What Kind of Data Throughput Should I Expect from an FP-3000?

[jws]

Execution Time

If you are running a PID control loop that is entirely contained on the FP-3000 (that is, the AI, PID, and AO blocks are all instantiated on the FP-3000) with no interaction with an HMI (Human-Machine Interface) software package (such as Lookout or BridgeVIEW), then you can expect a minimum 50 Hz control loop. Add the worst-case function block execution time for each function block in the loop to this minimum. A PID control loop has one AI, one PID, and one AO block. Together they take 20 ms. (The maximum execution time of a function block is a function block parameter you can find in the NI-FBUS Configurator under the Block Information tab for the individual function block.)

For example, if you want to run 10 PID control loops at 20 ms each, this will be a total of 200 ms of execution time (one-fifth of a second). You could run 10 loops as fast as 5 times per second. Be aware that the Expression block takes significantly longer to execute than other function blocks and is dependent on the downloaded expression. For specific information on the rate at which function blocks can execute entirely within one FP-3000, see the related link How many PID Loop Iterations per Second Can the FP-3000 Run? The document uses a PID loop as an example.

Bandwidth
The bandwidth of the bus affects communications between devices, including the interface board on the host machine. This is important if you are running a loop with function blocks from different devices. For example, a PID loop with the AI block on one device and the PID and AO blocks on another will not run as fast as a PID control loop contained entirely within one device. This is due to the time necessary for the data from the AI to cross the bus. The scheduler in the NI-FBUS Configurator software will take bus traffic into account when scheduling function block execution. The amount of information that can cross the bus in a given period of time depends on several factors including:


Whether you have adjusted the bus timing parameters to accommodate slower devices
How many devices are connected to the bus
Whether you are using OPC or traditional reads and writes (for host communications)

Note that if you want to monitor values with an HMI package on the host machine, you may have to slow down the execution rate. This is because HMI communications are considered unscheduled communications by the specification. Scheduled, publisher-subscriber Fieldbus communications are of higher priority than nonscheduled HMI communications. If you do not leave enough unscheduled time in your macrocycle, the HMI communications will never get through.

Rules of Thumb
Concerning Unscheduled Time:

Start with 30-40% of the total schedule (macrocycle) available for unscheduled communications. With slow devices (NOT the FP-3000) you can often include a large part of the time that function blocks are executing in your calculation of this percentage. This is because on slow devices only a small part of the execution time involves broadcasting information over the bus. During the rest of the execution time, there is no scheduled communication on the bus. With the FP-3000, the function block execution times are so short that there is usually not enough time to send an unscheduled message between the scheduled broadcasts of the function blocks. In this case, you need to include extra time in your macrocycle for unscheduled communication.

Concerning Interaction With the Host:
NI-FBUS Communications Manager version 2.3 -- expect roughly four reads per second from typical third-party transmitters.
NI-FBUS Communications Manager version 2.3 -- expect (as a best case) to be able to read 30 FP-3000 "channels" per second. (The FP-3000 is a very fast Fieldbus device.)
Using OPC with NI-FBUS Communications Manager version 2.3.5 -- expect (as a best case) to be able to read 400 analog or 1000 discrete FP-3000 "channels" per second. Note that this is divided among all the FP-3000s on the bus.