I wonder what requires all that processor power? Maybe I haven't pushed the envelope yet but it seems like my SLC 5/05 was pretty fast and I only seem to recall one processor on it?
Controllogix is a completely different animal from a SLC. My last project, 1 Controllogix processor, 8 DeviceNet cards (124 - 32 bit words of I/0 vs the SLC's 31 - 16 bit words, the I/O equivalent of 64 SLC DeviceNet Cards), two ethernet cards, 4 discrete I/O cards. 52 programs, ~10 routines per program, ~75 rungs per routine for ~39,000 rungs of code. At any scan rate, a single SLC processor isn't even in the ball park. With the real time communications and I/O update capabilities of the Controllogix, I suspect that the second processor is for communications.
The 5/05 is pretty puny compared to the ControlLogix. If your writing 5/05 type programs for your CLX you haven't pushed the envelope, much less even licked it. The CLX OS is quite a bit more sophisticated, runs multiple tasks, can run Function block, Structured Text, and Sequential Function chart programs, multi-tasks, and even allows multiple programmers with multiple online edits simultaneously. The CLX also stores your tag names and comments on PLC rather than on your PC hard disk like 5/05 does, so someone without the program on their hard disk can still upload and have the documented program available. Because of this the PLC logic solver has to do alot more interpreting to execute the programs. But it still gives you blinding speed.
The PLC-5 has a "boolean" processor in addition to an "everything else" processor. I don't know if this was true of the SLC-500 or not. The weird thing is, the ControlLogix boolean processor is part of the same silicon as the ARM processor. So I don't know what the second ARM processor you're seeing is. Perhaps it handles the backplane bus, like a PCI-bridge chip would in a PC.
There's a nice article in the November, 2002 A-B Journal titled "A Bit Twiddler's History of the PLC", by Dr. Ernst Dummermuth. Supposedly, the article is available online at http://www.abjournal.com
It won't answer the question specifically, I'm afraid.
You thought a SLiCk was fast? I consider the SLC underpowered.
However, I'm not surprised a ControlLogix has two processors. There is quite a bit going on, alot more than with a SLC. For starters, it has built in the capabilities needed to directly control servo systems through the MO2AE and MO8SE (and 16) modules. In addition, it has logic scheduling capabilities.
To give a point of reference, we built a machine using a 5/04. The scan time was on the order of 20ms. A second machine using ControlLogix with added servos and PID loops, and 50% more logic, was on the order of 8-10ms.
Just to clarify a few things, plus, I guess I am exposing my non-AB bias again....
>>>>>
-------- Original Message --------
From: Anonymous
The 5/05 is pretty puny compared to the ControlLogix. If your writing 5/05 type programs for your CLX you haven't pushed the envelope, much
less even licked it. The CLX OS is quite a bit more sophisticated,
<<<<<
The CLx is indeed an entirely new platform than the SLC. This is good and bad, in that it is a break from legacy equipment, so you get all the
benefits of a new platform. The downside, of course, is absolute incompatibility with any of your installed base of SLCs.
>>>>>
runs
multiple tasks, can run Function block, Structured Text, and Sequential
Function chart programs, multi-tasks,
<<<<<
Omron CS1 / CJ1 platform, my system of choice, has not yet released the FBD, ST, or SFC programming tools. They will be out soon, but I honestly wish that they would have come out with them faster tho.
>>>>>
and even allows multiple
programmers with multiple online edits simultaneously.
<<<<<
Omron does have you covered here, though. Since a project can consist of multiple processors, you have the added benefit of being able to have
multiple users online at once to not only the same processor, but you can be online with several processors at the same time, and monitoring program/data in all of them simultaneously. A great benefit to anyone who
has tried to debug communication troubles.
>>>>>
The CLX also
stores your tag names and comments on PLC rather than on your PC hard
disk like 5/05 does, so someone without the program on their hard disk
can still upload and have the documented program available.
<<<<<
CS1/CJ1 does this as well
>>>>>
Because of
this the PLC logic solver has to do alot more interpreting to execute
the programs. But it still gives you blinding speed.
<<<<<
Also, the format in which they store their tag database means that they have to do a lot more overhead for each instruction that it executes. This may explain the second processor. Omron allows the entire project file to be stored on the processor, but keeps it's memory allocation in a format that is more native to the processor. Additionally, the CS1/CJ1 supports
FTP connections, so if you need to pull the program up for archiving from a machine that doesn't have CxProgrammer on it (depends on situation, of course, but I have had times that I needed to do so, and was not able with the AB stuff), you can make an FTP connection and pull out the project file using standard networking protocols. This requires advance planning, and you need to save the project onto the compact flash memory card if you are going to use FTP, but you can get compact flash for a good price at
WalMart, so it's not a big deal. Another benefit to this is that you can create text files from ladder onto the memory card. It is a breeze to
drop some operational data into a CSV file on the card, and let Oracle pull this up and integrate it into the production database. No SCADA
required.
Blinding speed is relative, of course. From their website, Rockwell claims 0.08 mSec / K instructions. Omron CS1/CJ1 runs 0.02mSec / K
instructions. to be honest, I have yet to find a general purpose PLC that runs faster than the CS1/CJ1 platform.
Finally, since the backplane of a CLx processor is a ControlNet network, all of your backplane communications need to work through a network stack. This inherently slows down the backplane speed. Also, I have heard of at least one plant that had several machines, each running a CLx processor, networked together for data collection, etc using ControlNet. A backplane
failure on one machine caused by inserting a module with the wrong firmware shut down not only the one machine, but the network shutdown
propogated through the rest of the network, shutting down the backplanes of every other attached machine. Think about that.........
