Hello Guest it is March 18, 2024, 11:17:38 PM

Author Topic: Strange backlash compensation  (Read 4951 times)

0 Members and 1 Guest are viewing this topic.

Strange backlash compensation
« on: June 23, 2007, 10:42:03 AM »
I am running the latest official Mach3 on a fast Win-XP and a DSLD 3000 Micromill..
With the backlash option disabled and running a program or jogging it behaves as normal. 

In metric mode I set backlash to 0.1 and I tried federates from 1 to 100% with the same problem behavior. 
It is most obvious if I jog in fast (continuous) mode.

The table accelerates as normally for maybe 20 mm and then slows down to a crawl (presumably doing the
backlash comp) then it accelerates up to normal speed again and continuous fine.  Another jog in the same
direction works normally.  If I jog back, it does the slowdown to take care of the backlash and then subsequent
 jogs in that direction works fine.  If I disable the backlash option all is back to normal.

Should it not be doing the few backlash comp steps first and then go full speed to the next point. 
The same thing happens when a program is running.
What am I overlooking or what should I do differently?
TIA,
Bertho
Re: Strange backlash compensation
« Reply #1 on: June 23, 2007, 03:47:48 PM »
HI Bertho,
I tried backlash as well and had exactly the same result, I thought I had developed some electronic/mechanical problem.
I switched off since I didn't like the action and didn't need the accuracy.
It seems a little disconcerting in use, I wonder if it could not be a little cleverer and slow down impercepably  rather than the stop start routine.
I was only looking for 5thou correction which isn't much.

tmsmith
Re: Strange backlash compensation
« Reply #2 on: June 23, 2007, 04:59:37 PM »
I am at least happy to hear that I am not the only one with this problem.
The backlash compensation really should be completely transparent. 
If a move is in the same direction as the previously move, do not do anything.
If the direction changes, add the backlash distance to the next requested
distance and then do a single move to that location.

I do not understand why there is a special “backlash speed” option.
Bertho

Offline Chaoticone

*
  • *
  •  5,624 5,624
  • Precision Chaos
    • View Profile
Re: Strange backlash compensation
« Reply #3 on: June 23, 2007, 08:41:55 PM »
Hey Guys,
    Backlash is a problem for machines that are weak. Especially in CV mode. I know it doesn't sound like much but here is how it works. On a machine that is already tuned to the max, backlash has to move the axis in the opposite direction, a certain distance in a certain amount of time + the backlash amount in the same amount of time. Raising your accel. will help this I think. Also, what is the backlash speed set to?

Brett
;D If you could see the things I have in my head, you would be laughing too. ;D

My guard dog is not what you need to worry about!
Re: Strange backlash compensation
« Reply #4 on: June 23, 2007, 09:15:33 PM »
The problem is there regardless of my backlash speed setting.  It is very obvious and disturbing. 
I push continues jog and the axis accelerates up to speed and then it slows down to a crawl
where it presumably adds the backlash and then it accelerates up to speed again.

Sort of:  Oops! I forgot the backlash, let’s stop, and do the backlash comp before proceeding.
I fail to understand, which would not be the first time, why there should be a speed or other issue. 
If the next move is 10mm back, why not just add the 0.1mm backlash comp and make it a 10.1mm move.
Bertho


Offline Graham Waterworth

*
  • *
  •  2,667 2,667
  • Yorkshire Dales, England
    • View Profile
Re: Strange backlash compensation
« Reply #5 on: June 24, 2007, 05:41:24 AM »
How will mach know what distance you are going to jog?

Graham.
Without engineers the world stops
Re: Strange backlash compensation
« Reply #6 on: June 24, 2007, 08:02:15 AM »
Graham,
In the case of continuous jog mode you are correct that Mach3 does not know but the first
move will be at least more than the backlash comp so in my case it would automatically add 0.1mm
to that first move in the new direction without including that extra distance in the DRO calculations.

Another way would be to mathematically subtract the backlash distance from the DRO and then
just jog as normal.  The first 30 steps or so in my case would not move the table but they would
be counted by the DRO so 30 steps later, the DRO would be back to the starting value and the
backlash would be taken up.

Please remember that I am new to Mach3 and the Micromill so I am probably doing something
wrong in the configuration or otherwise but the backlash is clearly not working correctly on my
system and on tmsmith’s .  Also the backlash distance is very small so it ought to be completely
transparent; tmsmith was using 5/1000 and I am using 8/1000 (0.1mm).

The part that I do not understand is why there is a separate backlash speed.  If a G code directs
a movement in the opposite direction so backlash is required, I believe that the backlash move
should (or could) just be part of a normal move that is mathematically handled in the background
and then the normal acceleration and speed setting would handle that combined move.

By the way, I am extremely happy to have found Mach3 and the related products and the support
through from Art and Brian and the user community is fantastic.  I specially like the feature that I
can customize Mach3 to fit my needs and taste.  I have spent yesterday playing with LazyCAM
Pro but I save that for another post.
Have a nice Sunday!
Bertho

Offline Graham Waterworth

*
  • *
  •  2,667 2,667
  • Yorkshire Dales, England
    • View Profile
Re: Strange backlash compensation
« Reply #7 on: June 24, 2007, 08:10:37 AM »
OK, I hear what you are saying, how would you implement the compensation on a circle?

Graham.
Without engineers the world stops
Re: Strange backlash compensation
« Reply #8 on: June 24, 2007, 08:31:08 AM »
Would cutting a circle not also work automatically the same way? 
When one axis changes direction of the movement, for example X, Y is moving at its
highest feed rate while X is basically standing still.  Eventually, there will be a request
for a “microscopic” movement in the other direction of X and at that time the backlash
steps would be added as part of that microscopic move.
Bertho
Re: Strange backlash compensation
« Reply #9 on: June 25, 2007, 09:23:12 AM »
For backlash comp and A axis cutting I use version 1.9 anything beyond that is suspect.

Dan