elasticai.preprocessor.filter.filtering#

Module Contents#

Classes#

FilterCoeffs

Dataclass with filter coefficients Attributes: a: List with filter coefficients a b: List with filter coefficients b

SettingsFilter

Configuration class for defining the filter processor Attributes: gain: Integer with applied amplification factor [V/V] fs: Sampling rate [Hz] n_order: Integer with number of filter order or delay values in FIR-allpass f_filt: List with filter frequencies [Hz] (low/high-pass: only one value - rest: two values) type: String with selected filter algorithm [‘iir’, ‘fir’] f_type: String with selected filter structure [‘butter’, ‘cheby1’, ‘cheby2’, ‘ellip’, ‘bessel’] b_type: String with selected filter type [‘lowpass’, ‘highpass’, ‘bandpass’, ‘bandstop’, ‘notch’, ‘allpass’, ‘simple_low’]

Filtering

Data#

DefaultSettingsFilter

API#

class elasticai.preprocessor.filter.filtering.FilterCoeffs[source]#

Dataclass with filter coefficients Attributes: a: List with filter coefficients a b: List with filter coefficients b

a: list#

None

b: list#

None

class elasticai.preprocessor.filter.filtering.SettingsFilter[source]#

Configuration class for defining the filter processor Attributes: gain: Integer with applied amplification factor [V/V] fs: Sampling rate [Hz] n_order: Integer with number of filter order or delay values in FIR-allpass f_filt: List with filter frequencies [Hz] (low/high-pass: only one value - rest: two values) type: String with selected filter algorithm [‘iir’, ‘fir’] f_type: String with selected filter structure [‘butter’, ‘cheby1’, ‘cheby2’, ‘ellip’, ‘bessel’] b_type: String with selected filter type [‘lowpass’, ‘highpass’, ‘bandpass’, ‘bandstop’, ‘notch’, ‘allpass’, ‘simple_low’]

gain: float#

None

fs: float#

None

n_order: int#

None

f_filt: list#

None

type: str#

None

f_type: str#

None

b_type: str#

None

elasticai.preprocessor.filter.filtering.DefaultSettingsFilter#

‘SettingsFilter(…)’

class elasticai.preprocessor.filter.filtering.Filtering(setting: elasticai.preprocessor.filter.filtering.SettingsFilter, use_filtfilt: bool = False)[source]#

Bases: elasticai.preprocessor._common_func.CommonDigitalFunctions

Initialization

Class for Emulating Digital Signal Processing on FPGA

Parameters:

  • setting – Class for handling the filter stage (using SettingsFilter)

  • use_filtfilt – Boolean for applying zero-phase filtering

Returns:

None

get_coeffs() → elasticai.preprocessor.filter.filtering.FilterCoeffs[source]#

Getting the filter coefficients

Returns:

dataclass FilterCoeffs with filter coefficients

get_coeffs_quantized(bit_size: int) → tuple[elasticai.preprocessor.filter.filtering.FilterCoeffs, elasticai.preprocessor.filter.filtering.FilterCoeffs][source]#

Quantize the coefficients with given bit fraction for adding into hardware designs

Parameters:

bit_size – Integer with total bitwidth

Returns:

dataclass FilterCoeffs with quantized filter coefficients

get_coeffs_verilog_string(bitwidth: int, only_half_fir: bool = False) → str[source]#
filt(xin: numpy.ndarray) → numpy.ndarray[source]#

Apply filter structure on transient input data

Parameters:

xin – Numpy array with transient input data

Returns:

Numpy array with filtered data

filt_quantized(xin: numpy.ndarray, total_bitwidth: int, fraction_width: int, is_signed: bool = True) → numpy.ndarray[source]#

Apply filter structure on transient input data

Parameters:

  • xin – Numpy array with transient data

  • total_bitwidth – Integer with total bitwidth

  • fraction_width – Integer with fraction width

  • is_signed – Boolean with whether to sign the coefficients

Returns:

Numpy array with filtered and quantized data

plot_freq_response(num_points: int = 1001, show_plot: bool = True, path2save: str = '') → None[source]#

Function for plotting the frequency response of desired filter type

Parameters:

  • num_points – Number of points to plot

  • show_plot – Boolean for showing plot

  • path2save – Path to save figure

plot_grp_delay(num_points: int = 1001, show_plot: bool = False) → None[source]#

Plotting the Group Delay of filter

Parameters:

  • num_points – Number of points to plot

  • show_plot – Boolean for showing plot

Returns:

None

create_design(target: str, bitwidth: int, id: str, path2save: pathlib.Path, signed: bool = True) → None[source]#

Creating the hardware design for executing on specific target

Parameters:

  • target – String with target name [“mcu”, “pc”, “fpga”]

  • bitwidth – Integer with total bitwidth

  • id – String with unique identifier of device (appended to the name)

  • path2save – Path to save the hardware files

  • signed – Whether generated C designs use a signed integer data type

Returns:

None