import numpy as np
from finesse.components.general import Connector
from finesse.components.workspace import ConnectorWorkspace
from finesse.components.node import NodeDirection, NodeType
from finesse.parameter import float_parameter
[docs]class FilterWorkspace(ConnectorWorkspace):
pass
[docs]@float_parameter("gain", "Gain")
class Amplifier(Connector):
[docs] def __init__(self, name, gain=1):
super().__init__(name)
self.gain = gain
self._add_port("p1", NodeType.ELECTRICAL)
self.p1._add_node("i", NodeDirection.INPUT)
self._add_port("p2", NodeType.ELECTRICAL)
self.p2._add_node("o", NodeDirection.OUTPUT)
self._register_node_coupling("P1_P2", self.p1.i, self.p2.o)
def _get_workspace(self, sim):
if sim.signal:
if self.p1.i.full_name not in sim.signal.nodes:
return
refill = sim.model.fsig.f.is_changing or any(
p.is_changing for p in self.parameters
)
ws = FilterWorkspace(self, sim)
ws.signal.add_fill_function(self.fill, refill)
ws.frequencies = sim.signal.signal_frequencies[self.p1.i].frequencies
return ws
else:
return None
[docs] def fill(self, ws):
if ws.signal.connections.P1_P2_idx > -1:
for _ in ws.frequencies:
with ws.sim.signal.component_edge_fill3(
ws.owner_id,
ws.signal.connections.P1_P2_idx,
0,
0,
) as mat:
mat[:] = ws.values.gain
[docs] def eval(self, f):
return float(self.gain)
[docs]@float_parameter("gain", "Gain")
class Filter(Connector):
"""This is a generic Filter element that encapsulates some of the Scipy signal
filter tools. The `sys` attribute is the filter object which can be ZPK, BA, or SOS.
Parameters
----------
name : str
Name of element in the model
gain : Parameter
Overall floating point value gain to apply to the filter.
"""
[docs] def __init__(self, name, gain=1):
super().__init__(name)
self.gain = gain
self._add_port("p1", NodeType.ELECTRICAL)
self.p1._add_node("i", NodeDirection.INPUT)
self._add_port("p2", NodeType.ELECTRICAL)
self.p2._add_node("o", NodeDirection.OUTPUT)
self._register_node_coupling("P1_P2", self.p1.i, self.p2.o)
def _get_workspace(self, sim):
if sim.signal:
if self.p1.i.full_name not in sim.signal.nodes:
return
refill = sim.model.fsig.f.is_changing or any(
p.is_changing for p in self.parameters
)
ws = FilterWorkspace(self, sim)
ws.signal.add_fill_function(self.fill, refill)
ws.frequencies = sim.signal.signal_frequencies[self.p1.i].frequencies
return ws
else:
return None
[docs] def fill(self, ws):
Hz = self.eval(ws.sim.model_settings.fsig)
if ws.signal.connections.P1_P2_idx > -1:
for _ in ws.frequencies:
with ws.sim.signal.component_edge_fill3(
ws.owner_id,
ws.signal.connections.P1_P2_idx,
0,
0,
) as mat:
mat[:] = Hz
[docs] def bode_plot(self, f=None, n=None, return_axes=False):
"""Plots Bode for this filter.
Parameters
----------
f : optional
Frequencies to plot for in Hz (Not radians)
n : int, optional
number of points to plot
Returns
-------
axis : Matplotlib axis for plot if return_axes=True
"""
import matplotlib.pyplot as plt
import scipy
import scipy.signal
if f is not None:
w = 2 * np.pi * f
else:
w = None
# Need to make sure we are converting any symbolics to numerics before
# handing over to scipy
sys = (np.array(_, dtype=complex) for _ in self.sys)
w, mag, phase = scipy.signal.bode(sys, n=n)
fig, axs = plt.subplots(2, 1, sharex=True)
axs[0].semilogx(w / 2 / np.pi, mag)
axs[0].set_ylabel("Amplitude [dB]")
axs[1].semilogx(w / 2 / np.pi, phase)
axs[1].set_xlabel("Frequency [Hz]")
axs[1].set_ylabel("Phase [Deg]")
fig.suptitle(f"Bode plot for {self.name}")
if return_axes:
return axs
[docs]@float_parameter("gain", "Gain")
class ZPKFilter(Filter):
"""A zero-pole-gain filter element that is used for shaping signals in simulations.
It is a two port element. `p1` is the input port and `p2` is the output port. Each
one has a single node: `p1.i` and `p2.o`.
Parameters
----------
name : str
Name of element in the model
z : array_like[float | Symbols]
A 1D-array of zeros. Use `[]` if none are required. By default these are provided
in units of radians, not Hz.
p : array_like[float | Symbols]
A 1D-array of poles. Use `[]` if none are required. By default these are provided
in units of radians, not Hz.
k : [float | Symbol], optional
Gain factor for the zeros and poles. If `None` then its value is automatically
set to generate a unity gain at DC.
fQ : bool, optional
When True the zeros and poles can be specified in a tuple of
(frequency, quality factor) for each pole and zero. This automatically
adds the complex conjugate pair.
gain : Parameter
Overall gain for the filter. Differs from `k` as this is a `Parameter` so
can be easily switched on/off or varied during a simulation.
