Ken,
First, if that lousy old TI-565 is gonna end up in the trash somewhere... CALL ME! before it gets discarded.
Second, a little story...
When I started at my current position the main process was controlled by two PLC's. One was "lousy, old and undependable!". The other was "brand new" with all the bells, whistles and a whole bunch of the pretty blinkin' lights!
When I started there, the "lousy, old and undependable" PLC was "charged" with more down-time than any other factor in the process. The "brand new" PLC had its' problems but, hey, there were far fewer problems on the "brand new" PLC than there were on that "lousy, old and undependable PLC".
There was a major, and I mean MAJOR, campaign underway to replace that "lousy, old and undependable" PLC with a copy of the "brand new" PLC.
My first task was to investigate the "lousy, old and undependable" PLC and see if there was something I could do to "prop-up its' performance until the new system could be installed".
Now... one of the really important things that a Process Developer needs to understand is that "users" have this Neanderthal trait of "killing the messenger". In terms of a process controlled by a PLC, this means that the "users" tend to blame (and kill, as soon as possible) "that damned computer!"
Those of us that are a little more open-minded realize that the computer will do what you tell it to do. And the complement to that is... GI-GO... Garbage In - Garbage Out.
What this implies is... if a so-called Process Developer can not get the "system" to respond as required by the users, then, "that damned computer" must be the cause of everyones' grief!
The user refrain becomes... GET A BETTER COMPUTER!
Those of us "in the know" know that the refrain should be... "HANG THAT DAMNED PROCESS DEVELOPER!"
So... as I said, "My first task was to investigate the "lousy, old and undependable" PLC and see if there was something I could do to "prop-up its' performance until the new system could be installed"."
I did investigate. When I suggested to a fellow Process Developer (one that was familiar with the process) that I would probably have to rewrite the entire process... on-line... live! He said, "Damn! You're gonna be busy for the next 10 years!"
Within a few months (spent learning all the details of the system) I installed a re-write of the most critical part of the process. This part of the process was responsible for more down-time than any of the others.
That was seven years ago. Since then, that part of the process has been "charged" with zero-down-time... that's a big "goose-egg", as in, "none", "nada", "zip"... get my drift?
I have since moved on to the remaining parts of the process. There are only a few modules left to deal with. When my friend said, "...You're gonna be busy for the next 10 years!" I thought he was crazy... little did I know.
So, at this point, there is now a concerted effort to replace the "brand new" PLC with one like that "lousy, old and undependable" PLC.
There are two reasons for this.
The first is that the manufacturer of the "brand new" PLC has quit making PLC's (Honeywell).
The second is that the "lousy, old and undependable" PLC is as SOLID AS A ROCK!!!
Now-a-days, if there is any down-time "charged" to a computer, it is only charged to the "brand-new" Honeywell. This is simply because that "lousy, old and undependable" PLC has been reprogrammed from a process-point-of-view as opposed to being a huge "switch-handler". The Honeywell has yet to be addressed.
Now... here is my conundrum... that "lousy, old and undependable" PLC, or rather, that SOLID AS A ROCK!!! PLC, is a TI-555. And, of course, much to my chagrin, TI has also quit making PLC's.
"Ohhh, What a World! What a World!" (Wicked Witch of the East?)
So now, despite efforts from forces beyond my control, I'm trying to replace the Honeywell with another, SOLID AS A ROCK, TI.
The point of that "little" story is that...
"Programming a PLC does not necessarily make a process!"
END_STORY:
Now, Ken, with respect to your question...
I can only assume that you want to be able to try the newer control system as long as it works but then be able to shift back to the older system if the newer system fails. This might mean switching between systems while the process is running.
If that is the case, then you should have all input and feed-back information available to both computers - all the time.
You should then, almost as Panic Mode suggested, use a switch to control the outputs.
Both computers know what's going on at all times but only one is actually controlling the outputs.
You should definitely be cautious about the conditions that might make you want to switch from the newer system back to the older system.
For example, if a move stalls under the newer system, switching back to the older system might cause a radical move.
In this case, you might want to monitor the status of that switch and provide for some kind of smoothing effect, or a move cancellation of some kind, when you switch from one to the other.
(49) (LU2E)