DeviceNet topology questions (teach me)

[industectron]

Devicenet is new in this plant since getting some new equipment in. We've had trouble with devicenet since launch and I don't know if the people that sold it to us installed it correctly since devicenet is new to me. Is it ok to have 5 trunk lines coming out of an 8-port DeviceBox. Each of these "trunk lines" goes to one or more Ultra3000 (daisied chain)and then ultimately to a remoteI/O where it is terminated.

 

[industectron]

any help?

 

[Bob O]

Sorry this reply took so long

Google comes to mind or look here

http://www.ab.com/networks/dnetfaq.html

 

[Ken Roach]

DeviceNet installation can be tricky because the topology is a very flexible trunk-and-drop system.

First, find the terminating resistors. Those electrically mark the ends of the "trunk". Only the location of the terminating resistors matters for designating the trunkline, not the type of media or use of junction boxes or tees.

Next read the first couple of chapters of the DeviceNet Media Design and Installation Guide from Rockwell Automation:

http://literature.rockwellautomation.com/idc/groups/literature/documents/um/dnet-um072_-en-p.pdf

Then go to work measuring your drop line length and total drop line budget, as well as carefully tracing out the continuity and grounding of the shield wires.

After checking for terminating resistors, I always look for the dedicated and appropriately chosen and installed DeviceNet power supply. About 80% of the time on troublesome systems, the power supply isn't a DeviceNet-rated one, is shared with non-DeviceNet loads, or is not grounded correctly.

 

[industectron]

Thanks Ken and Bob, I will finsih those documents you mentioned. From the first few paragraphs I sense something is very wrong. I came into this position as installation was finishing. I have no previous DeviceNet experience. And since the supplier is from a Mid- Eastern Bloc country, the language issue is a problem in itself.
And Ken, from my brief readings (manuals) so far, they put this together as if there is more than one "trunk", which I believe is incorrect (there is only one scanner). This is one of the reasons I think we have a major,,umm problem. They have 8 resistors on the DN. One at the plc and one at each remote i/o, those 4 in between I think, should be considered drops. Which creates other issues. It seems to me that there should be only one trunk, so 2 resistors. The cable they used I think is wrong(thin round) about 700 ft. The reason the length is that much, is that they ran a separate cable from the plc panel to each (of 6)remote i/o.
I really thank you for your comments.

 

[LJBMatt]

This is one of the reasons I think we have a major,,umm problem. They have 8 resistors on the DN. One at the plc and one at each remote i/o, those 4 in between I think, should be considered drops. Which creates other issues. It seems to me that there should be only one trunk, so 2 resistors. The cable they used I think is wrong(thin round) about 700 ft. The reason the length is that much, is that they ran a separate cable from the plc panel to each (of 6)remote i/o.

Ouch... Ken is correct in stating you should only have one trunk line (at that length you would want the thicknet cable) per devicenet scanner (master) with a total of 2 trunk terminating resistors. I'd recommend the daisy chain and drop method as opposed to separate runs to each of your remote I/O. Doesn't sound like the manufacturer was sure of the wiring. Were you provided schematics with a Devicenet layout? You can use that to help put 2 and 2 together to see if everything was installed to print.

 

[industectron]

Hey Matt, thanks for the note,
And yes, I have determined that the prints match the incorrect world out there. And, by the way, those sheets are some of the few that do match the field.

They come from the plc to an 8 port Device Box, and then split into 6 cables, each one going directly to a group(6) of ultra3000 drives, all of this is in the control cabinets room. From the last drive, each cable continues to a remote i/O (which is on the plant floor) about 150-200 ft. by way of trench

 

[LJBMatt]

Each secondary port off the tap box are for drop purposes but with the lengths they are running the drop cables I can see why you are having issues.

 

[Ken Roach]

8 is too many.

Yeah, I've seen systems like that where it seemed like somebody read the installation guidelines then said "I reject your reality, and substitute my own !".

They come from the PLC to an 8 port Device Box, and then split into 6 cables, each one going directly to a group(6) of ultra3000 drives, all of this is in the control cabinets room. From the last drive, each cable continues to a remote i/O (which is on the plant floor) about 150-200 ft. by way of trench

This layout might be made to work, just maybe. If you place one termination resistor at the DeviceBox, each of those six cables going to the group of Ultra 3000 drives can be a drop, so you have up to 20 feet of allowable drop distance between the DeviceBox and the last Ultra 3000 servocontroller.

That last daisy-chain connector that heads out through the trench will be considered the route of the trunk line, and the second termination resistor should be out at that field I/O box.

I would be amazed if the network functioned at all with eight resistors on it. Take off all but two.

The trunkline distance also concerns me; thin trunkline is good only for about 100 meters.

