def test_down_conversion(args):
"""
@brief test to see if input signal of known frequency has been down converted
after mixing with known frequency
"""
output_steps = 2**args.freq_bits
# define the timestep for the 2 time series
dt_ddc = output_steps / args.clk
dt_data = 1. / args.clk
# read in the data from the ddc_bram and data_bram files
t_ddc, d_ddc = digital_utils.process_ddc_bram(args.ddc_file,
timestep=dt_ddc)
t_data, d_data = digital_utils.process_data_bram(args.data_file,
timestep=dt_data)
# get the power spectra of the signals
f_ddc, p_ddc = digital_utils.power(d_ddc, dt_ddc)
f_data, p_data = digital_utils.power(d_data, dt_data)
# plot the input signal on first subplot
pl.subplots_adjust(hspace=0.3)
pl.subplot(211)
pl.plot(f_data / 1e6, p_data)
# add the necessary info to the subplot
pl.figtext(0.2, 0.85, 'input tone at %.3f MHz' % (args.input_freq / 1e6))
pl.xlabel('frequency (MHz)', fontsize=16)
pl.ylabel('power', fontsize=16)
pl.xlim(-1.5 * args.input_freq / 1e6, 1.5 * args.input_freq / 1e6)
pl.subplot(212)
# determine the expected down-converted frequency, as mixer frequency - input freq
f_expected = abs(args.lof_int * args.clk / output_steps - args.input_freq)
# plot the down-converted spectra
pl.plot(f_ddc / 1e3, p_ddc)
pl.figtext(0.2, 0.4, 'mixed tone at %.3f kHz' % (f_expected / 1e3))
pl.xlabel('frequency (kHZ)', fontsize=16)
pl.ylabel('power', fontsize=16)
pl.xlim(-1.5 * f_expected / 1e3, 1.5 * f_expected / 1e3)
pl.show()
return
def test_down_conversion(args):
"""
@brief test to see if input signal of known frequency has been down converted
after mixing with known frequency
"""
output_steps = 2**args.freq_bits
# define the timestep for the 2 time series
dt_ddc = output_steps/args.clk
dt_data = 1./args.clk
# read in the data from the ddc_bram and data_bram files
t_ddc, d_ddc = digital_utils.process_ddc_bram(args.ddc_file, timestep=dt_ddc)
t_data, d_data = digital_utils.process_data_bram(args.data_file, timestep=dt_data)
# get the power spectra of the signals
f_ddc, p_ddc = digital_utils.power(d_ddc, dt_ddc)
f_data, p_data = digital_utils.power(d_data, dt_data)
# plot the input signal on first subplot
pl.subplots_adjust(hspace=0.3)
pl.subplot(211)
pl.plot(f_data/1e6, p_data)
# add the necessary info to the subplot
pl.figtext(0.2, 0.85, 'input tone at %.3f MHz' %(args.input_freq/1e6))
pl.xlabel('frequency (MHz)', fontsize=16)
pl.ylabel('power', fontsize=16)
pl.xlim(-1.5*args.input_freq/1e6, 1.5*args.input_freq/1e6)
pl.subplot(212)
# determine the expected down-converted frequency, as mixer frequency - input freq
f_expected = abs(args.lof_int * args.clk / output_steps - args.input_freq)
# plot the down-converted spectra
pl.plot(f_ddc/1e3, p_ddc)
pl.figtext(0.2, 0.4, 'mixed tone at %.3f kHz' %(f_expected/1e3))
pl.xlabel('frequency (kHZ)', fontsize=16)
pl.ylabel('power', fontsize=16)
pl.xlim(-1.5*f_expected/1e3, 1.5*f_expected/1e3)
pl.show()
return
def __plot_data(self, time_ax, fft_ax, data):
"""
@brief plot the timeseries and power spectrum of data
@param time_ax: the Axes object corresponding to timseries subplot
@param fft_ax: the Axes object corresponding to power subplot
@param data: the data to plot
"""
# clear the axes to remove old data
fft_ax.cla()
time_ax.cla()
# get the time array (in microseconds)
ts = np.arange(0, self.dt*self.max_to_plot, self.dt)*1e6
# plot the real and if data is DDC, the imaginary parts of timeseries
time_ax.plot(ts, data.real, c='b')
if self.dtype == 'ddc':
time_ax.plot(ts, data.imag, c='r' )
# compute the power spectrum of timeseries and plot
freqs, power = digital_utils.power(data, timestep=self.dt)
fft_ax.plot( freqs, power, c='k')
# label the axes and set limits
fft_ax.set_xlabel('frequency (Hz)', fontsize=16)
fft_ax.set_ylabel('power', fontsize=16)
