Machsupport Forum

Mach Discussion => General Mach Discussion => Topic started by: Greolt on March 08, 2008, 06:16:15 PM

Title: Backlash compensation for a rotary axis ?
Post by: Greolt on March 08, 2008, 06:16:15 PM

I've always been of the opinion that backlash is something you deal with via mechanical design and not software.

However I have a worm drive gearbox that I would like to use as the basis for a rotary axis.

It has a very small amount of backlash which I can not eliminate by mechanical means.

I am interested in any ones experiences in a similar situation using backlash compensation in Mach.

The main driven shaft which would become the rotary axis spindle is quite stiff and for engraving
on a cylinder and other light jobs I think it would be OK.  It won't slop back and forth.

But I still have to account for the small amount of movement in the driven shaft before the main shaft will change direction.

Anyone with an informed opinion?

Greg
Title: Re: Backlash compensation for a rotary axis ?
Post by: Overloaded on March 08, 2008, 07:36:05 PM
Hello Greg,
How bout' semi informed ?
I had a similar setup once but it was not motorized. My gearbox had dual output shafts. I mounted a chuck to one end, and a 6" dia. pulley to the other and a handwheel to the input. (would have preferred a motor)
Then I attached a 1/16" aircraft cable to the pulley and wrapped it around a couple of times, then through a small winch pulley to a hanging weight sufficient enough to keep the driven gear permanently against the worm regardless of the direction of feed.
It was sort of crude, but it eliminated ANY possible backlash and worked perfectly for light cutting.
I'm sure there is a software solution too. There's been quite a bit of discussion about that here on the forum.
Just thought I'd share a "Backwoods" technique.
RC
Title: Re: Backlash compensation for a rotary axis ?
Post by: Greolt on March 08, 2008, 08:04:31 PM
G'day RC

I get what your saying with loading the spindle in one direction.

In this case I would not go that far. 

Unless I see a fairly good prospect of this working well I will go off in some other direction.

I suspect that some of the drawbacks with Mach3's backlash compensation would not apply in this limited situation.

However I would really like to hear some feedback from those who have tried it.

Greg
Title: Re: Backlash compensation for a rotary axis ?
Post by: vmax549 on March 08, 2008, 10:00:42 PM
Greg you still may be able to tighten up the worm. Does the main axle have bearing caps? If so you can shim the plates into the worm decreasing the backlash. I have seen it done.

 The problem with backlash in the worm is there is little resistance to movement on the spindle and it can cause the axis to jitter from the loose play when you are machining inline of the axis and cross the centerline.

(;-) TP
Title: Re: Backlash compensation for a rotary axis ?
Post by: Chip on March 08, 2008, 10:16:44 PM
Hi, Greg

I setup an  A axis rotational with backlash enabled, Using a sherline rot. table that doesn't have much backlash.

Simulated it with 1/2 handle turn of backlash, Motor tuning setup for Deg moves.

Works Fine Hear.

Just as TP stated, Caution, If it's as you stated, "final output is tight" Engraving should work ok.

Hope this Helps, Chip

   
Title: Re: Backlash compensation for a rotary axis ?
Post by: Greolt on March 09, 2008, 01:10:54 AM
TP

I wish I could. 

This is a small industrial reduction box and was not designed for precision application.  Final drive is fairly stiff.

Chip

Thanks for trying that. I suspected compensation could work OK in this situation.

I might go ahead and turn up the input adapter and a spindle shaft and give it a "Temp setup" try.

A couple of wood turning chucks that I have should work well.

Greg
Title: Re: Backlash compensation for a rotary axis ?
Post by: jimpinder on March 09, 2008, 03:53:55 AM
I cannot understand your problem with backlash. Mach3 deals with backlash perfectly. I have a cheap lathe which has quite a lot of backlash. This summer I will deal with it, and probably put in ball screws - but until then Mach 3 deals with it and it is accurate to the thousanth of an inch.

You would be able to, I am sure, deal with backlash on a manual machine, I think we have all done that at one time or another - on Mach3 it is just the same, but the computer does it for you.

All backlash is is the non-movement of the "table" whilst the gears, belts etc etc settle down to pulling the other way.

On any system, to check it - use typed gCode commands, switch off backlash compensation -  move one way - stop, set measure to 0 - move same way for some distance ( the distance does not matter,because backlash has nothing to do with distance, but you can say move one inch and check the table is moving accurately) - stop - reverse back to start.

