Sign in to follow this  
Followers 0
Penko Mitev

Develop a fieldbus for modular IO

12 posts in this topic

Hello guys! I am in front of a big dilemma which way to go. We are currently writing a quote for a conveyor-like machine. It's transporters, which will carry out the transport between the warehouse and other parts of a factory. The transporters themselves are divided into modules - 1 meter each. Each module will have an asynchronious motor, controller by a variable-frequency drive, additional I/O(some pneumatics and sensors). If I have to hard-wire everything, there would be cables having length of 100 metres and above. Everyone knows this is not a good idea in some cases. Instead, I am thinking of developing a fieldbus, but as I don't have much experience with them, I am not sure what I need. Here is how I imagine the concept: - each mechanics module has an installed device on it where all the IO from the module is connected. So, 100 modules like this. Then, from the PLC, we connect all modules in a daisy-chain manner. The protocol is not so important - it could be Modbus, CAN(DeviceNet), Ethernet, whatever. My question is - what implementation would you use in this case? Are there such ready-to-use devices? If possible, I am looking for low-cost or average solutions in order to be able to prepare a competitive quote. Thank you in advance!

Share this post


Link to post
Share on other sites
I know there are other systems available, but my personal experience with Siemens and IFM Efector ASi bus interfaces to a Woodhead card in AB SLC racks has been nothing but positive.

Share this post


Link to post
Share on other sites
Whenever you are dealing with very low density IO, you really need to think about the cost of the interface module itself. This usually pushes you towards very low cost field buses. DeviceNet has a lot of great options and so does ASi. For all practical purposes the best approach is to just figure out the cost of each "station" and look at which one works out best. There is one huge disadvantage of any of the "line" type buses (basically everything except Ethernet). Whenever you have a problem on the bus, since the bus itself is something of a "party line", it can be extremely difficult to track down the culprit. This is where it can be very useful to break the bus up into zones and is the primary advantage of Ethernet for instance. The downside is of course that doing this drives costs up as you are adding active devices to the network fabric.

Share this post


Link to post
Share on other sites
With 100 nodes or more, I think that the serial fieldbuses will soon be stretched beyond their limits. That goes for Profibus DP, Devicenet, CAN, ASI. Of course, you can split the plant into several groups, so each group does not consist of so many nodes. If this is a very large plant, with long distances, and/or really many nodes, but with a relatively modest requirement for update speed, then maybe a more proces oriented fieldbus will be suited. That would be Foundation Fieldbus or Profibus PA. Apart from that, I think it is possibly a clear case for an Ethernet type fieldbus. Think Profinet or Ethernet/IP. You should absolutely arrange a "backbone" in the shape of a ring of swithces connected by fiber. Then connect from the switches to local nodes by means of shorter copper drop cables. That will mean you get a flexible and error-resistant system. I do not aggree with the notion that since there are many nodes, one should try to squeze the cost down. It is the other way around. With such a large plant with so many nodes, a small error can have much bigger consequences than on a single small machine. So it pays dividends to think in how to prevent one fault from bringing the whole system down, and how to pinpoint an error to fix it quickly.

Share this post


Link to post
Share on other sites
I forgot: In order to bring the cost down, while still using ethernet, think about using Beckhoff in stead of Siemens or Rockwell for the i/o modules. Beckhoff cost approximately half of Siemens modules, and approximately a third of Rockwell modules. And Beckhoff has just recently released a low-cost adapter module for Profinet. It does not have an integral switch, so you cannot daisy-chain from one adapter to the other, but maybe that should be avoided anyway.

Share this post


Link to post
Share on other sites
Interesting topic here. I now use CompoNet for all remote I/O. It is a CIP network as well and administered by ODVA, as is Device Net and Ethernet IP. The I/O modules are inexpensive compared to Device Net, it is much faster, more complex networks can be formed, twisted pair screened cable - nothing fancy - have to watch the resistance though as with all networks as there is a 120 ohm terminating resistor on the end. Works a treat, easy to set up. Biggest problem at the moment is that only Omron and some Japanese companies appear to have taken it up. The big push appears to be for Ethernet IP and I here that Ethercat is on the way for peer tp peer and more remote I/O. That is really intersting as Ethercat can be configured as a redundant wired ring - YOU BEAUTY!! Schneider have gone their own way naturally with their Advansys system as well. Have a really good look at the options and pick something to suit your application. Just bear in mind that some systems require very expensive cable and have other problems with voltage drop etc - Device Net is a really good example - I no longer use it if I can help it - way too expensive for my liking.

Share this post


Link to post
Share on other sites
Guys, sorry for not answering immediately about what happened :). In fact, we have a much bigger reduction in price from Schneider than the other PLC brands we work with. Therefore, I prefer using Schneider for this project too. I contacted Schneider Bulgaria and I was given offer about the components. We have to use about 50 Remote IO modules, 6 inputs/6 outputs each. The sales agent also included a Modicon M340 PLC because he told me Twido can't support so much IOs. I say Twido, because it is still the only Schneider PLC I have experience with. He knows this and was fast enough to tell me before asking the question :). Actually, in the inquiry, I told him there will be 35 modules. Unfortunately, after estimating precisely and after the customer added a request for additional functionality, we stopped at 50. The sales agent told me it would be much better if we create two MODBUS networks so there are almost equally big. Additionally, we will have a Variable Frequency Drive at every 6 meters(this is one mechanical module). So, this means around 45 motors with 45 VFDs. The guy told me it's good to create a separate CANopen network for the VFDs. So, this is the way he created the quote. We talked on the phone and I was told there is some limitation for the maximum devices to be addressed at the same time over a MODBUS network. It was something like 8. That's why we added some Ethernet device which handles the entire communication and I don't have to add communication logic(checks, error handling, etc) to my PLC logic. It converts the Modbus and is connected to the Ethernet port of the PLC. Everything is on the stage of QUOTE for now. Soon we will get notified about the customer's plans. That's for now. Penko

