Internet PLC Forum

What are these data? Are they 16-bit data in DM or 32-bit data A to Z? Are you certain that you have read from the correct register address using READMODBUS? The addresses used in READMODBUS are zero offset from 40001. E.g. To read from address 40011, you have to read from address 10.

Note that READMODBUS only return 16-bit data from the slave. You may want to check

I setup two PLCs, slave set to ID = 5, then initialized with the data as follow:

DM[1] = 1332
DM[2] = 2992
DM[3] = 2660
DM[4] = 1400

On the master, I have an input trigger a single custom function as follow:

READMB2 3,5,1000,DM[51],8

After execution, I got the following results on the master:

  DM[51] = 1332
  DM[52] = 2992
  DM[53] = 2660
  DM[54] = 1400
  DM[55] = 0
  DM[56] = 0
  DM[57] = 0
  DM[58] = 0

So the data are read correctly from the slave as expected. I did not observe the negative results you mentioned. Did you really execute the instruction:

  READMB2 3,5,1000,DM[51],8

Or did you use the RTU form READMB2 13,5,1000,DM[51],8 ?

If you are using MODBUS RTU mode only for COMM3, you may want to use the SETPROTOCOL command to set the COM port ino RTU mode only. This will eliminate some of the guest work involved by the slave CPU when it is in Auto protocol mode because the CPU has to try to determine if an RTU command is valid at any one time.

In your original post you seem to suggest that the READMB2 does not read the data correctly even between one master and slave. But now you reveal that you are using multiple master/slave PLCs, then it seems more likely that there are problem in your setup of master/slave communications between multiple PLCs.

We have a number of customers using multiple slaves (10 or more) in RS485 network, mostly using the NETCMD$ command but some also uses the READMODBUS or READMB2 command for communications. There were no  similar report that the DMs were not correctly read. There are many ways for the data in DM to change, most often the changes is caused by the PLC program itself (assignment of DM to other values in the other part of the program) either in the master or the slave? Do you have more than 1 PLC using the same ID? You should try to isolate where the problem occur by letting the master PLC monitor when the data change and pause  for you to examine where thing goes wrong.

Do you mean that when an error occur, it affects multiple PLC simultaneously? i.e. before the first error is observed, all PLCs runs fine but when an error occur, a few PLCs go into different error mode? Sounds like a spike of electrical interference could have occurred and cause various PLCs to enter fault state

1) Is there large frequency inverters in your environment, or large motor contactors? The RS485 should not be bundled into the same trunkings or conduit that carry very noisy power system such as inverter or motor current. If your environment comprises a lot of large noise generator then you should try to run a separate metal trunking for the RS485 and ground the trunking at one point to the earth ground to create a Faraday cage effect to shield the interference.

2) Secondly, each PLC's own 24V power supply wire must not be bundled with the motor or inductive load current in the same control panel. Although they may still work when run standalone, the problem caused by electrical noise entering the power supply to the PLC may get amplified when the PLC s are linked together in a large network. Power supply to the PLC is like blood in the body and contamination to the power supply is like poisoning of the blood and can lead to serious problems.

3) Of the 31 PLCs, are there any unit that never enter error state? If so, how many of them actually never fail? How many  units experience different fault state?

4) If you keep the RS485 wire connected to all the PLC but stop the master PLC from sending commands, does the PLC still experience fault state after several hours? This test is to determine if corrupted host link commands are the problem or just mere connection of them to RS485 itself is already a problem.

5) Did you connect the RS485 to two biasing resistors that are pull to a +9V and 0V separate power supply? This helps to keep the RS485 line in known state when no master is driving the RS485 line and can be effective in reducing communication errors caused by floating RS485 picking up noise voltages.

If the RS485 driver are also damaged, then you are getting more than noise interference. In fact they are receiving enough energy to kill the RS485 IC driver. Please check your RS485 wiring to ensure that there is no possible leakage to other high voltage sources

If you can't shield/isolate the PLCs from the electrical damage, then you may want to consider using Fibre optic cable to link the PLCs together. To do so you need RS485 to fibre cable converter. An example is shown at our webpage at:

http://www.tri-plc.com/ioboard.htm

Scroll down to the end of he page and see a picture of RS485 converter made by RE Smith (www.rs485.com, model FC485). We tested this unit previously and work flawlessly.

           

If using Fibre optic cable to link the PLCs together, how many units RS485 to fibre cable converter is needed for 31 PLCs. How about the cost?

Other them replacement of Fibre optic cable, do you have any more experiences to shield/isolate the PLCs from the electrical damage?

To kill the RS485 IC driver, how much energy or noise interference is needed?
How much energy or noise interference is needed to affects multiple PLC simultaneously as follow:
1) Some PLC RTC indicator LED turn off. (without setting the real time clock)
2) Some PLC Pause indicator LED turn on. (message display "Pause by user" in On-line Monitoring but user didn't do that.)
3) Some PLC Run Error indicator LED turn on. (message display "Divide by Zero" or "Illegal Opcode - Please inform manufacturer!")
4) Some PLC cannot detect the ID address.
5) Some PLC ID address was changed.
6) Some PLC no respond.
7) Some PLC no error.

What difference between one master / slave PLCs and multiple master / slave PLCs?
How does the setup of master / slave communications between multiple PLCs make this problem?
Can you show me the example of correct setup  of master / slave communications between multiple PLCs?

Thank for your supports.

It is quite rare that induced voltage from electrical applicances have enough energy to kill RS485 driver (typically only surge induced by a direct lightning strike to the building could cause such damage). The first thing you want to investigate should be the possibility of some short/leakage of the RS485 wire to high voltages. Also make sure that you don't  bundles the RS485 wires with power cable that are carrying high energy motor current. Why not start off by removing the existing RS485 cable, then run a separate shielded twisted pair cable to link up 6 PLCs at a time, forming five different regions. Without any communication going on, check if the PLC runs into any fault state after some time.  Then eventually link up the five regions together into a large network and connect up the master to run the test.

It is important that all the PLCs that are linked together share the same 0V. Typically, you connect the 0V of each PLC power supply to the shield to provide a common reference.

Fibre optic does cost more but they are free of electrical interference and would be the only solution if you can't resolve the electrical problem in your system.