Re: [CAD_CAM_EDM_DRO] true closed loop vs. hobby closed loop

snagglexr650 wrote:

>Hello,
> I work in a machine shop and we have several retrofitted
>lathes/mills using OpenCNC and various motors/drives including
>Yaskawa servo motors/drives, we also have a Haas VF4 VMC. These are
>all high dollar, industrial level cnc controls, with "true" closed
>loop feedback from the encoders.
>
> It seems to be widely considered that true closed loop actually
>allows the computer control to self-adjust the motors if error in
>positional accuracy has occured. However, all the machines we have,
>including the Haas (not a retrofit) have an allowable following error
>(FE) on each axis. If that FE is exceeded, the drive faults out and
>cuts power to the servo motor. If the positional error falls within
>that FE range, then inaccuracy does occur, it's happened to me.
>Furthermore, this FE is used in axis crash situations. Say you crash
>Z into your vice/part/table, as the axis stops moving due to hitting
>a solid object, FE is exceeded and the drive faults. Attempted
>positional compensation via the control would simply drive the axis
>into the part harder and cause more damage to the machine.
>
> Now, with that in mind, compare the +/- 128 counts 'FE' on the
>Gecko drives. If positional error exceeds 128 counts, then the drive
>faults out and servos stop.
>
> In either scenario, it appears that the same thing happens, and
>that even in the case of 'true' closed loop controls, there is no
>attempt to correct the positional accuracy, just a fault.
>
> My question is where is the difference?
>

The Gecko drives are known to be within +/- 128 counts of commanded
position, but you don't know where. With software closed loop, the
CNC program DOES know what the error is at any time. EMC allows you
to have 2 terms for following error. One is a lower limit at standstill,
and can be set VERY small, perhaps 2-5 encoder counts. The other limit
is proportional to velocity. This allows you to have a larger following
error limit when moving fast, and especially at rapid feed rates, while
keeping a small limit at the low cutting feedrates, where it matters.
EMC gives you not only the traditional P,I and D loop comensation terms,
but you get feed-forward terms, as well, which allow for VERY fine
tuning of the servo response. It also can graph the servo performance,
so if you have questions about whether it is performing properly NOW,
you can just take a graph while running it through the suspect move, and
know whether it is doing it right or not.

> I know software is the
>key to closing the loop. Is it simply that with Geckos, your
>step/direction software will happily continue to run the program
>without knowledge of the drive faults,
>

You really need to hook up the fault indication from the Gecko drives so
it can stop ALL axes and alert the CNC program.

> and that on the Haas
>control/OpenCNC software control the software is aware of the fault
>and stops running the program and displays an 'excessive following
>error exceeded' message?
>
> Is this true? If not, then please explain to me the difference
>between hobby closed loop via Gecko Rutex
>

Note that Rutex has a fault limit of 20,000 to 30,000 encoder counts,
depending
on the model! Even with high encoder resolution, that could be over ONE
INCH!

Another advantage of true closed-loop is that the machine can be
switched back
and forth between closed-loop CNC and manual DRO mode, with the screen
always showing the true position. If your machine has handwheels, this can
be a nice feature.

Jon