Using Simulink

Typecasting Parameters

In general, during simulation, Simulink silently converts parameter data types to signal data types (if they differ) when computing block outputs that are a function of an input signal and a parameter. The following exceptions to this rule occur:

• If the signal data type cannot represent the parameter value, Simulink halts the simulation and signals an error.

• Consider, for example, the following model.

  

  This model uses a Gain block to amplify a constant input signal. Computing the output of the Gain block requires computing the product of the input signal and the gain. Such a computation requires that the two values be of the same data type. However, in this case, the data type of the signal, uint8 (unsigned 8-bit word), differs from the data type of the gain parameter, int32 (signed 32-bit integer). Thus computing the output of the Gain block entails a type conversion.

  When making such conversions, Simulink always casts the parameter type to the signal type. Thus, in this case, Simulink must convert the Gain block's gain value to the data type of the input signal. Simulink can make this conversion only if the input signal's data type (uint8) can represent the gain. In this case, Simulink can make the conversion because the gain is 255, which is within the range of the uint8 data type (0 to 255). Thus, this model simulates without error. However, if the gain were slightly larger (for example, 256), Simulink would signal an out-of-range error if you attempted to simulate the model.

• If the signal data type can represent the parameter value but only at reduced precision, Simulink optionally issues a warning message and continues the simulation (see Parameter precision loss).

• Consider, for example, the following model.

  

  In this example, the signal type accommodates only integer values, while the gain value has a fractional component. Simulating this model causes Simulink to truncate the gain to the nearest integral value (2) and issue a loss-of-precision warning. On the other hand, if the gain were 2.0, Simulink would simulate the model without complaint because in this case the conversion entails no loss of precision.

Note    Conversion of an int32 parameter to a float or double can entail a loss of precision. The loss can be severe if the magnitude of the parameter value is large. If an int32 parameter conversion does entail a loss of precision, Simulink issues a warning message.

Enabling Strict Boolean Type Checking

Working with Data Objects

© 1994-2005 The MathWorks, Inc.