siriusfox/ps_plotter
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Major refactor to ease duplicate computations and plotting

Browse source
This commit is contained in:
Luke 2018-07-20 19:07:01 -07:00
parent d6d75b804f
commit 539c2f7481
4 changed files with 326 additions and 120 deletions
Download patch file Download diff file Expand all files Collapse all files

62
FreqClass.py Normal file
View file

@ -0,0 +1,62 @@
#!/usr/bin/env python3
import numpy as np
class FreqClass:
def __init__(self, steps, f0, bw):
self.f0 = f0
self._bw = bw
self._steps = steps;
self._update_delta()
def _update_delta(self):
self._delta = self._bw/self.f0*np.linspace(-1/2,1/2,self._steps)
def __repr__(self):
return self.__str__()
def __str__(self):
return "%gGHz, %gGHz BW sweep [%d points]" % \
(self.f0, self._bw, self._steps)
@property
def hz_range(self):
return (np.min(self.hz), np.max(self.hz))
@property
def delta(self):
return self._delta
@property
def bw(self):
return self._bw
@bw.setter
def bw(self, bw):
self._bw = bw
self._update_delta()
@property
def steps(self):
return self._steps
@steps.setter
def steps(self, steps):
self._steps = steps
self._update_delta()
@property
def hz(self):
return self.f0*(1+self._delta)
@property
def f(self):
return self.hz
@property
def rad(self):
return 2*np.pi*self.f0*(1+self._delta)
@property
def w(self):
return self.rad
@property
def jw(self):
return 2j*np.pi*self.f0*(1+self._delta)
@property
def delta(self):
return self._delta

