary tables are just brothers of the same relationship.......spring loading the worm would I think solve the backlash problem without breaking the bank.

It certainly is not a hit and miss solution with a dubious outcome solution that I proposed as some have indicated.......probably because the simplicity is not exotic enough.

This is like digging a grave with a gold plated shovel for a deceased person of high office, as opposed to a common spade you can buy at the local hardware store........it may not fit the occasion but the outcome is the same.

My hat's off to those that produce the harmonic drive and get it to work, even if not with 100% results, but for the record, it will not be something that will be common practice.
Ian.

[RCaffin]

Hi Ian

The DIY harmonic drive manufacture is tantamount to going to the Moon in a garbage bin propelled by the power of a bundle of fire crackers

Possibly so - but if you get there and back safely .... garbage cans become interesting.

If the outcome is not predictable, then the method is hit and miss at best.

Here I will take serious issue with you. If no-one ever tried something which was not totally predictable in advance, we would not have machine tools or even steam power. Trying something new is what has got us here. It is called Research and Development ... and for me it is FUN.

This is like digging a grave with a gold plated shovel for a deceased person of high office, as opposed to a common spade

But pollies use gold-plated spades all the time in their photo-ops.
And if you can get Tony Abbott into the hole, I'll even buy the spade for you. (Apologies to US readers: this is an Australian thing.)

Cheers
Roger

[daniellyall]

yep your PM is a bit of a nob bit better than the last LOL

[handlewanker]

Hi, the hit an miss thing is a true analogy of the inventor's tool kit or when you put a few blocks of steel together on a bench just to see if the "design" is as you envisage it......this is the same as doodling on the back of an envelope purely to see if the gem of an idea, that has been gnawing at your brain, would work.........until you actually make it, the outcome is not predictable.

Tell me that a Formula one race car will win the race it is designed to run in.....even with the best FEA and a ouijia board you can't do that.

I frequently delve into the hit and miss method, based purely on the knowledge that with past experience, learned at the coal face, the outcome will be as I envisage it.........mostly not quite so would be more correct, but in the absence of extensive R&D with FEA and other exotic methods, you give it a go with state of the art materials that you're familiar with and correct as you need.

Treading outside of your knowledge base does lead to unpredictable outcomes.

Even though the harmonic drive has been applauded as an ideal way to rotate a table with no or practically no backlash, the complexity of it's production relegates it to the too hard basket for most enthusiasts.

Previously I deplored the use of a worm dive for rotary motion without backlash, but with the realisation that a spring loaded worm could solve the problem and make it work, for pure simplicity, that is the design I would subscribe to.

Practically most, if not all, of the parts are off the shelf items and require minimal further machining, and provided the worm and worm wheel were not old and worn out, the outcome would be predictable.

Reductions in the realms of 1:100 are easily achievable with a worm and worm wheel drive, and if you unship the worm from engagement you can get high speed rotation to do CNC turning on a mill too.
Ian.

[erosnicolau]

Hey, gents,

Wow, a lot of new messages that I didn't get email alerts for So let me chip in again...

1. The grease I'm using is not (I believe) moly grease - simply because the box says "Unsoare grafitata - bazata pe calciu si grafit" (Graphite grease - based on calcium and graphite). Unless the labeling is wrong, this particular grease is molybdenum-free. But your replies really opened my eyes to a whole new world of considerations - those on the type of grease to use.

2. @Ian, thank you for your blunt honesty - it's both a chilling shower and a confirmation that I'm getting somewhere despite pretty bad odds. I'll give you a bit of background info on me, to give you a better idea of where I come from, what I learned in 2 years and why I do what I do:
- I'm no machinist, no engineer, no automation graduate, nothing related to this field - I'm a linguist turned graphic designer with a slope towards 3d, passionate about photography. This low backlash personal quest started off one year and a half ago, more as a personal ambition, while researching for a DIY portable time-lapse photography motion controller project, itself started off two and a half years ago. So, to me, hit and miss has been, by excellence and forced by my lack of an engineering background, the only (HARD) way of learning.
- In these two years I learned from scratch how to turn a lathe, a milling machine, a flat grinding machine - at least without breaking them, how to dress and balance a grinding stone, how to sharpen a drill bit or a lathe cutting tool at least to usable standards, how to program and operate a CNC router. It's been a steep curve, you might consider it unnecessary, but to me it's been more about the learning process than about a commercially predictable precision result.
- Where I didn't have have a conventional tool, I designed and built it myself - see the panto-grinder for precision-grinding the HD cams for my prototypes. I'm still learning the painful way (for instance, only last week did I finally understand that you CAN'T do any precision grinding without flow cooling) and I'm SO happy about all the discoveries I make on my own or by researching, asking around or reading from experienced guys like you.

