Note
Click here to download the full example code
Make a Rectangular AWG¶
This simple example illustrates how to make a rectangular AWG and how to run a Caphe simulation on it. For this example, we use the IPKISS AWG Designer module.
Getting started¶
We start by importing the technology with other required modules:
from si_fab import all as pdk # noqa
from ipkiss3 import all as i3
import numpy
import pylab as plt
import awg_designer.all as awg
from si_fab_awg import all as awg_pdk
Next, create a template for the free propagation region. This defines the layers, slab modes, etc.
The aperture¶
Make a virtual aperture
ap = awg_pdk.SiRibAperture(
slab_template=slab_t,
aperture_core_width=W,
aperture_cladding_width=1.0,
)
ap_lo = ap.Layout()
ap_sm = ap.FieldModelFromCamfr()
ap_cm = ap.CircuitModel(simulation_wavelengths=[1.55])
The input star coupler¶
Make a multi-aperture for the arms consisting of N apertures like these, arranged in a circle and get the transformations of the individual apertures.
angle_step = i3.RAD2DEG * (W + 0.2) / R
angles_arms = numpy.linspace(-angle_step * (N - 1) / 2.0, angle_step * (N - 1) / 2.0, N)
ap_arms_in, _, trans_arms_in, trans_ports_in = awg.get_star_coupler_apertures(
apertures_arms=[ap] * N,
apertures_ports=[ap],
angles_arms=angles_arms,
angles_ports=[0],
radius=R,
mounting="confocal",
input=True,
)
Make the input star coupler
sc_in = awg.StarCoupler(aperture_in=ap, aperture_out=ap_arms_in)
sc_in_lo = sc_in.Layout(
contour=awg.get_star_coupler_extended_contour(
apertures_in=[ap],
apertures_out=[ap] * N,
trans_in=trans_ports_in,
trans_out=trans_arms_in,
radius_in=R,
radius_out=R,
extension_angles=(10, 5),
)
)
sc_in_lo.visualize()

/bbotworker/DEPLOY_IPKISSDOC/build/_ftb/awg_designer/components/star_coupler/layout.py:425: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.
front_shapes = np.array([fs[::-1] for fs in front_shapes])
<Figure size 640x480 with 1 Axes>
The output star coupler¶
Make the multi-apertures for the outputs and get the transformations for the individual apertures.
angle_step = i3.RAD2DEG * (4.7) / R
angles_ports = numpy.linspace(angle_step * (M - 1) / 2.0, -angle_step * (M - 1) / 2.0, M)
ap_arms_out, ap_out, trans_arms_out, trans_ports_out = awg.get_star_coupler_apertures(
apertures_arms=[ap] * N,
apertures_ports=[ap] * M,
angles_arms=angles_arms,
angles_ports=angles_ports,
radius=R,
mounting="rowland",
input=False,
)
Make the output star coupler
sc_out = awg.StarCoupler(aperture_in=ap_arms_out, aperture_out=ap_out)
sc_out_lo = sc_out.Layout(
contour=awg.get_star_coupler_extended_contour(
apertures_in=[ap] * N,
apertures_out=[ap] * M,
trans_in=trans_arms_out,
trans_out=trans_ports_out,
radius_in=R,
radius_out=R / 2,
extension_angles=(10, 10),
)
)
sc_out_lo.visualize()

/bbotworker/DEPLOY_IPKISSDOC/build/_ftb/awg_designer/components/star_coupler/layout.py:425: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray.
front_shapes = np.array([fs[::-1] for fs in front_shapes])
<Figure size 640x480 with 1 Axes>
The rectangular waveguide array¶
Make the RectangularWaveguideArray
<RectangularWaveguideArray.Layout view 'RectangularWaveguideArray_1:layout'>
The Arrayed Waveguide Grating¶
Make an AWG with the 3 building blocks
rect_awg = awg.ArrayedWaveguideGrating(
star_coupler_in=sc_in,
star_coupler_out=sc_out,
waveguide_array=waveguide_array,
)
awg_lo = rect_awg.Layout()
awg_lo.visualize()

