Hello Guest it is March 02, 2021, 07:33:06 PM

Author Topic: Problem with Backlash...  (Read 11618 times)

0 Members and 1 Guest are viewing this topic.

Re: Problem with Backlash...
« Reply #10 on: November 10, 2010, 10:21:16 PM »
Thanks Rich,
But yes I already try to running at slower speed and the problem is always here because at my knowledge when the X make a backlash adjust the Y seem not use slope for stop and start .Also any speed I enter do not change the result...motor skip when restart...;-(((

Claude T.

Offline RICH

*
  • *
  •  7,419 7,419
    • View Profile
Re: Problem with Backlash...
« Reply #11 on: November 11, 2010, 04:50:48 AM »
Quote
use Machmill for this machine because with Machturn the drawing is not translate correct

Curious about the above. Can you post the dxf drawing.
What program are you using to generate the code?

RICH
Re: Problem with Backlash...
« Reply #12 on: November 12, 2010, 10:17:05 AM »
On other lathe I use Deskcnc before and now use Mach3 and the drawing are import from deskcnc and all work fine on other lathe but this lathe have no backlash and a very light carriage

Thanks for help Rich

Claude
Re: Problem with Backlash...
« Reply #13 on: November 18, 2010, 07:49:58 AM »
Claude,

I have exaclty the same issue with my mill. I'm using the latest lockdown vn...40. I am cutting a profile using X axis and Z axis with a round part in the A axis. The X axis moves along the part and Z goes up and down to give the right profile on the part. A then rotates half a degree for the next part of the profile and x moves back the other way.

I have 0.05mm backlash in X and Z that is not normally an issue but for this part it is, so I turned on backlash comp and set up each axis distance, shuttle time and velocity and all is fine with each axis.

BUT, as you say, when one axis is moving along smoothly (x) at 0.3M/min and then the other (z) changes direction the X axis suddenly stops to wait for the Z axis to complete its compensation, then suddenly X carries on at the same rate.

From what I have read in other threads, this sudden stop is possibly because the backlash comp is not within the Mach3 "trajectory planner" - it just stops motion, adds the compensation regarding the adjustments set and starts it up again. This is very hard for stepper motors as, with a heavy X axis, it will possibly/probably loose steps. There is no respect for the (de)acceleration or any other setting for the axis that is having to wait for the other axis compensation.

So, it is possible to adjust the way the compensated axis behaves (as Rich explains well) and to get that smooth and not loose steps but other axis are abruptly interupted. Luckily my mill is small and it just about copes with it, but it is painful to watch and I'm sure will loose steps under heavy load one day.

Rich, Hood, Brian, Art, may I ask if my understanding is correct or have I missed a magic setting somewhere? If correct, are there any plans to get backlash inside the trajectory planner so it can work with all axes smoothly and be more stepper friendly?

Or, any advise to improve the situation for the "not compensated" axis motion?

Thanks!

Woody.

Offline Jeff_Birt

*
  •  1,107 1,107
    • View Profile
    • Soigeneris
Re: Problem with Backlash...
« Reply #14 on: November 18, 2010, 09:35:03 AM »
When using backlash comp every axis has to stop when any axis is changing directions. If you don't do that then your just changing the position of the backlash error. Think about travel around a circle counter clockwise (anti-clockwise for any Brits) as you hit the 3 O'Clock position the X-axis will need to reverse, if Y does not stop you will have a small straight line while Y keeps moving and X is trying to 'catch up' by taking out the backlash.

open loop backlash comp is not a cure for a loose machine, it can help in some cases but will cause problems in others...
Happy machining , Jeff Birt
 
Re: Problem with Backlash...
« Reply #15 on: November 18, 2010, 04:29:27 PM »
Jeff,

Yes, completely agree, but there are different ways of stopping. ;)

It knows it has to stop the non-affected axis. It should ramp down to a stop, then the other axis gets compensated, then they both should start off with acceleration. Of course, this means CV is not possible, but that may be better than loosing steps?

