I work in a big Company that work With modbus systems. Modbus that Controls lights, heating, VAV/HVAC, Fire Dampers, UPS, repeaters for modbus etc.
we're having trouble finding out which is the trouble and have been trying several type of Methods to Ensure that we have eliminated all ends.
- reverse polarity
- combined ground
- eliminating noise on the bus
Fire dampers go via 240v and the rest go via 24v AC. The problem is that the fire dampers are the problem here and we can't figure out which is the problem. there are several floors that has the same system which we combined the ground for eliminating noise on the bus. but on the last floor we haven't had any Luck so far. The system go via Niagara and we've put a JACE up at each repeater to check how the communication on the bus was after each repeater (there are 2 Modbuses in this floor and each BUS has 3-4 repeaters). One bus is of course better than the other but it isnt good enough for delivery, but for Modbus 1. It's like from every repeater you go to and the further you get to the end of the modbus the better it gets.
But for the other it's more like the opposite. The main problem here is the signal. we are having interference on the buss line so Our Components that run on 230VAC are not able to have a good and stable communication. So we are hoping that you could give us some inputs on what to do here.
1. I'd read the thread here
2. I'd re-read the app note "Ten ways to Bulletproof RS-485 Interfaces"
I'd figure it's time belt and suspenders all around:
3. Cabling: foil and braid shielding over twisted pairs, like this Belden cable.
I'd use one pair for the A/B drivers and the use both conductors in the other pair for the signal ground.
4. I'd get the RS-485 comm cables out of the trays/conduit that carry power or other switching loads. When you have a problem, it's when the pocket book opens up to pay for the separate cable runs away from the noise generating cabling.
5. I'd pay for the RS-485 isolator vendor's next vacation with all the isolators I'd put install, one for every node/drop on the network.
Great post David_2... this is our community!! Nice job!! Appreciated:)
Cheers from: Jeremy Pollard, CET The Caring Canuckian!
Crisis, necessity, change
Integrator, Educator, Consulting, Columnist - Control Design
Another thought - lack of bias to voltage supply or ground on the RS-485 driver lines can cause problems.
ICP-DAS makes a small module that provides a wiring terminal block for connecting +V, ground and switching the bias to the driver lines.
The trick is finding the +V and GND on the RS-485 printed circuit boards and bringing those 'out' to the real world as connections.
Almost all the problems I've fixed were very low level. Often with this type of chronic problem you have to get down to the wire level and first find out what exactly the signals should look like. Then get a scope and actually look at them. As mentioned, termination, excessive loading, stubs, even connections, bad wiring, poor wire routing, isolation, it's often a few of these, especially with long runs. Take a look at the signal integrity and fix as you go along. I've seen Modbus work with truly horrible looking signals, the problem is sometimes it doesn't. If you fix where the signal starts looking bad and keep going, pretty soon it makes a big difference. It's not just wire, it's a differential transmission line and when it looks like one, it works well. Repeaters can help, but if the input is crappy enough they aren't much help. You'll learn a lot looking at the waveforms. And you'll often be amazed that it works as well as it does.