Speed in MB+


Thread Starter

Beatriz Morte

The speed in our net is about 22-24ms (it is very big).
Our MB+ Network has 500m of length. We have 15 PLC's in the net (Modicon). We have 1 Scada System.
The kind of programming is with MSTR (instruction of Modicon). We know that, Modicon has other posibilities, like Peer Cop o Global Data Base (but the instalation is done).
The Scada System use communication blocks. Each scada points have a communication block.
We need to reduce the big time of the net.
Your comments and feedback will be appreciated.
A 22-24 msec network update time does not seem to me to "very big", it is on same order of magnitude as the PLC program scan time. Using global data or peer cop might be slightly more
efficient, but I wonder if you could specify the process problems you are having with the present installation.

Jay Kirsch
Macro Automatics
2985 E. Hillcrest Drive, Ste 101
Thousand Oaks, CA 91362
[email protected]
List Management Account wrote:

> Hi,
> The speed in our net is about 22-24ms (it is very big).
> Our MB+ Network has 500m of length. We have 15 PLC's in the net (Modicon).
> We have 1 Scada System.

It is actually possible to make RS485 'hubs' to do a star formation, which can often allow considerable improvements where large 2D/3D areas have to be coverd as opposed to strictly linear. They are not suitaable for all protocols, but should work fine with MODBUS RTU in master slave arrangements. MB+, with a ring architecture probably would not work. But, if you got you SCADA to act as a master and poll the slaves rather than use a ring (MB+) you should be fine.

Trouble is, nobody sells them! I have on occasions made my own, anybody know of anybody selling them?

(BTW, I am guessing at the ring architecture of MB+, never actually used it, the limitation of the hub is that slaves only see the master messages, so they cannot reply in turn. You could actually do a bidirectional affair, but it is messy. I did this a few years ago with a 1Mb CAN hub, but a needed quite a few limped LC lines for nansecond delays to resolve the meta-stability issues. Definitely not for the faint hearted!).
Dear Beatriz,

We have same problem before, but now we upgrade our software to Concept 2.1 from 2.0 and change all programming from MSTR to only DFB function ( do by load the LOADER function to DFB only ), so we increase free space of memory in our PLC to increase/ optimize our MB+. Of course you can do this during shutdown period only.
Hope this info can help you.

MB+ on Copper is spec-ed at 450 end to end without repeaters.

How are your MSTR blocks setup. Are they tied to the power rail?, Scheduled? or event driven?

I assume when you mention "The speed in our net is about 22-24ms" you are use MBPSTAT to get this info. What are your retries, bad CRC count etc
showing you.

Rob E.

Darold Woodward

In my experience (and I have done a fair amount of performance measurement of Modbus Plus networks), 24 ms to read all data from all 15 PLCs (if that is what you mean) is very good. I would expect something much slower. You must have
very little data moving between PLCs.

When you designed the network and used the "Modbus Plus Network Planning and Installation Guide," what did you estimate for network performance?

Global data can increase performance. If you have less than 32 words of data from each PLC, you can echange the data with the token rotation. Token rotation times for 15 nodes with 32 registers of global data should be around 10-20ms according to the Modicon's graph in the guide.

If you are collecting useful data from 15 nodes, you should see response times of 100-200ms. I suspect that you are possibly reporting the token rotation time rather than data acquisition times. There are also other factors. For example, if you are using a PC, the PC, HMI software, drivers, Mb+ card, and other devices all contribute delay.

The easiest fix is to determine your actual requirements and see if Mb+ can deliver them. Then you might be able to put a router in the middle of your network (or even more than one). The rare messages that need to traverse the whole network will be slower, but performance for each half will be much better.

Darold Woodward PE
SEL Inc.
[email protected]
MB+ is a token rotation peer to peer communication setup (that is everyone a master, multiole master) and NOT one master multiple slave
MB+ will only work on fibre in a ring format using the 490NRP25400 copper to fiber convertors.

Rob E.