When Networks go bad ...

Has anyone experienced an LCN crash due to traffic overload?
What caused the overload?
Was there any warning?
What happened?
How did you recover?
How long did it take to recover?
Thanks, Walter.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Are you refering to Local Control Networks over Ethernet? I have found that those give plenty of warning as one can see the number of collisions and missed packets go up as devices are added. I have found that most PLCs are too slow to overload the network. The Control Logix and Ethernet I/P is another story, but there is documentation provided by Rockwell that tells one how to calculate the system through put in packets per second.
Recovery can be done by reducing the packet rate on offending modules. The most common are HMIs that request data for items that are not on the screen and at too fast a rate. Recovery can be instantaneous.
Another problem is trying to communicate with device that aren't on the bus. This can use up connections and queue space that cause problems. It is best to disable connections to devices that aren't responding to avoid this.
Peter Nachtwey
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

excuse. I

they say

put
the
process
an
was
each
The
I have seen a Ethernet NIC fail in such a way that it continually tried to transmit on a network which existed for HMI's/data collection. Obviously this "broadcast storm" brought the network to a halt. It's the only time I've seen a NIC (any kind of NIC, be it DH+, Modbus, or Ethernet) fail. And it was sudden, with no warning.
How many nodes were on the LCN? How does the network run? Is it CSMA/CD? Or deterministic? If we're going to be able to help you you'll have to be able to give details about this Honeywell network. I'd guess your problem was a "screamer" (bad NIC), but I've also seen sloppy HMI projects that were looking for non-existent nodes or bad data locations dramatically increase network traffic. Were any changes to the network or program(s) made prior to the storm?
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Walter,
I have seen when the bandwidth got too high for LCN's. The data just never got from one space to another, leaving lots of question marks on the screens as everything timed out. It was a mess. I can provide some references in private to you if you would like to chat with the fellows who worked it out.
I have also seen one other problem with LCN's. It seems that they hum. Nothing musical, but they hum at the same frequency a female rat hums when she is in estrus. Male rats will mount her and bite her behind the ear during the encounter. At any rate, rats, especially, um, excited ones, aren't too picky, and quickly mounted the cable and bit it. Since the sound was then loudest where he bit it, he returned several times until he had chewed completely around one cable and almost completely around the backup. That was one busy rat. The LCN cable was chewed down to the central wire (it's really just a big coax cable). Once he got far enough on the backup, everything went down.
Michael

collisions
that
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here. All logos and trade names are the property of their respective owners.