Source code for skcomponents.processing.adc

import numpy as np
from skcomponents.processing.processor import SignalProcessor


class ADC(SignalProcessor):

    def __init__(self, bits: int, max_value: float = None, min_value: float = None, rescale: bool = True):
        """
        Used to digitize signals.

        Parameters
        ----------
        bits : int
            Number of bits in ADC
        max_value : float
            maximum possible measurement value. Optional, if None, value will be determined from data.
        min_value : float
            minimum possible measurement value. Optional, if None, value will be determined from data.
        rescale : bool
            If true data is converted back to original units. If False, data is left as counts.
        """
        self.bits = bits
        self.max_value = max_value
        self.min_value = min_value
        self.rescale = rescale

    @property
    def adu(self):
        """
        Analog to Ditigit Units.
        """
        return (self.max_value - self.min_value) / (2 ** self.bits)

    def process_signal(self, values: np.ndarray):

        if self.min_value is None:
            min_value = np.min(values)
        else:
            min_value = self.min_value
        if self.max_value is None:
            max_value = np.max(values)
        else:
            max_value = self.max_value

        values[values > max_value] = max_value
        values[values < min_value] = min_value
        bins = np.linspace(min_value, max_value, 2 ** self.bits)
        bins = bins + (bins[1] - bins[0]) / 2
        idx = np.digitize(values, bins, right=True)

        bins = bins - (bins[1] - bins[0]) / 2
        if self.rescale:
            return bins[idx]
        else:
            return idx

    def noise_estimate(self, signal: np.ndarray) -> np.ndarray:
        return np.ones_like(signal) * self.adu * 0.289  # quantization noise for rounded signal = 0.289LSB