Sample Times in Systems and Subsystems

Purely Discrete Systems

A purely discrete system is composed solely of discrete blocks and can be modeled using either a fixed-step or a variable-step solver. Simulating a discrete system requires that the simulator take a simulation step at every sample time hit. For a multirate discrete system—a system whose blocks Simulink^® samples at different rates—the steps must occur at integer multiples of each of the system sample times. Otherwise, the simulator might miss key transitions in the states of the system. The step size that the Simulink software chooses depends on the type of solver you use to simulate the multirate system and on the fundamental sample time.

The fundamental sample time of a multirate discrete system is the largest double that is an integer divisor of the actual sample times of the system. For example, suppose that a system has sample times of 0.25 and 0.50 seconds. The fundamental sample time in this case is 0.25 seconds. Suppose, instead, the sample times are 0.50 and 0.75 seconds. The fundamental sample time is again 0.25 seconds.

The importance of the fundamental sample time directly relates to whether you direct the Simulink software to use a fixed-step or a variable-step discrete solver to solve your multirate discrete system. A fixed-step solver sets the simulation step size equal to the fundamental sample time of the discrete system. In contrast, a variable-step solver varies the step size to equal the distance between actual sample time hits.

The following diagram illustrates the difference between a fixed-step and a variable-step solver.

In the diagram, the arrows indicate simulation steps and circles represent sample time hits. As the diagram illustrates, a variable-step solver requires fewer simulation steps to simulate a system, if the fundamental sample time is less than any of the actual sample times of the system being simulated. On the other hand, a fixed-step solver requires less memory to implement and is faster if one of the system sample times is fundamental. This can be an advantage in applications that entail generating code from a Simulink model (using Simulink Coder™). In either case, the discrete solver provided by Simulink is optimized for discrete systems; however, you can simulate a purely discrete system with any one of the solvers and obtain equivalent results.

Consider the following example of a simple multirate system. For this example, the DTF1 Discrete Transfer Fcn block Sample time is set to [1 0.1] [], which gives it an offset of 0.1. The Sample time of the DTF2 Discrete Transfer Fcn block is set to 0.7 , with no offset. The solver is set to a variable-step discrete solver.

Running the simulation and plotting the outputs using the stairs function

set_param(bdroot,'SolverType','Variable-Step','SolverName','VariableStepDiscrete','SaveFormat','Array');
simOut = sim(bdroot,'Stoptime','3');
stairs(simOut.tout,simOut.yout,'-*','LineWidth',1.2);
xlabel('Time (t)');
ylabel('Outputs (out1,out2)');
legend('t_s = [1, 0.1]','t_s = 0.7','location','best')

produces the following plot.

(For information on the sim command. see Run Simulations Programmatically. )

As the figure demonstrates, because the DTF1 block has a 0.1 offset, the DTF1 block has no output until t = 0.1. Similarly, the initial conditions of the transfer functions are zero; therefore, the output of DTF1, y(1), is zero before this time.

Hybrid Systems

Hybrid systems contain both discrete and continuous blocks and thus have both discrete and continuous states. However, Simulink solvers treat any system that has both continuous and discrete sample times as a hybrid system. For information on modeling hybrid systems, see Modeling Hybrid Systems.

In block diagrams, the term hybrid applies to both hybrid systems (mixed continuous-discrete systems) and systems with multiple sample times (multirate systems). Such systems turn yellow in color when you perform an Update Diagram with Sample Time Display Colors turned 'on'. As an example, consider the following model that contains an atomic subsystem, “Discrete Cruise Controller”, and a virtual subsystem, “Car Dynamics”.

Car Model

With the Sample Time option set to All, an Update Diagram turns the virtual subsystem yellow, indicating that it is a hybrid subsystem. In this case, the subsystem is a true hybrid system since it has both continuous and discrete sample times. As shown below, the discrete input signal, D1, combines with the continuous velocity signal, v, to produce a continuous input to the integrator.

Car Model After an Update Diagram

Car Dynamics Subsystem After an Update Diagram

Now consider a multirate subsystem that contains three Sine Wave source blocks, each of which has a unique sample time — 0.2, 0.3, and 0.4, respectively.

Multirate Subsystem After an Update Diagram

An Update Diagram turns the subsystem yellow because the subsystem contains more than one sample time. As shown in the block diagram, the Sine Wave blocks have discrete sample times D1, D2, and D3 and the output signal is fixed in minor step.

In assessing a system for multiple sample times, Simulink does not consider either constant [inf, 0] or asynchronous [–1, –n] sample times. Thus a subsystem consisting of one block that outputs constant value and one block with a discrete sample time will not be designated as hybrid.

The hybrid annotation and coloring are very useful for evaluating whether or not the subsystems in your model have inherited the correct or expected sample times.

Sample Time in Subsystems

Subsystems fall into two categories: triggered and non-triggered. For triggered subsystems, in general, the subsystem gets its sample time from the triggering signal. One exception occurs when you use a Trigger block to create a triggered subsystem. If you set the block Trigger type to function-call and the Sample time type to periodic, the SampleTime parameter becomes active. In this case, you specify the sample time of the Trigger block, which in turn, establishes the sample time of the subsystem.

There are four non-triggered subsystems:

Virtual
Enabled
Atomic
Action

Simulink calculates the sample times of virtual and enabled subsystems based on the respective sample times of their contents.

The atomic subsystem is a special case in that the subsystem block has a SystemSampleTime parameter. Moreover, for a sample time other than the default value of –1, the blocks inside the atomic subsystem can have only a value of Inf, –1, or the identical (discrete) value of the subsystem SampleTime parameter. If the atomic subsystem is left as inherited, Simulink calculates the block sample time in the same manner as the virtual and enabled subsystems. However, the main purpose of the subsystem SampleTime parameter is to allow for the simultaneous specification of a large number of blocks, within an atomic subsystem, that are all set to inherited. To obtain the sample time set on an atomic subsystem, use this command at the command prompt:

get_param(AtomicSubsystemBlock,‘SystemSampleTime’);

Finally, the sample time of the action subsystem is set by the If block or the Switch Case block.

For non-triggered subsystems where blocks have different sample rates, Simulink returns the Compiled Sample Time for the subsystem as a cell array of all the sample rates present in the subsystem. To see this, use the get_param command at MATLAB^® prompt.

get_param(subsystemBlock,'CompiledSampleTime')