def compute_noise(minfreq, maxfreq, data_parent_folder, meas_folder, en_fig):
# variables to be input
# data_parent_folder : the folder for all datas
# meas_folder : the specific folder for one measurement
# en_fig : enable figure
file_name_prefix = 'dat_'
data_folder = (data_parent_folder + '/' + meas_folder + '/')
fig_num = 200
# variables from NMR settings
(param_list, value_list) = data_parser.parse_info(data_folder,
'acqu.par') # read file
adcFreq = int(data_parser.find_value('adcFreq', param_list, value_list))
nrPnts = int(data_parser.find_value('nrPnts', param_list, value_list))
total_scan = int(
data_parser.find_value('nrIterations', param_list, value_list))
# parse file and remove DC component
# data = np.zeros(nrPnts)
for m in range(1, total_scan + 1):
file_path = (data_folder + file_name_prefix + '{0:03d}'.format(m))
# read the data from the file and store it in numpy array format
one_scan = np.array(data_parser.read_data(file_path))
one_scan = (one_scan - np.mean(one_scan)) / \
total_scan # remove DC component
# data = data + one_scan
spectx, specty = nmr_fft(one_scan, adcFreq, 0)
fft_range = [
i for i, value in enumerate(spectx)
if (value >= minfreq and value <= maxfreq)
] # limit fft display
# standard deviation of the fft
print('\t\tNOISE RMS = ' + '{0:.5f}'.format(np.std(specty[fft_range])))
if en_fig:
plt.ion()
fig = plt.figure(fig_num)
fig.clf()
ax = fig.add_subplot(2, 1, 1)
line1, = ax.plot(spectx[fft_range], specty[fft_range], 'r-')
# ax.set_ylim(-50, 0)
ax.set_xlabel('Frequency [MHz]')
ax.set_ylabel('Amplitude [a.u.]')
ax.set_title("Noise spectrum")
ax.grid()
ax = fig.add_subplot(2, 1, 2)
line1, = ax.plot(one_scan, 'r-')
ax.set_xlabel('Time')
ax.set_ylabel('Amplitude [a.u.]')
ax.set_title("Noise amplitude")
ax.grid()
fig.canvas.draw()
fig.canvas.flush_events()
def compute_iterate(nmrObj, data_parent_folder, meas_folder, en_ext_param,
thetaref, echoref_avg, direct_read, datain, en_fig):
data_folder = (data_parent_folder + '/' + meas_folder + '/')
# variables from NMR settings
(param_list, value_list) = data_parser.parse_info(data_folder,
'acqu.par') # read file
SpE = int(data_parser.find_value('nrPnts', param_list, value_list))
NoE = int(data_parser.find_value('nrEchoes', param_list, value_list))
en_ph_cycle_proc = data_parser.find_value('usePhaseCycle', param_list,
value_list)
tE = data_parser.find_value('echoTimeRun', param_list, value_list)
Sf = data_parser.find_value('adcFreq', param_list, value_list) * 1e6
Df = data_parser.find_value('b1Freq', param_list, value_list) * 1e6
total_scan = int(
data_parser.find_value('nrIterations', param_list, value_list))
file_name_prefix = 'dat_'
# en_ext_param = 0
# thetaref = 0
# echoref_avg = 0
if (direct_read):
(a, a_integ, a0, snr, T2, noise, res, theta, data_filt, echo_avg,
t_echospace) = compute_multiple(nmrObj, data_parent_folder,
meas_folder, file_name_prefix, Df, Sf,
tE, total_scan, en_fig, en_ext_param,
thetaref, echoref_avg, direct_read,
datain)
else:
(a, a_integ, a0, snr, T2, noise, res, theta, data_filt, echo_avg,
t_echospace) = compute_multiple(nmrObj, data_parent_folder,
meas_folder, file_name_prefix, Df, Sf,
tE, total_scan, en_fig, en_ext_param,
thetaref, echoref_avg, 0, datain)
# print(snr, T2)
return a, a_integ, a0, snr, T2, noise, res, theta, data_filt, echo_avg, Df, t_echospace
def compute_stats(minfreq, maxfreq, data_parent_folder, meas_folder, plotname,
en_fig):
# variables to be input
# data_parent_folder : the folder for all datas
# meas_folder : the specific folder for one measurement
# en_fig : enable figure
# compute settings
process_sum_data = 1 # otherwise process raw data
file_name_prefix = 'dat_'
data_folder = (data_parent_folder + '/' + meas_folder + '/')
fig_num = 200
# variables from NMR settings
(param_list, value_list) = data_parser.parse_info(data_folder,
'acqu.par') # read file
adcFreq = data_parser.find_value('adcFreq', param_list, value_list)
nrPnts = int(data_parser.find_value('nrPnts', param_list, value_list))
total_scan = int(
data_parser.find_value('nrIterations', param_list, value_list))
# parse file and remove DC component
nmean = 0
if process_sum_data:
file_path = (data_folder + 'asum')
one_scan_raw = np.array(data_parser.read_data(file_path))
nmean = np.mean(one_scan_raw)
one_scan = (one_scan_raw - nmean) / total_scan
else:
for m in range(1, total_scan + 1):
file_path = (data_folder + file_name_prefix + '{0:03d}'.format(m))
# read the data from the file and store it in numpy array format
one_scan_raw = np.array(data_parser.read_data(file_path))
nmean = np.mean(one_scan_raw)
one_scan = (one_scan_raw - nmean) / \
total_scan # remove DC component
# compute fft
spectx, specty = nmr_fft(one_scan, adcFreq, 0)
specty = abs(specty)
fft_range = [
i for i, value in enumerate(spectx)
if (value >= minfreq and value <= maxfreq)
] # limit fft display
# compute std
nstd = np.std(one_scan)
if en_fig:
plt.ion()
fig = plt.figure(fig_num)
# maximize window
plot_backend = matplotlib.get_backend()
mng = plt.get_current_fig_manager()
if plot_backend == 'TkAgg':
# mng.resize(*mng.window.maxsize())
mng.resize(800, 600)
elif plot_backend == 'wxAgg':
mng.frame.Maximize(True)
elif plot_backend == 'Qt4Agg':
mng.window.showMaximized()
fig.clf()
ax = fig.add_subplot(311)
line1, = ax.plot(spectx[fft_range], specty[fft_range], 'b-')
# ax.set_ylim(-50, 0)
ax.set_xlabel('Frequency (MHz)')
ax.set_ylabel('Amplitude (a.u.)')
ax.set_title("Spectrum")
ax.grid()
ax = fig.add_subplot(312)
x_time = np.linspace(1, len(one_scan_raw), len(one_scan_raw))
x_time = np.multiply(x_time, (1 / adcFreq)) # in us
x_time = np.multiply(x_time, 1e-3) # in ms
line1, = ax.plot(x_time, one_scan_raw, 'b-')
ax.set_xlabel('Time(ms)')
ax.set_ylabel('Amplitude (a.u.)')