About predictable results: after going through too many prototype iterations, I've learned that predictability also relies heavily on the right tools and expertise - so maybe the regular, life-long-experience machinist can relegate my hack attempts at the HD to the waste bin as being far too complex to justify billing time - but to me this isn't billing time, it's hobby time. And, I've learned that it's not a shame, but a privilege to outsource the finishing touches to people with both the tools and the experience. Keep in mind that I live in a rather technologically backward country, where asking anything below 0.02mm is met with loud laughter, so some of my trials and tribulations also relate to my ambition to overcome the lack of available precision work that can be sourced here, locally. Such an ambition was, for instance, to build myself an (off-topic) wire-race bearing. Everyone I asked, here, laughed in my face. But I did it (with adjustable preload, lightweight, strong and working to satisfactory results):



I hope this rather long reply doesn't sound too whiny or something - on the contrary, I'm rather proud of what I achieved so far, and this gives me confidence to push further.

Cheers,
Eros

[handlewanker]

Hey Eros, you are a definite achiever.......self teaching with the yen to be perfect has to be the mark of a craftsman.....that's no BS.......I spent 5 years learning my Fitting and Turning trade in the late 50's.......where you get to do only what the guy who pays you wants, so if you can progress from your non engineering beginnings at the rate you have, you'll be a genius before long.

If I came across as some say...arrogant and intolerant, put it down to my age as I don't get better as I ripen.....LOL.

The harmonic related posts are interesting as there is always a better or different way to do something, and as they say, the proof is in the pudding........the worm drive 4th axis I proposed and started building is still in the mixing bowl yet to stand trial.....more irons in the fire than I can handle at the moment.

My opinion of the harmonic drive is......like a ball race, if you want it to function at peak efficiency, you have to use peak efficiency methods, and that means highly sophisticated manufacturing methods......anything that falls short of that is sure to be second rate at best.

Nobody would dream of attempting to make his own deep groove or angular contact ball race with the methods available in even the most highly equipped DIY workshop.......the same applies even more so to the linear rail and bearing block, and the ball screw and nut etc etc.

On the topic of sending out items to be finished by firms with the right equipment.....that is how to achieve a better job if the price is within your budget.....nothing shameful about that....why reinvent the wheel with tons of machinery to do just a small job once.........hardening, grinding and wire cutting falls into this category......sometimes you have to do it yourself and invent or build the equipment when it's hard to come by commercially.
Ian.

[erosnicolau]

Hi, guys,

( @Ian - thank you for your encouraging words. I'm just your average guy, but with enough passion and commitment, I guess... )

Sorry for the delay, I've been caught up on various other stuff.

First, I did a backlash test at home, with the same setup (laser mounted on the output), at a 11.35m distance (the longest available space, from the study to the bedroom) with somewhat disheartening results: 11 arc-minutes. After some investigation, I found out that the servo wasn't actually tuned up, and it was behaving very badly, as illustrated in this incomplete 360*s motions test:



Today I managed to tune it up so that the servo errors now linger around 0.11 degrees - around 6% of a complete step. After the reduction, this translates to ~1.08 arc-minutes of motor-generated backlash. Much better:



With this confirmed, I set out for a second backlash test, and the average error at 11.35m is around 5.33mm, resulting in a backlash of around 1.61 arc-minutes. Extracting from there the motor-generated backlash, I am left with an incredible ~0.53 arc-minutes.



If you spot any flaw in my math / logic, please point it out...
Cheers,
Eros

P.S. I know the videos are as entertaining as watching flies get it on, and that the elegance of my fixtures is paramount - sorry about that, please ignore...

[RCaffin]

Hi Eros

The figures sound about right - similar to what I was getting with a second-hand (but in quite good condition imho) HD.
Now we need to know HOW! :-)

Cheers
Roger

[handlewanker]

Hi, I'd say that if you use a harmonic drive, when it's new and commercially made with the proper materials, hardened and ground rolling surfaces etc, then you probably have a good, very reduced backlash drive.....probably not quite 100%, but as near as dammit for all average needs.....provided you like the cost factor.