Your measure should now read 0 - but it will not - this is backlash. Enter it in the table and switch on backlash compensation.

It is simple and accurate and all mechanical systems must have it, otherwise they would seize up - it is only when it gets big enough to notice it becomes a problem.

The only problem I can see with a rotary table is measuring the backlash - I have things to measure to a thou, but nothing that will measure to a degreee (accurately). I suppose you could convert to thous and then convert back - it depends on how you set up the axis.



Title: Re: Backlash compensation for a rotary axis ?
Post by: Greolt on March 09, 2008, 04:07:01 AM
Thanks Jimpinder

I know what backlash is.

I would not consider Mach's software compensation a viable option for a linear axis.

But that's just my opinion.  Each to his own hey.  :)

Greg
Title: Re: Backlash compensation for a rotary axis ?
Post by: Hood on March 09, 2008, 04:28:14 AM
Greg
 I have never used Machs Backlash because I have never needed to but I have heard there were big improvements made to it a  while back. Have you tried it recently or are you just going on what it was like a year or so ago?
 Hood
Title: Re: Backlash compensation for a rotary axis ?
Post by: Greolt on March 09, 2008, 04:51:29 AM
No mine is just a philosophical viewpoint.

I am not really anti  Mach's compensation, just software compensation in general. (Although I have read a lot of users hassling with it)

In the type of CNC machines which I am interested in I believe mechanical design is the place to tackle backlash.

Just my opinion, not something I want to debate.  ;D ;D

Having said that, on this gearbox, mechanical measures are not practical so I might try it. :P

Maybe I will be so impressed as to change my opinion.  ;D ;D

Greg
Title: Re: Backlash compensation for a rotary axis ?
Post by: Hood on March 09, 2008, 05:18:47 AM
I whole heartedly agree with you and so far I have been lucky enough not to have to resort to software compensation.
 There are some settings I think you have to mess around with, just in case you havent seen them look here http://www.machsupport.com/forum/index.php/topic,5523.msg38212.html#msg38212
Hood
Title: Re: Backlash compensation for a rotary axis ?
Post by: docltf on March 09, 2008, 10:22:12 AM
greolt

      go ahead and build it .i believe you will love it.will it have double chucks,front and rear.make sure when you do the rear chuck that you engineer a space for a drag rotor in case you
      have to install one later.how large is the thru bore, is it over 1.5 ,if it is when you are finished messin with it send it me.

bill
Title: Re: Backlash compensation for a rotary axis ?
Post by: vmax549 on March 09, 2008, 11:21:06 AM
Jim normal backlash on the linear axis can be compensated for because when you apply the load of the cut it loads up the axis and stacks the backlash up in the cut direction.

But on a rotary when you are cutting on the centerline of a cylinder that cutting load is not constant as the bit works the radius of the cylinder and it can cause the axis spindle backlash to work back and forth as the cutter moves across the centerline and changes the load direction of the bit.

Greg you could always machine in two concentric bearing holders so you could easiely adjust the backlash??? Just like the big boys do.(;-) Or offset bore the case and shim the bearings over.

Just a thought.  (;-) TP
Title: Re: Backlash compensation for a rotary axis ?
Post by: Greolt on March 10, 2008, 02:21:14 AM

Bill

I went ahead and made a spindle for it.

1" - 10 tpi  which is the size of my wood lathe spindle so my existing chucks etc will fit.

Only one ended though.  I will build a tail stock to suit if it works out.

I have not measured it yet but I think the backlash is going to be a lot better than I thought.


TP

I would not go to so much effort as to rebuild the bearing housings etc. 

For that much work I would go in a different direction. (Harmonic drive)

I'm definitely  not one of the big boys.  Just a carpenter.  ;D


Hood

I have not delved into the backlash compensation settings yet but I will get to that.  :)

Greg

.
Title: Re: Backlash compensation for a rotary axis ?
Post by: jimpinder on March 10, 2008, 08:24:54 AM
Greg - I'm sorry about the diatribe for backlash - and I am quite sure you understand it. I just get carried away sometimes.

Do I understand that this circular table, using the angular axis, will move to a new position, and then stand still whilst the other axis go to work on it.

If so - I know this sounds crude - but I would put a parking brake on it.

