T3-886 MODBUS TCP issue

I have a T3-886 that I am polling roughly once a second over a radio link. It will work perfectly for roughly a month then stop responding to polls. The LED’s indicate the T3-886 is seeing network traffic. When I plug into a local network switch T3000 does not find the T3-886 when doing a search. Powering down and restarting the T3-886 solves the problem temporarily. Has anybody else seen similar issues?

I would update the firmware to get started. If you can do a block poll rather than register by register you can get more data with less polling and therefore less load on the T3.

Unfortunately, the client’s site is several hours travel to service, so I have not been able to update the firmware in the current onsite yet. I did swap the hardware last trip and I can verify that the problem is not specific to a specific unit because the replacement just failed. I did update the firmware on the unit I removed, and the change is from 5.8 to 5.9. I did not see any changes listed that look like they will affect the MODBUS. I am already a block read (the only transaction at this point). Interestingly there is a 22i in the same cabinet also being accessed with a block read that does not appear to be going offline.

We have done a ton of work on communications over the years and knock on wood things are pretty solid. That said, we’re always on the lookout for unexpected situations.

One thing you could try is reducing the polling to once per two seconds. A large block read at once per second is a fair bit of traffic right.

When all the obvious has been tried then we’d need a wireshark capture of the data to dig deeper.

I will pull together the equipment to set up wireshark capturing. I suspect we are dealing with a memory leak of some kind or looping problem with the software. When the failure occurs the 886 will not even respond to a ping request unless powered down forcing a reboot. This is a remote unmanned site and fortunately this client is reasonably tolerant so I am hoping we can identify what is going on before I am forced to different hardware.

During fairly heavy polling the communications just stops and the device has to be reset, that sounds like something we have been through from a long time ago. I urge you to do a firmware update and if that doesn’t help make sure you are using the block reads, and if that doesn’t help back off on the polling rate. These devices have been around a long time in hundreds of different environments. We always keep our clients happy.

I am not unhappy but I am using the 886 as part of an operating water supply system so uptime is mandatory. Both the 22i and the 886 at the site are currently being polled every 2 seconds. I am only requesting 16 words of data from the 886 (44 from the 22i) in a single block read so I lean towards a problem caused by communication latency rather than overloading.

So once you have exhausted all the usual culprits it’s time to haul out the wireshark tool. The logs will tell you what’s going on.