Share this post


Link to post
Share on other sites
Couple of things to consider 1) Look at the total cost of doing the job (hardware, software, commissioning, screens etc) not just the hardware discount. 2) Schneider in Australia are lousy with discounts. 3) End prices can be put up easily so that bigger discounts can be given. 4) Quite often that means you pay more for hardware anyway. 5) Modbus has its good and bad points. I do not like it for remote I/O as it has to generally poll from software and is too slow - plus the limtations of course. 6) Have not used Modbus TCP so cannot comment. 7) Software is important - if you like FBs then the M340 has a pretty good FB implimentation - I do not like FB for many reasons but the main one is if you delete online and then try to insert it is an absolute pain - go offline - change program - go online - stop PLC and download program. I use ladder with very few exceptions as I can really easily change programs online without all the nonsense. 8) I would seriously consider using Omron and CompoNet for the remote I/O. 9) Using the CRT series of remote I/O you can have up to 1024 inputs and 1024 outputs on 1 card. 10) Additionally there are bit I/O analogues and you can also use the Smart Slice series GRT. 11) 16 input aqnd output blocks available and 8 input/8output as well. 12) You can then easily configure the CompoNet master card and the remote I/O in the software configurator. 13) You do not then have to pool the remote I/O - it just happens. 14) The inputs and outputs are then just numbers - no other addressing or polling required. 15) The software is a complete suite to program all Omron devices, PLCs, screens, motion etc. 16) If you use an Omron screen they have 'Smart Active Parts' that you can employ on the screen to monitor the remote I/O. 17) There is a simulator in the suite (CX-One) for the PLC and the screen and both simulators work automatically together. 18) This equates, to me, to faster development, reliable and fast remote I/O network, faster programming, faster commissioning and I make more money. 19) My time is my most valuable resource and if something is going to save me time so I can get on to the next job I use it. The hadrware cost is not really that relevant, good networks, fast software development, fast commissioning, good software tolls are where I make my money - not from chasing discount from suppliers - provided the hadware is not exhorbitantly expensive of course. 20) I used the Twido on 2 jobs only by the way and then gave it away completely - to me it ios in the 'toy' range of PLCs but lots of people like it. The biggest bug with it was the software - it was awful - I will have a look at the latest software as I am hearing it is quite good. Something else about CompoNet, you can obtain extenders for longer distances. The extenders can also be used for changing wire types and spedds between segments if you wish. Edited by BobB

Share this post


Link to post
Share on other sites
Bob, thank you for the detailed reply. In fact, I will go for a quote from OMRON for the components you specified. Of course, what I want most is fast programming, reliable fieldbus, high speed. This is a large project, so adding several thousand bucks is not big problem. That's why we did a general estimation of what would such a fieldbus network cost(the maximum) and we included it into the quote. Now, there is flexibility to choose our provider. Twido with the TwidoSuite software is perfect for up to 100 I/Os. Take a look, you won't be sorry.

Share this post


Link to post
Share on other sites
AARGHHH!!! Used the dang things twice and have not used them since!!! Software was the most ghastly heap of c**p I have ever seen in my life. Even when they finally released a version that allowed online programming it was ghastly. I hear there is a new version that is supposed to be quite good but have not looked at it - once bitten twice shy!!! I do large and small jobs - multiple high powered networked PLCs to 40-50 I/O. Power stations to swimming pools. I now use Omron CS1 for the biggies including the redundant processor and power supply unit, CJ1 or 2 for medium stuff and CP1 for the smaller stuff. You can get 180 I/O on the CP1 shoe box PLC. Best thing is great software and same instruction set for all the PLCs including up to 64 bit floating point. Great function set with stuff you never see in other PLCs but other PLCs have instructions I would like to have too. We PLC people are never satisfied - particularly when we see something we like that is not available in our favourite brand.

Share this post


Link to post
Share on other sites
I second you :P

Share this post


Link to post
Share on other sites
If I understand correctly: You need 50 i/o nodes + 45 VFDs. That is not a small project, but one where the networking will be a critical backbone for the entire plant. Would never chose Modbus for such a project. With so many nodes, the likelyhood that a problem may occur is relatively high, and the consequences will be relatively expensive too. So dont be penny-pinching on this part of the project. I would chose an ethernet-based fieldbus (that does not include Modbus TCP), and I would arrange it in a ring with automatic switch-over. Think about that you most probably must be able to maintain the system including the networking without shutting down the plant. If this is possible with Modicon, I dont know, but the idea to use a mix of Modbus and CAN sounds as if the Modicon salesguy is trying to squeze in the lowest possible price because that is what he think is what you want. That you are even considering a Twido for such a job, somehow I think you are not the right person for this job.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now
Sign in to follow this  
Followers 0