There are several options - a belt round the table that can wind on like a garrot, an electric disc brake that might be better because you could control it through software, or my son ( who weighs 22 stone and loves going down mountains on his bike) puts his faith in bicycle brakes - the latest disc brakes are awsome in their power - I now use them on miniature railway engines. They are compact and easy to fit and have hydraulic or cable operation. Or, in the picture you have posted, a drilled and taped hole for a screw onto the main shaft of the gearbox.
Title: Re: Backlash compensation for a rotary axis ?
Post by: Greolt on March 10, 2008, 09:23:47 AM
Jimpinder

"Do I understand that this circular table, using the angular axis, will move to a new position, and then stand still whilst the other axis go to work on it."

That is not how I envisage using it. I'm thinking that it will rotate as one horizontal axis (X or Y) and one vertical axis (Z) move about.

However I will keep that bicycle idea in mind. :)

Greg

Title: Re: Backlash compensation for a rotary axis ?
Post by: docltf on March 10, 2008, 02:32:08 PM
Greolt
       i see you decided to fill the bore up with an adapter shaft rather than bore that keyway out and adapt from the outside.im sad now!


bill
Title: Re: Backlash compensation for a rotary axis ?
Post by: jimpinder on March 10, 2008, 03:41:26 PM
Greg -

In that case, get the thing built, with as little backlash in it as you can - but then, I think, you will have to use mach3 backlash compensation. It should be OK because the holding force will always be in the direction of the cut.
Title: Re: Backlash compensation for a rotary axis ?
Post by: rustyolddog on March 13, 2008, 11:13:30 AM
My .02

Even a rotary table that has adjustable backlash will still have backlash. Software compensation will work fine as long as you are dealing with degree movements. If you convert your rotary axis to act as a linear axis, then you'll have issues since the backlash will vary for each radius that you are trying to compensate for.

All that being said, your best solution esp. since that gearbox doesn't have a way to adjust the backlash is to adapt an electric clutch or brake on the output shaft and then set up Mach to disengage the clutch when you rotate. I think this can be done with an output on your controller and a relay or SSAC and perhaps in your post setup. The second best option is to rig a constant brake that puts drag on the system.  The only problem is if you exceed the brake capacity in a heavy cut, it will move on you and cause your cut to be out of tol. or break tools etc.

If this is something you might consider, look for a lawnmower or automotive electric clutch. Surplus Center is a good source.
Title: Re: Backlash compensation for a rotary axis ?
Post by: Greolt on March 29, 2008, 05:37:32 AM

I finally got some shed time today so I knocked up a motor mount for the little stepper I temporarily pinched of another machine.

Keen to give it a go I clamped it very dodgy like to the table and "What do you know....It goes!!!!!"

Had to dial in 0.225 degree back lash compensation in Mach3. That amounts to about 0.1mm or 0.004" backlash at 50mm diameter.

Too much backlash to live with but the compensation seems to work OK so far.

I've never done rotary axis stuff before so for a test I used a great little program that bloke called Andrew wrote.

See this link, http://vectric.com/forum/viewtopic.php?t=2920

It takes a gcode file that you have created in 2D and converts it to a rotary axis. Works great.

Anyway as you can see in the pics it all sort of worked as expected. Not bad for a first try.

So I think on the strength of that first try the gearbox should be OK and I will go ahead and set it up properly and make a tail stock for it.

I also need to make another screen set to suit rotary axis work.  Maybe this one will be in flash.  ;D

Greg

.
Title: Re: Backlash compensation for a rotary axis ?
Post by: jimpinder on March 29, 2008, 06:21:48 AM
Nice job, Greg.

I can see where you are going now.

As far as backlash is concerned, I don't know if Andrews program keeps the axis as a rotary axis or converts it to a linear one.

If it keeps it as a rotary axis, then I cannot see you having any trouble with backlash compensation, since it clearly auto compensates for the diameter you are working at. The "thrust" of the tool will be taken by the gearbox, and if it reverses to "write" the other way, backlash comes on first and takes up the movement.

I wish I could get down to 4 thou of backlash - I dare not tell you how much my lathe has ( I am working on it, with new ball type leadscrews on the horizon) but I find Mach 3 compensation both accurate and reliable (although to listen to it working is sometimes a bit concerting. I was cutting brass handrails - a pole with a ball at the top and centre and half ball at the bottom - something you cold not do manually - as the z axis came to the ball, the x axis had to come out and back, describing the arc. When it changed direction at the apex, you could not see any distortion in the brass. This, I thought, was excellent.

Best of luck.
Title: Re: Backlash compensation for a rotary axis ?
Post by: Chaoticone on March 29, 2008, 06:36:23 AM
Good job Greg. I like it. I want to build a 4th axis soon (relative term you know :) ) for the Hurco.

