Re: Relation of a DRO to a CNC modification of a machine?

>
> Thank you for a very articulate explanation of the last paragraph.

I'm aware of the differences between Steppers and Servos as far as
performance. I struggled to find suitable drives for some big
unipolar skip-o-matic Slo-syns for my Bridgeport and it was a dead-
end everywhere I turned and experimenting can get expensive. So, I
86'd them and went with big servos instead. Does Mach2 perform the
way you outline above with a Gecko 320? Someone mentioned that they
thought Geckos monitor the return from the encoder and will fault
after a 128 step discrepancy. Will they actually correct themselves?
I know in playing with the motor tuning in Mach2 I've adjusted the
acceleration and velocity way high and the Z axis driving the knee
will start and stall at like 80 ipm. To my knowledge the Gecko didnt
fault. Unless it did and it reset itself as I was resetting the
acceleration and velocity. I attributed the stall to the10 or 12 amp
Power supply not keeping up with the 30 amp stall torque Servo

> (40 in-lb)
>
> Ron

Ron: here is the way the Gecko (and Rutex) work: You send a pulse
from the PC. The controller software knows (because you told it)
how many pulses per inch/mm it needs to send to move the axis along
a path.

The Gecko takes the pulse stream and applies DC to the Servo motor
of the correct polarity (determined by the Direction signal) and
then reads the feedback from the encoder (which is also pulses).
They are not particularily a 1:1 ratio. What happens is that an
error count is maintained (it actually is an analog voltage) who's
magnitude is porportional to the difference between the commanded
position and the actual position determined by the encoder. The
resolution of the Gecko is 128 bits so based on the count ratio you
could have that much error. It's selfcorrecting so there are
no "missed" steps. The actual motor position in real time may lag
slightly behind the commanded position but it will catch up. As
long as the error does not exceed the 128 count it will just adjust
itself and catch up. The gain and dampening adjustments control
what the response curve of that error count is. Higher gain means
the circuit is more aggressive in its attempt to keep the two counts
equal but that can lead to oscillation so the dampening is put in to
help counter that. The two controls are inner active so if you
adjust one you often have to adjust the other.

If you notice, in this that at no time is the controller software
told about how much the error is (or even that there is one) and it
takes no part in interactively speeding up or slowing down the pulse
speed to help the compensation. Under most circumstances this does
not matter since you "tune" the controller software to not exceed
the capabilites of the motors to respond.

In the even the error exceeds the max count of the Gecko the gecko
fires a fault signal. Depending on how you have it hooked up it can:

Stop
Cause all the Gecko's to stop (if you tie the fault pins together)
Cause the controller software to stop using the e-stop function.

While this sounds drastic you do not want the machine to try and
correct for conditions where the error count is too wide. That
would normally indicate you are attempting to operate the machine
outside it's normal operational envelope (i.e. the bit is jammed in
the work or the gantry has hit something).


While the ultimate closed loop control is done by the controller
software in direct communication with the table position the more
components living inside the loop the more complex the control
issues. The only hobby grade software closed loop system is the one
Jon sells at PICO and that works with EMC.

Tom