Re: Re: GeckoDrives.. Can anyone Help?

Hi,

I have been following this servo debate on the virtues of a
tachometer versus discrete steps at low speeds. Unless I have
misaprehended a basic concept (not uncommon for me), I think it is a
distinction without a difference.

A tachometer feedback loop is an interior or minor loop, while the
encoder feedback forms the major feedback loop. Digital incremental
encoders are just that; there is no feedback info between encoder
edges. It is this encoder property that takes a lot of the
theoretical luster off of tach feedback schemes.

The G320/G340 does not use tach feedback and the motion at low speeds
is incremental; the motor advances one encoder edge for every step
pulse, then stops until the next step is sent. On a diagonal XY move
this results in a "staircase" motion path. If a 500 line encoder and
a 5 TPI leadscrew is used, the "staircase" dimension is 0.0001". The
other way to look at it is there are only 4 degrees of motion
freedom; +X, -X, +Y and -Y.

If velocity feedback (tach) is used, then in an ideal world the
motion path would be a series of "connect the dots" diagonals between
the "staircase" locations. The promise is unlimited degrees of motion
freedom.

Where does the velocity command come from? It is based on the
position error of the previous encoder location and is a prediction
of what velocity will cause the motor to arrive at the next encoder
location at the right time. The problem is position is the first
integral of velocity with time.

An error in velocity translates into a position error over time. As
speed approaches zero, the time between encoder edges increases.
Since there is no position feedback between encoder edges this error
will become large.

Comparing the two, tach feedback will result true diagonal moves
(monotonic, or continuous motion) at very low speeds but there is no
increase in accuracy. That is still determined by
encoder "granularity".

What matters is what happens when the chili meets the cheese; the
practical differences are between the two methods. If the velocity
feedback method was perfect (it's not), then its advantage over the
G320/G340 would be .707 times the encoder's linear resolution
deviation from the ideal motion path. This is only a 71 micro-inch
difference when using a 500 line encoder and 5 TPI leadscrew.

Anyway, that's my 2 cents worth. I will be glad to support the above
description with graphs if anyone is interested.

Mariss

--- In CAD_CAM_EDM_DRO@y..., "Tom Eldredge" <harmonyt@r...> wrote:
> Jon Elson,
>
> Thanks for clarifying yesterday's email. So the tach is the real
missing
> link.
>
> Understanding what you do about this situation and the Gecko
drives, do you
> think Mariss could add tach feedback to his amps also? If he, say
for
> example, added DC tach feedback from the motor to the G340 amp,
would we
> then be able to say we have true velocity control? Would this
configuration
> completely bridge the gap?
>
> It seems I remember Mariss saying something about making a
velocity drive
> sometime in the future. I wonder if this is what he is thinking
of. I am
> going to do a search on his emails to see if my memory is showing
its
> weakness again.
>
> I was under the impression that the multiplier on the G340 drives
only
> allowed one to use the higher count encoders found in retrofit
situations.
> It does not do anything for us in the way of smoothing out the
motion does
> it?
>
> Thanks again for your valuable help, Jon.
>
> Tom Eldredge