180
TankGlobals.py
View file

@ -1,50 +1,164 @@
#!/usr/bin/env python3
import numpy as np
import sys
################################################################################
# BEWARE, FOR BEYOND THIS POINT THERE BE DRAGONS! THIS IS ONLY FOR EASE OF
# GENERATING ACADEMIC PUBLICATIONS AND FIGURES, NEVER DO THIS SHIT!
################################################################################
def g1_map_default(system):
# compute correction factor for g1 that will produce common gain at f0
g1_swp = system.g1 * np.sin(np.pi/2-system.phase_swp) / system.alpha_swp
return g1_swp
# Operating Enviornment
#####
f0 = 28
bw0 = 8 # assumed tuning range (GHz)
bw_plt = 4 # Plotting range (GHz)
fbw = bw0/f0 # fractional bandwidth
frequency_sweep_steps = 101
gamma_sweep_steps = 8
gamma = 1 - np.power(f0 / (f0 + bw0/2),2)
gamma_limit_ratio = 0.99 # how close gamma can get to theoretical extreme
phase_limit_requested = (1-1/gamma_sweep_steps)*np.pi/2
class ampSystem:
"""define global (hardware descriptive) variables for use in our system."""
def __init__(self, quiet=False):
self.f0 = 28 # design frequency (GHz)
self.bw0 = 8 # assumed extreme tuning range (GHz)
self.bw_plt = 4 # Plotting range (GHz)
# Configuration Of Hardware
#####
q1_L = 20
q1_C = 7
l1 = 180e-3 # nH
gm1 = 25e-3 # S
self.q1_L = 25
self.q1_C = 8
self.l1 = 140e-3 # nH
self.gm1 = 25e-3 # S
# Compute frequency sweep
#####
w0 = f0*2*np.pi
fbw_plt = bw_plt/f0
delta = np.linspace(-fbw_plt/2,fbw_plt/2,frequency_sweep_steps)
w = w0*(1+delta)
f = f0*(1+delta)
jw = 1j*w
self._gamma_steps=8
self._gamma_cap_ratio = 0.997
self.alpha_min=1
if not quiet:
## Report System Descrption
print(' L1 = %.3fpH, C1 = %.3ffF' % (1e3*self.l1, 1e6*self.c1))
print(' Rp = %.3f Ohm' % (1/self.g1))
print(' Q = %.1f' % (self.Q1))
self._gamma_warn = False
self._g1_map_function = g1_map_default
@property
def w0(self):
return self.f0*2*np.pi
@property
def fbw(self): # fractional bandwidth
return self.bw0/self.f0
@property
def phase_max(self):
return np.pi/2 * (1 - 1/self.gamma_len)
##################
# Compute system
#####
c1 = 1/(w0*w0*l1)
g1_L = 1 / (q1_L*w0*l1)
g1_C = w0 * c1 / q1_C
g1 = g1_L + g1_C
@property
def c1(self):
return 1/(self.w0*self.w0*self.l1)
@property
def g1(self):
g1_L = 1 / (self.q1_L*self.w0*self.l1)
g1_C = self.w0 * self.c1 / self.q1_C
return g1_L + g1_C
@property
def Q1(self):
return np.sqrt(self.c1/self.l1)/self.g1
@property
def gamma_len(self):
return self._gamma_steps
@property
def gamma(self):
gamma = 1 - np.power(self.f0 / (self.f0 + self.bw0/2),2)
phase_limit_requested = (1-1/self.gamma_len)*np.pi/2
# Verify gamma is valid
#####
gamma_max = g1 * np.sqrt(l1/c1)
if gamma > (gamma_limit_ratio * gamma_max):
print("==> WARN: Gamma to large, reset to %.3f (was %.3f) <==" % \
(gamma_limit_ratio * gamma_max, gamma))
gamma = gamma_limit_ratio * gamma_max
gamma_max = 1/(self.alpha_min*self.Q1)
if gamma > (self._gamma_cap_ratio * gamma_max):
if not self._gamma_warn:
self._gamma_warn = True
print("==> WARN: Gamma to large, reset to %.1f%% (was %.1f%%) <==" % \
(100*self._gamma_cap_ratio * gamma_max, 100*gamma))
gamma = self._gamma_cap_ratio * gamma_max
return gamma
@property
def alpha_swp(self):
range_partial = np.ceil(self.gamma_len/2)
lhs = np.linspace(np.sqrt(self.alpha_min),1, range_partial)
rhs = np.flip(lhs,0)
swp = np.concatenate((lhs,rhs[1:])) if np.mod(self.gamma_len,2) == 1 \
else np.concatenate((lhs,rhs))
return np.power(swp,2)
@property
def gamma_swp(self):
return np.cos(np.pi/2-self.phase_swp) / self.Q1 / self.alpha_swp
@property
def phase_swp(self):
#def phaseSweepGenerate(g1, gamma, c, l, phase_extreme, phase_steps):
# Linear PHASE gamma spacing
# First compute the most extreme phase given the extreme gamma
# if gamma is tuned to the limit, and we want to match the gain performance,
# then this is the required tuned g1 value.
gamma = self.gamma
g1_limit = np.sqrt(np.power(self.g1,2) - np.power(gamma,2)*self.c1/self.l1)
# This implies a Q in that particular setting
Q_limit = self.Q1*self.g1/g1_limit
# given this !, I compute the delta phase at that point.
phase_limit = np.pi/2 - np.arctan(1/(Q_limit*gamma))
phase_swp = np.linspace(-1,1,self.gamma_len) * self.phase_max
if phase_limit < self.phase_max:
print( "==> ERROR: Phase Beyond bounds. Some states will be ignored")
print( " %.3f requested\n"
" %.3f hardware limit" % \
(180/np.pi*self.phase_max, 180/np.pi*abs(phase_limit)))
print( " To increase tuning range, gamma must rise or native Q must rise")
phase_swp = np.where(phase_swp > phase_limit, phase_swp, np.NaN)
# This gives us our equal phase spacing points
return phase_swp
@property
def c1_swp(self):
return self.c1 * (1 + self.gamma_swp)
def set_g1_swp(self, g1_swp_function):
self._g1_map_function = g1_swp_function
@property
def g1_swp(self):
return self._g1_map_function(self)
def compute_block(self, f_dat):
g1_swp = self.g1_swp
c1_swp = self.c1_swp
y_tank = np.zeros((self.gamma_len,f_dat.steps), dtype=complex)
tf = np.zeros((self.gamma_len,f_dat.steps), dtype=complex)
for itune,gamma_tune in enumerate(self.gamma_swp):
c1_tune = c1_swp[itune]
g1_tune = g1_swp[itune]
y_tank[itune,:] = g1_tune + f_dat.jw*c1_tune + 1/(f_dat.jw * self.l1)
tf[itune,:] = self.__class__.tf_compute(f_dat.delta, gamma_tune, g1_tune, self.gm1, self.l1, self.c1)
tf = tf.T
return (y_tank, tf)
def compute_ref(self, f_dat):
y_tank = self.g1 + f_dat.jw*self.c1 + 1/(f_dat.jw * self.l1)
tf = self.__class__.tf_compute(f_dat.delta, 0, self.g1, self.gm1, self.l1, self.c1)
return (y_tank, tf)
@classmethod
def tf_compute(cls, delta, gamma, gx, gm, l, c):
Q = np.sqrt(c/l)/gx
return gm / gx \
* 1j*(1+delta) \
/ (1j*(1+delta) + Q*(1-np.power(1+delta,2)*(1+gamma)))

