Filter Design Toolbox

Quantized FFT Properties Reference

To implement an FFT or inverse FFT (IFFT) algorithm, you specify a quantized FFT, along with its property values. When you create a quantized FFT, you are creating a MATLAB object. You specify the FFT quantization parameters as values assigned to the quantized FFT properties. With these property values, you specify the quantized FFT:

• Data formats
• Length
• Radix number (2 or 4)
• Scaling values for each stage

For a quick reference to properties, see A Quick Guide to Quantized FFT Properties. Details of all of the properties associated with quantized FFTs are described in the following sections in alphabetical order.

CoefficientFormat

The CoefficientFormat property values specify how FFT coefficients (twiddle factors) are quantized in the FFT algorithm. You specify these values with a quantizer. You set them according to the quantizer property values:

• Format (except when the Mode property value is set to 'double' or 'single')
• Mode
• OverflowMode
• RoundMode

Default value: quantizer('fixed','round','saturate',[16,15])

InputFormat

The InputFormat property values specify how inputs are quantized in the FFT algorithm. You specify these values with a quantizer. You set them according to the property values of a quantizer's:

• Format (except when the Mode property value is set to 'double' or 'single')
• Mode
• OverflowMode
• RoundMode

Default value: quantizer('fixed','floor','saturate',[16,15])

Length

The Length property value is a scalar integer indicating the length of the FFT. Specify the length as a power of the radix number.

Default value: 16

NumberOfSections

The value of this read-only property is a scalar that specifies the number of sections (stages) in your FFT algorithm. This number is computed from the Length and the Radix property values as

• log₂(Length)/log₂(Radix)

Default value: 4

MultiplicandFormat

Products in quantized FFTs always involve two types of multiplicands:

• Inputs (data)
• Coefficients

The MultiplicandFormat property values specify how inputs that are multiplied by coefficients are quantized in the FFT algorithm. You specify these values with a quantizer. You set them according to the property values of a quantizer's:

• Format (except when the Mode property value is set to 'double' or 'single')
• Mode
• OverflowMode
• RoundMode

Default value: quantizer('fixed','floor','saturate',[16,15])

OutputFormat

The OutputFormat property values specify how outputs are quantized in the FFT algorithm. You specify these values with a quantizer. You set them according to the property values of a quantizer's:

• Format (except when the Mode property value is set to 'double' or 'single')
• Mode
• OverflowMode
• RoundMode

Default value: quantizer('fixed','floor','saturate',[16,15])

ProductFormat

The ProductFormat property values specify how the results of multiplication are quantized in the FFT algorithm. You specify these values with a quantizer. You set them according to the property values of a quantizer's:

• Format (except when the Mode property value is set to 'double' or 'single')
• Mode
• OverflowMode
• RoundMode

Default value: quantizer('fixed','floor','saturate',[32,30])

Radix

The Radix property indicates the form of the FFT algorithm you want to apply. The Radix property value can be either:

• 2 (default)
• 4

ScaleValues

The ScaleValues property values are specified as a scalar (or vector) that introduces scaling for inputs (and the outputs from each FFT section in the vector case) to the FFT algorithm:

• Specify the ScaleValues property value as a scalar if you only want to scale the input to the FFT algorithm.
• Specify the ScaleValues property as a vector of length L if you have L sections in your FFT, and you want to scale:
• The input to the first section (with the first entry in the vector you supply)
• The input to each subsequent section (with each corresponding entry in the vector you supply)

Default value: 1

Remarks: The value of the ScaleValues property is not quantized. Data affected by the presence of a scaling factor within the FFT algorithm is quantized according to the appropriate data format.

It is good practice to choose values for this property that are positive or negative powers of two.

SumFormat

The SumFormat property values specify how the results of addition are quantized in the FFT algorithm. You specify these values with a quantizer. You set them according to the property values of a quantizer's:

• Format (except when the Mode property value is set to 'double' or 'single')
• Mode
• OverflowMode
• RoundMode

Default value: quantizer('fixed','floor','saturate',[32,30])

A Quick Guide to Quantized FFT Properties

Function Reference