Re: I'm curious

--- In CAD_CAM_EDM_DRO@yahoogroups.com, "caedave" <caedave@b...> wrote:

> During this discussion someone asked for a closed loop definition
> (I seem to have deleted it). I dug out my old lecturing notes

(Robotics)

> and hereby (changed to suit this forum) is my definition.
>
> Position A (The required position)
> is compared with Position B (The current position)
> Any resulting discrepancy is applied to
> the current position B drive.
>
> This process is continuously repeated
> until such time as Position B is equal to Position A.
>
> In a CNC content the permitted rate of change
> of position B will be controlled
> by the currently defined feed rate
> (variable for arc moves)
> and each driven axis will have
> it's own comparator circuit.
>
> In a fully closed loop control system
> the actual current position must be
> known in relationship to a known
> reference start (datum) point.
>
> This will depend on your controller
> program being Abs or Rel where
> Abs references a fixed start point
> and Rel references the position
> at the end of the last move command line.
>
> Labouring the point.
> Move X 100, Move Y 100 is relative while
> Move X100.Y0,
> Move X100.Y100 is the absolute
> to get to the same point
> assuming both start at X0,Y0.
>
> Dave M.

Theoretically, if the G-Code instructs the Y axis to move from 0.0 to
1.000 and the X to move from 0.0 to 1.000, the two axes should move in
unison, drawing a straight line.

but, in an applied operation, one axis could encounter more resistance
and so does not track in a perfect 1:1 with the other, but winds up
recovering and winds up at 1.000.1.000

Some error is to be expected and to be allowed, whereas there is a
point where the error is too great and a fault would be signaled. In
actuality, there must be an error or the unit would not move. example
is the section.

> Position A (The required position)
> is compared with Position B (The current position)
> Any resulting discrepancy is applied to
> the current position B drive.

So, there would be an accuracy component to the APPLIED deffination,
vs. the THERORETCIAL one.

Interestingly, a rough cut would (should) be expected to have more
tollerance to variations to the commanded path. A finish cut is taken
to more exacting levels so as to attain the desired result.

Ideally the system would allow a programable deviation. A PID or some
other control alogrythm would monitor the deviation and correct for it.

The discussion is often where the monitoring and correction to
deviation occurs. in the controller section of the driver section. I
am curious why one is better than the other, and are we talking angles
dancing on the head of a pin or orders of magnitude better ?

Dave