Examples
--------
Below are a few examples of using a ZPK filter in a simple simulation and
plotting the output.
>>> import finesse
>>> finesse.init_plotting()
>>> model = finesse.Model()
>>> model.parse(\"\"\"
... # Finesse always expects some optics to be present
... # so we make a laser incident on some photodiode
... l l1 P=1
... readout_dc PD l1.p1.o
... # Amplitude modulate a laser
... sgen sig l1.amp
...
... zpk ZPK_unity [] []
... link(PD.DC, ZPK_unity)
... ad unity ZPK_unity.p2.o f=fsig
...
... zpk ZPK_1 [] [-10*2*pi]
... link(PD.DC, ZPK_1)
... ad zpk1 ZPK_1.p2.o f=fsig
...
... zpk ZPK_2 [-10*2*pi] []
... link(PD.DC, ZPK_2)
... ad zpk2 ZPK_2.p2.o f=fsig
...
... # Using symbolics
... variable a 20*2*pi
... zpk ZPK_symbol [] [-1j*a, 1j*a] -1
... link(PD.DC, ZPK_symbol)
... ad symbol ZPK_symbol.p2.o f=fsig
...
... # Using gain parameter instead of k keeps the unity response at DC but
... # just flips the sign
... zpk ZPK_symbol2 [] [-1j*a, 1j*a] gain=-1
... link(PD.DC, ZPK_symbol2)
... ad symbol_gain ZPK_symbol2.p2.o f=fsig
...
... # Symbolics for an RC low pass filter
... variable R 100
... variable C 10u
... zpk ZPK_RC [] [-1/(R*C)]
... link(PD.DC, ZPK_RC)
... ad RC ZPK_RC.p2.o f=fsig
...
... fsig(1)
... xaxis(fsig, log, 0.1, 10k, 1000)
... \"\"\")
>>> sol = model.run()
>>> sol.plot(log=True)
"""
[docs] def __init__(self, name, z, p, k=None, *, fQ=False, gain=1):
super().__init__(name, gain)
import cmath
if k is None:
k = np.prod(np.abs(p)) / np.prod(np.abs(z))
root = lambda f, Q: -2 * np.pi * f / (2 * Q) + cmath.sqrt(
(2 * np.pi * f / (2 * Q)) ** 2 - (2 * np.pi * f) ** 2
)
if fQ:
self.z = []
for f, Q in z:
r = root(f, Q)
self.z.append(r)
self.z.append(r.conjugate())
self.p = []
for f, Q in p:
r = root(f, Q)
self.p.append(r)
self.p.append(r.conjugate())
else:
self.z = z
self.p = p
self.k = k
@property
def sys(self):
"""The scipy `sys` object.
In this case it is a tuple of (zeros, poles, k). This does not convert any
symbolics used into numerics.
"""
return (self.z, self.p, self.k * self.gain)
[docs] def eval(self, f):
"""Calculate the value of this filter over some frequencies.
Parameters
----------
f : array_like
Frequencies in units of Hz
Returns
-------
H : array_like
Complex valued filter output
"""
import scipy.signal as signal
return (
float(self.gain)
* signal.freqs_zpk(
np.array(self.z, dtype=complex),
np.array(self.p, dtype=complex),
float(self.k),
2 * np.pi * f,
)[1]
)
[docs]@float_parameter("gain", "Gain")
class ButterFilter(ZPKFilter):
[docs] def __init__(self, name, order, btype, frequency, *, gain=1, analog=True):
super().__init__(name, [], [], [], gain=gain)
self.__order = order
self.__btype = btype
self.__analog = analog
self.__frequency = frequency
self.set_zpk()
[docs] def set_zpk(self):
import scipy.signal as signal
z, p, k = signal.butter(
self.order,
2 * np.pi * np.array(self.frequency),
btype=self.btype,
analog=self.analog,
output="zpk",
)
self.z = z
self.p = p
self.k = k
@property
def frequency(self):
return self.__frequency
@frequency.setter
def frequency(self, value):
self.__frequency = value
self.set_zpk()
@property
def order(self):
return self.__order
@order.setter
def order(self, value):
self.__order = value
self.set_zpk()
@property
def btype(self):
return self.__btype
@btype.setter
def btype(self, value):
self.__btype = value
self.set_zpk()
@property
def analog(self):
return self.__analog
@analog.setter
def analog(self, value):
self.__analog = value
self.set_zpk()
[docs]@float_parameter("gain", "Gain")
class Cheby1Filter(ZPKFilter):
def __init__(self, name, order, rp, btype, frequency, *, gain=1, analog=True):
import scipy.signal as signal
zpk = signal.cheby1(
order,
rp,
2 * np.pi * np.array(frequency),
btype=btype,
analog=analog,
output="zpk",
)
super().__init__(name, *zpk, gain=gain)