Modbus Communications

[macmet]

Hello All,

First time posting in this particular forum. I'm a mech/matls guy but I've been giving the task of improving our Modbus communications in our small boiler plant.

Part of this project included utilizing existing Modbus communications from our gas burners to our SCADA system. The data from these burners is inconsistent and we often go long periods of time without any data transferred to the SCADA. However, during this time of no data transfer, we continue to communicate with the burner in terms of set-points, on/off, etc with no effect on system operation. The problem with this is that we're trying to determine effects of other parameters on emissions and although we know our burners are running, we're not getting any real time emissions data, nor are we confident in the data we do get.

In investigating why we're having issues with the Modbus, I noticed that the Modbus wire is 18 gauge, but the one specified by the manufacturer is 24 gauge. Could this be the cause of the issue?

Note, we are planning on doing a trial with the proper wire but we do not have any at our plant right now. I'd just like to develop a little bit better understanding so I will have an idea of what to expect during this trial.

 

[MikeHalloran]

More copper in the conductor cannot be the problem.

Whether and how it's twisted and/or shielded could be a problem.
Length, baud rate, voltage or current signaling could be a problem.

You haven't told us enough about your particular installation for us to be of much help, so you might need to help yourself.
Start here:

http://en.wikipedia.org/wiki/Modbus

Mike Halloran
Pembroke Pines, FL, USA

 

[JG2828]

I agree with Mike. 18 gauge is better than 24 gauge.
I'm guessing it is 2-wire RS485?
If so, you will want to use (2) twisted pair wiring with a shield.
Use one pair for the + and - and use one of the wires in the other pair for the signal GND. Never use the shield drain wire for the signal ground.

Take a look at this paper for information on troubleshooting RS485 communications.

http://www.bb-elec.com/tech_articles/rs485_basics.asp

 

[macmet]

Thanks both of you for your replies. I will check out both of the links provided.

As far as I know, the baud rate, address, etc is correct.

Lenght of the wire is ~25 ft and ~ 50 ft (there are two burners both showing same issues).

Wire installed is shielded.

JG, you're correct, it is RS485 communications. I'm intrigued by your comment stating "Never use the shield drain wire for the signal ground". I've attempted to attach a photo showing the wiring in the burner's communication panel. Note the shielded part of the wire. Is this what you mean when you say "never use the shield drain wire as ground". Could that be our problem?

 

[macmet]

Further to my last post, the terminals relevant in this case are 11,12 and 13, 14.

For both burners the shield is run to the "S" terminal.

 

[MikeHalloran]

If your baud rate is above 4800, you've got too much wire flying around untwisted and unshielded. You could shorten those leads and still leave room to reterminate them twice, and have a neater but not 'tight' installation.





Mike Halloran
Pembroke Pines, FL, USA

 

[JG2828]

That could very well be your problem. What does the wiring look like where it actually gets wired to the communication device? Is the shield wire connected to the signal ground input?

http://www.bb-elec.com/guide/RS485_Trouble_Shooting.pdf

http://www.bb-elec.com/bb-elec/literature/tech/485appnote.pdf

----------------------------
Number of Conductors
The signal ground conductor is often overlooked when ordering cable. An extra twisted pair must be specified to have enough conductors to run a signal ground. A two-wire system then requires two twisted pair, and a four-wire system requires three twisted pair.
----------------------------

 

[macmet]

Please see attached picture of other end of the connection.

This brings up another potential problem in my opinion... It looks to me like the wires are crossed. (e.g. you can see that the black is in 47-12, while the red is in 48-11, but the wiring diagram shows that 47-11 are to be connected and 48-12). I've asked the vendor about this and was told that if they were crossed we'd get no signal at all. I don't know enough to argue this point, but this does not seem right to me.

Appreciate all the feedback.

 

[JG2828]

The vendor is correct. If they were crossed you wouldn't get any communication.
You have sent two pictures of how the RS485 cables are wired to terminal blocks. Do you have any pictures showing how they are actually wired to the communication equipment?

 

[danw2]

Spur wiring?

RS-485 is intended to be daisy chain wired: device to device to device.

RS-485 doesn't like spurs. Spurs are lengths of cable wire that are not direct connections to the instrument/device wiring terminals, but lead wiring from main line at a terminal strip/junction block to a device. Reflections are created in spurs which induce errors.

The image below shows the difference between spur and no spur wiring.

Are your devices wired with spur wiring, or is the a cable routed to the device where it connects and cabling then proceeds to the next device?

 

[VEBill]

If there seems to be a problem with a signal, then look at the signal.

Get a suitable oscilloscope and look at the signal. At these data rates (slow) you should be able to see exactly what is going on.

 

[macmet]

The wiring diagram I have shows that the two separate burners are to be wired completely independently. Which I believe is what Dan refers to as a spur, I know it as a stub.

If I remove the shield from the ground terminals, should that help my communications? I do have some spare wire running between all the panels, so I may be able to use one of those wires to connect the grounds.

Something doesn't seem right. I will try to do some continuity testing today to get a little more information.

 

[JG2828]

I agree with VE1BLL. If you can get your hands on an oscilloscope you should be able to see whats going on. The link below shows some good and bad RS485 signals as seen with an oscilloscope and explains how to correct them.

http://www.maxim-ic.com/app-notes/index.mvp/id/763

 

[danw2]

Independent wiring means that each burner controller has its own serial port on the SCADA system?

The burner controllers are not sequential drops on a single serial port?

Are there resistors across the serial comm terminals at the burner controllers?

 

[macmet]

Sorry, Dan. When I posted my last comment your image in your earlier post wasn't there for whatever reason. And no, there are no resistors. At least not that I've seen.

The way our wiring diagram shows how it is to be wired, it is wired to a common data transfer interface in a stub format, and then goes into our SCADA.

Unfortunately yesterday I did not get any further on this as I was dragged into a long meeting. I hope to get more info today.

Oscilloscope is not really in the cards right now. I can see if our controls contractor can bring one next time he's here though.

 

[macmet]

I was just able to confirm that our units are wired in the spur fashion.

I also tried using a separate wire for the ground wiring, but that did not seem to change anything. Some of the communications seemed to work, but others didn't, and it seemed to be intermittent data transfer again.

Tomorrow I will try new wire and I will also attempt to daisy chain it to see if that helps.

I appreciate all the feedback.

 

[macmet]

Hi Everyone,

I just wanted to send out an update on what is going on here. Today I disconnected one of ends of the shielding wire and it appears to have corrected the problem.

It's still too early to say for sure, but I should know more on the after weekend.

 

[MacGyverS2000]

Circulating current in the shield?

Dan - Owner

http://www.Hi-TecDesigns.com

 

[2dye4]

macmet

You must eliminate the spur wiring configuration and wire as shown with 'no spurs' in above drawing.

Also you must keep not mix the pairs. Meaning inside the cable the wires are twisted in pairs. If one conductor from a pair and another single conductor from another pair is used as a TX or RX port the system will not function.

Also each RS485 device will likely have a terminator resistor option or an included resistor to be attached at the terminals of each RX or TX pair. The rule is that if a TX or RX line lands on a terminal and another pair leave from it then the termination resistor should NOT be applied. If only a single RX or TX line lands there the termination resistor should be applied.
In the previous drawing from danw2 the termination resistor is only installed on the last device on the right.

And also minimize the length of lead outside the overall cable.

You should wire it correctly even if you manage to get it working because it is likely marginal and will fail again at a very unfortunate time..