Brett
Title: Re: Backlash compensation for a rotary axis ?
Post by: Greolt on March 29, 2008, 07:27:37 PM


As far as backlash is concerned, I don't know if Andrews program keeps the axis as a rotary axis or converts it to a linear one.



Yes Andrew's little conversion program uses degrees.  It does not simply convert to  linear code.  So as you say the backlash setting stays constant at different diameters.

In my one test so far I set Backlash Speed at 100% and Shuttle Accel. at 0.01 seconds.  I have no idea why it is called "Shuttle Wheel Setting"

Greg
Title: Re: Backlash compensation for a rotary axis ?
Post by: Greolt on April 01, 2008, 07:01:46 PM
Just for fun I made a little video.  ;D

http://www.youtube.com/watch?v=Kd2jyn8l0I4

I don't know how to make it show on this post like others do.
Title: Re: Backlash compensation for a rotary axis ?
Post by: Greolt on February 23, 2009, 04:19:32 AM
Finally got off my butt and made a mount and tailstock for this 4th axis.   :D

Greg
Title: Re: Backlash compensation for a rotary axis ?
Post by: RICH on February 23, 2009, 06:21:56 AM
Greg,
Seems like a year has passed by. Your doing good. Got a few here that i have even forgot
why the project was  started in the first place!  ;)
RICH
Title: Re: Backlash compensation for a rotary axis ?
Post by: olf20 on February 23, 2009, 07:37:25 AM
Very nice! I'm also working on one. Slow but sure.
What bit did you use to cut your work??
Thanks olf20
Title: Re: Backlash compensation for a rotary axis ?
Post by: docltf on February 23, 2009, 07:46:47 AM
greg
what program did you use to cut that wooden figure.

thanks bill
Title: Re: Backlash compensation for a rotary axis ?
Post by: Greolt on February 25, 2009, 06:44:33 AM
Rich

Yeah I usually get things finished eventually.  Sometimes takes a while.  :)

Olf20

1/4" and 1/8" ballnose endmills.

Bill

Therein lies my problem.  I don't have CAM software to do these sort of toolpaths.

A friend provided this one.  He does have some industrial CAM.  I can't afford it.  :(

All I can do so far is convert 2D toolpaths to 4th axis with a Wrapper program.  A bit limited.

Greg
Title: Re: Backlash compensation for a rotary axis ?
Post by: docltf on February 25, 2009, 08:08:14 AM
Greg

I would like to have a copy of that figure you were cutting and give it a go.could you post the g-code.

thanks bill
Title: Re: Backlash compensation for a rotary axis ?
Post by: Greolt on September 04, 2009, 07:37:30 AM
I have been posting some 4th axis stuff on the Vectric forum to encourage users to consider the possibilities of adapting Aspire to a 4th axis.

Anyway I made this little youtube video.  ;D

Greg

http://www.youtube.com/v/yWFSdogFx_g&hl=en&fs=1&
Title: Re: Backlash compensation for a rotary axis ?
Post by: RICH on September 04, 2009, 08:15:49 AM
Greg,
Not sure if you saw the posting by someone selling software for doing work on the 4th axis.
May be worth looking into. 

RICH
Title: Re: Backlash compensation for a rotary axis ?
Post by: RICH on September 04, 2009, 08:25:21 AM
Sorrry about that Greg
 I found the posting and  see you looked at it.
"New Wizards For Indexers"

RICH
Title: Re: Backlash compensation for a rotary axis ?
Post by: Greolt on September 04, 2009, 08:51:01 AM
Rich

Yeah there are a few utilities for doing bits and pieces on the cheap with a 4th axis.

The Indexer Wizards

Virtual Indexer Tool, for shopbot

D2nc

and my favourite, Wrapper, which is free.

Greg
Title: Re: Backlash compensation for a rotary axis ?
Post by: Greolt on October 19, 2009, 12:38:23 AM
I had ago at making lithophanes on a rotary axis.

Great little utility program this CNCWrapper, best twenty bucks I ever spent,   http://www.cncwrapper.com/

As i sometimes do, I made a Youtube video.   ::)

My partner saw this video and laughingly said "You sound really daggy!"

That's Australian slang, and most of you may not have heard that expression.

So to hear what it means to be a DAG or DAGGY, then watch the video.  ;D

Greg

http://www.youtube.com/watch?v=dblNnF4JJGk&hl=en&fs=1&