View Single Post
  #34   Report Post  
Posted to rec.crafts.metalworking
Ned Simmons Ned Simmons is offline
external usenet poster
 
Posts: 1,803
Default long term reliablity computer boards

On Wed, 29 Apr 2009 14:29:32 -0500, "Pete C."
wrote:


Ned Simmons wrote:

On Wed, 29 Apr 2009 12:35:02 -0500, "Pete C."
wrote:



What functionality? It's still closed loop, still flags an error and
shuts down the control if the servo is overloaded and can't hold
position, you can still tune the parameters, etc. I can't think of any
feature you loose that is relevant for the class of control.


The first thing you'd lose is the Galil controller itself, unless it's
a newer card than what I imagine it is. Some of Galil's newer
controllers can be configured for either step & direction or +/-10V,
but not the older generations.

But setting that aside, the first couple things that occur to me are,
since the gains are now set in the amp rather than the controller,
you'd give up: the auto-tuning routines available for the controller;
the ability to replace a drive without going thru a manual tuning
process(with the Galil board the setup parameters can be stored in the
motion program); and the ability to change the loop gains on the fly
in software. Also, the Gecko is missing integral gain.

Unless you were able to set up a second feedback loop from the encoder
to the controller board, you wouldn't get the "digital 'scope"
utilities that Galil supplies, and the motion program wouldn't have
access to real position and velocity numbers.


Ok, and exactly how relevant are those features to the day to day
operation of this class of machine? We're talking old used retrofitted
iron here, not the latest $1M machining centers. Nearly all of what you
indicated is only relevant to one time setup.


If you care to keep the machine running and making money, ease of
setup is a big deal.

The lack of a derivative term in the loop is not merely a one-time
convenience issue. It can have a big impact on positioning
performance, and may be more important on an older machine where the
control may be called on to compensate for some of the machine's
shortcomings.

I believe Mach3 can also
read encoder positions in some configurations so it does get to see the
"real" position, but even so, that makes little difference for this
class of control and machine. You're trying to put a Ferrari engine in a
Chevy truck.


I don't consider an obsolete motion controller to be a Ferrari, and
don't see the point in replacing something that's up and running and
familiar with a system that has less functionality.

--
Ned Simmons