First of all, I mounted a 17" monitor on the pendant, and brought in a
wireless keyboard. That made the pendant look less like an orphan.

Then I went on to deal with the annoying issue of PID tuning. First of
all, I know little about tuning. Second, there is two loops, one is
velocity loop in the servo drives, and also, there is a positioning
loop in EMC. Third, the drives have a test/offset pot that is very
hard to adjust to be anywhere close to no creep.

Of the above reasons, the first is the worst, of course.

So first I tuned the AMC servo drives using their standard method that
Gunner mentioned: I increased gain until the motors started to
vibrate, and backed out two turns.

Lack of tuning evidenced itself in large following errors, such as
0.005" or so.

After I increased P and slight messed with I, following error is
always under 0.002" on all axes, and usually is under 0.001". I do not
consider myself done, but this is a start and I can move to other
things for now.

i
Then I went through every axis, and by manipulating P and I, reduced

Ignoramus30076 wrote:
First of all, I mounted a 17" monitor on the pendant, and brought in a
wireless keyboard. That made the pendant look less like an orphan.

Then I went on to deal with the annoying issue of PID tuning. First of
all, I know little about tuning. Second, there is two loops, one is
velocity loop in the servo drives, and also, there is a positioning
loop in EMC. Third, the drives have a test/offset pot that is very
hard to adjust to be anywhere close to no creep.

Of the above reasons, the first is the worst, of course.

So first I tuned the AMC servo drives using their standard method that
Gunner mentioned: I increased gain until the motors started to
vibrate, and backed out two turns.

Lack of tuning evidenced itself in large following errors, such as
0.005" or so.

After I increased P and slight messed with I, following error is
always under 0.002" on all axes, and usually is under 0.001". I do not
consider myself done, but this is a start and I can move to other
things for now.

i
Then I went through every axis, and by manipulating P and I, reduced



ability to continue typing?? G




--
Steve Walker
(remove wallet to reply)

Ignoramus30076 wrote:
First of all, I mounted a 17" monitor on the pendant, and brought in a
wireless keyboard. That made the pendant look less like an orphan.

Then I went on to deal with the annoying issue of PID tuning. First of
all, I know little about tuning. Second, there is two loops, one is
velocity loop in the servo drives, and also, there is a positioning
loop in EMC. Third, the drives have a test/offset pot that is very
hard to adjust to be anywhere close to no creep.

Forget the creep, it doesn't matter, the positioning loop takes care of
it. At the very final stage, you can tweak the offset pot to zero out
standing-still position error.
Of the above reasons, the first is the worst, of course.

So first I tuned the AMC servo drives using their standard method that
Gunner mentioned: I increased gain until the motors started to
vibrate, and backed out two turns.

Lack of tuning evidenced itself in large following errors, such as
0.005" or so.
That is not really all that large.

After I increased P and slight messed with I, following error is
always under 0.002" on all axes, and usually is under 0.001". I do not
consider myself done, but this is a start and I can move to other
things for now.

i
Then I went through every axis, and by manipulating P and I, reduced
Yes, yes??? You should be able to get the error down to well under
..001" up to 30 IPM or so, and that should cover any cutting move.

At the final part of tuning, you do jog moves, they have a trapezoidal
velocity profile, and you want all parts of the move, ramp up, cruise
and ramp down to have low error. Adjusting FF1 and FF2 greatly help there.

Jon

On 2010-08-04, Jon Elson wrote:
Ignoramus30076 wrote:
First of all, I mounted a 17" monitor on the pendant, and brought in a
wireless keyboard. That made the pendant look less like an orphan.

Then I went on to deal with the annoying issue of PID tuning. First of
all, I know little about tuning. Second, there is two loops, one is
velocity loop in the servo drives, and also, there is a positioning
loop in EMC. Third, the drives have a test/offset pot that is very
hard to adjust to be anywhere close to no creep.

Forget the creep, it doesn't matter, the positioning loop takes care of
it. At the very final stage, you can tweak the offset pot to zero out
standing-still position error.
Of the above reasons, the first is the worst, of course.

So first I tuned the AMC servo drives using their standard method that
Gunner mentioned: I increased gain until the motors started to
vibrate, and backed out two turns.

Lack of tuning evidenced itself in large following errors, such as
0.005" or so.
That is not really all that large.

After I increased P and slight messed with I, following error is
always under 0.002" on all axes, and usually is under 0.001". I do not
consider myself done, but this is a start and I can move to other
things for now.

i
Then I went through every axis, and by manipulating P and I, reduced
Yes, yes??? You should be able to get the error down to well under
.001" up to 30 IPM or so, and that should cover any cutting move.

At the final part of tuning, you do jog moves, they have a trapezoidal
velocity profile, and you want all parts of the move, ramp up, cruise
and ramp down to have low error. Adjusting FF1 and FF2 greatly help there.

OK, thanks. I will look again at following error when doing G1 F20 or
something instead of G0.

i