ax.set_title("Amplitude. std=%0.2f. mean=%0.2f." % (nstd, nmean))
ax.grid()
# plot histogram
n_bins = 200
ax = fig.add_subplot(313)
n, bins, patches = ax.hist(one_scan, bins=n_bins)
ax.set_title("Histogram")
plt.tight_layout()
fig.canvas.draw()
fig.canvas.flush_events()
# fig = plt.gcf() # obtain handle
plt.savefig(data_folder + plotname)
# standard deviation of signal
print('\t\t: rms= ' + '{0:.4f}'.format(nstd) +
' mean= {0:.4f}'.format(nmean))
return nstd, nmean
def compute_multiple(nmrObj, data_parent_folder, meas_folder, file_name_prefix,
Df, Sf, tE, total_scan, en_fig, en_ext_param, thetaref,
echoref_avg, direct_read, datain):
# variables to be input
# nmrObj : the hardware definition
# data_parent_folder : the folder for all datas
# meas_folder : the specific folder for one measurement
# filename_prefix : the name of data prefix
# Df : data frequency
# Sf : sample frequency
# tE : echo spacing
# total_scan : number_of_iteration
# en_fig : enable figure
# en_ext_param : enable external parameter for data signal processing
# thetaref : external parameter : rotation angle
# echoref_avg : external parameter : echo average reference
# datain : the data captured direct reading. data format: AC,averaged scans, phase-cycled
# direct_read : perform direct reading from SDRAM/FIFO
data_folder = (data_parent_folder + '/' + meas_folder + '/')
# variables local to this function
# the setting file for the measurement
mtch_fltr_sta_idx = 0 # 0 is default or something referenced to SpE, e.g. SpE/4; the start index for match filtering is to neglect ringdown part from calculation
# perform rotation to the data -> required for reference data for t1
# measurement
perform_rotation = 1
# process individual raw data, otherwise it'll load a sum file generated
# by C
proc_indv_data = 0
# put 1 if the data file uses binary representation, otherwise it is in
# ascii format
binary_OR_ascii = 1
ignore_echoes = 0 # ignore initial echoes #
# simulate decimation in software (DO NOT use this for normal operation,
# only for debugging purpose)
sim_dec = 0
sim_dec_fact = 32
# variables from NMR settings
(param_list, value_list) = data_parser.parse_info(data_folder,
'acqu.par') # read file
SpE = int(data_parser.find_value('nrPnts', param_list, value_list))
NoE = int(data_parser.find_value('nrEchoes', param_list, value_list))
en_ph_cycle_proc = data_parser.find_value('usePhaseCycle', param_list,
value_list)
# tE = data_parser.find_value('echoTimeRun', param_list, value_list)
# Sf = data_parser.find_value(
# 'adcFreq', param_list, value_list) * 1e6
# Df = data_parser.find_value(
# 'b1Freq', param_list, value_list) * 1e6
# total_scan = int(data_parser.find_value(
# 'nrIterations', param_list, value_list))
fpga_dconv = data_parser.find_value('fpgaDconv', param_list, value_list)
dconv_fact = data_parser.find_value('dconvFact', param_list, value_list)
echo_skip = data_parser.find_value('echoSkipHw', param_list, value_list)
# account for skipped echoes
NoE = int(NoE / echo_skip)
tE = tE * echo_skip
# compensate for dconv_fact if fpga dconv is used
if fpga_dconv:
SpE = int(SpE / dconv_fact)
Sf = Sf / dconv_fact
# ignore echoes
if ignore_echoes:
NoE = NoE - ignore_echoes
# time domain for plot
tacq = (1 / Sf) * 1e6 * np.linspace(1, SpE, SpE) # in uS
t_echospace = tE / 1e6 * np.linspace(1, NoE, NoE) # in uS
if fpga_dconv: # use fpga dconv
# load IQ data
file_path = (data_folder + 'dconv')
if binary_OR_ascii:
dconv = data_parser.read_hex_float(
file_path) # use binary representation
else:
dconv = np.array(
data_parser.read_data(file_path)) # use ascii representation
if ignore_echoes:
dconv = dconv[ignore_echoes * 2 * SpE:len(dconv)]
# normalize the data
# normalize with decimation factor (due to sum in the fpga)
dconv = dconv / dconv_fact
dconv = dconv / nmrObj.fir_gain # normalize with the FIR gain in the fpga
# convert to voltage unit at the probe
dconv = dconv / nmrObj.totGain * nmrObj.uvoltPerDigit
# scale all the data to magnitude of sin(45). The FPGA uses unity
# magnitude (instead of sin45,135,225,315) to simplify downconversion
# operation
dconv = dconv * nmrObj.dconv_gain
# combined IQ
data_filt = np.zeros((NoE, SpE), dtype=complex)
for i in range(0, NoE):
data_filt[i, :] = \
dconv[i * (2 * SpE):(i + 1) * (2 * SpE):2] + 1j * \
dconv[i * (2 * SpE) + 1:(i + 1) * (2 * SpE):2]
if en_fig: # plot the averaged scan
echo_space = (1 / Sf) * np.linspace(1, SpE, SpE) # in s
plt.figure(1)
for i in range(0, NoE):
plt.plot(((i - 1) * tE * 1e-6 + echo_space) * 1e3,
np.real(data_filt[i, :]),
linewidth=0.4,
color='b')
plt.plot(((i - 1) * tE * 1e-6 + echo_space) * 1e3,
np.imag(data_filt[i, :]),
linewidth=0.4,
color='r')
plt.title("Averaged raw data (downconverted)")
plt.xlabel('time(ms)')
plt.ylabel('probe voltage (uV)')
plt.savefig(data_folder + 'fig_avg_raw_data.png')
# raw average data
echo_rawavg = np.mean(data_filt, axis=0)
if (en_fig):
if en_fig: # plot echo rawavg
plt.figure(6)
plt.plot(tacq, np.real(echo_rawavg), label='real')
plt.plot(tacq, np.imag(echo_rawavg), label='imag')
plt.plot(tacq, np.abs(echo_rawavg), label='abs')
plt.xlim(0, max(tacq))
plt.title("Echo Average before rotation (down-converted)")
plt.xlabel('time(uS)')
plt.ylabel('probe voltage (uV)')
plt.legend()
plt.savefig(data_folder + 'fig_echo_avg_dconv.png')
