Quantizing Floating-Point Filters (Digital Filter Design Toolkit)

After selecting a filter structure, you must quantize the coefficients of the reference floating-point filter. Quantizing the filter coefficients is the process of approximating each floating-point value with a fixed-point value that you use in a fixed-point mathematical computation or hardware implementation. Using the DFD FXP Quantize Coef VI, you can configure the filter coefficients quantizer Q_C and convert the reference floating-point filter to a fixed-point filter.

If you use the Easy instance of the DFD FXP Quantize Coef VI to quantize the filter coefficients, you must complete the following steps:

1. Specify appropriate word length values for coefficients a/k word length and coefficients b/v word length.
2. Specify the appropriate gain processing target.
3. Specify the appropriate gain word length value. This VI automatically calculates the integer word lengths for coefficients a/k, coefficients b/v, and gain. The VI then uses the resulting values to quantize the filter coefficients.

If you use the Advanced instance of the DFD FXP Quantize Coef VI to quantize the filter coefficients, you must complete the following steps:

1. Configure the coefficients a/k quantizer and coefficients b/v quantizer.
2. Specify the appropriate gain processing target.
3. Specify the appropriate gain word length value. This VI then automatically calculates the integer word length of gain and uses the resulting value to quantize the gain if you set the gain processing input to On Target.

Specifying the Gain Processing Target

A filter gain implies a multiplication operation. If you process the filtered signal on a target, for example, an NI Reconfigurable I/O (RIO) target, the filtering process requires hardware resources on the target for the multiplication operation that the filter gain introduces. However, if you want to process the filtered signal on a host machine, you can move the filter gain operation to the host machine and save resources on the target. Use the gain processing input of the DFD FXP Quantize Coef VI to specify an appropriate gain processing target.

Refer to the Lowpass_Step 3_Analyze Quantized Lowpass VI in the labview\examples\Digital Filter Design\Case Studies\Single-Rate Filter directory for an example that demonstrates how to quantize a floating-point lowpass filter and analyze the quantized lowpass filter.

Open example