RS485 With Termination or not?

[John_R]

Hey All,

I'm looking for opinions on using RS485 With Termination or not? (MODBUS RTU on V570).

On past projects using MODBUS to talk to one or more A-B VFD's, I have turned on the termination at the PLC and let it be the "end unit". In these cases the VFD's were fairly close to the PLC.

Now I'm working on a project to collect data from multiple Mass Flow Sensors via MODBUS, and the last unit will be a hundred feet or so from my PLC. I'm wondering if I should set my PLC port to no termination, and put a termination resistor on the unit furthest away?

To me, this seems like the correct way to do things.

(I also plan to use 3-cond w/shield MODBUS cable, and only connect shield to ground at the PLC)

Right? Wrong? Indifferent?

Let's hear some opinions on running long MODBUS runs.....

JohnR

[Ausman]

Hi John,

485 seems to be very forgiving, and depending what you read, there are myriad "best methods" out there.  Some say to connect the earth at each drop, others that is a big no-no.  Do "this".....whilst other say "don't do this" on the same topic. And often the variations are encountered when reading literature from very big names in control gear.

My usage which has never given trouble and routinely run at 9600.....

My 485 runs have maxed out at about 200m.  Yes....by the time it meanders around a building it adds up!  I mostly use 2 x twisted pairs with foil shield and a drain wire, with the pairs 1.5mm².  One pair is the 485 one, and the other is power for the device.  This is purely a convenience type thing and what is readily available, but perhaps having the power inside the shield also helps keep it "pure".  If the power is already there in sufficient size, then the run is done in a single pair type, and I often change "back and forth" along the entire run.  All my runs are only earthed at the plc end, into a terminal block that is only used for comms and has a separate leaded connection to the main earth point as well as being a rail clamp type.  Stub drops are done as short as possible with bootlace pins.  Earth/drainlines at stub drops are sheathed and bootlaced together for tidiness, with a quick solder at the exposed end to stop connection deterioration, then hot melt type heatshrunk back onto the "inwards" cable at the breakout.  I try to keep things uniform in physical layout to aid later troubleshooting if necessary.

Resistors....I always have both ends on.  But there is a host of info out there about how resistors need to be carefully matched to cable impedance, but also usage with only the far drop having it.  My Shentek usb adapter happily works on all my runs without termination on at the adapter, so there is only the far end point on.  I tried to find a previously found online reference which was great, in discussing a "not functioning at all" comms issue and how they solved it with the scope pointing to just a few ohms difference on the resistors, but alas couldn't. Will post it up if I find it eventually.  Everyone will no doubt have links...another good one that I did have bookmarked is this:

https://www.maximintegrated.com/en/app-notes/index.mvp/id/763

Hope it helps.

cheers, Aus

[kratmel]

It's hard to add something, but the main thing to say is that the V570 panel has an optically isolated port. In my experience,

it's important to keep this isolation when creating a longer network link.
That is, use a separate power supply and properly ground the cable screen (if present).
Many times I additionally put an optical insulator (or use RS485 isolated add-on card) for this network

because without the isolation from time to time the drivers of the signal transmitter failed.
And the main thing to pay attention to the advice of forum participants

 

[Gabriel Franco]

I always use terminal resistors at both ends of an RS485 network, even in point-to-point short cables. Also, I use shielded cables ground at a single point. Rarely had problems with this.

[Isakovic]

Quote

When connecting to RS485 it is not just important but absolutely imperative that you do not connect wires to pins 2,3,4 and 5. Even if the opposite end is connected to nothing at all (except 1 and 6). What testing has shown is that if any of the other 4 wires touch each other (because you cut them short) they will make an otherwise quite communication line unbelievably noisy.  Also even if the ends are not touching the wires themselves act as little antennas and noise becomes worse.

This was in kratmel's link. I did not know that.

[Ausman]

11 hours ago, Isakovic said:

This was in kratmel's link. I did not know that.

In trying to help resolve another issue on the forum, but mainly for my own interest and knowledge, when I can I have been experimenting with pin usage under certain conditions, and what defaults etc always happen.  I haven't finished this yet, and will post when I do...which could be a while.  Why I am bringing this up here is Walker's reference to the 485 socket's other terminals.  In theory 232 is still meant to work whilst using 485, so that is possibly the source of the noise, but I have yet to finalise all results.

All my 485 connections to date have always been only a "dedicated single pair build into a 6 terminal plug", and I have never had trouble.

cheers, Aus

[kratmel]

At a time when the optical fiber only appeared in automation,

I sometimes must to replace it with a pair of twisted wires  as  wrote Aus (not even shielded) and the 485 receiver + transmitter.

There was also an optical isolation and all this works until today.

Now I understand that I did not even think about how fast the connection was.

But terminators were needed at both ends of the line. Otherwise there were communication errors that arose when something powerful turned on nearby.

[John_R]

Hey All,

Thanks for the suggestions/opinions, nice to see that I'm not the only one missing any hard-fast rules.

I think I'm going to follow this topology which comes from a recent (non Unitronics) project that I was involved in.

We are using Belden 3106A cable, which is listed as "1.5 pair with shield", meaning there is a pair for the signal (ORG/wht, WHT/org), a single conductor for DC common (BLU/wht), which is connected to the DC common on each device, and the Shield, which is connected cable to cable at each device but only earthed at the PLC end of the circuit. And a 120 ohm terminating resistor at the last device.

JohnR

 

 

 

[Ausman]

And the termination resistor switched on at the plc.  In my situations I'd only be needing the single twisted pair, without the .5.

cheers, Aus