IP Address Selection

I’m after some advice on a couple of networking questions, we are putting in a fibre network using two stratix 8000 one in production and one in packing and a 8300 which will sit in the control room, from the 8000 we will then use cat 6 to connect to all our equipment, which will be around about 200+ nodes, my plan is to break it up into two different IP ranges for each of the switches and do the routing with the 8300, is this an efficient way to do it? Also I want to assign an IP range which will allow me to once finished gain access remotely to the network and allow others to gain access over the internet, this is a new line being installed so I am to specify the addresses to the suppliers and would like to get it right first time so that I don’t have to change them all at a later date.

The other question is one of the pieces of kit is a large proofer and retarder system and this has 42 22-COMM-E, 11 1734-AENTR, and a PV+1500 connected to one 1756-EN3TR I will be putting in a ENBT to connect to our network, now what I was wondering is how do I assign the IP addresses of the drives and point I/O as it’s on its own network but I still want to have full access to whatever is on it from my main fibre network for drilling down/pinging a drive, I haven’t used a controllogix with multiple NIC’s so I’m not sure on the routing capability or best practices so any help or advice would be greatly received.

Rob

Rob

Do you have a drawing of what you want to do? if not that needs to be done first if you are serious about getting it right the first time.

Once you have that drawing post it here and we can help in a more detailed way.

IP address range should not matter as long as you are using a private range and it's different from other equipment at your site and /or remote sites on a WAN. I say IP addresses don't matter much in your case becase you are using a 8300 which is a router.

Are you planning to connect your 200 nodes in a ring? The part numbers you listed here will not cover that alone if so and a stratix 8000 with max expansion sets won't cover 200 nodes.

You also speak of internet access is this for remote support / monitoring? How will you remote into the network? VPN? On you corporate LAN or on your machine LAN? Either way i would recommend at least 1 LAN firewall in front of your machine network just in case IT makes a Boo Boo in the future.

You really need to draw this first before you go any further. Draw it in Visio, Smart Draw or Autocad, etc.

I am starting to put together a drawing in RSLogix architect and wanted to add the IP Addresses as I added the components, I will put together a simple drawing together to show the layout, it will only be an example as the machines are not here but I have had conversations over the phone so I have some info.

Our other two lines which will soon integrate into our new line are on the 192.168.1.xx range but I will change them as and when they migrate over to the network.
The 200 nodes are connected through a 5700 switch and from there to the 8000, What we will be having is the three 8000 series in a fiber ring and from each of the 8000 will drop to each machine which will have a the 5700 in the enclosure with the PLC, HMI, VFD connected into that.

Yes this will be for remote monitoring/support as far as how we will be doing this then we have not nailed down an exact way we want to do this, we are not going to be using the corporate network we have our own dedicated ADSL line, but we have been to Rockwell HQ and they showed us vantage point and would like the ability to use this to allow the corporate network computers to see data like OEE, KPI’s from our dedicated network using a thin client like firefox, I.E.

Yes I have started drawing up like I said in RSLogix Architect but to get the ball rolling sooner I will use word/visio and post as soon as I get it, do you need to have every exact component or just a general idea of the layout which we will be having, as drawing 200+ nodes in visio might get a bit complex

Regards

Rob

a quick example of the topology of the network

Rob

From your drawing is each 5700 supposed to be handling a line? Or is each 8000 considered a line?

Most peole have HMI's and Historian,etc on one network at the machine level and Drives and IO on another. Depending on the amount of drives and IO you have you may want 3 networks per line so a detailed layout is best.

If you are doing a layout you would be better served to do it in IAB which will allow you to export it to visio. The viso drawing can be made into a pdf for you staff and also imported into your autocad or other cad drawings. You can import a IAB file into architect but as AFAIK architect will not export to visio.

IAB is for layout and it builds a BOM for you. Architect is more for multiple controllers on the same machine, line or project and is primarly used to organize exchanged tags, Connections,etc.

