On Apr 30, 2008, at 11:15 AM, Janos Dosa wrote:
<snip>
At the speed of 2100 mm/min the error from the theoretical path is
huge around 50mm (2 inches) or so. Same problem in smaller scale if
the machine is slower.
Your machine should fault with a following error long before it gets
to 2" of error.
This sounds like a servo tuning problem: start up halscope and plot
axis position, axis commanded and
following error. Set your following error in the ini at 1.0 and 10 to
give you room to play without halting.
Do short moves (at max velocity) and just long enough to get to
commanded velocity: and watch the following error.
Unless you have some tuning software tuning a servo is both an art
and a science.
I usually start by setting P high but still stable (no oscillation).
Then add FF1 and FF2 in small increments adjusting the parameters in
the direction that minimizes following error. Then go back and add a
small amount
of D. Finally add some I.
Now work with small adjustments P, I, D, FF1, FF2 and even FF0, if
you need it, until you can't get it following error
better. D has a narrow band where it works so treat it kindly.
I can get following error down to 0.001-0.005" at 5 m/min on my
Mazak; at normal machining speeds for me; i.e.
250 mm/min my following error is about 0.00005" in cruise mode with
accel and decel bumps that are larger.
Everyone has their own method of tuning and anything that works is OK.
I must say that tuning machines with glass scales is more difficult
than a machine with encoders. Having a machine with very low
backlash will make tuning easier.
After the tuning is satisfactory then set your min and max following
error as small as you can make work reliably in with your machine.
You really do want it to stop if following error gets very large.
Now you can again change G61/G64 modes and see if they give you the
desired results.
Hope this helps.
Dave
Changing the G64 mode to G61 or G61.1 or G64P0.01 mode does not
change anything.
The version of the EMC2 is 2.2.4.
The machine has digital servos capable to operate at 1MHz. (step/dir)
The machine if it operates at slow speed and there is a Pause
command issued at the endpoints, than works better in fact is very
accurate at the end points.
There is a feedback line from the glass scale(s) and there is a
reader/display unit that show independently the actual position
from the glass scale.
After a 96 hour continuous milling job the machine returns to
0.0000 with no misalignment at all both the internal and external
counter shows the same.
Since the theoretical path and the actual movement significantly
differs - see the XYZ values and the cursor in the black square -
and the G61,G61.1 command does not change that it is possible that
there is a bug in the trajectory planner.
Any idea how can we force the EMC2 to execute either the G61 or the
G61.1 command and switch the sw. to that mode?
Thanks.
JD.
----------------------------------------------------------------------
---
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Don't miss this year's exciting event. There's still time to save
$100.
Use priority code J8TL2D2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://
java.sun.com/javaone_______________________________________________
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users
-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Don't miss this year's exciting event. There's still time to save $100.
Use priority code J8TL2D2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users