# simulate additional decimation (not needed for normal operqtion). For
# debugging purpose
if (sim_dec):
SpE = int(SpE / sim_dec_fact)
Sf = Sf / sim_dec_fact
data_filt_dec = np.zeros((NoE, SpE), dtype=complex)
for i in range(0, SpE):
data_filt_dec[:, i] = np.mean(
data_filt[:, i * sim_dec_fact:(i + 1) * sim_dec_fact],
axis=1)
data_filt = np.zeros((NoE, SpE), dtype=complex)
data_filt = data_filt_dec
tacq = (1 / Sf) * 1e6 * np.linspace(1, SpE, SpE) # in uS
else:
# do down conversion locally
if (direct_read):
data = datain
else:
if (proc_indv_data):
# read all datas and average it
data = np.zeros(NoE * SpE)
for m in range(1, total_scan + 1):
file_path = (data_folder + file_name_prefix +
'{0:03d}'.format(m))
# read the data from the file and store it in numpy array
# format
one_scan = np.array(data_parser.read_data(file_path))
one_scan = (one_scan - np.mean(one_scan)) / \
total_scan # remove DC component
if (en_ph_cycle_proc):
if (m % 2): # phase cycling every other scan
data = data - one_scan
else:
data = data + one_scan
else:
data = data + one_scan
else:
# read sum data only
file_path = (data_folder + 'asum')
data = np.zeros(NoE * SpE)
if binary_OR_ascii:
data = data_parser.read_hex_float(
file_path) # use binary representation
else:
# use ascii representation
data = np.array(data_parser.read_data(file_path))
dataraw = data
data = (data - np.mean(data))
# ignore echoes
if ignore_echoes:
data = data[ignore_echoes * SpE:len(data)]
# compute the probe voltage before gain stage
data = data / nmrObj.totGain * nmrObj.uvoltPerDigit
if en_fig: # plot the averaged scan
echo_space = (1 / Sf) * np.linspace(1, SpE, SpE) # in s
plt.figure(1)
for i in range(1, NoE + 1):
# plt.plot(((i - 1) * tE * 1e-6 + echo_space) * 1e3, data[(i - 1) * SpE:i * SpE], linewidth=0.4)
plt.plot(((i - 1) * tE * 1e-6 + echo_space) * 1e3,
dataraw[(i - 1) * SpE:i * SpE],
linewidth=0.4)
plt.title("Averaged raw data")
plt.xlabel('time(ms)')
plt.ylabel('probe voltage (uV)')
plt.savefig(data_folder + 'fig_avg_raw_data.png')
# raw average data
echo_rawavg = np.zeros(SpE, dtype=float)
for i in range(0, NoE):
echo_rawavg += (data[i * SpE:(i + 1) * SpE] / NoE)
if en_fig: # plot echo rawavg
plt.figure(6)
plt.plot(tacq, echo_rawavg, label='echo rawavg')
plt.xlim(0, max(tacq))
plt.title("Echo Average (raw)")
plt.xlabel('time(uS)')
plt.ylabel('probe voltage (uV)')
plt.legend()
plt.savefig(data_folder + 'fig_echo_avg.png')
# filter the data
data_filt = np.zeros((NoE, SpE), dtype=complex)
for i in range(0, NoE):
data_filt[i, :] = down_conv(data[i * SpE:(i + 1) * SpE], i, tE, Df,
Sf)
# simulate additional decimation (not needed for normal operqtion). For
# debugging purpose
if (sim_dec):
SpE = int(SpE / sim_dec_fact)
Sf = Sf / sim_dec_fact
data_filt_dec = np.zeros((NoE, SpE), dtype=complex)
for i in range(0, SpE):
data_filt_dec[:, i] = np.sum(
data_filt[:, i * sim_dec_fact:(i + 1) * sim_dec_fact],
axis=1)
data_filt = np.zeros((NoE, SpE), dtype=complex)
data_filt = data_filt_dec
tacq = (1 / Sf) * 1e6 * np.linspace(1, SpE, SpE) # in uS
# scan rotation
if en_ext_param:
data_filt = data_filt * np.exp(-1j * thetaref)
theta = math.atan2(np.sum(np.imag(data_filt)),
np.sum(np.real(data_filt)))
else:
theta = math.atan2(np.sum(np.imag(data_filt)),
np.sum(np.real(data_filt)))
if perform_rotation:
data_filt = data_filt * np.exp(-1j * theta)
if en_fig: # plot filtered data
echo_space = (1 / Sf) * np.linspace(1, SpE, SpE) # in s
plt.figure(2)
for i in range(0, NoE):
plt.plot((i * tE * 1e-6 + echo_space) * 1e3,
np.real(data_filt[i, :]),
'b',
linewidth=0.4)
plt.plot((i * tE * 1e-6 + echo_space) * 1e3,
np.imag(data_filt[i, :]),
'r',
linewidth=0.4)
plt.legend()
plt.title('Filtered data')
plt.xlabel('Time (mS)')
plt.ylabel('probe voltage (uV)')
plt.savefig(data_folder + 'fig_filt_data.png')
# find echo average, echo magnitude
echo_avg = np.zeros(SpE, dtype=complex)
for i in range(0, NoE):
echo_avg += (data_filt[i, :] / NoE)
if en_fig: # plot echo shape
plt.figure(3)
plt.plot(tacq, np.abs(echo_avg), label='abs')
plt.plot(tacq, np.real(echo_avg), label='real part')
plt.plot(tacq, np.imag(echo_avg), label='imag part')
plt.xlim(0, max(tacq))
plt.title("Echo Shape")
plt.xlabel('time(uS)')
plt.ylabel('probe voltage (uV)')
plt.legend()
plt.savefig(data_folder + 'fig_echo_shape.png')
# plot fft of the echosum
plt.figure(4)
zf = 100 # zero filling factor to get smooth curve
ws = 2 * np.pi / (tacq[1] - tacq[0]) # in MHz
wvect = np.linspace(-ws / 2, ws / 2, len(tacq) * zf)
echo_zf = np.zeros(zf * len(echo_avg), dtype=complex)
echo_zf[int((zf / 2) * len(echo_avg) -
len(echo_avg) / 2):int((zf / 2) * len(echo_avg) +
len(echo_avg) / 2)] = echo_avg
spect = zf * (np.fft.fftshift(np.fft.fft(np.fft.ifftshift(echo_zf))))
spect = spect / len(spect) # normalize the spectrum
plt.plot(wvect / (2 * np.pi), np.real(spect), label='real')
plt.plot(wvect / (2 * np.pi), np.imag(spect), label='imag')
plt.xlim(10 / max(tacq) * -1, 10 / max(tacq) * 1)
plt.title("FFT of the echo-sum. " +
"Peak:real@{:0.2f}kHz,abs@{:0.2f}kHz".format(
wvect[np.abs(np.real(spect)) == max(
np.abs(np.real(spect)))][0] / (2 * np.pi) *
1e3, wvect[np.abs(spect) == max(np.abs(spect))][0] /
(2 * np.pi) * 1e3))
plt.xlabel('offset frequency(MHz)')
plt.ylabel('Echo amplitude (a.u.)')