52
tankComputers.py Normal file
View file

@ -0,0 +1,52 @@
#!/usr/bin/env python3
import numpy as np
################################################################################
# Define my helper functions.
def dB20(volt_tf):
"""Describe signal gain of a transfer function in dB (i.e. 20log(x))"""
return 20*np.log10(np.abs(volt_tf))
def ang(volt_tf):
"""Describe phase of a transfer function in degrees. Not unwrapped."""
return 180/np.pi*np.angle(volt_tf)
def ang_unwrap(volt_tf):
"""Describe phase of a transfer function in degrees. With unwrapping."""
return 180/np.pi*np.unwrap(np.angle(volt_tf))
def dB10(pwr_tf):
"""Describe power gain of a transfer function in dB (i.e. 10log(x))"""
return 10*np.log10(np.abs(pwr_tf))
def dB2Vlt(dB20_value):
return np.power(10,dB20_value/20)
def wrap_rads(angles):
return np.mod(angles+np.pi,2*np.pi)-np.pi
def atand(x):
return 180/np.pi*np.arctan(x)
def rms_v_bw(err_sig, bandwidth_scale=1):
"""compute the rms vs bandwidth assuming a fixed center frequency"""
# First compute the error power
err_pwr = np.power(np.abs(err_sig),2)
steps = len(err_pwr)
isodd = True if steps%2 != 0 else False
# We want to generate the midpoint to the left, and midpoint to the right
# as two distinct sets.
pt_rhs_start = int(np.floor(steps/2))
pt_lhs_stop = int(np.ceil(steps/2))
folded = err_pwr[pt_rhs_start:] + np.flip(err_pwr[:pt_lhs_stop],0)
# Now, we MIGHT have double counted the mid point
# if the length is odd, correct for that
if isodd: folded[0]*=0.5
# Now we need an array that describes the number of points used to get here.
# this one turns out to be pretty easy.
frac_step = np.arange(int(not isodd),steps,2)/(steps-1)
ind = 2*np.arange(0,frac_step.shape[0])+1+int(not isodd)
# Now actually compute the RMS values. First do the running sum
rms = np.sqrt(np.cumsum(folded,0) / (ind*np.ones((folded.shape[1],1))).T )
return (frac_step*bandwidth_scale, rms)