As far as I know, so correct me if I'm wrong.....harmonic drives have a cyclic motion that is not 100% accurate compared to the rotational divisional accuracy of a worm and worm wheel set-up, but that might be just an extremely small occurence in the design.

The crunch comes when you get one, pre loved......that has a smidgeon of backlash and cannot be adjusted to get the backlash out.

That is the realisation I subscribe to, and to that end is why I advocated a spring loaded worm and worm wheel set-up that is infinitely adjustable, and flexible enough for DIY manufacturing tolerances to not cost an arm and a leg.....even the basic components are off the shelf, and can be as high or low a reduction as you need.

Making a harmonic drive could be quite straight forward in the making once you know how, but to get it to a working state that remained working as needed is another thing.

For CNC work you need to have a drive that is 100% all the time.......the Holy Grail etc.
Ian.

[RCaffin]

Hi Ian

My test figures and graphs have been published here at CNCZone.

'Second hand' can mean different things of course. the one i bought seemed only lightly used.

I was not able to detect any cyclic wobble in the output myself (see graphs), but then again, I was not really looking for it. There may be some, but if so it is at a very low level imho.

Cheers
Roger

[handlewanker]

Hi Rog.....I'm probably thinking of one that has been well used and sold because it's starting to show wear etc..........brand new price off the shelf is probably beyond the average user for 4th axis hobby work as opposed to the popular belt drive that most 4th axes users and makers employ.

Going back to the belt drive design, I think that as long as there are no long sections of belt not actually on the drive or driven pulley, then you wouldn't get as much "spring" , if any, from such a design.

I doodled a design with a timing belt, and applied a pair of jockey wheels that pushed the middle of the belt around the pulleys on opposite sides more, and up between the two toothed pulleys so that practically all of the belt at one time was wrapped around the pulleys to the max, leaving just enough clearance between the belt top and bottom when the belt is under tension.

It does space the pulleys apart a bit more, but it also prevents having unsupported belt material between the pulleys out in the air, and is a means to apply tension to the belt......a bit more work instead of just forcing the two pulleys apart and clamping them,

I think something like this has already been designed on a rotary table where you have a large diam pulley as big as the table itself.

I totally agree that belt drive is neat and simple, but not able to give a high reduction unless compounded......reduction being as the user needs, but 1:10 seems to be the max that you could get.

You might not get backlash if the belts are tight enough to reduce springing, but you might get some lost motion with the fit of the belt teeth to the pulley teeth when doing reversals of rotation.

In that case I imagine that if a modified pulley tooth design is used that wedges the belt tooth profile on both sides to the pulley tooth profile..... purely for accuracy of movement as opposed to a drive means....... then lost motion on reversal might not happen as heavy drives causing belt stretch aren't a way of life for 4th axis work.

I don't believe a modified pulley tooth profile would infringe the patent for such an application, even if the belt tooth profile is patented and commercial timing pulley profiles that work with them too.
Ian.

[RCaffin]

Hi Ian

harmonic drives have a cyclic motion that is not 100% accurate compared to the rotational divisional accuracy of a worm and worm wheel set-up

A NASA paper from ~1991 did claim that there was a cyclic error, but it was clear that any such cyclic error would depend a lot on the exact tooth profile. They mentioned up to 0.008 degrees, which is just under 0.5 arc-minutes. One wonders whether any worm and wheel combo has been made to that accuracy. Theoretically possible of course, but could you afford it?

HD do not tell you what the profile for the teeth in a modern HD is exactly, but I would be inclined to suspect that in the 24 years since that NASA paper they might have made some progress in minimising any such error.

There is also a patent by Shimadzu Corp, 2004, 0090624 on correcting the input rotation to minimise the output error in a spectrophotometer. Great theory, but the experimental results looked a bit rough after a quick scan. Patent wars.

The crunch comes when you get one, pre loved......that has a smidgeon of backlash and cannot be adjusted to get the backlash out.

Oh, true, for sure. If there is backlash, forget it. You cannot correct it out - unlike the possibilities with a worm&wheel.
But, you know, a harmonic drive which was so worn out that it displayed visible backlash ... that is going to require a very long hard life. The one I bought did not show any sign of anything like that sort of mileage. As far as i could see, it still had the original grease.