Rob

well let me explain abit more about it, basically we have about 11 peices of kit from 11 different suppliers which come together to form one production line, mixers, makeup line, proofer/retarder system, cooling spiral, product conveyors, bagging/packaging machines, box errectors, packing conveyors, box closers, stretch wrappers.

we only have one controllogix in the line and that is proofer/retarder system, that has a seperate NIC which has the drives, I/O and one HMI on, and another one will be installed to connect onto our network.

what i am wanting to do is have a L71 back in the control room and use this to do product tracking and generally tie all the comms of each machine together.

it maybe pushing it to have more than one network i had to push to get this fiber one installed so to go back and ask for another 1 or 2 they would say no, as for the fiber ring we are installing a 6 core and the panduit fiber enclosures allow upto 2 stratix 8000 so if in the future we are using alot of bandwidth then we have the ability to break it down.

thats not a problem i actually find IAB way easier to use, ok i will get on with it as soon as i get back into work

could you recommend a good IP Range or a series of different ranges to start off with?

192.168.x.x is considered a private 'Class C' addressing scheme, and as such is limited to 254 nodes per subnet... which is fine, if you plan on routing each subnet from a 'source' (backbone) network.

I recommend using a private 'Class B' network scheme which is 172.16.x.x through 172.31.x.x

That way, you can have greater flexibility with the numbering of each type of device on either a physical or geographic location.

~Mike

forgot to answer your first question, the 8000 are situated in one downstairs which has all the production equipment from flour system, mixers, to cooling spiral, and then the other 8000 is upstairs and is covering packaging conveyors to stretch wrappers.

the 5700 are located inside each of the OEM's panel and port one of these are then connected back using CAT 6 to a port on the 8000.

one question i want to ask as time is short and the OEM is requesting the IP addresses but the proofer retarder system which has the 42 drives, 11 Point I/O and the screen, does this IP Range matter as in does it need to use the same 3 octets as the LAN to allow routing between NIC's as bearing in mind that i still want my engineering PC to be able to route from the LAN over to the Device NIC to say... download setting from the PC to a VFD if one goes down in the night.

thanks for that info mike, any chance you could maybe expand abit on the difference between the two and what the extra flexibility would be.

what i planned was to have the first piece of kit in the line as say...

xxx.yyy.zzz.5 - PLC
xxx.yyy.zzz.6 - I/O
xxx.yyy.zzz.7 - VFD
xxx.yyy.zzz.8 - HMI

then the next part of the line

xxx.yyy.zzz.15 - PLC
xxx.yyy.zzz.16 - I/O
xxx.yyy.zzz.17 - VFD
xxx.yyy.zzz.18 - HMI

bear in mind these are seperate pieces of kit from different suppliers not just one OEM's line split into sections controlled by multiple PLC's

this is why i was thinking of using the above IP address style instead of having all the PLC's on the LAN taking up the first say 30 assigned IP address then the next 30 as I/O then VFD, HMI etc. unless this is a better way, and im sure it would make fault finding easier for the guys if the IP address of each machine are grouped together not grouped together as a whole production line.

I hope this all made sense

Rob

robw53 said:

thanks for that info mike, any chance you could maybe expand abit on the difference between the two and what the extra flexibility would be.

what i planned was to have the first piece of kit in the line as say...

xxx.yyy.zzz.5 - PLC
xxx.yyy.zzz.6 - I/O
xxx.yyy.zzz.7 - VFD
xxx.yyy.zzz.8 - HMI

then the next part of the line

xxx.yyy.zzz.15 - PLC
xxx.yyy.zzz.16 - I/O
xxx.yyy.zzz.17 - VFD
xxx.yyy.zzz.18 - HMI

bear in mind these are seperate pieces of kit from different suppliers not just one OEM's line split into sections controlled by multiple PLC's

this is why i was thinking of using the above IP address style instead of having all the PLC's on the LAN taking up the first say 30 assigned IP address then the next 30 as I/O then VFD, HMI etc. unless this is a better way, and im sure it would make fault finding easier for the guys if the IP address of each machine are grouped together not grouped together as a whole production line.

I hope this all made sense

Rob

Click to expand...

Well... without trying to describe 2 years worth of IT training, the Class B network simply allows for more IP addresses to fall under one 'subnet'... the part that isn't clear in my previous post is the 'subnet mask', which really makes all the difference. Normally you have a mask of 255.255.255.0, but to allow for more addresses in a Class C network, you have to expand the mask to 255.255.252.0

since you already are planning on 'routing' your network, you could split up each 'line' with a range of addresses that are in different subnets.

For example: The first line would have addresses setup like this... 172.16.1.x

The second line would have an address range of 172.16.2.x; each with a subnet mask of 255.255.255.128, and so on...