144
tankPlot.py
View file

@ -10,25 +10,6 @@ sys.path.append("./pySmithPlot")
import smithplot
from smithplot import SmithAxes
################################################################################
# Define my helper functions.
def dB20(volt_tf):
"""Describe signal gain of a transfer function in dB (i.e. 20log(x))"""
return 20*np.log10(np.abs(volt_tf))
def ang(volt_tf):
"""Describe phase of a transfer function in degrees. Not unwrapped."""
return 180/np.pi*np.angle(volt_tf)
def ang_unwrap(volt_tf):
"""Describe phase of a transfer function in degrees. With unwrapping."""
return 180/np.pi*np.unwrap(np.angle(volt_tf))
def dB10(pwr_tf):
"""Describe power gain of a transfer function in dB (i.e. 10log(x))"""
return 10*np.log10(np.abs(pwr_tf))
def atan(x):
return 180/np.pi*np.arctan(x)
################################################################################
# Override the defaults for this script
rcParams['figure.figsize'] = [10,7]
@ -36,76 +17,57 @@ default_window_position=['+20+80', '+120+80']
################################################################################
# Operating Enviornment (i.e. circuit parameters)
from TankGlobals import *
import TankGlobals
from FreqClass import FreqClass
from tankComputers import *
S=TankGlobals.ampSystem()
f=FreqClass(501, S.f0, S.bw_plt)
################################################################################
# Now generate the sweep of resonance tuning (gamma, and capacitance)
# Linear based gamma spacing
#gamma_swp = np.linspace(-gamma,gamma,gamma_sweep_steps)
# Linear PHASE gamma spacing
# First compute the most extreme phase given the extreme gamma
g1_limit = np.sqrt( g1*g1 - (gamma*gamma) * c1/l1 )
K_limit = np.sqrt(c1/l1)*1/g1_limit
phase_limit = np.mod(np.pi/2 - np.arctan( -1/K_limit * 1/gamma ),np.pi) - np.pi
if abs(phase_limit) < phase_limit_requested:
print("==> WARN: Phase Beyond bounds, leaving at limits. <==")
print("==> %.3f requested, but hardware limit is %.3f <==" % \
(180/np.pi*phase_limit_requested, 180/np.pi*abs(phase_limit)))
sys.exit(-1)
else:
phase_limit = phase_limit_requested
# This gives us our equal phase spacing points
phase_swp = np.linspace(-1,1,gamma_sweep_steps) * phase_limit
# Then use this to compute the gamma steps to produce arbitrary phase given
# our perfect gain constraint.
gamma_swp = np.sign(phase_swp)/np.sqrt(np.power(np.tan(np.pi/2 - phase_swp),2)+1) * g1 / np.sqrt(c1/l1)
# We want a smooth transition out to alpha. So For now assume a squares
# weighting out to the maximum alpha at the edges.
gain_variation = -8*0 # dB
S.alpha_min = dB2Vlt(gain_variation)
# compute correction factor for g1 that will produce common gain at f0
g1_swp = np.sqrt( g1*g1 - (gamma_swp*gamma_swp) * c1/l1 )
# this is defined as the class default
g1_swp = S.g1_swp
# and compute how much of a negative gm this requres, and it's relative
# proportion to the gm of the assumed main amplifier gm.
g1_boost = (g1_swp - g1)
g1_ratio = -g1_boost / gm1
g1_boost = (g1_swp - S.g1)
g1_ratio = -g1_boost / S.gm1
c1_swp = c1 * (1 + gamma_swp)
## Report System Descrption
print(' L1 = %.3fpH, C1 = %.3ffF' % (1e3*l1, 1e6*c1))
print(' Rp = %.3f Ohm' % (1/g1))
print(' Max G1 boost %.2fmS (%.1f%% of gm1)' % \
(1e3*np.max(np.abs(g1_boost)), 100*np.max(g1_ratio)))
y_tank = np.zeros((len(gamma_swp),len(f)), dtype=complex)
tf = np.zeros((len(gamma_swp),len(f)), dtype=complex)
for itune,gamma_tune in enumerate(gamma_swp):
c1_tune = c1_swp[itune]
g1_tune = g1_swp[itune]
K = np.sqrt(c1/l1)/g1_tune
