Modbus RTU wiring

[aallen]

Hello all,
I have a relatively simple question regarding RS485 modbus RTU.
When wiring for multidrop, or daisychain, what is the proper way to do it?

Everywhere I can look just says to daisy chain at the instrument, but what if it is not possible to double land wires in the terminal?

https://imgur.com/a/mFgaJQK

Is something like this acceptable? If not, can you show me including any T's or terminal blocks that would be required?

Sorry for the noob question!

[the_msp]

Don't do it that way.

Use belden cable s/stp whch is like 0.3mm sq. Two pairs go in a 0.5 or 0.75 crimp.

[aallen]

Originally Posted by the_msp

Don't do it that way.

Use belden cable s/stp whch is like 0.3mm sq. Two pairs go in a 0.5 or 0.75 crimp.

Will solid wire F/UTP with impedance 100 ohm work? I know it is not optimal, but is the foil shield good enough?

[OkiePC]

Ideally, you want shielded twisted cable designed for use with RS-485 properly terminated with no extra terminations or "star" wiring, strictly daisy chained, with termination resistors at only the end nodes.

In real life, you can get away with some compromise. How much you can get away with depends on several factors, the two most important ones, in my experience, are the number of nodes and total cable length.

I would recommend against compromising on cable quality, and would opt for really short jumps from terminal blocks as in your first post...if you have to choose between crappy cable and extra terminations, go with the extra terminations.

Try it, run some tests to see how many (if any) communication errors you have, and if there are devices that generate electrical noise in the (physical or electrical) vicinity, have them running when you test for worst case performance.

I had a system with one PLC and two VFDs and about 800' of RS485 wiring which was installed by an electrician who failed to route the cable all the way to my panel, so I had to run about 40' of cable to a terminal block where it split off to the two drives, one about 30' away and the other 750' away. It worked fine, but I did have about a 5% error rate. This was acceptable in that case just controlling two water pumps where as long as I could message both of them within 10 seconds, the drives were happy, and performance was fine.

[aallen]

It's currently wired up with two 18 AWG wires in the topology similar to my crude drawing in OP.

It works.. sort of.. Probably 10% of the polls result in a timeout or some other error that causes me not to receive the data requested.

But I would like to redo it correctly in the future. (not my original install, just inherited it)

[Lare]

Actually it can work if you keep distances short (some 30cm to couple meters), but it is not best.


Scheider have these documents.
https://www.se.com/ww/en/faqs/FA221785/
https://www.se.com/nz/en/faqs/FA275097/


Also Schneider have modbus splitter LU9GC3 which have inner schematic as your own wiring. It have some 3m long cables sold as seperatly.

https://ckm-content.se.com/ckmConten...0DA0000000abSm

And as we know Schneider (Modicon) have invented modbus so they would't make splitter if it won't work well.

[OkiePC]

How many drops (nodes including the master)?

How long are the RS-485 cable runs?

I have seen Cat-5 cable used for RS-485 comms work just fine for a few feet even without terminating resistors. It will certainly do better than just two individual wires!

[aallen]

Including the master there are 11 nodes. As far as distance goes, I couldn't tell you.
The trunk line is probably in the realm of 200 ft, but as far as I can tell there's multiple backbone with stub/star topologies.

I have 4 serial channels, the one I described is the biggest one though. One has 3 nodes, one has 1 node technically because it is a HART > TCPIP converter, the other channel also has 3 nodes.

[Lare]

Originally Posted by aallen

It's currently wired up with two 18 AWG wires in the topology similar to my crude drawing in OP.

It works.. sort of.. Probably 10% of the polls result in a timeout or some other error that causes me not to receive the data requested.

But I would like to redo it correctly in the future. (not my original install, just inherited it)

There should be end terminators (120-150ohm) still on both ends (longest line. Usually one on master side and other on longest distance slave)

[aallen]

There is a termination resistor on the master because I enabled it, but due to the wacky topology I am not sure where to even place the other termination resistor.

[Lare]

Originally Posted by aallen

Including the master there are 11 nodes. As far as distance goes, I couldn't tell you.
The trunk line is probably in the realm of 200 ft, but as far as I can tell there's multiple backbone with stub/star topologies.

Star is probably biggest problem.

[Lare]

Originally Posted by aallen

There is a termination resistor on the master because I enabled it, but due to the wacky topology I am not sure where to even place the other termination resistor.

You need to draw it to paper first. Then mark all line lenghts for checking where you locate antother resistor.

For 200ft (~61m) another resistor is needed on other end.
Or simply test it by locating resistor to different places. RS485 on copper should work 98-100% if done well.

[OkiePC]

+1 to what Lare just said.

Sketch out the network, and since it sounds like you have 3 (or 4?) of them, pick one and conquer it first.

[Peter Nachtwey]

There is no info on the devices. The devices should have serial buffers so they aren't all driving the lines a once or looking a load all the time. If there are buffers then they should be in a high impedance state until the sending program enables the RTS ( request to send ) to enable the driver. I would look at the documentation for the devices.



Ditto what Lare said about the terminating resistors. A long time ago I used to drive 32 slaves with one master. That was before Ethernet and just about everything I have done was a long time ago.