petersonra
Senior Member
- Location
- Northern illinois
- Occupation
- engineer
Modbus/tcp is a completely different animal from modbus physically and speed wise.
Modbus RS485 Ground vs Drain
- Thread starter TwoBlocked
- Start date
zbang
Senior Member
- Location
- Roughly 5346 miles from Earls Court
I'd say about the only "modbus" part is the actual commands (and responses)- no target ID, no checksums, no multi-drop, etc.
(z goes hunting around the office for some term resistors I need for a project.... might not make a difference for a 5' long connection but then it might.)
(z goes hunting around the office for some term resistors I need for a project.... might not make a difference for a 5' long connection but then it might.)
petersonra
Senior Member
- Location
- Northern illinois
- Occupation
- engineer
zbang
Senior Member
- Location
- Roughly 5346 miles from Earls Court
Well, in a TCP packet in an IP datagram in an Ethernet frame. The limit isn't the number of ethernet addresses, the limit will be the number of reachable IP address minus one (for the initiator/server); that gets into IP addressing and routing.)
herding_cats
Senior Member
- Location
- Kansas
- Occupation
- Mechanical Engineer
this is fine. What you don't want is grounding on both ends. That can cause weird transient issues. If you can ground at the center point of the ground wire run, that is technically the best practice. Always use shielded wire and avoid high volt wires in runs. Never run parallel to even a 120v wire.
zbang
Senior Member
- Location
- Roughly 5346 miles from Earls Court
That's an old wives tale, it matters little for RS-485 (a balanced line) and especially one that's shielded (which is why the shield is there in the first place). Likewise, since RS485 is a balanced line it doesn't need a "ground" for reference (it's optional).
Birken Vogt
Senior Member
- Location
- Grass Valley, Ca
Kohler uses RS485 comm to their transfer switches. I have seen it ignorantly run on individual THHN all mixed with the generator power and battery charger power lines in one conduit. Working fine. I think I mentioned that earlier. Not recommended but that is what I saw.
brycenesbitt
Senior Member
- Location
- United States
Yeah, kind of. Optically isolated RS-485 works great without a ground.
--
In summary under the 2020 NEC it appears that running RS-485 with power conductors for associated equipment is fine, as long as the cable is Type TC which indirectly calls out the required 600V rating for the interior conductors. Must be 16 or 18 gauge. Apparently, a sheath is optional but I'd never put anything other than sheathed shielded cable in given that they are 33 cents/foot or less.
---
The nonsense about running next to 120V circuit is just that. You'd have to manufacture a situation combining poor RS485 practices with a particularly nasty motor, chopper laser or some of piece of equipment doing violence to the 60Hz sine wave. And at that point you might have the RS485/Modbus packet loss anyway.
Last edited:
rlundsrud
Senior Member
- Location
- chicago, il, USA
Much like rs485 signal transmission, modbus rtu uses a differential signal which rejects common mode noise. It operates at very low baud rates (9,600-19,200) and is very robust. I do building automation for schneider electric and we do a lot of data center work where this protocol is used regularly. We use rs485 low capacitance (typically <14 pF/ft) tinned 24 awg twisted pair with shield) cable but I could run modbus on a pair of coat hangars if I wanted to.
We never use terminating resistors as it just adds more load. The reflected signal only becomes an issue with distances greater than 3,000 ft at 19,200 and 6,000 ft at 9,600. We exclusively use 9,600 as the data density is very low so running at higher baud rates isn't necessary.
As far as using the shield as a ground, you can use it if you have equipment that requires a zero volt reference but I have never come across that with modbus. I do see that with bacnet ms/tp and we do use the shield for this purpose. The reason you can do this goes back to the way differential signaling functions. There is lots of info available online that spells this out how the differential signal rejects common mode noise.
The traditional method is to terminate the shield at the origin (where the RTU daisy chain originates) and twist the shield wires together at each controller then tape it back rather than terminate it at the field controller. With that said, many field controllers do have a terminal for the shield termination. It is not connected to anything internally and is just for the purpose of landing the two shield wires. This is how I have my techs terminate the shield, I would still check your product literature to confirm this is the case for you.
As far as the A- and B+, I have seen it labeled both ways but A- is more common. In every case I have seen, the minus is nearer to the common or shield termination on the field controller. Always use the white wire as the plus (+) and the black as the minus (-). We have had installers use white as the minus because electricians think of white as a neutral. It will work but it can cause confusion in the future.
One last thing, always use a daisy chain, never use a star topology. When people T-tap a daisy chain it can create terrible signal reflections. You might get away with a short stub T-tap that is less than 50 feet. The problem with this is you may not know when you pass the threshold of operability. It might work initially then fail a week or even a couple years later as the transceivers age. I have a saying for this, you can sin a little and you might be forgiven but if you sin a lot you will go to hell. The moral is don't sin, follow the standards. This way you won't have to worry about the cabling if you have issues getting things online.
