Systemsoftware
2-20
System- and communication configuring D7-SYS - SIMADYN D
Edition 03.2001
NOTE
If time-critical functions are processed on the CPUs of a subrack, then
please observe the following rules:
•
Limit the number of $ signals to a minimum.
•
Select the L bus for the $ signals, which are configured in interrupt
tasks (alarm tasks).
•
Select the C bus for the $ signals, which are not configured in
interrupt tasks (alarm tasks).
•
If possible, configure all of the communication links of the rack
coupling to one or a maximum of two CPUs of the subrack.
•
Configure the CPUs with the configured communication links of the
rack coupling so that, if possible, there are no additional CPUs
between these CPUs and the rack coupling module.
2.1.6.5 Data transfer between interrupt tasks of several CPUs
A fast $ signal must always be configured if the signal is generated or
used in an interrupt task. This is the because an interrupt event can occur
at any instant in time and therefore the consistency mechanisms must be
bypassed in order to prevent data loss. In this case, a conflict could occur
between the demand for data consistency and low deadtimes. A decision
must now be made depending on the particular application.
NOTE
It should always be checked as to whether problems could occur if there
is no data consistency (data consistency mechanism bypassed).
The data consistency can be achieved by looping the signals through a
cyclic task on the CPU module which is used to calculate the interrupt
task. The deadtime computation is illustrated in the following table.
Time interval
Computation
Minimum value
1 * Tx
Maximum value
2 * Tx + 1 * Ty + 1 * T_interrupt
Table 2-8
Deadtime computation
•
Tx = sampling time of the cyclic tasks through which the signals are
looped,
•
Ty = sampling time of the source/destination (target) CPU and
•
T_alarm = maximum interrupt repeat time of the interrupt task.
Fast $ signal
