Moku Laser Lock Box implements a filter upstream of the setpoint and before the signal is split into fast and slow paths. In addition to lowpass and bandstop shapes, it is possible to realise a custom filter with user-provided coefficients.

The custom IIR filter is implemented as four cascaded, direct-form I, second-order sections, with a final output gain stage. The total transfer function is given by:

To specify a filter, you must supply a text file containing the filter coefficients. The file should have six coefficients per line, with each line representing a single stage. If output scaling is required, this should be given on the first line:

Entries should be comma-separated, e.g.

7.8357416974,
1.0000000000, 0.0044157, 0.0088314, 0.0044157, -1.669291,  0.969226,
1.0000000000, 0.0472217, 0.0944434, 0.0472217, -1.898858,  0.9341904,
1.0000000000, 0.0375275, 0.0750551, 0.0375275, -1.9259771, 0.9311308,

Each coefficient must be in the range [-4.0, +4.0). Internally, these are represented as signed 48-bit fixed-point numbers, with 45 fractional bits. The output scaling can be up to 8,000,000.

The filter sample rate depends on the device in use:

Moku Pro - 78.125 MHz

Moku Lab - 31.25 MHz

Moku Go - 7.8125 MHz

Filter coefficients can be computed using signal processing toolboxes. For example, we can use the scipy package to generate the coefficients for a second-order Butterworth filter for Moku Lab as follows:

from scipy import signal
import numpy

def gen_butterworth(corner_frequency):
    """
    Generate coefficients for a second order butterworth low-pass filter.

    Corner frequencies should be in the range:
    1 kHz < corner frequency < 31.25 MHz for Moku Lab.
    """
    sample_rate = 31.25e6
    b, a = signal.butter(2, corner_frequency, 'low', analog = False, fs = sample_rate)

    coeffs = [[1.0, b[0], b[1], b[2], a[1], a[2]]]
    numpy.savetxt("coeffs.txt", coeffs, delimiter=",")

Filter coefficients may also be set using API calls, see set_custom_filter.