plt.legend()
plt.savefig(data_folder + 'fig_echo_A.png')
# matched filtering
a = np.zeros(NoE, dtype=complex)
for i in range(0, NoE):
if en_ext_param:
a[i] = np.mean(
np.multiply(data_filt[i, mtch_fltr_sta_idx:SpE],
np.conj(echoref_avg[mtch_fltr_sta_idx:SpE]))
) # find amplitude with reference matched filtering
else:
a[i] = np.mean(
np.multiply(data_filt[i, mtch_fltr_sta_idx:SpE],
np.conj(echo_avg[mtch_fltr_sta_idx:SpE]))
) # find amplitude with matched filtering
a_integ = np.sum(np.real(a))
# def exp_func(x, a, b, c, d):
# return a * np.exp(-b * x) + c * np.exp(-d * x)
def exp_func(x, a, b):
return a * np.exp(-b * x)
# average the first 5% of datas
a_guess = np.mean(np.real(a[0:int(np.round(SpE / 20))]))
# c_guess = a_guess
# find min idx value where the value of (a_guess/exp) is larger than
# real(a)
# b_guess = np.where(np.real(a) == np.min(
# np.real(a[np.real(a) > a_guess / np.exp(1)])))[0][0] * tE / 1e6
# this is dummy b_guess, use the one I made above this for smarter one
# (but sometimes it doesn't work)
b_guess = 0.01
# d_guess = b_guess
# guess = np.array([a_guess, b_guess, c_guess, d_guess])
guess = np.array([a_guess, b_guess])
try: # try fitting data
popt, pocv = curve_fit(exp_func, t_echospace, np.real(a), guess)
# obtain fitting parameter
a0 = popt[0]
T2 = 1 / popt[1]
# Estimate SNR/echo/scan
f = exp_func(t_echospace, *popt) # curve fit
noise = np.std(np.imag(a))
res = np.std(np.real(a) - f)
snr_imag = a0 / (noise * math.sqrt(total_scan))
snr_res = a0 / (res * math.sqrt(total_scan))
snr = snr_imag
# plot fitted line
plt.figure(5)
plt.cla()
plt.plot(t_echospace * 1e3, f, label="fit") # plot in milisecond
plt.plot(t_echospace * 1e3, np.real(a) - f, label="residue")
except:
print('Problem in fitting. Set a0 and T2 output to 0\n')
a0 = 0
T2 = 0
noise = 0
res = 0
snr = 0
if en_fig:
# plot data
plt.figure(5)
# plot in milisecond
plt.plot(t_echospace * 1e3, np.real(a), label="real")
# plot in milisecond
plt.plot(t_echospace * 1e3, np.imag(a), label="imag")
# plt.set(gca, 'FontSize', 12)
plt.legend()
plt.title(
'Matched filtered data. SNRim:{:03.2f} SNRres:{:03.2f}.\na:{:03.1f} n_im:{:03.1f} n_res:{:03.1f} T2:{:0.2f}msec'
.format(snr, snr_res, a0, (noise * math.sqrt(total_scan)),
(res * math.sqrt(total_scan)), T2 * 1e3))
plt.xlabel('Time (mS)')
plt.ylabel('probe voltage (uV)')
plt.savefig(data_folder + 'fig_matched_filt_data.png')
plt.show()
print('a0 = ' + '{0:.2f}'.format(a0))
print('SNR/echo/scan = ' +
'imag:{0:.2f}, res:{1:.2f}'.format(snr, snr_res))
print('T2 = ' + '{0:.4f}'.format(T2 * 1e3) + ' msec')
return (a, a_integ, a0, snr, T2, noise, res, theta, data_filt, echo_avg,
t_echospace)
def compute_wobble(nmrObj, data_parent_folder, meas_folder, s11_min, S11mV_ref,
useRef, en_fig, fig_num):
# S11mV_ref is the reference s11 corresponds to maximum reflection (can be made by totally untuning matching network, e.g. disconnecting all caps in matching network)
# useRef uses the S11mV_ref as reference, otherwise it will use the max
# signal available in the data
# settings
# put 1 if the data file uses binary representation, otherwise it is in
# ascii format. Find the setting in the C programming file
binary_OR_ascii = 1 # manual setting: put the same setting from the C programming
data_folder = (data_parent_folder + '/' + meas_folder + '/')
(param_list, value_list) = data_parser.parse_info(data_folder,
'acqu.par') # read file
freqSta = data_parser.find_value('freqSta', param_list, value_list)
freqSto = data_parser.find_value('freqSto', param_list, value_list)
freqSpa = data_parser.find_value('freqSpa', param_list, value_list)
nSamples = data_parser.find_value('nSamples', param_list, value_list)
freqSamp = data_parser.find_value('freqSamp', param_list, value_list)
spect_bw = (freqSamp / nSamples) * 4 # determining the RBW
file_name_prefix = 'tx_acq_'
# plus freqSpa/2 is to include the endpoint (just like what the C does)
freqSw = np.arange(freqSta, freqSto + (freqSpa / 2), freqSpa)
S11mV = np.zeros(len(freqSw))
S11_ph = np.zeros(len(freqSw))
for m in range(0, len(freqSw)):
# for m in freqSw:
file_path = (data_folder + file_name_prefix +
'{:4.3f}'.format(freqSw[m]))
if binary_OR_ascii:
# one_scan = data_parser.read_hex_int16(file_path) # use binary
# representation
one_scan = data_parser.read_hex_int32(
file_path) # use binary representation for 32-bit file
else:
one_scan = np.array(
data_parser.read_data(file_path)) # use ascii representation
os.remove(file_path) # delete the file after use
# find voltage at the input of ADC in mV
one_scan = one_scan * nmrObj.uvoltPerDigit / 1e3
spectx, specty = nmr_fft(one_scan, freqSamp, 0)
# FIND INDEX WHERE THE MAXIMUM SIGNAL IS PRESENT
# PRECISE METHOD: find reflection at the desired frequency: creating precision problem where usually the signal shift a little bit from its desired frequency
# ref_idx = abs(spectx - freqSw[m]) == min(abs(spectx - freqSw[m]))
# BETTER METHOD: find reflection signal peak around the bandwidth
ref_idx = (abs(spectx - freqSw[m]) <= (spect_bw / 2))
# S11mV[m] = max( abs( specty[ref_idx] ) ) # find reflection peak
# compute the mean of amplitude inside RBW
S11mV[m] = np.mean(abs(specty[ref_idx]))
S11_ph[m] = np.mean(np.angle(specty[ref_idx])) * (360 / (2 * np.pi))
if useRef: # if reference is present
S11dB = 20 * np.log10(np.divide(S11mV,
S11mV_ref)) # convert to dB scale
else: # if reference is not present
S11dB = 20 * np.log10(S11mV / max(S11mV)) # convert to dB scale
S11_min10dB = (S11dB <= s11_min)
minS11 = min(S11dB)
minS11_freq = freqSw[np.argmin(S11dB)]
try:
S11_fmin = min(freqSw[S11_min10dB])
S11_fmax = max(freqSw[S11_min10dB])
except:
S11_fmin = 0
S11_fmax = 0
print('S11 requirement is not satisfied...')