Do you have a 1756-DNB Series C scanner ? That module has some neat network health/diagnostic functions that can help tell you about the network signal conditions. A scope capture of the blue and white wire signal relative to the black (DC Common) wire would be an excellent tool as well.

 

[industectron]

Thanks to Matt and Ken,
okay I,m back now,,,been reading, measuring, and putting out other fires at the same time.
Matt,,, the point you make about the device box is one of the things that has been bugging me also. From what i read, these usualy serve to distribute nodes or devices, not to "split up trunks" as it seems they have done.
Ken you may have answered a question that I have not been able to find a definitive answer yet, although I have not had time to read all the documentation yet, and that is, can you have multiple nodes along a single drop line, such as you said, several drives daisy chained on a drop line as long as you keep in the length spec. But then again, as you mentioned, and what little I'v read so far, I am still worried about that long run of thin cable from those daisy chained groups. I've got 6 thin cables fom the cabinet control room to the remote i/o's out there on the process 150 ft. plus

Right now, with the good points you both have made I'm leaning on abandoning the thin cables and pull in one thick cable from the plc's, bypass the device box, daisy chain all the drives in the cabinet and continue to the remote i/o's and daisy chain them also. A rough estimate is 400 ft. That's all the turns back and forth in the cabinets, everything. What do you guys think? Would this be the way to do it right? Do you think this is what should have been done in the first place?

 

[industectron]

OHH Ken,, i almost forgot,,,yes we have a 1756-DNB scanner ,,, i'll check what series,,, that is a neat idea about the scope ,, i'll check it out

 

[Alan Case]

Do a search on ABs website for the "Devicenet Book Of Knowledge" This will explain everything you need.
Regards Alan Case

 

[mellis]

Let me clarify your network a bit so we are all working on the same problem.

You have one cable from the PLC to an eight position device port.

From the device port you have six cables and each one of them does the following:

Daisy chains 6 drives located close together and then continues for 150 ft plus to a remote I/O block.

Is this correct?

What Ken suggested would work if ONLY ONE of those 6 cables continued on to a remote I/O block. If you have 6, there is no way that is going to work the way it is.

A sketch of the network would help a lot.

However, all that aside... Your idea of starting over and pulling a new cable and making it one continuous daisy chain will definitely work. What you are proposing is 100% trunk with no drops at all. It is the simplest DeviceNet you can make. If you can get it done within the recommended maximum length for the cable type and you can keep the power within spec for the entire length, you've got it. If you really need 400ft of thick cable, you're over the limit for 500Kb/s operation. If you can accept 250 or 125, you're ok.

However, if the cable from the device port to the drives meets the spec for drop cables, I see no need to change it. Each drop must be no longer than 6m or 20ft. That's from the device port to the last daisy-chained drive (it's all one drop). Total accumulated drop length for 500Kb/s is 39m or 128ft. You may be close on that. Once again, if you lower the speed, you're fine.

Run your new trunk from the PLC and daisy chain the device port, and the remote I/O blocks. You can even use short drops to the I/O blocks if you like (and you may want to make it easier to disconnect a remote I/O block with minimum disruption to the network).

On the subject of speed... My personal opinion is: If 125Kb/s meets your needs, use it. Your process will determine how fast the network needs to be. A 125Kb/s network that is rock solid beats even a slightly flakey 500Kb/s in my book. But if you need the speed, you got to have it.

Bottom line. If you do indeed have 6 runs of 150ft plus thin cable, you HAVE to redo the trunk. The max for thin cable as trunk is 100m or 328ft. You're clearly over that.

If Ken's interpretation of your network was right and there is only one 150ft run of thin... Never mind.

Good luck and let us know how it turns out.

 

[LJBMatt]

I'll be honest i've never been a fan of tab boxes. AB and other vendors make and inline trunk "T" with quick connects on all ends. The top of the "T" continues the trunk and the bottom you use for your drop. If you need multiple drops in an area you can just connect as many "T's" together as you need and then continue your trunk line cable. The advatange is that if any of your cables are cut (trunk or drop) with quick disconnects it can be replaced faster and takes the guesswork out of reconnection.

 

[industectron]

Melis, thank you for the response, I appreciate it,

So to clarify, yes Mellis we have 6 thin cables leaving our panel in the control room. Each cable comes off a drive and heads out to the plant floor to one of the 6 remote i/o's, where the cable run ends.
You make a great point about keeping the 8 port device box and making each of those groups of drive(s) a drop.(For clarity some "groups" have 6 drives, some have less) I checked the distances in the panel and the longest drop with all the daisy chaining would be 16 ft.
Also the speed we are using now, 125k, seems to be working ok. If we can make improvemnets on the speed (reliably), that would be a plus.
So should I expect performance and maintainability to be equal in either configuration? i.e. solid trunk or trunk with drops