Skip to Content

Register Login

Industry Online Support

Technical Forum

Navigation

Time delays in MPI network

Created by: gumis at: 6/14/2009 3:41 PM (6 Replies)

Rating (0)

Thanks 3

7 Entries

6/14/2009 3:41 PM	Rate (0)
gumis Posts: 45 Rating: (0)	Dear all, In my application CPU314 communicates with CPU315-2DP in MPI network. I use X_GET function in CPU315-2DP to send one byte of information from IB0 of CPU314 to QB0 of CPU315-2DP. I connected first probe to I0.0 of CPU314 signal module and second probe to Q0.0 of CPU315-2DP signal module. I change the logic state of signal on I0.0 and I measure time delays between rising and falling edge. I have done ten tests for rising edge and ten tests for falling edge for 19.2Kbit/s. Also,Ihave done ten tests for rising edge and ten tests for falling edge for 19.2Kbit/s. For 19.2Kbit/s polar values are like following:70ms,116ms. For 187.5Kbit/s polar values are like following:13ms,26.6ms. The scan cycle time in CPU315-2DP varies between 2 and 3ms. The problem is, why these values differ so much? I know part of them is contributed by signal module delays. As far as I know it can be up to a few ms. Am I right that main impact is done by retransmissions due to damged frames? ----------------------------- I have done test for Global Data definition instead of X_GET. For 19.2Kbit/s the polar values are like following:64.4ms,126ms (based on ten tests). Best regards, gumis
gumis Posts: 45 Rating: (0)
Suggestion To thank Quote Answer

6/14/2009 4:00 PM	Rate (0)
Darius Platinum Member Joined: 9/28/2005 Last visit: 12/10/2024 Posts: 2957 Rating: (243)	Hello, I think it is protocol and speed dependant. Don't you think ? If your 314 has DP communication capability the I think better way is to use Profibus on 187.5 kbps connection. Theoretically it must behave like ET module (I mean speed and protocol).
	Best regards, Darius
Suggestion To thank Quote Answer

6/14/2009 4:36 PM	Rate (0)
gumis Posts: 45 Rating: (0)	My purpose is learning as much as possible. Making the system as fast as possible is not necessary for me. The problem is, why delays values for specific speed ratediffer from each other so much? Also, why these differences are lower for higher speed rate? My CPU314 has not Profibus capability but CPU315-2DP was speaking with Simoreg via Profibus DP during measurements. Regards, gumis
gumis Posts: 45 Rating: (0)	Last edited by: gumis at: 6/14/2009 4:39 PM
Suggestion To thank Quote Answer

6/14/2009 5:22 PM	Rate (0)
Ayman Elotaify Platinum Member Joined: 5/28/2008 Last visit: 1/2/2025 Posts: 4518 Rating: (841)	Hello, First of every think I would like you to see that thread. I think we were talked more there about the same problem. Second try to use GD communication method and let's know if the delay become lower or longer. Third. You send a byte (IB) from CPU to the other CPU (QB). So let's imagine the following senario. 1. You make a DI high. and start measuring the time. 2. The CPU will consider this DI as high in the next PII update. 3. The CPU will execute the program in OB1, and also interrupts if present. Here send function will be performed to the second CPU 4. The 2nd CPU recieves from the first CPU these data and move it to outputs. I tried to think like both CPUs, May be Iam wrong, but I thik you have got my point now and understod where's the delay. Again, I tried as I could to clarify this phenomina and in case of any member has another ide, he is welcome. Regards,
	Ayman Elotaify
Suggestion To thank Quote Answer
This contribution was helpful to 1 thankful Users gumis

6/14/2009 5:46 PM	Rate (0)
jklm Platinum Member Joined: 12/22/2006 Last visit: 1/25/2022 Posts: 1742 Rating: (309)	Hello, If you would analyse what happens ineach MPI bus cycle, you would detect that some administrative eventsappear on regular times. One of them is finding out if some new partners were not connected in the mean time to the bus. There is a parameter in the MPI network properties called "highest station address", set by default on 31. So, if you have only station 2 and 3 for instance, station 3 must check at regulat interval the gap between 3 an 31 (and 31 to 2). Consequenceis that the bus cycle is not constant. Ifat the same time you configuration PG is connected to MPI, you have a device more, and depending on what runs on the PG (online functions), you will also have an impact. Another point is that the transmission is not synchronized with the periphery changes, as explained by Ayman Elotaify.
	Best regards jklm Suggestion was helpful? Feel free to rate it.
Suggestion To thank Quote Answer
This contribution was helpful to 1 thankful Users gumis

6/15/2009 6:06 AM	Rate (0)
Paresh Posts: 116 Rating: (1)	Hi Gumis, It is really nice question. I think every person who is working with PLC must come across such condition. According to me PLC scan time is a time which CPU takes to execute logic blocks which are defined by user along with some defined administrative tasks which are always mentioned in PLC manual. As far as communication between external device is concerned, it all depends on implementation scheme of PLC. You are having 2-3 ms of PLC scan time which means your logic blocks are getting executed within 2-3 ms. It do not indicate external PLC's ouptuts are getting updated within 2-3 ms. Since in PLC, there might be some backgroung thread handling communication between external devices and updaing outputs of external device will completely depend on this thread. For this reason, you may have noticed such timing differences. Thanks and Regards Paresh
Paresh Posts: 116 Rating: (1)
Suggestion To thank Quote Answer
This contribution was helpful to 1 thankful Users gumis

6/15/2009 10:47 AM	Rate (0)
gumis Posts: 45 Rating: (0)	Making summary: 1. Factors influencing on values of delays: - time of data transmission from signal module to PI memory (CPU314) - time of data transmission from PQ memory to signal module (CPU315-2DP) - execution of function responsible for MPI communication and any other instruction used in OB1inCPU315-2DP (as far as I know output of signal module are always refreshed at the end of PLC cycle) Above three factors do not make the main impact on delay values. I will check what is the transfer time from signal module to PI memory and from PQ to signal module. In my opinion, the main impact is made administrative events. As Mr. edr said, one from these PLC or both of them check if the new communication partners were connected in the mean time to the bus. My PG is not connected to any of these PLC during measurements. According to my measurements, using of global data available from NetPro does not speed up transmission. 2. Factors influencing differences in values of delays: - cycles of both PLC are not synchronized - I always change the state of bit in signal module in the random instant in CPU314 cycle Is the searching for new communication partner the only administrative task making by PLC working in MPI network? Have you got some information about MPI? I have not found so much about this. Best regards, gumis
gumis Posts: 45 Rating: (0)	Last edited by: gumis at: 6/15/2009 11:05 AM
Suggestion To thank Quote Answer

|

Share this page:

Share this page on...

Follow us on