Homebrew STG card - CRAZY-3

["Arne Chr. Jorgensen" <instel-at-online.no>]

Hi,

More comments:

If this "crazy-stuff"  would be interresting,  then here is a few more things:

I think many would like screw terminals.  It looks like it should be possible
to make a board "protrude" the enclosure - let it stick outside with screw
terminals.  If one used some old AT enclosures,  you may srew a board across
the backside of the ISA ports,  as there should be enough space.  You could
also make a new front on these, and install terminals on this side.  Why ?

I just think it would be nice to keep most cables inside,  and just have a
simple hookup for motors etc. If you could use these old enclosures,  then I
don't think anyone would hesitate to cut/screw and modify them.  There is
enough of them around for experimentations.  

And I would like to make a few things clear:

1.  This would be an active subsystem where the cpu could do a lot of nice
things.  Like arranging interrupts, io, - etc.  into well suited buffers.

2.  Some modules could run realtime on this box.  ( I could make a crude
realtime system - but I think people at NIST would already know what would be
needed to do so, while I probably don't )

3.  A different approach to #2,  would be to just have an API.  The motion
controllers, etc. - run as before on the Linux box,  but you address the io via
an API to this other machine.  

4. The bridge between them can be done in many different ways.  DMA would be
the fastest,  serial,  par - port.  ( Linux can use a laplink cable - and a
device driver exist in DOS for a mounting a box as a device )  You could use a
different read/write cable with a small card.  Maybe a scsi connection could
work ?  

5.  Using an API - then a lot of stuff could be made.  All it would take, would
be to remap IO to this defined API.

6.  If you used DOS ( there is also a free DOS version - I think )  you have a
lot of tools available,  else I think there is a small linux port that works.

7.  You have all BIOS calls available, say - write messages, handle keyboard,
etc.  And all of it would be on the "realtime" side ( If anyone remember my
comment on this some time ago - "back side entry" )

8.  Some could also use this with HD and some other programs,  but use the EMC
with the interpreter.  So some of these "stepster" programs could be used for
output ( Sure you had to hack it a bit - but it could maybe be of value for the
experimentator )

9.   This computer could do work in the time frame that the RTLinux is not
addressing it,  so a lot could be handling - let say a slow ADC system,
channel multiplexing and other stuff.  

10.  A small RT system could be used, - using the clock interrupt.  This could
be set up to a different interrupt rate.  No one would care about the time of
day, etc.  So you could freely use it.  Using semaphores and other things, -
it should be possible to "sync"  activities in a pretty good manner.  

The main idea was just to have a bus system,  that nobody would feel any
hesitation to play with,  and would be cheap an easy for everyone to get hold
of.

The things that would have to be done, - is to discuss what link would be best,
and how to set up a standard API that "Fred et al."  would not reject to.  

//ARNE