Again, that's probably much too simplified, so here's a link that might help http://www.techrepublic.com/article/ip-subnetting-made-easy/6089187

~Mike

Rob

For all intents and purposes network IP classes really don't exist anymore. It's a old and outdated concept.

My preference for a private range is the 10.x.x.x range as the 192.168.x.x ranges is frequently used in consumer level device and the 172.x.x.x range is typically used by corporate IT for LAN devices. IMHO I would stay away from both.

BTW what schema is you site using on your current equipment?

I do something like this

Line # 1

10.1.1.0-254 Core Connections
10.1.2.0-254 PLC and IO
10.1.3.0-254 Drives
10.1.3.0-254 HMI, Historian, Etc.
10.1.4.0-254 NA

Line # 2

10.2.1.0-254 Core Connections
10.2.2.0-254 PLC and IO
10.2.3.0-254 Drives
10.2.3.0-254 HMI, Historian, Etc.
10.2.4.0-254 NA

For clarity the 10.x.x.x "Class A" network gives you the most possible nodes weighing in at 16,777,216 possible nodes.

In my example "Core connection" is your connection to the rest of the network and how you would connect to manage the system.

By using masking you could manage all the sub networks. Not the prefered way but it will work. Having something like a 8300 per line to do the routing would be the ideal situation.

This would keep routing nice and clean also.

the other line doesn't have a network as such, the only bit that it does is between two mixers and the flour delivery system, the stratix 6000 is in the flour system panel and then in the mixers they just have dumb switches.

Stratix 6000 192.168.1.5

Flour System:
PLC-192.168.1.10
HMI-192.168.1.15

Mixer 1
PLC-192.168.1.20
HMI-192.168.1.25

Mixer 2
PLC-192.168.1.30
HMI-192.168.1.35

all set to 255.255.255.0


in your example....

Line # 1

10.1.1.0-254 Core Connections
10.1.2.0-254 PLC and IO
10.1.3.0-254 Drives
10.1.3.0-254 HMI, Historian, Etc.
10.1.4.0-254 NA

does this mean that with the right mask that everything above is visible to one another, even though the third octets are different?

with the above example what benefit would it give over say assigning a group of addresses as per my example with the flour system and mixers? as each one may only have about 6 nodes and the above example maybe a good example for a large node count line it may well confuse people.

i think i need to keep going over the subnetting examples and videos on the web untill it sinks in as im still not getting my head around it, but it being 3am doesn't help.

I have a slightly similar system. 3 EN2Ts on 192.168.211.XXX, 192.168.212.XXX and192.168.213.XXX all with a mask of 255.255.255.0
These 3 networks talk to multiple VFDs and IO units.
I also have a scada network on 192.168.210.XXX.
For programming the main PLC processor I usually connect to the scada network. To access the other networks for faultfinding etc was a pain. You can utilise linx to bridge networks but that seems counterproductive. In theory I was told a layer 3 router could do the bridging for me but the cost didnt justify the result. In the end I ran all 4 networks up to where the programming PLC will sit. I have ordered 3 USB to ethernet adaptors. I will plug into the scada network with my real ethernet connection and to the 3 field networks with the USB adaptors.
Regards Alan Case

Rob

In your example your flour system and Mixer 1 and Mixer 2 are all on the same network. Is this ok or not? Answer: It depends

It depends on the amount of I/O the RPI, The type of traffic (Unicast, Broadcast,Multicast) and that is largely determined by the controller and firmware you are using. Small networks are normally not a issue but larger netwoks can be.

As your flour system and 2 mixers are setup now will work but that /24 subnet mask (255.255.255.0)means everything is on the same network unless it goes to different adapters and switches and does not connect anywhere.

Things being on the same network is mostly not a problem until something goes wrong like a bad cable, port, etc then you have a very difficult time finding the issue depending on the size and complexity of your network.

Example. On your flour and mixer system can you run and maintain partial production if 1 mixer goes down? If so it might be work being on seperate VLAN networks as a problem on mixer 1 won't cause a issue on mixer 2 and the problem will be easier to find Vs what you have now.

Segmented Network = Higher cost,more robust to faults,easier to troubleshoot

Single Network = Less complex, Easier to implement, more prone to traffic issues / performance issues, Can be more difficult to find problmes / troubleshoot.

A lot of these are questions you will have to answer /decide what's best for your site and there are other varibles to affect your decision such as time,cost,experience,maintenance,etc.

I like seperate networks in each line but then it's best to have routing to make things easy to maintain remotly.