Echelon Neuron C - Manual (Page 76)

64 How Devices Communicate Using Network Variables
poll(nviSetpoint);
}
when(nv_update_occurs(nviSetpoint))
when(nv_update_occurs(nviCurrent)) {
nvoValve = control_algorithm(nviCurrent, nviSetpoint);
}
While this straight-forward solution is adequate for many devices, it is not
without problems:
• The multi-NV polling approach still needs to ensure that all input
network variables have been updated prior to computing a new output
value. The last example also needs to track arrival of all input network
variable updates, as demonstrated in the previous example.
• Polling network variables immediately after reset (or power-up) can lead
to network congestion for a site-wide power-up, and can cause the
network stabilization to take some time after power has been applied.
While reset-time polling is not a problem if only a small number of devices issue
only a small number of poll requests, a single device (or its developer) does not
normally know the nature and application algorithm of the other devices in the
network.
When using power-up polls, combined with a suitable technique to monitor input
network variable updates, it is strongly recommended that you insert a random
delay of a few seconds between the reset and the initiation of these poll requests.
This delay helps to spread the power-up peak traffic demand in the network, and
improves overall startup performance.
Example:
network input SNVT_temp nviCurrent;
network input SNVT_temp nviSetpoint;
network output SNVT_volt nvoValve;
mtimer powerupDelay;
when(reset) {
powerupDelay = 16ul * random() + 500ul;
// 0.500 to 4.5s delay
}
when(timer_expires(powerupDelay)) {
poll(nviCurrent);
poll(nviSetpoint);
}
when(nv_update_occurs(nviSetpoint))
when(nv_update_occurs(nviCurrent)) {
nvoValve = control_algorithm(nviCurrent, nviSetpoint);
}
The best option, where appropriate, is to use poll-free techniques. Tracking of
incoming updates, keeping records of last known-good-values or use of
heartbeating sensors are all good tools to solve the problem of initial network
variable updates for many devices.