On Mon, Apr 05, 2010 at 07:56:51AM -0700, Peter C. Wallace wrote:
>
> A good Estop system is to disconnect servo primary power and at the
> same time discharge the servo filter capacitors with a resistor for .5
> - 1 second decay time. This will brake the motors so the voltage decay
> should not be too fast.
Many thanks for that solution. It needs only one SPDT relay, while motor
disconnection and shorting needs four, plus 4 power resistors.
> >
> > B) What about a charge-pump? i.e. Estop if the controller goes down or
> > loses the plot. Or does the 5i20 do that in the FPGA?
>
> There is a watchdog in the FPGA. It will reset all I/O to the power up state
> if not accessed at the preset interval. I would not depend too much on this
> as a software/hardware failure of some kind might let the watchdog petting
> thread continue while the rest is going crazy.
Yes, especially if that thread is time-sliced, like the other threads
that we can specify invocation intervals for, then the watchdog is
petted even if everything has ground to a halt. =8-((
Hmmm, if we were to toggle the pet output in e.g. the path planner
thread, and only output the current state in the device driver which
talks to the FPGA, after passing the toggling value along the chain of
motion controlling threads, much like an Estop chain, then all vital
threads have to be scheduled, and execute a vital section of code, or
the watchdog bites.
It is probably easier to have each motion-related thread notify the
petting thread, via RTAPI, that it has run. The watchdog is then only
petted if no needed thread has gone incommunicado. (Passing a toggling
value between threads is more programming effort, and subject to greater
timing variation, due to scheduling vagaries.)
Even on embedded systems which controlled only telephone calls, and
thereby money I'll admit, I've always been careful never to allow a
single thread to pet the watchdog, because it's as unreliable as you've
warned.
> Its watchdog will disable all I/O, (turn off all 7I37 outputs for example) so
> this will happen automatically on a watchdog bite but an output in series
> with
> the Estop chain is also good so EMC has an input to the chain for less severe
> faults/user-input
Oooh, that's nifty bonus, that I hadn't expected.
Thanks.
Erik
--
Meskimen's Law:
There's never time to do it right, but always time to do it over.
------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users
