RE: [CAD_CAM_EDM_DRO] EMC: why do I have so many problems?

In a normal type servo you would be right, but these are a little different.
That is probably why there is such a miss-conception that using G340s and
high line counts will actually give you an improvement of some sort.

The Gecko G3X0 series only works on position and AFAIK does not consider
velocity at all. The incoming steps drive an accumulator (with a G340 they
drive the multiplier circuit which drives the accumulator). The encoder
drives another accumulator. A differentiator looks at the spread between the
2 accumulators and generates an error signal that the amp section looks at.
As the divergence between the 2 accumulators increases the error signal
moves a fixed amount from the center point for each step of divergence. No
consideration of velocity at all. As the error signal moves further from the
center point the amp section sends more juice to the motor. The drives have
a range of about 115 encoder counts (in 4X mode) that they can be off from
the commanded position before they fault.

Now with a G340 if you send it 1 step the input accumulator is driven to
think 10 steps. The encoder has to move 10 counts to get the error signal
back to center. Acts exactly like a setup with a 10X lower resolution
encoder with the single exception that position is held a 10X tighter value
while sitting between steps, but who cares as this has no effect on the step
increment which is what will really decide how much stair step you get on
long tapers and circles. In my mind it even gives a downside in that your
range before you fault in a 10X mode would be about 12 steps. Seems to me
that you would have to tone your acceleration down quite a bit in comparison
to a 1X mode (G320) where you have about a 115 step range before a fault.
Sure a lot easier for the software to give a nice ramp in 115 steps compared
to 12.

Now I am sure Mariss can correct my terminology and point out where I have
the specifics a little off, but from a conceptual basis this is pretty much
how it works.

Tim
[Denver, CO]

> -----Original Message-----
>
> Actually, I think there is a reason to go with the higher encoder
> resolution and step multiplying. In a system without a tachometer,
> the encoder must be differentiated to find the velocity value to feed
> into the servo loop. A fine encoder resolution gives a smoother
> position derivitive, and hence a better velocity number. This can in
> theory mean better servo loop response (ie, make a fast step move
> without ringing).
>
> I do not know how complete the PID loop in the geckos is, ie, if they
> will really take advantage of it.
>
> Chris
>
> Addresses:
> FAQ: http://www.ktmarketing.com/faq.html
> FILES: http://groups.yahoo.com/group/CAD_CAM_EDM_DRO/files/