Thanks Jeff.  I was using a dial indicator and moving back and forth before and 
after setting "BACKLASH = 0.0005" and found no difference.  I think I was homed 
but will verify next time I am at the machine and will also try your method to 
see what the hal pins believe.  Is the "+" and "-" optional?
-Tom

> On Oct 30, 2015, at 10:25 PM, Jeff Epler <[email protected]> wrote:
> 
> For the following testing, I used the 2.7 branch at ref
> v2.7.1-26-g21ae480 (26 commits past the most recent release, 2.7.1)
> 
> I tested backlash compensation by modifying configs/sim/axis/axis.ini to
> add backlash to the X axis:
> 
>   [AXIS_0]
>   ...
>   -BACKLASH = 0.000
>   +BACKLASH = 0.100
> 
> I started with that configuration, homed all axes, and went into MDI
> mode.
> 
> I approached X0 first from the positive direction, and then used halcmd
> to look at the relationship between the joint command and the motor
> command:
>   [MDI in axis]
>   g0 x1
>   g0 x0
> 
>   [in a terminal window]
>   $ halcmd show pin axis.0.motor-pos-cmd axis.0.joint-pos-cmd
>   Component Pins:
>   Owner   Type  Dir         Value  Name
>       23  float OUT             0  axis.0.joint-pos-cmd
>       23  float OUT       0.93955  axis.0.motor-pos-cmd ==> Xpos
> 
> And then I approached from the negative direction:
>   [MDI in axis]
>   g0 x-1
>   g0 x0
> 
>   [in a terminal window]
>   jepler@babs:~$ halcmd show pin axis.0.motor-pos-cmd axis.0.joint-pos-cmd
>   Component Pins:
>   Owner   Type  Dir         Value  Name
>       23  float OUT             0  axis.0.joint-pos-cmd
>       23  float OUT       1.03955  axis.0.motor-pos-cmd ==> Xpos
> 
> In this case, I see that the motor positions differ by .1 inch depending
> on the direction of approach.  So the result of my testing is that
> backlash compensation seems to operate as expected in at least one
> situation.
> 
> I got similar results when using incremental and continuous jogs -- the
> difference between joint-pos-cmd and motor-pos-cmd was either 1.03955 or
> 0.93955 depending on the direction of the last motion.
> 
> (The specific values 1.03955 and 0.93955 depend on the details of how
> the machine was homed and would differ from run to run at least on a
> real machine)
> 
> And now idle speculation as to why backlash compensation might be seen
> not to operate at all for some users: I believe backlash compensation is
> not activated until all axes are homed. So if you are using a
> configuration that permits MDI & AUTO mode without homing, you might see
> this happen (but this is just a memory of how linuxcnc used to work at
> some time in the past and may not be accurate information for 2.7)
> 
> Jeff
> 
> ------------------------------------------------------------------------------
> _______________________________________________
> Emc-users mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/emc-users

------------------------------------------------------------------------------
_______________________________________________
Emc-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-users

Reply via email to