y_tank_tmp = g1_tune + jw*c1_tune + 1/(jw * l1)
y_tank[itune,:] = y_tank_tmp
tf_tmp = gm1 / g1_tune * \
1j*(1+delta) / \
( 1j*(1+delta) + K*(1 - (1+gamma_tune)*np.power(1+delta,2)) )
tf[itune,:] = tf_tmp
tf = tf.T
################################################################################
# Generate a reference implementation
(y_tank, tf) = S.compute_block(f)
(_, tf_ref) = S.compute_ref(f)
# double to describe with perfect inversion stage
tf = np.column_stack((tf,-tf))
ref_index = int(gamma_swp.shape[0]/2)
tf_r = tf / (tf[:,ref_index]*np.ones((tf.shape[1],1))).T
y_tank = y_tank.T
# compute the relative transfer function thus giving us flat phase, and
# flat (ideally) gain response if our system perfectly matches the reference
tf_r = tf / (tf_ref*np.ones((tf.shape[1],1))).T
# We will also do a direct angle comparison
tf_r_ang_ideal = wrap_rads(np.concatenate((-S.phase_swp, -np.pi - S.phase_swp)))
tf_r_ang = np.angle(tf_r)
tf_r_ang_rms = np.sqrt(np.mean(np.power(tf_r_ang-tf_r_ang_ideal,2),0))
y_tank = y_tank.T
################################################################################
# Compute RMS phase error relative to ideal reference across plotting bandwidth
(bw_ang, rms_ang_swp)=rms_v_bw(tf_r_ang-tf_r_ang_ideal, S.bw_plt)
(bw_mag, rms_gain_swp)=rms_v_bw(tf_r, S.bw_plt)
print(ang(tf[f==28,:]))
################################################################################
h1 = pp.figure()
h2 = pp.figure(figsize=(5,7))
h3 = pp.figure(figsize=(5,7))
mgr = pp.get_current_fig_manager()
################################################################################
ax1 = h1.add_subplot(2,2,1, projection='smith')
@ -115,14 +77,19 @@ ax4 = h1.add_subplot(2,2,4)
ax1.plot(y_tank, datatype=SmithAxes.Y_PARAMETER, marker="None")
ax2.plot(np.angle(tf), dB20(tf))
ax3.plot(f,dB20(tf))
ax4.plot(f,ang_unwrap(tf))
ax3.plot(f.hz,dB20(tf))
ax4.plot(f.hz,ang_unwrap(tf))
################################################################################
ax8 = h2.add_subplot(2,1,1)
ax9 = h2.add_subplot(2,1,2)
ax8.plot(f,dB20(tf_r))
ax9.plot(f,ang_unwrap(tf_r.T).T)
ax6 = h2.add_subplot(2,1,1)
ax7 = h2.add_subplot(2,1,2)
ax6.plot(f.hz,dB20(tf_r))
ax7.plot(f.hz,ang_unwrap(tf_r.T).T)
ax8 = h3.add_subplot(2,1,1)
ax9 = h3.add_subplot(2,1,2)
ax8.plot(bw_mag,dB20(rms_gain_swp))
ax9.plot(bw_ang,rms_ang_swp*180/np.pi)
ax1.set_title('Tank Impedance')
ax2.set_title('Transfer Function')
@ -131,20 +98,31 @@ ax3.set_title('TF Gain')
ax3.set_ylabel('Gain (dB)')
ax4.set_title('TF Phase')
ax4.set_ylabel('Phase (deg)')
ax8.set_title('TF Relative Gain')
ax8.set_ylabel('Relative Gain (dB)')
ax9.set_title('TF Relative Phase')
ax9.set_ylabel('Relative Phase (deg)')
for ax_T in [ax3, ax4, ax8, ax9]:
ax6.set_title('TF Relative Gain')
ax6.set_ylabel('Relative Gain (dB)')
ax7.set_title('TF Relative Phase')
ax7.set_ylabel('Relative Phase (deg)')
for ax_T in [ax3, ax4, ax6, ax7]:
ax_T.grid()
ax_T.set_xlabel('Freq (GHz)')
ax_T.set_xlim(( np.min(f), np.max(f) ))
ax_T.set_xlim(f.hz_range)
ax8.set_title('RMS Gain Error')
ax8.set_ylabel('RMS Gain Error (dB)')
ax9.set_title('RMS Phase Error')
ax9.set_ylabel('RMS Phase Error (deg)')
for ax_T in [ax8, ax9]:
ax_T.grid()
ax_T.set_xlim((0,S.bw_plt))
ax_T.set_xlabel('Bandwidth (GHz)')
################################################################################
h1.tight_layout()
h2.tight_layout()
h3.tight_layout()
mgr.window.geometry(default_window_position[0])
h1.show()
mgr.window.geometry(default_window_position[1])
h2.show()
h3.show()