Why is Ehternet Slow?

ceilingwalker

Lifetime Supporting Member
Join Date
Mar 2010
Location
Phoenix, AZ
Posts
1,586
Hello. I would like to know what happens to a small network to cause updates to be slow? What I mean by this is, I have two clients, a PLC, and a server all connected to one workgroup, using static IP's. The two clients are equidistant to the server. On the graphics page are push buttons that allow one to turn on and off valves, for example. At the one client, within 3-5 seconds after pushing the pb the valve reacts and updates the display. On the second client when you push the same pb, it takes 45 seconds for it to display on the screen. One cable runs from server to a network switch and then the PLC and two clients also connect to that same network switch. I have been comparing and contrasting the two clients and can find nothing different. Might someone have an idea as to why this is happening? Thank you
 
Without knowing any of the specifics of your equipment, it's difficult to guess. I have many small networks in my facility and none have the latency you describe. My feedbacks are rarely longer than a second.
 
No idea what hardware you're using, but have a look at this thread regarding auto-negotiation:

http://www.plctalk.net/qanda/showthread.php?t=74607&highlight=half+duplex

Hub or switch? I'm having a problem on one of my networks right now (hopefully to be fixed tomorrow) with a Dell 2724 switch. It has an odd frame size that causes lots of frame fragmentation and re-trys. We're replacing it tomorrow.

Hubs are a mess when it comes to industrial networks; they don't route packets, they just re-broadcast to all ports and that causes collisions and data timeouts.

Are your Ethernet cables in good shape; any physical damage that could be causing intermittent problems?

Is your hardware configured correctly? I've had weird issues with equipment in that past when multiple or unneeded communication protocols were enabled.

Has someone plugged something into your network (possibly without your knowledge) that's burying it with traffic? Some hardware is notoriously poorly behaved.

Has someone plugged an Ethernet cable from one port on the switch back into the switch? That will create a packet storm and bury coms with other devices.

If you can provide us with more details about your network, we can probably be more helpful. The crystal ball is a little hazy right now.


-rpoet
 
No idea what hardware you're using, but have a look at this thread regarding auto-negotiation:

http://www.plctalk.net/qanda/showthread.php?t=74607&highlight=half+duplex

Hub or switch? I'm having a problem on one of my networks right now (hopefully to be fixed tomorrow) with a Dell 2724 switch. It has an odd frame size that causes lots of frame fragmentation and re-trys. We're replacing it tomorrow.

Hubs are a mess when it comes to industrial networks; they don't route packets, they just re-broadcast to all ports and that causes collisions and data timeouts.

Are your Ethernet cables in good shape; any physical damage that could be causing intermittent problems?

Is your hardware configured correctly? I've had weird issues with equipment in that past when multiple or unneeded communication protocols were enabled.

Has someone plugged something into your network (possibly without your knowledge) that's burying it with traffic? Some hardware is notoriously poorly behaved.

Has someone plugged an Ethernet cable from one port on the switch back into the switch? That will create a packet storm and bury coms with other devices.

If you can provide us with more details about your network, we can probably be more helpful. The crystal ball is a little hazy right now.


-rpoet

I will sound silly but my experience with networking is very VERY limited. I put a ringer on all cables and it appears to be fine. I did pull new wire myself so I know it is in good condition. This system was working fine. The only real change that was made was I removed one of the network "switches". It is NetGear. I don't know who changed the IP address but they made it the same as the server, at the client. So, at first it wouldn't even receive from the server. Now it does but with errors from time to time. Thank you for all the things to check, I will do it tomorrow.
 
You said you made the cables. The Ethernet cables? Did you pair the conductors properly when you put on the RJ45 connectors?
 
It is NetGear. I don't know who changed the IP address but they made it the same as the server, at the client. So, at first it wouldn't even receive from the server. Now it does but with errors from time to time. Thank you for all the things to check, I will do it tomorrow.

You should be clear what the IP addresses on the network are and also make 100% sure that every IP address is unique. Having two things assigned the same IP is death to the network.
 
ceilingwalker,

several questions.
1. is your switch managed or unmanaged. an unmanaged switch is constantly polling each port to see if they need data. the managed switch waits for a data request and their faster.
2. what is the refresh rate for the pb's on each screen on your hmi terminals? the faster the refresh rate, the slower the network.
3. what is your hmi interface? any data acquisition?
you may have to add another comms interface to your terminals for the hmi terminals.
4. on my ab slc 5/05 terminals, i have found that lots of online edits will slow the comms speed way down. i save the program, put the machine in manual, make a change, erase the change, conpile it and download the program again into the plc and its much faster. Laugh all you want, i have had to do it on several systems.
5 any other items on the network communicating to the system?

regards,
james
 
Some things that will cause this are..

Ethernet Cables run in parallel with AC Power Lines, without sufficient separation/shielding between the two you will get signal bleed from the power lines into the Ethernet creating noise that forces Retries of the corrupted/noisy datapackets.
120V single phase 15amp you want about 3ft with unshielded wiring.