S11_bw = S11_fmax - S11_fmin
if en_fig:
plt.ion()
fig = plt.figure(fig_num)
fig.clf()
ax = fig.add_subplot(211)
line1, = ax.plot(freqSw, S11dB, 'r-')
ax.set_ylim(-35, 10)
ax.set_ylabel('S11 [dB]')
ax.set_title("Reflection Measurement (S11) Parameter")
ax.grid()
bx = fig.add_subplot(212)
bx.plot(freqSw, S11_ph, 'r-')
bx.set_xlabel('Frequency [MHz]')
bx.set_ylabel('Phase (deg)')
bx.set_title(
'incorrect phase due to non-correlated transmit and sampling')
bx.grid()
fig.canvas.draw()
fig.canvas.flush_events()
plt.savefig(data_folder + 'wobble.png')
# write S11mV to a file
with open(data_folder + 'S11mV.txt', 'w') as f:
for (a, b, c) in zip(freqSw, S11dB, S11_ph):
f.write('{:-8.3f},{:-8.3f},{:-7.1f}\n'.format(a, b, c))
# print(S11_fmin, S11_fmax, S11_bw)
if useRef:
return S11dB, S11_fmin, S11_fmax, S11_bw, minS11, minS11_freq
else:
return S11mV, S11_fmin, S11_fmax, S11_bw, minS11, minS11_freq
def compute_gain(nmrObj, data_parent_folder, meas_folder, en_fig, fig_num):
data_folder = (data_parent_folder + '/' + meas_folder + '/')
# settings
# put 1 if the data file uses binary representation, otherwise it is in
# ascii format. Find the setting in the C programming file
binary_OR_ascii = 1 # manual setting: put the same setting from the C programming
(param_list, value_list) = data_parser.parse_info(data_folder,
'acqu.par') # read file
freqSta = data_parser.find_value('freqSta', param_list, value_list)
freqSto = data_parser.find_value('freqSto', param_list, value_list)
freqSpa = data_parser.find_value('freqSpa', param_list, value_list)
nSamples = data_parser.find_value('nSamples', param_list, value_list)
freqSamp = data_parser.find_value('freqSamp', param_list, value_list)
spect_bw = (freqSamp / nSamples) * 4 # determining the RBW
file_name_prefix = 'tx_acq_'
# plus freqSpa/2 is to include the endpoint (just like what the C does)
freqSw = np.arange(freqSta, freqSto + (freqSpa / 2), freqSpa)
S21 = np.zeros(len(freqSw))
S21_ph = np.zeros(len(freqSw))
for m in range(0, len(freqSw)):
# for m in freqSw:
file_path = (data_folder + file_name_prefix +
'{:4.3f}'.format(freqSw[m]))
if binary_OR_ascii:
# one_scan = data_parser.read_hex_int16(file_path) # use binary
# representation
one_scan = data_parser.read_hex_int32(
file_path) # use binary representation for 32-bit file
else:
one_scan = np.array(
data_parser.read_data(file_path)) # use ascii representation
os.remove(file_path) # delete the file after use
'''
plt.ion()
fig = plt.figure( 64 )
fig.clf()
ax = fig.add_subplot( 111 )
ax.plot( one_scan )
fig.canvas.draw()
fig.canvas.flush_events()
'''
# find voltage at the input of ADC in mV
one_scan = one_scan * nmrObj.uvoltPerDigit / 1e3
spectx, specty = nmr_fft(one_scan, freqSamp, 0)
# FIND INDEX WHERE THE MAXIMUM SIGNAL IS PRESENT
# PRECISE METHOD: find reflection at the desired frequency: creating precision problem where usually the signal shift a little bit from its desired frequency
# ref_idx = abs(spectx - freqSw[m]) == min(abs(spectx - freqSw[m]))
# BETTER METHOD: find reflection signal peak around the bandwidth
# divide 2 is due to +/- half-BW around the interested frequency
ref_idx = (abs(spectx - freqSw[m]) <= (spect_bw / 2))
# compute the mean of amplitude inside RBW
S21[m] = np.mean(abs(specty[ref_idx]))
# compute the angle mean. Currently it is not useful because the phase
# is not preserved in the sequence
S21_ph[m] = np.mean(np.angle(specty[ref_idx])) * (360 / (2 * np.pi))
S21dB = 20 * np.log10(S21) # convert to dBmV scale
maxS21 = max(S21dB)
maxS21_freq = freqSw[np.argmax(S21dB)]
if en_fig:
plt.ion()
fig = plt.figure(fig_num)
fig.clf()
ax = fig.add_subplot(111)
line1, = ax.plot(freqSw, S21dB, 'r-')
ax.set_ylim(-30, 80)
ax.set_ylabel('S21 [dBmV]')
ax.set_title("Transmission Measurement (S21) Parameter")
ax.grid()
# bx = fig.add_subplot( 212 )
# bx.plot( freqSw, S21_ph, 'r-' )
# bx.set_xlabel( 'Frequency [MHz]' )
# bx.set_ylabel( 'Phase (deg)' )
# bx.set_title( 'incorrect phase due to non-correlated transmit and sampling' )
# bx.grid()
fig.canvas.draw()
fig.canvas.flush_events()
plt.savefig(data_folder + 'gain.png')
# write gain values to a file
with open(data_folder + 'S21.txt', 'w') as f:
for (a, b, c) in zip(freqSw, S21dB, S21_ph):
f.write('{:-8.3f},{:-8.3f},{:-7.1f}\n'.format(a, b, c))
return maxS21, maxS21_freq, S21
'''
Created on Mar 30, 2018
@author: David Ariando
'''
import nmr_std_function.data_parser as nmr_info
data_folder = "D:/NMR_EECS397_PCBv3_2017/DATAS/nmr_2018_03_07_19_32_23/"
file_name = "CPMG_iterate_settings.txt"
(param_list, value_list) = nmr_info.parse_info(data_folder, file_name)
# echo_spacing = value_list[[i for i, elem in enumerate(param_list) if 'echo_spacing_us' in elem][0]]
echo_spacing = nmr_info.find_value('echo_spacing_us', param_list, value_list)
print(echo_spacing)
# put the name into the "filename", e.g. dat_001
import matplotlib.pyplot as plt
from nmr_std_function import data_parser
import numpy as np
data_folder = 'Z:/NMR_DATA'
fold1 = '2020_05_18_23_23_49_cpmg'
fold2 = '2020_05_18_23_24_28_cpmg'
filename = 'dat_001'
# variables from NMR settings
# load acquisition setting 1
( param_list, value_list ) = data_parser.parse_info(
data_folder + '/' + fold1 + '/', 'acqu.par' ) # read file
dwellTime1 = data_parser.find_value(
'dwellTime', param_list, value_list )
dconvFact1 = data_parser.find_value(
'dconvFact', param_list, value_list )
# load acquisition setting 2
( param_list, value_list ) = data_parser.parse_info(
data_folder + '/' + fold2 + '/', 'acqu.par' ) # read file
dwellTime2 = data_parser.find_value(
'dwellTime', param_list, value_list )
dconvFact2 = data_parser.find_value(
'dconvFact', param_list, value_list )
# load data 1
file_path = ( data_folder + '/' + fold1 + '/' + filename )
