Modbus cabling?

[TwoBlocked]

I have a question about wiring a "daisy chain" architecture between Modbus capable devices. Of course you need two pairs for each device. One pair coming from a previous device, and another pair going to the next device. At one end of this "daisy chain" is some sort of contol module and the other end is an end of line resistor.

As far as the actual cable runs, the cables could physically go from one device to the next, to the next, and so on. Or they could all go to one location and be daisy chained together there with jumpers. Physically running all the cables to one physical location is what I have a question about, particularly about the cables themselves.

I do not see a problem if you have two twisted pair in one cable, individually shielded, one designated as going "in" and the other going "out". But what if the cable has four conductors, and rather than individually shielded, are over-all shielded, and rather than two twisted pair all four are twisted together? Two wires would be designated as "in" and the other two designated as "out". Would this cause an interference or cancellation problem? Have any of you fine folks seen this and has it caused problems?

 

[coop3339]

It is best for each pair to be individually twisted and shielded. Having all 4 twisted together will probably cause issues. Twisting the pairs keeps them in close proximity to each other allowing the magnetic lines to cancel eachother out to help eliminate noise and inference. The end of line resistor is more of a dampening load to block the signal from reflecting. The modbus masters may also have internal resistors for the same purpose.

Depending on the speed of the comms and distance between devices, what you are proposing might work. The higher the baud rate and the longer the runs are will make it more likely to fail. The shield should also only be bonded to ground in one location. I usually ground it at the master side, with the rest spliced together at each location but not grounded. You want the shield to act like a wick or antenna to pull the interference to ground. If you ground it in more than one location it can allow a current to flow making your problem worse.

 

[herding_cats]

Use #22 shielded cable. Red/Black as A&B.

Ideally with grounding you would want a “drain” in the middle of the run. So connect all grounds on the daisy chain, but don’t try to ground them individually. When finished with your device install, find the midpoint of the run (if you had 10 devices, go to #5) and then ground that drain point. Use a separate pair for power (12-24vac/vdc). You don’t have to use shielded cable for that. Use something of a different color all together. Like control wire pair, blue/white for example. Or thermostat wire.

If you have 2 pair with shield/ground, being 4 conductors and ground, do not use one pair for data and the other pair for power. There is special wire out there with one pair shielded and the other not, but it’s stupid expensive. Run two leads separate.

 

[TwoBlocked]

Since the drain wire is not insulated, I find it risky to continue the drain wire from one run to the next. Too much of a chance of it becoming grounded at both ends. Instead, I ground the shield at the master and tape off the other end. When the next run is continued, in a daisy chain, I ground in the panel where the first run was taped off and so on. I do this with analog wiring when continuing in a remote panel to the field devices, too.

What I run into is experienced guys saying "Oh, I know what should be done, but I've never had a problem doing it my way." I prefer to do it professionally.

 

[herding_cats]

The ground drain is insulated from the sheathing.

If the daisy chain grounds are not connected then the shielding isn’t grounded.

Only the installer controls if the ground is grounded at both ends. What’s risky here?

 

[TwoBlocked]

The ground drain is insulated from the sheathing.

If the daisy chain grounds are not connected then the shielding isn’t grounded.

Only the installer controls if the ground is grounded at both ends. What’s risky here?

What I mean is:

Cable A goes from the master to a grounded terminal junction box. Cable B goes from the grounded terminal junction box to the field device. Rather than landing the bare wire drains of both cables A & B on a terminal strip in the junction box, I prefer to tape off the drain on cable A, and land the drain of cable B on a ground in the junction box. If I were to do the former, land the uninsulated drains of both cables to a terminal strip, one or both could come into contact with ground in the junction box permitting a ground loop to develop. How could this happen? Some other electrician could get in there and do who knows what with other uninsulated drains. Also, extra terminals for the drain are usually not available, or what is available are grounded terminal blocks. Those of course would defeat the purpose. Sure, I could do the spaghetti tubing thing, which I like to do for appearances, but to do so to continue the ground doubles the work.

 

[zbang]

Sure, I could do the spaghetti tubing thing, which I like to do for appearances, but to do so to continue the ground doubles the work.

Adding seconds to each termination; if it's the right thing to do, just do it.

 

[TwoBlocked]

Adding seconds to each termination; if it's the right thing to do, just do it.

That sounds judgmental.

 

[zbang]

How so? We do many things that might not be required but lead to a more workmanlike job, like using fork or ring terminals under some screws or dressing the wire forms. Sleeving on bare wires is just another part of that.

 

[TwoBlocked]

I am checking out of this Topic. Happy Thanksgiving.

 

[brycenesbitt]

But what if the cable has four conductors, and rather than individually shielded, are over-all shielded, and rather than two twisted pair all four are twisted together? Two wires would be designated as "in" and the other two designated as "out". Would this cause an interference or cancellation problem? Have any of you fine folks seen this and has it caused problems?

Why would this come up?
The whole goal of a twisted pair is to have opposite voltages exposed to the same noise, so the difference is zero.
Adding wires is just making the system more complex for no apparent noise margin benefit.