Image Processing Reference
In-Depth Information
If management traffic is directly mingled with application data, such as in CAN, LonWorks, or
PROFIBUS PA, care must be taken that this management traffic does not influence the primary con-
trol application. his is typically achieved by analyzing network traffic and leaving enough bandwidth
headroom. For complex systems and safety-critical systems that require certain guarantees on system
behavior,thisanalysiscanbecomeverydiicult.
22.8.1 Monitoring and Diagnosis
In order to perform passive monitoring of the communication of the application, it usually suffices
to listen on the bus. However, the monitoring device must have knowledge of the communication
scheme used in the network, in order to be able to understand and decode the data traffic. If this
scheme is controlled by the physical time, as it is the case in time-triggered networks, the monitoring
node must also synchronize itself to the network.
Advanced field devices often have built-in self-diagnostic capabilities and can disclose their own
status to the management system.
he handling of the control flow of diagnostic information varies for
different fieldbus systems
. Typically, a diagnosis tool or the diagnosis part of the management frame-
work will regularly check the information in the nodes. his method is called
status polling
.Insome
ieldbusprotocols(e.g.,FF),devicescanalsotransmitstatusmessagesbythemselves(alertreporting).
In general, the restrictions from the implementation of the management interface of a fieldbus
protocol also apply to monitoring, since in most fieldbus systems the monitoring traffic is transmitted
using the management interface.
For systems that do not provide this separation of management from application information at the
protocol level, other means must be taken to ensure that monitoring does not interfere with the field-
bus application. Since status polling usually is performed periodically, it should be straightforward
to reserve adequate communication resources during system design, so that the control application
is not disturbed. In case of alert reporting, the central problem without adequate arbitration and
scheduling mechanisms is how to avoid overloading the network in case of “alarm showers,” where
many devices want to send their messages at once. It can be very difficult to give timeliness guaran-
tees (e.g., the time between an alarm occurs and the time it is received by the respective target) in
such cases. he typical approach to deal with this problem (e.g., as taken in CAN) is to provide much
bandwidth headroom.
For in-depth diagnosis of devices, it is sometimes also desirable to monitor operation and inter-
nals of individual field devices. his temporarily involves greater data traffic which cannot be easily
reserved a priori. herefore, the management interface must provide some flexibility on the diagnosis
data in order to dynamically adjust to the proper level of detail using some kind of “Pan and Zoom”
approach [].
22.8.2 Calibration
The calibration of transducers is an important management function in many fieldbus applica-
tions.hereissomeambiguityinvolvedconcerningtheuseofthisterm.Bergestrictlydistinguishes
between calibration and range setting:
“Calibration is the correction of sensor reading and physical outputs so they match a standard”
[, p. ]. According to this definition, calibration cannot be performed remotely since the device
must be connected to a standardized reference input.
Range setting is used to move the value range of the device so that the resulting value delivers the
correctly scaled percentage value. It does not require applying an input and measuring an output,
thus it can be performed remotely. In the HART bus this operation is called calibration, whereas
calibration is called
trim
.