My RT is working nicely these days. Photo in the thread devoted to my RT.

Cheers
Roger

[handlewanker]

Hi, just thinking about how difficult it is to machine the rotors and housings for a HD, all that machining and then hardening and grinding, and I came back to thinking that if the rotor and housing(s) were cut from pre-hardened steel blanks by wire cutting, then the profile would be extremely accurate and not need any further work on the rolling faces.

Being pre-hardened means when the rotors and matching housings are cut out, they are as accurate as the CNC process can be and are not affected by distortion from secondary hardening as would happen if they're machined from tool steel and then hardened and ground.

The CNC wire cutting process is a relatively cheap method compared to machining soft tool steel, hardening it and the grinding to size......the first is a one off operation the second a multi one......time and expertise being money etc.

If the rotor blanks are then mounted on spindles with Loctite, the DIY part is at assembly to the already finished housings and the only send out is the wire cutting, no secondary hardening or grinding needed.

I think if this path was chosen then it would be a viable build method as opposed to buying in ready made or even contemplating machining, hardening and grinding from tool steels.

The last firm I worked for had key profile gauges made from pre-hardened tool steel without any secondary finishing operations, it was that accurate.

There is a cost involved, but it's mainly the CNC wire cutting part, and if you have wire cutting facilities and CNC it's almost free.

If I had the CAD knowledge to produce the CAD drawings of the parts for wire cutting, then I think this would be the path I would choose.

BTW, wire cutting hardened steels works better than with soft steel, so that is a win win all the way to the bank......anyone see a problem with that method?
Ian.

[RCaffin]

Wire cutting - good idea but awfully slow.
Using formed carbon electrodes in the shape of the whole gear for the electro-erosion might be much much faster. Do the whole blank in one go.
Just a thought.

Cheers
Roger

[handlewanker]

Hi.......one thing that is of prime importance when you want to get to the end result with the most accuracy and that is.....minimal intermediate steps between the raw material and the finished object.

That's where CNC comes into it's own......and I think I could say without contradiction.... by using state of the art machining methods......I can't think of another method that supersedes a wire cut process where accuracy is paramount and cost is most effective.

Possibly nothing can beat a wire cut profile for accuracy, provided the machine it's cut on is also as accurate as you can get which is probably the easiest part to arrive at with the linear rails and ground ball screws now available.

We're able to move the axes in micron steps very accurately, so why corrupt the accuracy of the linear rails, ball screws and G code by producing a secondary less accurate thing to do the final cutting.

The electrode for the EDM sinker method would have to be cut by CNC methods anyway and then mounted very precisely if the concentricity and form accuracy of the rotor and housing is to be held to close tolerances.

Tolerance means it's only as good as you can get it, whereas with a wire cutting method under CNC control, the accuracy built into the wire cutting machine is the final outcome.....eliminating the Human contact takes out all of the tiny hit and miss factors.

I think a gear can be very precisely cut with CNC wire cutting and have a profile that doesn't need any further work......the surface finish, going by the results that I mentioned with wire cut profile gauges, would mean the surface finish is good to go right out of the box and hot from the press etc.

No final grinding or polishing needed to "correct" surface blemishes that you'd normally get if it was machined from soft tool steel, hardened and then ground.

That is the only way I would enter into a make it with hardened materials method.......going down the machinable plastics route is also a possibility as the parts would have a certain resilience to mate very closely without binding.

A HD, for a 4th axis or rotary table reduction solution, is lightly loaded and moves relatively slowly anyway, so wear would only occur in the hundreds of years category if ever.
Ian.

[dlang]

Here is some math and CNC info about gear profile... In short, it can't be described by two arcs
AWAY REALTY | ?????? ????????? ?????????? ????????????

I know this is an old post, but this domain seems to have gone away, does anyone have a copy of the info

[Brian_vz]

aystarik, I can't download the .pdf, the page a appears to me in blank.
I would like to read it.

[RCaffin]

URL for the PDF?

Cheers
Roger

[phennessey]

aystarik, I can't download the .pdf, the page a appears to me in blank.
I would like to read it.

URL for the PDF?

Cheers
Roger

http://web.archive.org/web/200501230...mata/2.139.pdf