| nmrObj.RX2_L_msk | nmrObj.RX2_H_msk
| nmrObj.RX_SEL1_msk | nmrObj.RX_FL_msk
| nmrObj.RX_FH_msk)
while True:
nmrObj.cpmgSequence(cpmg_freq, pulse1_us, pulse2_us, pulse1_dtcl,
pulse2_dtcl, echo_spacing_us, scan_spacing_us,
samples_per_echo, echoes_per_scan,
init_adc_delay_compensation, number_of_iteration,
ph_cycl_en, pulse180_t1_int, delay180_t1_int)
meas_folder = parse_simple_info(data_folder, 'current_folder.txt')
data_fold = (data_folder + '/' + meas_folder[0] + '/')
file_path = (data_fold + 'asum')
(param_list, value_list) = data_parser.parse_info(data_fold,
'acqu.par') # read file
SpE = int(data_parser.find_value('nrPnts', param_list, value_list))
NoE = int(data_parser.find_value('nrEchoes', param_list, value_list))
data = np.zeros(NoE * SpE)
data = np.array(data_parser.read_data(file_path))
ppval = np.max(data) - np.min(data)
print("peak-to-peak : %d" % ppval)
# print("rms : %f" % ppval / (2 * np.sqrt(2)))
plt.ion()
fig = plt.figure(1)
fig.clf()
plt.plot(data)
fig.canvas.draw()
def compute_multiple(data_parent_folder, meas_folder, file_name_prefix, Df, Sf,
tE, total_scan, en_fig, en_ext_param, thetaref,
echoref_avg, direct_read, datain):
# variables to be input
# data_parent_folder : the folder for all datas
# meas_folder : the specific folder for one measurement
# filename_prefix : the name of data prefix
# Df : data frequency
# Sf : sample frequency
# tE : echo spacing
# total_scan : number_of_iteration
# en_fig : enable figure
# en_ext_param : enable external parameter for data signal processing
# thetaref : external parameter : rotation angle
# echoref_avg : external parameter : echo average reference
# datain : the data captured direct reading. data format: AC,averaged scans, phase-cycled
# direct_read : perform direct reading from SDRAM/FIFO
data_folder = (data_parent_folder + '/' + meas_folder + '/')
# variables local to this function
# the setting file for the measurement
mtch_fltr_sta_idx = 0 # 0 is default or something referenced to SpE, e.g. SpE/4; the start index for match filtering is to neglect ringdown part from calculation
# perform rotation to the data -> required for reference data for t1
# measurement
perform_rotation = 1
# process individual raw data, otherwise it'll load a sum file generated
# by C
proc_indv_data = 0
proc_dconv = 1 # process dconv in python, otherwise use fpga dconv
dconv_factor = 1 # decimation factor for downconversion
# receiver gain
pamp_gain_dB = 54
rx_gain_dB = 0
totGain = 10**((pamp_gain_dB + rx_gain_dB) / 20)
# ADC conversion
voltPerDigit = 3.2 * (10**6) / 16384 # microvolt
# variables from NMR settings
(param_list, value_list) = data_parser.parse_info(data_folder,
'acqu.par') # read file
SpE = int(data_parser.find_value('nrPnts', param_list, value_list))
NoE = int(data_parser.find_value('nrEchoes', param_list, value_list))
en_ph_cycle_proc = data_parser.find_value('usePhaseCycle', param_list,
value_list)
# tE = data_parser.find_value('echoTimeRun', param_list, value_list)
# Sf = data_parser.find_value(
# 'adcFreq', param_list, value_list) * 1e6
# Df = data_parser.find_value(
# 'b1Freq', param_list, value_list) * 1e6
# total_scan = int(data_parser.find_value(
# 'nrIterations', param_list, value_list))
# compensate for dconv_factor if fpga dconv is used
if not proc_dconv:
SpE = int(SpE / dconv_factor)
Sf = Sf / dconv_factor
# time domain for plot
tacq = (1 / Sf) * 1e6 * np.linspace(1, SpE, SpE) # in uS
t_echospace = tE / 1e6 * np.linspace(1, NoE, NoE) # in uS
if proc_dconv:
# parse file and remove DC component
if (direct_read):
data = datain
else:
if (proc_indv_data):
# read all datas and average it
data = np.zeros(NoE * SpE)
for m in range(1, total_scan + 1):
file_path = (data_folder + file_name_prefix +
'{0:03d}'.format(m))
# read the data from the file and store it in numpy array
# format
one_scan = np.array(data_parser.read_data(file_path))
one_scan = (one_scan - np.mean(one_scan)) / \
total_scan # remove DC component
if (en_ph_cycle_proc):
if (m % 2): # phase cycling every other scan
data = data - one_scan
else:
data = data + one_scan
else:
data = data + one_scan
else:
# read sum data only
file_path = (data_folder + 'asum')
data = np.zeros(NoE * SpE)
data = np.array(data_parser.read_data(file_path))
data = (data - np.mean(data)) / \
total_scan # remove DC component
# compute raw data before gain stage
data = data / totGain * voltPerDigit
if en_fig: # plot the averaged scan
echo_space = (1 / Sf) * np.linspace(1, SpE, SpE) # in s
plt.figure(1)
for i in range(1, NoE + 1):
plt.plot(((i - 1) * tE * 1e-6 + echo_space) * 1e3,
data[(i - 1) * SpE:i * SpE],
linewidth=0.4)
# raw average data
echo_rawavg = np.zeros(SpE, dtype=float)
for i in range(5, NoE):
echo_rawavg += (data[i * SpE:(i + 1) * SpE] / NoE)
if en_fig: # plot echo rawavg
plt.figure(6)
plt.plot(tacq, echo_rawavg, label='echo rawavg')
plt.xlim(0, max(tacq))
plt.title("Echo Average")
plt.xlabel('time(uS)')
plt.ylabel('amplitude')
plt.legend()
# filter the data
data_filt = np.zeros((NoE, SpE), dtype=complex)
for i in range(0, NoE):
data_filt[i, :] = down_conv(data[i * SpE:(i + 1) * SpE], i, tE, Df,
Sf)
else: # use fpga dconv
# in-phase data
file_path = (data_folder + 'dconvi')
dconvi = np.array(data_parser.read_data(file_path))
# quadrature data
file_path = (data_folder + 'dconvq')
dconvq = np.array(data_parser.read_data(file_path))
# combined IQ
data_filt = np.zeros((NoE, SpE), dtype=complex)
for i in range(0, NoE):
data_filt[i, :] = \
dconvi[i * SpE:(i + 1) * SpE] + 1j * \
dconvq[i * SpE:(i + 1) * SpE]
# normalize data
data_filt = np.divide(data_filt, NoE * 25e3)
# scan rotation
if en_ext_param:
data_filt = data_filt * np.exp(-1j * thetaref)
