Hello Guest it is March 28, 2024, 04:25:29 PM

Author Topic: Backlash...  (Read 11532 times)

0 Members and 1 Guest are viewing this topic.

Offline RICH

*
  • *
  •  7,427 7,427
    • View Profile
Re: Backlash...
« Reply #10 on: September 29, 2016, 04:31:01 PM »
Yes, the compensation can help.

Here is a link to some info worth looking at and play with the settings to see how things work for
your machine:

http://www.machsupport.com/forum/index.php/topic,29149.msg204717.html#msg204717

I will assume that you checked the backlash at different feedrates and different places along the axis travel.
So the "stable" numbers you mention reflect the above.
You may be able to optimise the adjustment of the table gibs and thrust bearings.... just note that you can't
have no backlash since the table won't move and as you take out play you also will use up motor torque to move
the axis.

For now,

RICH

Offline Davek0974

*
  •  2,606 2,606
    • View Profile
Re: Backlash...
« Reply #11 on: September 29, 2016, 04:40:05 PM »
Thanks,

no , i only used micro-jog to check the readings, i did check at 3 points on each axis though.

I'm not sure how much can be got rid of but i do know the thrust bearings need replacing, however the cost id holding that off for now.

On larger work, 0.1mm won't matter but on a 3mm milled hole its a big difference.

I'll read the linked thread.

Offline RICH

*
  • *
  •  7,427 7,427
    • View Profile
Re: Backlash...
« Reply #12 on: September 29, 2016, 04:59:55 PM »
For the time being can you clean / lube / and re pre-load the thrust bearings?
See how much they  contributor to backlash.........

Load up the bearings so movement requires some torque / hard to turn the screw ( one can call that zero play). Measure with torque wrench. then reduce pre-load by 1/2 the torque value and see what the difference the bearing contributes to the total  backlash.

Just a thought,

RICH

Offline Davek0974

*
  •  2,606 2,606
    • View Profile
Re: Backlash...
« Reply #13 on: September 29, 2016, 05:07:55 PM »
They were cleaned and repacked when i assembled it all, the original bearings are face to face and extra torque on the nut makes no difference, the new crappy ones i got with the new end plate were just bog standard bearings so i made some shims to enable me to tighten the nut solid. they were quite firm but not tight.

I would not want to pull them apart unless i had the new ones to go in really.

I will put the dial indicator on the screw-ends and measure end float - that can only be attributed to bearings. i have a feeling it will be probably 90% of my backlash figure.

Offline RICH

*
  • *
  •  7,427 7,427
    • View Profile
Re: Backlash...
« Reply #14 on: September 30, 2016, 11:01:49 AM »
I quess you have what you have and until you upgrade components BC can help improve accuracy.
The best approach is finding just what each component adds to the backlash.
One needs to isolate the component, and test each, adding the numbers should approx the total backlash.

Usual mechanical items that cause some degree of backlash:

- Motor coupling - anything other than hard coupled will add to the BC, it's just a matter of how much
                   torque is applied to it.
 
- Thrust bearings - Can approach zero play, but will never be "zero"

- Screw - lost motion can be anywhere, vary, and linearity different along the screw and may not be
          the same in the oposite direction

- Screw Nut - it is what it is

- Older Machine - One could fix / replace all of the above but conditon of the different machine surfaces
                  can affect accuracy.

Just some comments:

I hard couple the motors to the screw shaft. Requires good alignment, but, have found that over time the
bellows and rubber type can cause problems in repeatability of movement.

Thrust bearings are something that a lot of folks just don't pay attention to. They need to be preloaded
properly such that they minimize play with the least force / torque required for movement. I measure
the installation to confirm.

I have never used the screw mapping in Mach so don't know how well it works for a worn screw.
To profile a screw can be a PITA, rather easy on my end because I have the equipment to do it
opticaly and very accurately in rather short order. What would you do if you found that linear tolerance
spec is not over the length and is all in a small section of the length? OR Accuracy is different in the
travel directions?

For "hobby" use adjustable ball screw nuts are an option such that it is easy to minimise any play.
You pay for good quality screws and nuts. The accuracy of your machine is only as good as the components
used. For a lathe I recommend the best quality screw one can rationalize buying for the X axis ( the smaller
the more expensive they seem to be ).

Practicaly speaking on a retrofit, one can replace the axis components and find that the a machine surface
causes inaccuracies or even installation problems.  

Just some thoughts,

RICH

Offline Davek0974

*
  •  2,606 2,606
    • View Profile
Re: Backlash...
« Reply #15 on: September 30, 2016, 01:49:06 PM »
Yep, all noted, thanks.

I'm not chasing sub-microns here, at the end of the day she's an old Bridgeport with the stability of a wet lettuce leaf;)

I did not spend the most on mega-quality ball-screws - it just did not seem worth it as the ones i got were pretty costy and better C rating ones just went astronomic!

The thrust bearings are most definitely crap, I fully admit that and they will be pulled as soon as I have the cash for the ground and matched angular contact ones - the ones in it are just regular radial groove ones, I was pretty surprised how good it actually works and was expecting a lot worse TBH.

The belts could maybe be tighter as that would be another source of lost motion, not sure how tight is too tight on a 20mm HTD belt ?

Overall I'm very impressed with it, but need to get the 3mm holes to a better size as these are a part i make regularly on the mini-mill which i hope to sell-on soon.

I will add 0.1mm BC and try another test this weekend just to see what happens :)

Offline RICH

*
  • *
  •  7,427 7,427
    • View Profile
Re: Backlash...
« Reply #16 on: September 30, 2016, 02:52:14 PM »
Quote
belts could maybe be tighter as that would be another source of lost motion

Timing belts are surprisingly accurate....even when home made timing gears are used and driven axis movement may be off a little but it's repeatable / rather consistent per revolution.

Been a long time since we did testing of timing belts.
Basis was measured computer and driver output pulses as compared to varing belt tensions to see the affect on an encoder output connected to the axis side to see the differences between the three pulses. Difference were small values of steps per unit as they  relate to axis movement.
 

RICH

Offline Davek0974

*
  •  2,606 2,606
    • View Profile
Re: Backlash...
« Reply #17 on: October 30, 2016, 03:58:29 PM »
Replaced both sets of bearings on X & Y - got a great price from BearingBoys of £60 per pair of Nachi graded bearings.

Amazing difference, really quiet when doing rapids now, backlash is sub 0.1mm on each axis.

A little bit of compensation should get rid of that i think :)

Offline Davek0974

*
  •  2,606 2,606
    • View Profile
Re: Backlash...
« Reply #18 on: October 31, 2016, 04:15:34 PM »
Whats teething with backlash compensation and the shuttle acceleration?

Is there a recommended setting for servos and a CSMIO controller?

Currently have it on 50% acceleration and 0.001 shuttle accel, but the shuttle thing does not seem to make much difference - what should it do?

All values are sub 0.1mm so its just a tiny but sharp knock when changing direction on jog, n to tried any code yet.

Offline RICH

*
  • *
  •  7,427 7,427
    • View Profile
Re: Backlash...
« Reply #19 on: October 31, 2016, 08:12:33 PM »
Never used servos, only steppers with BC a long, long time ago.
BC at that time never worked with the SS and would think an external controller would need to support BC.

My experience is expressed in  my link posted in the other reply I made.
Reread the info and try starting very slowly and observe the motion.
The faster BC is implemented  you will get a knock / thump sound as the slack in the axis is taken up very rapidly. 

Sorry can't be of much more help,

RICH