Version 6.8 (R2008b) Signal Processing Blockset™ Software Release Notes

Version 6.8 (R2008b) Signal Processing Blockset Software

This table summarizes what's new in Version 6.8 (R2008b):

New features and changes introduced in this version are

• Code Generation Support for Signals Up to 128 Bits

• New Biquad Filter Block

• Increased N-Dimensional Support

• New Data Type Support Table

• Rectangular ROI Support for Minimum and Maximum Blocks

• Autocorrelation Block Provides Additional Fixed-Point Support

• QR Factorization Block Supports Full-Sized Outputs

• Counter Block Enhancements

• FFT Block Correctly Applies Divide-By-Two on Butterfly Outputs

• Matrix Sum Block Is Now the Simulink Sum Block

Code Generation Support for Signals Up to 128 Bits

Signal Processing Blockset™ software now supports C code generation and the Simulink Accelerator and Rapid Accelerator modes for fixed-point and integer word lengths up to 128 bits. This enhancement enables you to generate code for anything you can simulate using Signal Processing Blockset blocks. The only exception is the Sort block, which only supports code generation up to 32 bits for fixed-point and integer signals.

New Biquad Filter Block

A new Biquad Filter block has been added to the Signal Processing Blockset Filtering / Filter Designs library. This block allows you to implement IIR filters with optimized numerics.

Increased N-Dimensional Support

The following blocks now support N-D signals:

• Downsample

• DCT

• FFT

• IDCT

• IFFT

• Window Function

New Data Type Support Table

You can now access the Signal Processing Blockset Data Type Support Table through the Simulink model Help menu. The table provides information about data type support and code generation coverage for all Signal Processing Blockset blocks. Select Help > Block Support Table > Signal Processing Blockset or Help > Block Support Table > All Tables. As always, you can also type showsignalblockdatatypetable at the MATLAB command line to bring up the table.

Rectangular ROI Support for Minimum and Maximum Blocks

The Minimum and Maximum blocks now support Rectangular ROI (region of interest). See the block reference pages for more information.

Autocorrelation Block Provides Additional Fixed-Point Support

The Autocorrelation block now supports the scaling of fixed-point signals. In previous releases, the block's Scaling parameter had to be set to None for fixed-point inputs. See the block reference page for more information.

QR Factorization Block Supports Full-Sized Outputs

The QR Factorization block now supports full-sized output matrices Q and R. In previous releases, the block only produced economy-sized outputs. This release adds the Output size parameter to enable you to select either Economy or Full outputs. Refer to the block reference page for more information.

Counter Block Enhancements

This release brings the following enhancements to the Counter block:

• The former Clk port label is now Inc when the block is counting up, and Dec when the block is counting down.

• The Inc/Dec port now supports 8-, 16-, and 32-bit signed and unsigned integer and fixed-point data types.

• A new Max port was added to enable you to control the Counter size via an input port.

• The Hit values parameter now accepts vectors as well as scalar values.

• A new check is in place to require that the Hit values are integers.

• A new check is in place to require Maximum count to be an integer.

• A new check is in place to require Maximum count to be representable by the Count data type.

Compatibility Considerations

Because of the new checking on the Maximum count and Hit values parameters, your models might produce new errors. Change these signals to data types supported by each parameter.

FFT Block Correctly Applies Divide-By-Two on Butterfly Outputs

The Skip divide-by-two on butterfly outputs for fixed-point signals check box on the Fixed-point pane of the FFT block dialog has been moved to the Main pane and renamed Divide butterfly outputs by two. This check box now correctly applies to both fixed-point and floating-point inputs. In releases prior to R2008b, the block always ignored this check box for floating-point inputs.

Compatibility Considerations

The FFT block can give results in R2008b and later releases that are scaled differently than in previous releases when all the following conditions are met:

• The block input is floating point.

• The Divide butterfly outputs by two check box is selected (or the Skip divide-by-two on butterfly outputs for fixed-point signals check box on the Fixed-point pane was NOT selected, for releases prior to R2008b) .

• The Logging mode parameter on the Fixed-Point Tool for the model is not set to Minimums, maximums and overflows.

Make sure that this check box is set correctly for your needs. Refer to the block reference page for more information.

Matrix Sum Block Is Now the Simulink Sum Block

The Matrix Sum block is now identical to the Simulink Sum block, with different defaults selected.

Compatibility Considerations

Your existing models that contain the old version of the Matrix Sum block will continue to work in this release, however, the old version of the block will be removed from the product in a future release. You should use the slupdate function to replace Matrix Sum blocks in your models with the new version. You might get slightly different results using the new version of the block when Inherit via internal rule is used.

Signal Processing Blockset Release Notes

Version 6.7 (R2008a) Signal Processing Blockset Software