Retransmits/failed downloads on Rockwell system

[CHRISC1024]

One of our clients are seeing retransmits on multiple segments at thier plant. They are also experiencing failed downloads to thier FFLD's. I have seen this in the past and most time if it's not a noise issue I have found the problem to be the macrocycle. I know that Rockwell's RSView sets the macrocyle for "optimum" performance but have found that many times this setting is too small. What we have been told is to set the background time to equal the forground time to give a 50/50 split and then try to shrink it down.

Am I correct that if a macrocycle is too short it can cause retransmit errors as it can miss a device on a pass.

Thanks
Chris

[ccarter]

Chris,

All the retransmits occuring with several devices or one device in particular?

I have seen a short macrocycle cause significant problems with download times, and increase the retransmit somewhat rate at least somewhat. Additionally, a short Macrocycle time can cause downloads to take FOREVER. This is especially true if scheduled communications (Compel Data or "CD", and also called Publisher-Subscriber) are spaced in a way that only short burst of time are availble between them for unscheduled communications (Client-Server and Sink-Source) to take place. Downloading is a Client-Server operation and since it is only allowed to happen between CDs, it attempts to perform a tiny piece of the load to the FFLD each time.

If it is safe to do so, you might consider lengthening the Macrocycle and then look to see if the retransmit rates drop accordingly. If that works, you can then experiment a little with shortening the macrocycle to achieve the best balance. The recommendation given in the latest rev of the FF Engineering Guidelines calls for at least 30% unscheduled time. I can see that recommendation working once everything has been downloaded to, but I can't promise it is long enough during the initial downloads. I have also seen recommendations that call for as much as 70% unscheduled time, so I guess it is somewhat a user's call (after various ratios of Scheduled to Unscheduled percentages have been tried). I am sure some of the others who visit this group and have significant experiences can add to this as well.

If lengthing the macrocycle does not lead to significant drop in the retransmit rate, then something else is going on. This can be especially true for new installations during commissioning when all the devices are getting their respective communications acts together. (It notably seems to occur with a few types of devices, especially if those devices have been put on a bench or off-host system and set up prior to plant installation.) If you sustpect this is a problem, just confirm all involved devices are correctly commissioned and stable.

Chuck Carter
Center Director
The Fieldbus Center at Lee College