comms analysis first go

[richerdick]

i am using some software that monitors packets of data to and from the comms port of a plc and displays it in hex.
the comms port on a working system is an rs485 link with 4 nodes on one branch and 4 different nodes on another parallel branch each branch has 120ohm terminating resistors across the end node and there is 2 120ohm resistors across the rx tx of the coms port plug itself.

The plc is set to control 4 nodes on one branch and 4 nodes on the other. each node gets its adress via links on the plug and socket that supply it.

I am trying to understand a bit about the protocol that the plc uses to communicate. I have set up a test rig with our spares using the same type of plc with two nodes one of each type on one leg with terminating resistors at each end.
looking at the comms monitoring program i can see the data being to transmitted to each node. however the reply packets are not as expected the plc adresses each of the 8 node it would normally address with packets as i would expect. Only 2 of the nodes reply but the reply data is random. perhapps 1 out of 5 reply packets from the connected nodes is in the expected format the other 4 will be complete nonsence but they are constant not erratic. i know it is wrong nbecause i can look at the acual machine working and all reply packets are as expected. am i missing something fundemental. i ddnt use twisted pairs as there is only about 2ft of cable. do i need to change termination resitors? what could i be issing this is my first venture into comms anaysis

 

[danw2]

Which protocol is being used? Yours? Proprietary? (if so, name?) Open CAN? Modbus? Profibus?

 

[richerdick]

Non standard, half duplex. All i know thanks

 

[mls1]

It would be helpful to see some of the packets and to get more information on devices addresses in the system.

Another problem that is often overlooked is bias resisters (not the same thing as terminating resisters). They maintain a voltage across the lines in the idle state. If the combined bias resistance from all devices is too low it can cause to much load on the drivers.

There is a free RS-485 troubleshooting guide here:

http://www.bb-elec.com/guide/RS485_Trouble_Shooting.pdf

 

[richerdick]

i have now got comms running in the test rig by removing all termination resistors. the original drawing stipulates that three pairs of termination resistors are used one at the plc serial port and one at each end of two branches of the comms link each branch goes off to a different type of node 4 nodes on each branch.
anyone shed a little light on why it might work without terminating resistors but no with?

 

[controlsdude]

If the distances are not too great, it has been my experience that you can get away from using the terminating resistors. Of course if your cable is of good quality this also helps. But not using terminating resistors on all networks is not recommended. Every comm network that I have dealt with had terminating resistors.

 

[richerdick]

i have investingated further and found that the drawings wre wrong in fact there are no terminating resistors present on the network ????

 

[EricRICPDAS]

I think you don't need terminating resistors with only 2 feet of cable. For example, at 115200 baud, I have only needed terminating resistors with 600+ feet of cable. Too many terminating resistors can cut bias voltage to too low a level.

Eric R

http://www.icpdas-usa.com