A wire connected from one port to another on the same hub will cause a broadcast storm, a switch will generally cut those off, but NOT when you run two wires from one switch to another as it bounces between those two. They don't understand that they had it JUST before the other. A Managed pair of switchs specially configured for failover can have two wires between them and not have a problem. Only one is used until it fails then the other is brought online.


PING the various devices and see what your response times are. You will need to know your devices IP's.

Ping XXX.XXX.XXX.XXX -t

will ping until you issue a break command in the Windows Command Line(or close the window).

double to low triple digit pings are good.
 
You should be clear what the IP addresses on the network are and also make 100% sure that every IP address is unique. Having two things assigned the same IP is death to the network.
Originally I could not get the client and the server to talk because someone changed the IP address at the client to the address of the server. After that was fixed the two talk but the updates are really slow.
 
ceilingwalker,

several questions.
1. is your switch managed or unmanaged. an unmanaged switch is constantly polling each port to see if they need data. the managed switch waits for a data request and their faster.
I don't know how to tell the difference. :unsure:
2. what is the refresh rate for the pb's on each screen on your hmi terminals? the faster the refresh rate, the slower the network.
I don't know. I have not needed to go to the Factory Talk Software for this project before and am unsure how to even check. Or is this going to be in the KEPServer Enterprise Software?
3. what is your hmi interface? any data acquisition?
you may have to add another comms interface to your terminals for the hmi terminals.
This system was rebuilt a year or so ago and nothing has changed since then, that I am aware of. I don't believe any data is being collected and stored but I don't know.
4. on my ab slc 5/05 terminals, i have found that lots of online edits will slow the comms speed way down. i save the program, put the machine in manual, make a change, erase the change, conpile it and download the program again into the plc and its much faster. Laugh all you want, i have had to do it on several systems.
I am not in a position to "laugh" at what anyone else does here. This is a PLC-5 and oddly enough, I go online with the processor and it is normal speed and the processor is connected to the same network switch.
5 any other items on the network communicating to the system?
james
PLC-5, laptop (client), Allen-bradley Industrial Computer (client), server (also a client).
 
This system was working fine. The only real change that was made was I removed one of the network "switches". It is NetGear. I don't know who changed the IP address but they made it the same as the server, at the client. So, at first it wouldn't even receive from the server. Now it does but with errors from time to time. Thank you for all the things to check, I will do it tomorrow.

The key phrase is "This system was working fine". Without a doubt, after that something changed to effect the performance of the system. The troubleshooting job is to find out what changed.


  • You mentioned taking out a network switch. What prompted you to do this? Did you replace it with something else? Is this when the slow down began? If so, then that switch was doing something for you and you took it away.
  • Can you isolate the clients one at a time from the system and see if performance improves?
  • Are the industrial computers running out of disk space?
  • Even if they are not communicating with the plant, is there any physical connection between your machine network and the plant network... maybe some kind person thought you should be connected and helped you out?
  • Did any software get updated?
Just some ideas but sort of hard without knowing more about the network or system.
 
PING the various devices and see what your response times are. You will need to know your devices IP's.

Ping XXX.XXX.XXX.XXX -t

will ping until you issue a break command in the Windows Command Line(or close the window).

double to low triple digit pings are good.

When I ping'd the plc-5 it was 32 bytes at 1mS and TTL=255.
 
I found my problem whilst performing a ping and I feel like a fool! I have another work group that is controlled by a SLC-5/05 and it was plugged into the switch and I didn't even realize it. DOPE! The IP addresses for both work groups are 90.0.0.9X, which includes two RedLion HMI's and the SLC Processor. I am not sure why but as soon as I unplugged that cable my laptop went back to normal. Thank you much every much all of you. Another Id10t alarm!
 
Ooooops, called it off too soon. It does update quickly however, when I select an action button from the HMI and am trying to open or close a valve, I get a message displayed "Failed to load data server 'RNA://$Global/Chromalloy:RSLinx Enterprise' on machine 'CHROMALLOYUSER'(result=0x8007274d)."
 

Similar Topics

Newbie question here, I have been trying to setup a Ethernet/IP network consisting of 3 AB Micrologix 1400 plc's, one Idec HG4G-CJT22MF-B OI, and...
Replies
9
Views
3,329
Hello. I would like to know if I could purchase an Ethernet module for my CLX-L31, as opposed to purchasing a whole new processor? The part number...
Replies
3
Views
1,583
Basically, when i am online with the processor via and its running, i try to make some online edits and it keeps booting me off. replaced the...
Replies
6
Views
2,195
A site I am working on is upgrading to ethernet radios. Has anybody worked with freewave radios? Are they reliable? The other choice is MDS...
Replies
3
Views
1,797
What are peoples opinions here on using either Ethernet or Controlnet to link seperate Control Logix chassis, HMI etc. Everyone seems to use...
Replies
16
Views
4,058
Back
Top Bottom