At the moment Mach just bangs the unaffected axis to a stop, compensates the problem axis gently then and bangs the other back to the velocity again.

Unless I'm wrong, reading other threads it seems there is a solution being considered and I'm wondering if / when that will be implemented.

Woody.

Offline RICH

*
  • *
  •  7,419 7,419
    • View Profile
Re: Problem with Backlash...
« Reply #16 on: November 18, 2010, 05:52:50 PM »
Woody,
I have seen the "banging" you describe. My friend was showing me how screwed up backlash was ( he has about 1/8" on one axis )and i even posted  that there was a bug ( THERE WASN'T). He was runnning the axes at 100 IPM and doing reverse movements  with one of the axes during two axis moves, and yes, there was some banging going on, but no missed steps.
What do you expect to happen at those velocities......!
You can't really see what's happening at fast velocities other than the banging. If one would start with slow settings and gradualy increase them ( backlash settings and velocity ) you will find a point where the stopping of one axis and backlash implementation
creates  the situation. Since weight of the table comes into play you could start to loose steps because the motor is not capable
for the axis travel. ( there are programs that evaluate motor capabilites )
So what is one to do?
Well i would experiment some and find the operating parameters that are appriate for your running condition and use of backlash. I may add that if there was no backlash and you changed direction that quickly you still have forces acting on the system
and the "unheard banging" is pounding the crap out of your machine elsewhere.

RICH
Re: Problem with Backlash...
« Reply #17 on: November 19, 2010, 03:30:14 PM »
Rich,

Yea, its painful and as Mach is, unless the uncompensated axis goes really slowly, there is no way to avoid the sudden stop in that axis. I've tried to draw what I mean. Does this make sense to you?



1) without backlash, X axis almost constant speed, slows a bit for the Z axis to get up to speed
2) with Mach3 backlash, when Z changes direction, X suddenly stops whilst Z changes and does comp, the X suddenly starts again
3) how it could be, X predicts it has to stop (look ahead) so ramps down, Z does its thing, X ramps up again. All done together so exact profile is followed and no instant acels required.

Woody.

Offline RICH

*
  • *
  •  7,419 7,419
    • View Profile
Re: Problem with Backlash...
« Reply #18 on: November 19, 2010, 07:19:51 PM »
Quote
1) without backlash, X axis almost constant speed, slows a bit for the Z axis to get up to speed

Maybe, i can't really see it  when  jogging on my lathe???? but , not during a G0 or G01 move that's active and i am not sure if compensation takes place if you do a G02 or G03 move greater than 90 degrees as I would consider that  a continous move.  Would need to test since i don't recall.

Quote
2) with Mach3 backlash, when Z changes direction, X suddenly stops whilst Z changes and does comp, the X suddenly starts again

If it's an X & Z move the X is intended to stop at the same point  / time as Z, as soon as the backlash is implemented ( and how fast that is done depends on the backlash settings), X should accelerate  per your settings and if that next move is an X & Z then move rate  is governed by the slower of the axes.

Quote
3) how it could be, X predicts it has to stop (look ahead) so ramps down, Z does its thing, X ramps up again. All done together so exact profile is followed and no instant acels required.

Would still be the same motion as in 2) there is an interruption in movement although at higher speeds it may seem instantaneous.

The Maunual states " Mach3 is not able to fully honor the axis acceleration parameters when compensating for backlash. So stepper
systems will generaly have to be detuned to avoid risk of lost steps". Which brings me back to what i posted before.
"Well i would experiment some and find the operating parameters that are appropriate for your running condition and use of backlash."

Just a user like yourself, ;)
RICH

Re: Problem with Backlash...
« Reply #19 on: November 22, 2010, 11:51:53 AM »
Rich and Woodycam,

Thanks for good explication of my problem, it's exactly my problem.
It's possible find a software patch for ""fully honor the axis acceleration parameters when compensating for backlash""
At my sense why include this parameter (backlash) if not operate with other ?

Thanks
Cturc