I hope you find this to be helpful. I know I went a little long on this, but I am getting old and I need validation now and then.
,' -)
We never use terminating resistors as it just adds more load. The reflected signal only becomes an issue with distances greater than 3,000 ft at 19,200 and 6,000 ft at 9,600. We exclusively use 9,600 as the data density is very low so running at higher baud rates isn't necessary.
As far as using the shield as a ground, you can use it if you have equipment that requires a zero volt reference but I have never come across that with modbus. I do see that with bacnet ms/tp and we do use the shield for this purpose. The reason you can do this goes back to the way differential signaling functions. There is lots of info available online that spells this out how the differential signal rejects common mode noise.
The traditional method is to terminate the shield at the origin (where the RTU daisy chain originates) and twist the shield wires together at each controller then tape it back rather than terminate it at the field controller. With that said, many field controllers do have a terminal for the shield termination. It is not connected to anything internally and is just for the purpose of landing the two shield wires. This is how I have my techs terminate the shield, I would still check your product literature to confirm this is the case for you.
As far as the A- and B+, I have seen it labeled both ways but A- is more common. In every case I have seen, the minus is nearer to the common or shield termination on the field controller. Always use the white wire as the plus (+) and the black as the minus (-). We have had installers use white as the minus because electricians think of white as a neutral. It will work but it can cause confusion in the future.
One last thing, always use a daisy chain, never use a star topology. When people T-tap a daisy chain it can create terrible signal reflections. You might get away with a short stub T-tap that is less than 50 feet. The problem with this is you may not know when you pass the threshold of operability. It might work initially then fail a week or even a couple years later as the transceivers age. I have a saying for this, you can sin a little and you might be forgiven but if you sin a lot you will go to hell. The moral is don't sin, follow the standards. This way you won't have to worry about the cabling if you have issues getting things online.
I hope you find this to be helpful. I know I went a little long on this, but I am getting old and I need validation now and then.
,' -)
Birken Vogt
Senior Member
- Location
- Grass Valley, Ca
How is t-tapping any different from not installing terminators?
rlundsrud
Senior Member
- Location
- chicago, il, USA
A t-tap creates a parallel impedance path because of the transceivers on that tap. This is significantly worse than not terminating a daisy chain in terms of signal reflections.
Birken Vogt
Senior Member
- Location
- Grass Valley, Ca
I'd like to see some kind of visualization of that idea, vs. reflections in an an unterminated daisy chain bus.
rlundsrud
Senior Member
- Location
- chicago, il, USA
I'm not sure what you mean by that, could you elaborate?
Birken Vogt
Senior Member
- Location
- Grass Valley, Ca
I am having difficulty understanding how reflections from an unterminated branch would be worse or different than an unterminated end of a purely linear bus.
zbang
Senior Member
- Location
- Roughly 5346 miles from Earls Court
I'll ask my friend ref'd in post #18 to try it and post the scope photos.
ModbusMan
Member
- Location
- Cleveland, OH
- Occupation
- Building Automation Engineer
Definitely watching this one... I was given the impression by one of our fire contractors' lead designers that addressable smoke detectors from Edwards (and probably others, his outfit just wasn't their local dealer) functioned on two-wire RS485 or something very similar to it, and our building was t-tapped worse than the tree in "A Christmas Story" yet still operated normally.
zbang
Senior Member
- Location
- Roughly 5346 miles from Earls Court
Issues with tapping will depend both on the length of tap and the data rate. A 20' tap at 9600bps isn't going to be much of a problem; 200' at 250kbps probably will.
SceneryDriver
Senior Member
- Location
- NJ
- Occupation
- Electrical and Automation Designer
This becomes an issue with DMX lighting control in the Entertainment industry. DMX is RS485 that operates at 250Kbaud, so there's an extremely limited tap length that has a possibility of working reliably. I have seen large lighting rigs brought to their knees for lack of a $0.10 terminating resistor. Every good lighting tech has at least a few male XLR connectors with 120ohm 1/2W terminating resistors soldered across pins 2 & 3 kicking around the bottom of their gig bags.
Newer lighting equipment seems more tolerant of "bad" cabling practices, and I've seen (and installed) several permanent architectural lighting rigs that absolutely violated the snot out of the DMX standard with T-tapping all over the place. With terminators installed at the end of each run, the rigs were rock solid.
The DMX standard further calls for ground-referenced transmitters, and isolated receivers to prevent ground loops (this is different than RS485, which allows for ground-referenced receivers as well). I've also seen what happens when devices are ground-referenced when they really shouldn't be, and it typically involves blowing the RS8485 transceiver chips clean off the PCB with a loud bang.
They've been predicting the death of DMX in the lighting industry in favor of Ethernet solutions like sACN and Artnet for over 20 years, but DMX is still around and still a huge part of our industry. It'll happen eventually, once your average stagehand learns to spell "IP address" and "subnet mask." /S
SceneryDriver