Re: EMC for windows

[Jon Elson <elson-at-pico-systems.com>]

ERIC KELLER wrote:

>There is one counter-example which shows that under certain circumstances
>windows will work in real-time: Mach 1, which is the twin of EMC hosted under
>windows.  My theory on why that works is that, by definition, a step and
>direction system is open loop stable.
>
No, it isn't.  If the computer is suddenly busy with other tasks, the 
smooth stream of
step pulses suddenly ends, without a smooth ramping down to a stop.  The 
motors will
be guaranteed to skip steps under these conditions.  Smaller gaps in CPU 
availability
may just cause rattling noises, or intermittent lost steps.  Some 
advanced interpolating
drives may be able to handle this under some conditions of modest 
speeds, large motors
and small machines.  Now, the step pulses output by the computer are 
correct, but sudden
variations in speed will likely exceed the ability of the motors to 
follow them.

Mach 1 does what it does by implementing the critical part of the motion 
routines as a
high priority VXD task, and calling for enough priority so it gets 
service when the timer
it has set demands the CPU resource.  (This information is not 
authoritative, but Art has
previously described some of the magic he did to make this work.)

Instead of using an expensive, 3rd party real-time scheduler, he 
implemented its own scheduler,
mostly using already present software and hardware components.  You need 
to be a subscriber
to expensive documentation services from Microsoft, or otherwise have 
access to the internal
developer documentation to do this, but all these features are in there.

>  Thus, when bill gates has internet
>explorer dump your hard drive over the internet to Redmond, nothing bad
>happens, because no step/dir pulses are output.  By this theory, if Mach 1 were
>modified to drive a servo-to-go based system, the system would probably "go
>boom" occasionally under heavy use.
>  
>
No, if he's done it right, and it really has to be right or people would 
have already noticed, then
the basic scheme could also run a servo system, if mach 1 was designed 
around a servo motion
model (which it apparently is not.)

Jon