Experimental Physics and
Industrial Control System

Emmanuel Mayssat <[email protected]> · Fri, 14 Sep 2007 15:58:48 -0700

Hello again,

I am currently using a PLC for hardware and personal protection.
You probably do too.
Are you IOC computers monitored by your PLC?

If a computer IOC goes down, you may lose control of key components of
your favorite synchrotron/telescope. 

At that point the operator has two options
1/ run to the closest emergency exit
2/ push the emergency big-red button (where is it located again?)

At my site, we are contemplating monitoring the health of the computers
by the PLC. Good idea but how to do it?

There are 3 different failure modes for computers:
1/ computer is off
2/ computer is frozen
3/ ioc is not running correctly

I am interested in detecting failure 1 and 2, 3 could potentially be
solved with a "caget-if fail-restart me" cron job.

What would you do to detect 1 and 2 and feed the input on a switch?
I found that using a networked watchdog device could be a solution.
Indeed that device would need to be pinged at regular interval.
If not, it closes/open a switch.
Are they equipment that do that?

I found that the cb-7017 from measurement computing does something like
that. But I need 1 per computer and I don't want to reprogram my PLC
each time an IOC is added, plus in the future I may want to have
microCPU IOC as described in an earlier email.

Ok, I can serially chain all the watchdog switches on a few PLC input
links. If one of my IOC goes down then take down part of the machine. I
don't necessarily care about knowing exactly which IOC went down at that
point. So that could work except that I need one cb-7017 per IOC so it
does scale so nicely.

Ultimately I would like a networked device that can be pinged at several
ports and that can close once or several switches based on which port
was not pinged. Is there such a black-box device?

Food for thoughts....

--
Emmanuel

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System