GE Rx3i Modbus TCP/IP Channel Dilemma

Greetings to the forum. Wondering what experience exists out there for this problem I'm currently encountering with an Rx3i controller. Have six different LD blocks set up for six Modbus TCP/IP capable rectifiers. Using the GE PACSystems* RX7i, RX3i and RSTi-EP TCP/IP Ethernet Communications User Manual GFK-2224W, I have structured six different channels to go out and communicate open and closing connections with these rectifiers. My issue is, one or two channels work at the same time and when I add more, they start failing. It should be noted that these commands are going out over a two port ETM001 card. One port goes out to separate switch that connects to my rectifiers, the other goes out to a series of NIU racks.

When the channels try to communicate they get failures CRS codes such as invalid channel #, comm discarded; command issued before previous status completed, and channel already open (even though not active). A few things and considerations I have pondered:

* Can a single ETM001 using both ports separately on one IP, not handle both Modbus TCP/IP and NIU rack communications at the same time without overloading?

* ETM001 Max FTP Server Connections in settings is set to 2 = modulus. I think this has to be set to 16 high limit for maximum allowed connections, but if I change the previous integrator's modulus 2 setting, the NIU's seem to stop communicating back to my HMI (separate issue there).

* Is it possible the NIU (even though I see no code requesting channels on main or NIU racks) with their symbolic activity are utilizing channels I'm unaware of causing overlap channel issues, thus issuing my invalid channel requests?

I'm trying to use channels 1-6 for my rectifiers. I got one working online now, adding multiples is when the trouble starts. I should be able to have up to sixteen channels operating separately all at once, each sequenced properly, but not in order with one another? It should be able to do that according to documentation. Sorry for the long write, its a complicated issue. If anyone has experienced this and has any ideas, I'm all ears and thank you for any input you can provide

You have to sequence the ModBus communications. Cannot talk at same time.
I am talking to 4 devices and pulling 16 registers from each. Using ETM0001.
I use a counter to sequence the comms at 0.30 mS (3 devices a second). You can probably push this.
If I remember..... you get a bit that makes on comms complete.
Are you reading and writing?

Multiple channels should be able to talk and operate individually at the same time right? That is what I believe I read, up to sixteen channels at once, but each channel must be sequenced properly to not have issues? Do channels affect one another?

I have six channels trying to operate at same time, but each one is sequenced properly I believe. I clear the CRS word, load my block, make the comm_req, then watch the CRS code for a 1 which triggers a one shot to the next structure.

Beyond that, I did see someone say firmware updates happened on both CPE305 and ETM001 due to being shaky on Modbus. These cards are original from GE, never had firmware updates, so I believe I should also go to the customer portal and look at updating these. Not sure if that is my issue, but possible?

And yes, reading and writing, but one comm_req at a time waiting to go to next between each command (waits for CRS good response to continue). The structure is much like what was advised in the GFK-2224 manual chapter 9 examples.

Yes...I would recommend you make sure the CPE305 and ETM001 have most current firmware. There were lots of updates relating to ModBus TCP.

I think you can only talk to one device (channel) at a time and you can only write or read (2 comm requests) per device. I would probably do all my reads then all my writes.

Did you read through section 7.3 in:
GFK-2224W

Yes, it was stating only one comm_req per channel can be pending at one time, so given that, I'm assuming I can have one Comm_req processing in each of the six channels, but not more than one in a given channel. That is how I have it structured, but it still fails. Also, I do not get any CPU or diagnostics reports indicating an Ethernet interface overload, so I know it is not that. GE's first question was "Has the firmware ever been updated?" to which my answer is no....so I'm hoping that is actually my issue. I will be updating my CPE305 and ETM001 first thing tomorrow morning and have the line down for testing all day. Thanks for all your feedback

If you do just one channel can you read and write?
Then try 2 channels and so on.
Where does it fail?
Have you tried talking to multiple channels independently? I have 5 channels, but pull data a different times, not at the same time (0.2 sec each). I am only reading, so I can leave the channels open and sequence the comm requests for the reads.

