Calling AB DH+ Specialists

CGS

Member
Join Date
Jan 2006
Location
Victoria
Posts
6
Have a problem with our DH+ Network.
Nodes have been dropping off and believe it is the number of nodes on the network.
I wish to split the network into two.
All PLCs are SLC 50/4s.
There will be five SLC 50/4s on the "new" network with PV 550 connected via the serial ports.
Our SCADA servers will be on the other network and minimal information is being transmitted, however there is some Critical plant control also needed by the servers etc.

What is the most practical way of connecting the two networks via an Ethernet/IP bridge, I don't want to start changing PLCs to Contologix or SLC 50/5s?



Thanks in advance
 
ControlLogix gateway.


Use a ControlLogix chassis and powersupply.
Use a ENBT - Ethernet bridge
Use a (or multiple) DHRIO - DH+ bridge
 
CGS, you do not describe exactly what the setup is before the change or what it will be after the change.

Currently how many nodes and what are they ?
Are the data being polled by the SCADA's on each PLC ? Maybe the amount of data is little, but if the SCADA's as setup to read the data frequently because they have to detect when it changes, then the traffic will significant. And if you have many SCADA's that all poll the same data, the traffic will be multiplied even if it is really the same data.
If the above is the case, you will see only a slight improvement by removing the SCADA PCs from the DH+ network.

You may consider 2 clever techniques:
1. Change from the SCADA polling the data cyclically, to letting the PLCs send the data via messaging upon event.
2. Let only one SCADA PC aggregate the data and then distribute the data to the other PCs (via ethernet). I think this is possible with RSLinx Gateway. Alternatively I believe there are 3rd party OPC servers that has this functionality.
 
I am no DH+ specialist but I thought I would give you my setup. Some think it is a bit much, but it works (although a little slow at times) and information exchange is very critical. I have a DH+ system consisting of: 17 SLC 5/04 each with a DH485 PV on channel 0, and 1 PC collecting data and providing information. There are approximately 70 data points (both bits and words) that are polled by the PC every 1 second (soon to be 2 seconds on many tags) on each SLC. There are 16 of these identical systems I take care of. Each one of these systems connect to the PC via a contrologix gateway so the PC can access via ethernet.

The most critical part of the DH+ system (IMHO) is the setup and proper termination. The only time I've ever had a node just drop off is when the terminating resistor was disconnected and not put back. Make sure that your DH+ is not in a star or token ring config and that there is the proper terminating resistor only at the end of each network.

I know I didn't answer your question, but I think that your system should work (we would need more details like JesperMP asked) before you go starting over.
 
CGS
Are all the data points in 1 continuous block or are they all over the data table?
This can make system from what you describe to a decently functioning system. It is amazing the difference with all the data in 1 block.

Drewcrew6
 
To add to the valuable data above
Go to Linx and reset the counters and after an hour or a day monitor your nodes and see if one particular node is troublesome or one end of the DH+

I second the Control Logix gateway for DH+ to ethernet
We had 20 nodes they ran well but were slow, I then split it 10 to 9 it still works well with a small increase in speed
 
Thanks for all the replies.

Number of nodes: 18 (9x SLC50/4, 2X Servers, Rest PVs)

Connections including EOL resistors are in very good condition
as I "belled" out cables, thinking it may have been a combination of bad terminating and/or cables.

Polling of servers: Changed time and way messaging was being handled earlier in the piece.(No difference)

Nodes would start dropping out after 6-8hrs, have been resetting values in RSLynx since day one, hoping to see something irregular.(This and another computer that was listening only didn't lead me anywhere definite)

"Scope" only showed very little noise that wouldn't really effect the network as anticipated.

Believe it or not, I have finally found in the darkest & deepest archives of AB information in the small-print section, that "NO MORE THAN 15 NODES ARE ALLOWED ON A DH+ NETWORK"!!

I had already thought about using the Contrologix idea and needed confirmation, this is being installed asap.

Thanks
 
I strongly recommend a more methodical investigative approach.

Define what you mean by nodes "dropping off" the network.

Do you mean that SLC-5/04 controllers are displaying blinking green DH+ LED's instead of solid green ? Do you mean that messages from one controller to another are failing with Link Layer Timeout or Application Layer Timeouts ? Do you mean that PanelView terminals are displaying communication-related error codes ? Do you mean that your SCADA computers' RSLinx browses are showing red X's in the RSWho window, or that they are logging communication timeouts in the event log ?

Protocol analyzers (like the one sold with Frontline Test Equipments FTS4Control) can tell you a lot about errors and loading and node-to-node messaging. They're really most effective in the hands of RA field service engineers.

