Number of devices on a segment which is having CIF loops

[manik]

Dear Experts,

I have a following queries on FF segment calculations.

In one of our project we have designed a segment with 8 devices (8 AI and 2 AO ).

This two PID loops have been controlled by CIC with 1000 ms macro cycle.

Now the customer is looking for a CIF solution.

Here my doubt is, if we are implementing the CIF with the same 1000ms, can we accommodate the same 8AI and 2 AO devices in the same segment?

As per my understanding, if we are implementing the CIF then the latency time will be reduced and the same time number of devices in the loop also will be reduced.

Kindly comment me on my above mentioned queries.

Thanks in advance
Mani

[agupta]

The CIF will reduce the bus traffic, assuming the cycle time remain 1000 ms. At present, for every loop, there are at least two data transfers per cycle. One from AI to Controller in the Host and another Controller to AO. This will be replaced by on transfer from AI to AO. The AI to AO will be used usind CD-DT publishing, which uses less bus capacity.

[ankurmishra2007]

CIF will use less publication then control on Host......because on for loop having control on Host configuration is using 1 Publication from AI to Host, 1 for Host to AO and 1 for feedback. But in case of Control in Field publication will reduce because there is not host in between for communication, hence this will reduce the traffic on Bus.

As far as No of device configuration on segment concerns, there will not be any impact on no. of device per segment even if you will configure CIF loops.

Also Control in field increases the response time.

Hope my answers clear your doubt.


http://yourforum.7olm.org/forum

[rezabejd]

Originally Posted by ankurmishra2007

Also Control in field increases the response time.

I am not sure about the meaning you are intending from the above statement.

For the example cited, the time between scheduling / solving a new AI > PID > AO will be less since at least one "compel data" or publish event will be eliminated - two if the PID is solved in the valve positioner (the typical case) - maybe 60 ms less busy time per loop. However the response time of the loop (PID control) will hinge more on the macrocylce time. It is conceivable the macrocycle can be less (shortened) when you implement CIF; there will be more idle / free time compared to CIC. In this way you have the ability, if needed, to improve disturbance rejection and response to setpoint change w/CIF.

The other "feature" of CIF you will gain is determinism - you will be assured that the PID will solve a new output based on a new input (AI) exactly once every macrocycle (1000 ms) within ±1 ms, relentlessly with no unknown latencies (mismatch of cycle time between host & FF).

I concur with the advice you are getting; FF was conceived w/ the capability to solve PID in field devices and it works better when you utilize it. Your macrocycle will be less constrained and your PID will work better.

[jberge]

manik: Yes. You can keep the 8 AI and 2 AO on the same bus using 1000 ms when doing CIF.

Yes, correct. Using CIF the latency time will be reduced.

No. The number of devices does not have to be reduced.

http://www.fieldbus.org/images/stori..._field_arc.pdf

I also have a technical white paper on CIF. Send me an email on jonas.berge@emerson.com