<Figure size 640x480 with 1 Axes>
Running the Caphe simulation¶
print("Running Caphe simulation (with wavelength-independent star couplers)")
sc_in.CircuitModel(simulation_wavelengths=[1.55])
sc_out.CircuitModel(simulation_wavelengths=[1.55])
awg_cm = rect_awg.CircuitModel()
wavelengths = numpy.linspace(1.52, 1.58, 401)
import time # noqa
t0 = time.time()
S = awg_cm.get_smatrix(wavelengths)
print("Calculation time:", time.time() - t0)
for i in range(1, M + 1):
plt.plot(wavelengths, 10 * numpy.log10(numpy.abs(S["in1", "out{}".format(i)]) ** 2), label="out{}".format(i))
plt.xlabel("Wavelength")
plt.ylabel("Transmission [dB]")
plt.legend()
plt.show()

Running Caphe simulation (with wavelength-independent star couplers)
/bbotworker/DEPLOY_IPKISSDOC/build/ipkiss/boolean_ops/boolean_ops_elements.py:132: ShapelyDeprecationWarning: The 'cascaded_union()' function is deprecated. Use 'unary_union()' instead.
shapely_polygon = cascaded_union(MultiPolygon(shapely_polygons))
/bbotworker/DEPLOY_IPKISSDOC/build/ipkiss/boolean_ops/boolean_ops_elements.py:227: ShapelyDeprecationWarning: The array interface is deprecated and will no longer work in Shapely 2.0. Convert the '.coords' to a numpy array instead.
shape = shapely_geom_to_shape(geom)
/bbotworker/DEPLOY_IPKISSDOC/build/dependencies/shapely_wrapper.py:112: ShapelyDeprecationWarning: The 'cascaded_union()' function is deprecated. Use 'unary_union()' instead.
up = cascaded_union(mp)
/bbotworker/DEPLOY_IPKISSDOC/build/ipkiss/plugins/vfabrication/geometry.py:144: ShapelyDeprecationWarning: The array interface is deprecated and will no longer work in Shapely 2.0. Convert the '.coords' to a numpy array instead.
shape = shapely_geom_to_shape(geom)
/bbotworker/DEPLOY_IPKISSDOC/build/ipkiss3/algorithms/shape_cutting.py:50: ShapelyDeprecationWarning: STRtree will be changed in 2.0.0 and will not be compatible with versions < 2.
tree = STRtree(polys)
/bbotworker/DEPLOY_IPKISSDOC/build/ipkiss3/algorithms/shape_cutting.py:53: ShapelyDeprecationWarning: The array interface is deprecated and will no longer work in Shapely 2.0. Convert the '.coords' to a numpy array instead.
interiors = [np.array(interior) for interior in g.interiors]
/bbotworker/DEPLOY_IPKISSDOC/build/ipkiss/boolean_ops/boolean_ops_shapely.py:77: ShapelyDeprecationWarning: The array interface is deprecated and will no longer work in Shapely 2.0. Convert the '.coords' to a numpy array instead.
s = shapely_geom_to_shape(poly)
Calculation time: 68.26406264305115
Using SpectrumAnalyzer to analyze the results¶
We finally use i3.SpectrumAnalyzer
to
detect the peaks and calculate the crosstalk.

<Figure size 640x480 with 1 Axes>
Let’s calculate the passbands for a cutoff of 25 dB:
bands = sa.cutoff_passbands(-25)
And then calculate the nearest neighbor crosstalk for each channel. The value reported for out1 is the crosstalk caused by the neirest channels out8 and out2:
print(sa.near_crosstalk(bands))
OrderedDict([('out1', -31.236477229166177), ('out2', -28.541374107192397), ('out3', -28.884413846498934), ('out4', -29.581950083269703), ('out5', -29.07117242393749), ('out6', -28.661616260100196), ('out7', -28.434425585437545), ('out8', -31.465196995693372)])
We also calculate the far neighbor crosstalk, which ignores nearest neighbors:
print(sa.far_crosstalk(bands))
OrderedDict([('out1', -34.56778563008565), ('out2', -34.66204499735494), ('out3', -35.45329942167298), ('out4', -36.85881897276704), ('out5', -35.37251774450588), ('out6', -35.2135305925463), ('out7', -34.478369146023596), ('out8', -34.594663877253055)])