CPU Switching in AB

Dear all,
I just need to desighn a AllenBradely system in such a way that there are 5 remote controller and a main controller. the operation scheme is the remote controller has to control the local process and main controller is used to monitor normally.
but, when the remote CPU fails, i need to switch the main controller to act like the local controller for that particular remote process and to monitor the rest of Remote controllers.

Is this possible in AB - compactlogix and controlllogix platform.
If possible kindly help me to succeed this..

Thank you

saranraj said:

Dear all,
I just need to desighn a AllenBradely system in such a way that there are 5 remote controller and a main controller. the operation scheme is the remote controller has to control the local process and main controller is used to monitor normally.
but, when the remote CPU fails, i need to switch the main controller to act like the local controller for that particular remote process and to monitor the rest of Remote controllers.

Is this possible in AB - compactlogix and controlllogix platform.
If possible kindly help me to succeed this..

Thank you

Click to expand...

Remote as in physically remote?

I think he means that he has has plcs all over the plant/ site and one plc as a master or data concentrator and if one of the cpu fail in the local rack then he wants the remote / master cpu to control the i/o of the local rack in a remote rack typ of fashion. This will work but it is the switchover that i am not sure about making a remote cpu the redundant backup for a processor is easy but making it the backup for multiple processors in mltiple racks i am not sure about. Sounds like a good concept though? I am curious also.

Sounds interesting, but ...

What about the time when a code change would be necessary. How would you maintain it and ensure that the two controllers have the equivalent code?

We have a remote system with 47 PLC's controlling locally and a central Data Concentrator in which the scada system retrieves data. We have handshaking and alarming when the remote PLC ceases communciations. At that time, the automation staff is alerted and responds to determine the fault and/or repair as appropriate.

This sounds to me like something you are doing to satisfy a functional specification dreamed up by somebody unfamiliar with logic controllers, rather than something you are actually doing to increase control system availability.

The ControlLogix system could be built to do something like this; the "Master" controller could have I/O connections created but inhibited to the remote I/O modules. If it detects a failure of the "Remote" controller, it un-inhibits the connections and takes over the control process.

This is not a proper "redundancy" system, and it cannot be done with CompactLogix. You would need to write code and procedures to be certain that the "Remote" and "Master" controllers do not compete for I/O resources.

Thanks all,
i could understand the basic stupidity and difficulty in this concept.
but when AB people told me that this is impossible i argue them that this could be done.this is not actual redundancy by hardware but i think it could be done through software.

is it not possible to store all the codings in main controller like modules and making particular codings on when particular module fails...

this concept is possible as inverse concept.
normally main controller runs and remote controller in sleep mode.if main fails all remote awakes and takes the controll.why not it is possible as reverse philosophy?

Help me to find the solution .....thanks all

It sounds quite interesting but doesnt sound logical because 1 redundant controller for 5 controllers ??? I dont know about the process or IO numbers or is it possible to control the general process with 1 controller but if it is according to me instead of using 5 controllers with 1 redundant why not try 1 controller with 1 redundant and remote IOs.

The envelope is decided by the process, instead of dealing with 5 controllers I would go with 1 more capable controller by means of IO. Like instead opf 5 SLC CPUs try PLC5 (Hope mentioned right controllers beacuse I am not an AB guy)

Have a nice day.

thanks erdem,
your solution is right and it is the most common architecture that 1+1 main controllers and 5 RIO's. i have solved the solution with 5 compaq logix I/O's and main ontrol logix (1+1).
My problem is solving this type of architecture (1+5 cpu switching) is possible or not?

thanks again

You could do an old-fashioned switchover. Have an output module in each of your remotes, with an output turned "on". If the cpu goes to program or fault, it drops the output. You'd monitor the output with an input card in your main PLC. Input drops, you take over the IO.

You'd have to make sure you got the software right on your remote PLCs so they don't take over again; that is the tricky part. I would be comfortable with backing up one, 5 seems a little hairy. This sounds like an old Provox spec I worked with, a "1 for n" specification that worked, but only with one controller at a time.

Like the others, I've confused as to why you don't run it all from one location, or use an installed backup CPU. You would definitely get better results. The hardwire way doesn't ensure uninterruptible operation, you will lose your process.

In my opinion...

If the CPU you are asking to take control of others which have failed has enough horsepower to handle two of the systems at once, then it is definitely possilbe.

It should even be possible to maintain the code quite easily. The program will need to be built around the concept of portability and the failover process, which will not be a bumpless transition, unless there is some status variable that can predict a CPU fault before the I/O switch to their fault state.

Using local variables and the techniques described above by others, you could do it, but I don't think it would be cost effective, when you can set up a flash card or pre-program one spare CPU and have instead of a minimum of a few seconds of downtime, a minimum of fifteen minutes to find the faulted CPU and plug in the backup.

I suppose you could have one backup CPU with all the programs from the potential subordinate machine(s), with a MAIN that had mutually exclusive JSRs to the continuous tasks of the five potential replac-ees, and an HMI button to select one of them...then you could pre-program the spare once and it would be like the "relief guy" who knew how to do everone's job and could fill in when they're on vacation 'n stuff.

Again, that'd be too much work for something that's is only going to happen, what maximum once a year?

If you have 10 Logix processors and more than one failure per year, something is wrong...We have had a handful (like three in 3 years) of CPU faults that were remedied with the proper installation of a backup memory card, and by following firmware updates that have affected some of the comms cards.

Now, just for practice and proving the theory, it could be done...

If the purpose of the exercise is truly to eliminate downtime, then use the real redundancy as designed for Controllogix.

That's my opinion anyway.

It is possible to do.

This is done by remote I/O mapping from the main controller through communication module which should be independent of remote CPU.but we need to update the registers of crtical I/Os continously with some time delay.then only the status of main controller and remote controller will be in same state.

My suggestion for this system is better to go with cpu hot redenancy in main controller and the remainig remote areas are remote I/Os.

The architecture I described with ControlLogix (Bulletin 1756) hardware is not possible with CompactLogix (Bulletin 1769) hardware because the CompactLogix does not perform controller-independent bridging of networked I/O.

I very strongly recommend against attempting to invent a backup or redundancy architecture that is recommended against by the manufacturer and by industry experts.

OkiePC described a very common mechanism for the stocking of "fast recovery" controllers.

The mechanism you choose depends on the requirements of your process.

"Impossible" is a strong word. I don't think it's impossible to implement a 1 for n redundancy system with hardware that is not specifically designed for it. However, I very strongly doubt that there is a good reason to do so. Or that it is cost effective. Or that it actually provides a robust control system that is more reliable.

I agree with OkiePC.the task is to prove that "THIS could be done".
i can build the same process architecture using siemens s7-400 and the inverse process (Normal MAIN , fault REMOTE) by using s7-300.
but when i try the same in AB , i got so many comments that this could not be possible.this makes me irritating that something sounds impossible in automation field.
with the comments of you people i can get the clear cut picture and i could solve it by using a high memory CPU at master and a low edition at remote making status indication for failure and using the same for activating the sub-routines and making the process run.

thanks all..

"Impossible" is not a strong word.

ken Roach- i agree with your point , can you tell me any other platform to do the controller-independent bridging of networked I/O.