The troubleshooting I start with when using SLC/PLC controllers is using the Retry counters in each controllers Channel Diagnostics data file to examine the controller's perspective on network health.

While you're checking out the waveform of the DH+ signal, be sure to also physically measure the resistance between blue and clear wires (with all the nodes unplugged) and measure the resistance between the shield and ground, verifying the network shield is only grounded in one location. Check the signal using AC coupling as well to see if the network is riding on top of a 50 or 60 Hz induced signal.

Too little wire can be a problem, too. I always make sure my DH+ cables are 10 feet minimum. Remember that if you're running at 230.4 kb/s you can only run 2,500 feet of cable.

It would be a shame to go installing network gateways to attack a perceived bandwidth problem without understanding the sources of the bandwidth shortage, or overlooking installation issues.
 
Is now:
Number of nodes: 18 (9x SLC50/4, 2X Servers, Rest PVs)
The "Rest PVs" are then 7.

Will be:
There will be five SLC 50/4s on the "new" network with PV 550 connected via the serial ports.
I dont get it. Are you reducing the number of SLCs from 9 to 5, and moving the PVs from DH+ to serial ?
Or are you splitting the DH+ into 2 networks ?
What happens to the SCADA servers ?
 
Didn't want to turn this into a documentary (Joking)
But here goes!!
Anwers to questions, followed by result.


Ken Roach

Do you mean that SLC-5/04 controllers are displaying blinking green DH+ LED's instead of solid green?
Yes, it seems that the network would slow, until 3 or up to 5 nodes (always SLC5/04s and no particular node) stop communicating across it, then the network would regain speed.

Do you mean that PanelView terminals are displaying communication-related error codes?
Yes, when the node/s it recieves/transmits data from "falls off" network.
RSLinx browses are showing red X's in the RSWho window
Yes, then of course any messaging/control from associated PLC will not occur.

While you're checking out the waveform of the DH+ signal, be sure to also physically measure the resistance between blue and clear wires (with all the nodes unplugged) and measure the resistance between the shield and ground, verifying the network shield is only grounded in one location. Check the signal using AC coupling as well to see if the network is riding on top of a 50 or 60 Hz induced signal.

"Belled out" cables.(Continuity Testing) Resistances of all cables all within manufactures specs (If in doubt,replaced it).

Too little wire can be a problem, too. I always make sure my DH+ cables are 10 feet minimum.
Totally agree, shorter cable runs are>10ft
Remember that if you're running at 230.4 kb/s you can only run 2,500 feet of cable.
Due to overall length of network being >2500 ft and <10,000 ft staying with 57.6kb/s



Jesper
Or are you splitting the DH+ into 2 networks?
Yes, have done so,
Now
Network A = 8 nodes (5X SLC5/04s 2 have always had PV550s serially connected, 1X PV550 DH+, 1X PV900 DH+, 1X Gateway)
Network B = 12 nodes (2X Servers, 5X SLC5/04s, 4X PV550s DH+, 1X Gateway)

Conclusion (I hope!!)

As stated above, have split the network into two with a Contrologix Gateway. This has been running for the last 48hrs and nil nodes have dropped off the network or erred. Network has not slowed
Traffic was never an issue as I believe we had our control, messaging, etc between the nodes correct.
Before/ Traffic = 34% - 40% After/ Traffic = 22% - 26%
Bad Packets being sent - Nil
A major concern was the Token Passing.
Before = 100% After <80%
The main culprits were the PV1400e and its associated SLC5/04 as this was almost constantly at 99-100%, this has dropped dramatically since replacing it with a PV550. (Yes, I tried extending its polling time and limiting messaging and other modifications to no avail)
Bandwidth, as thought, vastly improved.
The max. number of nodes on a DH+ Network was a vital clue and took some searching to find.
I'll stay with my decision of splitting the network.

QED
 
Out of curiosity, how much cross messaging is occurring? Is it conditioned? Is it in continuous data stream (nicely laid out in the data table)?
 

Similar Topics

I am looking to get an AB series 1774 (PLC1) from the 1970's up and running. I have all the parts, and it powers up just fine without errors if I...
Replies
12
Views
2,288
Hello all, I’ve recently begun using Automation Studio on my own time to boost my knowledge of controllers beyond AB/Siemens. To ease myself into...
Replies
0
Views
871
Hello I'm new to PLC programming and I'm programing in ST. Programming envirement is GX works 3. I can't figure out how to jump to subprogram or...
Replies
1
Views
1,413
I have some code i need to block until a super genie for data entry has completed (by operator input)as per below. FUNCTION New_Setpoint()...
Replies
5
Views
2,248
I have a display that I can't find out where it is being called by. is there a "Display tree" that shows where each display is being called by? I...
Replies
1
Views
2,135
Back
Top Bottom