time_ax.set_xlabel(r'time ($\mu$s)', fontsize=16)
time_ax.set_ylabel('signal amplitude', fontsize=16)
time_ax.set_xlim(0, self.dt*self.show*1e6)
# draw and then we are finished
pl.draw()
return
def __plot_data(self, time_ax, fft_ax, data):
"""
@brief plot the timeseries and power spectrum of data
@param time_ax: the Axes object corresponding to timseries subplot
@param fft_ax: the Axes object corresponding to power subplot
@param data: the data to plot
"""
# clear the axes to remove old data
fft_ax.cla()
time_ax.cla()
# get the time array (in microseconds)
ts = np.arange(0, self.dt * self.max_to_plot, self.dt) * 1e6
# plot the real and if data is DDC, the imaginary parts of timeseries
time_ax.plot(ts, data.real, c='b')
if self.dtype == 'ddc':
time_ax.plot(ts, data.imag, c='r')
# compute the power spectrum of timeseries and plot
freqs, power = digital_utils.power(data, timestep=self.dt)
fft_ax.plot(freqs, power, c='k')
# label the axes and set limits
fft_ax.set_xlabel('frequency (Hz)', fontsize=16)
fft_ax.set_ylabel('power', fontsize=16)
time_ax.set_xlabel(r'time ($\mu$s)', fontsize=16)
time_ax.set_ylabel('signal amplitude', fontsize=16)
time_ax.set_xlim(0, self.dt * self.show * 1e6)
# draw and then we are finished
pl.draw()
return
parser.add_argument('signal_file', type=str, help='name of file containing true input signal')
parser.add_argument('convolved_file', type=str, help='name of file containing convolved signal')
parser.add_argument('--dt', type=float, default=5e-9, help='timestep of data file')
parser.add_argument('--keepDC', action='store_false', help='do not remove DC bias from input signal')
parser.add_argument('--smoothing', type=int, default=0, help='kernel of gaussian smoothing function to apply to power spectra')
parser.add_argument('--fitGaussian', action='store_true', help='whether to fit a gaussian to filter shape')
parser.add_argument('--fitSinc', action='store_true', help='whether to fit a sinc to filter shape')
args = parser.parse_args()
# proces the data from the convolved and true input signal files
t_conv, d_conv = digital_utils.process_data_bram(args.convolved_file, timestep=args.dt)
t_true, d_true = digital_utils.process_data_bram(args.signal_file, timestep=args.dt)
# get the power spectra
f_conv, p_conv = digital_utils.power(d_conv, args.dt, smoothing=args.smoothing, keepDC=args.keepDC)
f_true, p_true = digital_utils.power(d_true, args.dt, smoothing=args.smoothing, keepDC=args.keepDC)
# restrict the filter to positive frequencies
filt = p_conv/p_true
inds = np.where(f_true > 0)[0]
filt = filt[inds]
f_true = f_true[inds]/1e6 # now in MHz
# plot the filter
pl.plot(f_true, filt, c='k', label='recovered filter')
# fit the filter to a gaussian and sinc and plot
if args.fitGaussian:
t, f_fitted = fitFilter(f_true, filt, model='gaussian', bandpass=np.amax(f_true))
pl.plot(t, f_fitted, c='b', label='best fit gaussian')
help='whether to fit a gaussian to filter shape')
parser.add_argument('--fitSinc',
action='store_true',
help='whether to fit a sinc to filter shape')
args = parser.parse_args()
# proces the data from the convolved and true input signal files
t_conv, d_conv = digital_utils.process_data_bram(args.convolved_file,
timestep=args.dt)
t_true, d_true = digital_utils.process_data_bram(args.signal_file,
timestep=args.dt)
# get the power spectra
f_conv, p_conv = digital_utils.power(d_conv,
args.dt,
smoothing=args.smoothing,
keepDC=args.keepDC)
f_true, p_true = digital_utils.power(d_true,
args.dt,
smoothing=args.smoothing,
keepDC=args.keepDC)
# restrict the filter to positive frequencies
filt = p_conv / p_true
inds = np.where(f_true > 0)[0]
filt = filt[inds]
f_true = f_true[inds] / 1e6 # now in MHz
# plot the filter
pl.plot(f_true, filt, c='k', label='recovered filter')