theta = math.atan2(np.sum(np.imag(data_filt)),
np.sum(np.real(data_filt)))
else:
theta = math.atan2(np.sum(np.imag(data_filt)),
np.sum(np.real(data_filt)))
if perform_rotation:
data_filt = data_filt * np.exp(-1j * theta)
if en_fig: # plot filtered data
echo_space = (1 / Sf) * np.linspace(1, SpE, SpE) # in s
plt.figure(2)
for i in range(0, NoE):
plt.plot((i * tE * 1e-6 + echo_space) * 1e3,
np.real(data_filt[i, :]),
'b',
linewidth=0.4)
plt.plot((i * tE * 1e-6 + echo_space) * 1e3,
np.imag(data_filt[i, :]),
'r',
linewidth=0.4)
# find echo average, echo magnitude
echo_avg = np.zeros(SpE, dtype=complex)
for i in range(0, NoE):
echo_avg += (data_filt[i, :] / NoE)
if en_fig: # plot echo shape
plt.figure(3)
plt.plot(tacq, np.abs(echo_avg), label='abs')
plt.plot(tacq, np.real(echo_avg), label='real part')
plt.plot(tacq, np.imag(echo_avg), label='imag part')
plt.xlim(0, max(tacq))
plt.title("Echo Shape")
plt.xlabel('time(uS)')
plt.ylabel('amplitude')
plt.legend()
# plot fft of the echosum
plt.figure(4)
zf = 8 # zero filling factor to get smooth curve
ws = 2 * np.pi / (tacq[1] - tacq[0]) # in MHz
wvect = np.linspace(-ws / 2, ws / 2, len(tacq) * zf)
echo_zf = np.zeros(zf * len(echo_avg), dtype=complex)
echo_zf[int((zf / 2) * len(echo_avg) -
len(echo_avg) / 2):int((zf / 2) * len(echo_avg) +
len(echo_avg) / 2)] = echo_avg
spect = zf * (np.fft.fftshift(np.fft.fft(np.fft.ifftshift(echo_zf))))
plt.plot(wvect / (2 * np.pi), np.real(spect), label='real')
plt.plot(wvect / (2 * np.pi), np.imag(spect), label='imag')
plt.xlim(4 / max(tacq) * -1, 4 / max(tacq) * 1)
plt.title("fft of the echo-sum")
plt.xlabel('offset frequency(MHz)')
plt.ylabel('Echo amplitude (a.u.)')
plt.legend()
# matched filtering
a = np.zeros(NoE, dtype=complex)
for i in range(0, NoE):
if en_ext_param:
a[i] = np.mean(
np.multiply(data_filt[i, mtch_fltr_sta_idx:SpE],
np.conj(echoref_avg[mtch_fltr_sta_idx:SpE]))
) # find amplitude with reference matched filtering
else:
a[i] = np.mean(
np.multiply(data_filt[i, mtch_fltr_sta_idx:SpE],
np.conj(echo_avg[mtch_fltr_sta_idx:SpE]))
) # find amplitude with matched filtering
a_integ = np.sum(np.real(a))
# def exp_func(x, a, b, c, d):
# return a * np.exp(-b * x) + c * np.exp(-d * x)
def exp_func(x, a, b):
return a * np.exp(-b * x)
# average the first 5% of datas
a_guess = np.mean(np.real(a[0:int(np.round(SpE / 20))]))
#c_guess = a_guess
# find min idx value where the value of (a_guess/exp) is larger than
# real(a)
# b_guess = np.where(np.real(a) == np.min(
# np.real(a[np.real(a) > a_guess / np.exp(1)])))[0][0] * tE / 1e6
# this is dummy b_guess, use the one I made above this for smarter one
# (but sometimes it doesn't work)
b_guess = 10
#d_guess = b_guess
#guess = np.array([a_guess, b_guess, c_guess, d_guess])
guess = np.array([a_guess, b_guess])
try: # try fitting data
popt, pocv = curve_fit(exp_func, t_echospace, np.real(a), guess)
# obtain fitting parameter
a0 = popt[0]
T2 = 1 / popt[1]
# Estimate SNR/echo/scan
f = exp_func(t_echospace, *popt) # curve fit
noise = np.std(np.imag(a))
res = np.std(np.real(a) - f)
snr_imag = a0 / (noise * math.sqrt(total_scan))
snr_res = a0 / (res * math.sqrt(total_scan))
snr = snr_imag
# plot fitted line
plt.figure(5)
plt.cla()
plt.plot(t_echospace * 1e3, f, label="fit") # plot in milisecond
plt.plot(t_echospace * 1e3, np.real(a) - f, label="residue")
except:
print('Problem in fitting. Set a0 and T2 output to 0\n')
a0 = 0
T2 = 0
noise = 0
res = 0
snr = 0
if en_fig:
# plot data
plt.figure(5)
# plot in milisecond
plt.plot(t_echospace * 1e3, np.real(a), label="real")
# plot in milisecond
plt.plot(t_echospace * 1e3, np.imag(a), label="imag")
#plt.set(gca, 'FontSize', 12)
plt.legend()
plt.title('Filtered data')
plt.xlabel('Time (mS)')
plt.ylabel('Amplitude')
plt.show()
print('a0 = ' + '{0:.2f}'.format(a0))
print('SNR/echo/scan = ' +
'imag:{0:.2f}, res:{1:.2f}'.format(snr, snr_res))
print('T2 = ' + '{0:.4f}'.format(T2 * 1e3) + ' msec')
return (a, a_integ, a0, snr, T2, noise, res, theta, data_filt, echo_avg,
t_echospace)
def compute_wobble(data_parent_folder, meas_folder, s11_min, en_fig, fig_num):
data_folder = (data_parent_folder + '/' + meas_folder + '/')
(param_list, value_list) = data_parser.parse_info(data_folder,
'acqu.par') # read file
freqSta = data_parser.find_value('freqSta', param_list, value_list)
freqSto = data_parser.find_value('freqSto', param_list, value_list)
freqSpa = data_parser.find_value('freqSpa', param_list, value_list)
nSamples = data_parser.find_value('nSamples', param_list, value_list)
freqSamp = data_parser.find_value('freqSamp', param_list, value_list)
spect_bw = (freqSamp / nSamples) * 8
file_name_prefix = 'wobbdata_'
freqSw = np.arange(freqSta, freqSto, freqSpa)
S11 = np.zeros(len(freqSw))
for m in range(0, len(freqSw)):
# for m in freqSw:
file_path = (data_folder + file_name_prefix +
'{:4.3f}'.format(freqSw[m]))
one_scan = np.array(data_parser.read_data(file_path))
spectx, specty = nmr_fft(one_scan, freqSamp, 0)
# FIND INDEX WHERE THE MAXIMUM SIGNAL IS PRESENT
# PRECISE METHOD: find reflection at the desired frequency: creating precision problem where usually the signal shift a little bit from its desired frequency
# ref_idx = abs(spectx - freqSw[m]) == min(abs(spectx - freqSw[m]))
# BETTER METHOD: find reflection signal peak around the bandwidth
ref_idx = (abs(spectx - freqSw[m]) <= spect_bw)
S11[m] = max(specty[ref_idx]) # find reflection peak
S11 = 20 * np.log10(S11 / max(S11)) # convert to dB scale
S11_min10dB = (S11 <= s11_min)
minS11 = min(S11)
minS11_freq = freqSw[np.argmin(S11)]
try:
S11_fmin = min(freqSw[S11_min10dB])
S11_fmax = max(freqSw[S11_min10dB])
except:
S11_fmin = 0
S11_fmax = 0
print('S11 requirement is not satisfied...')
