Richard,
Everyone seems to be addressing the question as asked without
considering other possible sources of error. If your encoders are
installed on the motor shaft, then that is where you should be
looking to see if the Rutex stopped in the correct location.
Mechanical backlash can cause false assumptions about the
results of commands from the computer and/or the PID loop.
Rutex drives close the PID loop within the drive, and the normal
setup does not bring the encoder feedback to the computer. In
this case the PID tuning is done in the Rutex drive. The PID
parameters on the computer affect the step pulse generator
only. While this can cause problems, it is a separate issue from
the PID in the Rutex drive that addresses machine dynamics.
Tuning Rutex drives on an EMC system is somewhat painful
because you must use a Rutex utility program that, as best
I know, is not available in a from that will run under Linux, and
also requires a cable modification to get the drives into SPI
tuning mode. Most users do this by having a separate computer
running Windows and temporary cable set up for tuning.
For what it is worth some quick definitions of terminology:
Backlash - Mechanical slop that causes apparent lost motion
when the direction of travel reverses.
Following error - Deviation from the commanded path. This refers
to the actual commands from the trajectory planner,
not the G-code. Acceleration limits are considered by
the planner and may result in the commanded path
not matching the G-code exactly while in motion
with constant velocity mode enabled, but should not
produce static errors at the end of motion travel. It is
normal to call the dynamic motion error a "following
error", but this term is not typically used to describe
errors that persist after motion stops. I guess one
could call those "travel errors".
P term - Proportional term of PID. Essentially a gain term for the
servo loop. Higher gain will result in tighter following,
stiffer control, but may result in overshoot or oscillation.
I term - Integral term of PID. Corrects for long term error. If P term
cannot be set high enough to force the end of travel to
match the command, then the error is continuously
"integrated" until there is sufficient evidence accumulates
that further motion is needed.
D term - Derivative term of PID. Seeks to compensate for the dynamic
effects such as the acceleration being greater or less than
the previous commands had requested. It helps to calm down
the system and prevent oscillation.
FF terms - Feed Forward terms. Not strictly part of PID, but often used
with it. Projects the need for correction before an error
even
happens. Computed directly from changes in command
without looking at feedback. Can dramatically improve servo
performance when applied to mechanical systems.
Hope this helps. Others welcome to correct and clarify my explanations.
Cheers,
Steve Stallings
-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] Behalf Of richard harris
Sent: Sunday, January 07, 2007 11:57 PM
To: [email protected]
Subject: [Emc-users] Following error
Hello,
I am attempting to tune my machine, running rutex drives and emc2.
I command a drive to move 8 inches one direction reverse and return to
zero. G1 X-8.0 F10 G1 x0.0 F10. When I test with a dial indicator if the
machine returned I see that I over traveled .005". I retuned the drive and
increased the "i" value and adjusted p and d to reflect the increased i. I
rerun my 8 inch test and find I returned to zero.
I then run the same test but with twenty inches of motion and find myself
2 thousands short. Just to see what would happen I increased my feed rate
from 10 to 15ipm. Suddenly the machine returns to zero. I increased the
feed to 30ipm and I overshoot.
It appears that I have a velocity dependent situation. I honestly have
little background in PID tuning, is this a drive tuning issue or is this an
EMC config issue? I am using the sample_inch.ini config and have only
changed the steps per inch, reduced the max acceleration, and max velocity
to reflect my machine.
Thanks,
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Emc-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-users
