Re: EMC compatable computers

[John Griessen <john_g-at-cibolo.com>]

I like the idea of industrial ethernet with no connection to your lan,
and only connections between your computer and motor controller.   Motor
controllers would benefit from some voltage isolation gained by plastic
fiber that carries switching signals, (with no protocol -- just on/off),
to a separate driver section on the motor.  The ethernet interface node
could take care of some number of axes generically, and you add more
nodes for more axes.  Ray H. said the emc software model had a concept
of non-realtime switching command data being loaded in to some registers
for the realtime part of the code to excute on time.  Those registers
could go to the controller card and not be anywhere in the PC box with
this approach.  

If you want to have measured sensor data coming back from your motors
and motor drivers and position sensors, a separate plastic fiber master
slave link could be made with sercos chips and those could be used for
switching signals too with no big loading problem and then you would
have one plastic fiber and only digital message signals and probably two
sercos chips to buy for each axis, (I have not read much about sercos or
its prices yet except that it has been used to do motor swtching driver
control for years reliably).

John Griessen



On Fri, 2003-01-10 at 23:14, Matt Shaver wrote:
 If you want to develop a high speed serial controller, I 
> think the best option is Ethernet. I don't mean a complete TCP/IP network 
> device, more like a dedicated 100BaseT NIC connected to an embedded ethernet 
> equipped servo controller through a crossover cable. The communications 
> protocol would be raw frames (there's integrated hardware CRC checking for 
> each frame). Inexpensive prototyping materials are widely available. As an 
> example:
> 
> http://www.edtp.com/
> 
> Comments? Phil? Anybody? Don't all speak at once... ;) ,
> Matt
>