S11_bw = S11_fmax - S11_fmin
if en_fig:
plt.ion()
fig = plt.figure(fig_num)
fig.clf()
ax = fig.add_subplot(1, 1, 1)
line1, = ax.plot(freqSw, S11, 'r-')
ax.set_ylim(-50, 0)
ax.set_xlabel('Frequency [MHz]')
ax.set_ylabel('S11 [dB]')
ax.set_title("Reflection Measurement (S11) Parameter")
ax.grid()
fig.canvas.draw()
fig.canvas.flush_events()
# old code, freeze after plotting
# plt.figure(fig_num)
# plt.plot(freqSw, S11)
# plt.title("Reflection Measurement (S11) Parameter")
# plt.xlabel('Frequency [MHz]')
# plt.ylabel('S11 [dB]')
# plt.grid()
# plt.show()
# print(S11_fmin, S11_fmax, S11_bw)
return S11_fmin, S11_fmax, S11_bw, minS11, minS11_freq
def compute_wobble(nmrObj, data_parent_folder, meas_folder, s11_min, en_fig,
fig_num):
data_folder = (data_parent_folder + '/' + meas_folder + '/')
(param_list, value_list) = data_parser.parse_info(data_folder,
'acqu.par') # read file
freqSta = data_parser.find_value('freqSta', param_list, value_list)
freqSto = data_parser.find_value('freqSto', param_list, value_list)
freqSpa = data_parser.find_value('freqSpa', param_list, value_list)
nSamples = data_parser.find_value('nSamples', param_list, value_list)
freqSamp = data_parser.find_value('freqSamp', param_list, value_list)
spect_bw = (freqSamp / nSamples) * 4 # determining the RBW
file_name_prefix = 'wobbdata_'
freqSw = np.arange(
freqSta, freqSto + (freqSpa / 2), freqSpa
) # plus half is to remove error from floating point number operation
S11 = np.zeros(len(freqSw))
S11_ph = np.zeros(len(freqSw))
for m in range(0, len(freqSw)):
# for m in freqSw:
file_path = (data_folder + file_name_prefix +
'{:4.3f}'.format(freqSw[m]))
one_scan = np.array(data_parser.read_data(file_path))
os.remove(file_path) # delete the file after use
# find voltage at the input of ADC in mV
one_scan = one_scan * nmrObj.uvoltPerDigit / 1e3
spectx, specty = nmr_fft(one_scan, freqSamp, 0)
# FIND INDEX WHERE THE MAXIMUM SIGNAL IS PRESENT
# PRECISE METHOD: find reflection at the desired frequency: creating precision problem where usually the signal shift a little bit from its desired frequency
# ref_idx = abs(spectx - freqSw[m]) == min(abs(spectx - freqSw[m]))
# BETTER METHOD: find reflection signal peak around the bandwidth
ref_idx = (abs(spectx - freqSw[m]) <= (spect_bw / 2))
# S11[m] = max( abs( specty[ref_idx] ) ) # find reflection peak
S11[m] = np.mean(abs(
specty[ref_idx])) # compute the mean of amplitude inside RBW
S11_ph[m] = np.mean(np.angle(specty[ref_idx])) * (360 / (2 * np.pi))
S11dB = 20 * np.log10(S11 / max(S11)) # convert to dB scale
S11_min10dB = (S11dB <= s11_min)
minS11 = min(S11dB)
minS11_freq = freqSw[np.argmin(S11dB)]
try:
S11_fmin = min(freqSw[S11_min10dB])
S11_fmax = max(freqSw[S11_min10dB])
except:
S11_fmin = 0
S11_fmax = 0
print('S11 requirement is not satisfied...')
S11_bw = S11_fmax - S11_fmin
if en_fig:
plt.ion()
fig = plt.figure(fig_num)
fig.clf()
ax = fig.add_subplot(211)
line1, = ax.plot(freqSw, S11dB, 'r-')
ax.set_ylim(-35, 0)
ax.set_ylabel('S11 [dB]')
ax.set_title("Reflection Measurement (S11) Parameter")
ax.grid()
bx = fig.add_subplot(212)
bx.plot(freqSw, S11_ph, 'r-')
bx.set_xlabel('Frequency [MHz]')
bx.set_ylabel('Phase (deg)')
bx.set_title(
'incorrect phase due to non-correlated transmit and sampling')
bx.grid()
fig.canvas.draw()
fig.canvas.flush_events()
plt.savefig(data_folder + 'wobble.png')
# write S11 to a file
with open(data_folder + 'S11.txt', 'w') as f:
for (a, b, c) in zip(freqSw, S11dB, S11_ph):
f.write('{:-8.3f},{:-8.3f},{:-7.1f}\n'.format(a, b, c))
# print(S11_fmin, S11_fmax, S11_bw)
return S11, S11_fmin, S11_fmax, S11_bw, minS11, minS11_freq