Do you have the firmware updates for the CPE305 and ETM001?
Are you monitoring the status bit of each Ethernet channel for each device?

Yes, I can one to two channels to work, multiples start failing beyond that. I can read and write in a sequence of one shots successfully on a single channel all day long no problem. When multiple channels are added in, all but the first start failing once I go for three channels or more. I use one shot sequencing with no timers involved. The more eyes the better, I'll post and if you want feel free to take a look. Perhaps you see something I do not that is causing a hiccup. I monitor each status bit on each channel and followed examples from Chapter 9 in the GFK-2224 manual. I think my revision is July 2018 2224W. I'm going out now this morning to extract ethernet data, upgrade firmwares, etc. Maybe I do need delays on separate channels? I do not get any CPU or ETM reports of overload or any errors at all in the CPU.

Some of my errors on the additional channels include CRS codes of:

8890 - Invalid Channel Number
B490 - Already Open (even thought the status bit is not)
AB90 - Comm Discarded, Issued Req Before previous status word set
AA90 - Connection to the TCP Server has failed.

That last one is alarming because all cables and power are good to rectifiers. I can ping them at all times just fine.

The one big thing that irks me in all of this, the ETM001 card settings has its "Max FTP Connections" set to 2 = modulus. I swear that represents the amount of channels I'm allowed and that needs to be 16. The previous integrator that uses the second port on the ETM001 to communicate to NIU racks (which I'm not very experienced on), set this value to 2. If I change it, it messes up the NIU racks and I do not get data back, but that is a separate issue. I feel this also a potential culprit. I'll test that today too. Anyway, if you see anything funny, feel free to comment and again, really really appreciate your input on this. I know continuous troubleshooting will get me through it. Thank you!

Also, do not see a way to attach a zip. I upgraded with a $20 donation that I think allows it? I don't see my account updated though, perhaps it takes a little bit. I'll keep an eye and post the file when I'm able to if you want to see it.

Have you called GE Tech Support to ask them about how many channels?
Have you upgraded the firmware?

SUCCESS!!! It works! So here is what I had to do:

1. I was not making full use of LAN status bits at the beginning of each sequence.

2. My close instruction was missing a single one-shot instruction and keeping some memory locations active which resulted in some channel errors.

3. I never updated the firmware. The ETM module was actually replaced and was up to date. The CPE305 CPU has an update, but I did not install it where everything is working perfectly. I would like to at one point, but today's goal was getting production lines running as quickly as possible.

I also moved my channels from 1-6 to 11-16 higher up the ladder in case the NIU racks were secretly using any channels, but I do not think they are. Anyway, I'm posting a copy of the completed file working for others that may come across this thread, it is a good example of sequencing MODBUS TCP/IP commands in a GE Rx3i in blocks labeled MODBUSRECT1 though 6.

I also added two reset MOVE WORD instructions at the end of each channel, not part of the communication, but resets HMI rectifier values on screens for operators, otherwise those registers are retentive and the line looks like it still has amperage and voltage when it does not.

Thanks for all your help Nathan, it got me thinking and going in the right direction. GE never got back to me on my case haha.

GREAT!
Glad you got it resolved.
It is always a little tricky to get the comm request sequence to work. But now that hurdle is over.

A note for future viewers, I had to add a 5 second delay to the beginning of my close operation. Apparently trying to initiate a close from open too fast does not allow all my open conditions to properly terminate quick enough for the close comm_req and can generate a 43920 register value which represents error code AB90 meaning:

A COMMREQ was discarded because the application program issued the COMMREQ
before the COMMREQ Status Word for the previous COMMREQ was set.

A five second timer to start the close operation alleviated this issue and gave the previous open commands enough time to finish. The close seems to happen fast and I have not needed to add a delay for re-initiating my open request. I can follow up immediately after a close without issue.

To both ZestyMozzarella and NetNathan regarding your experience with tech support. Have you noticed any difference since Emerson has taken over?