Lab 6 Sample Notebook: RTL-SDR¶

As Lab 6 moves into the section Writing Your Own Demodulator Algorithms you will need to add two Python packages beyond the installation of the RTL-SDR drivers and making use of module lab6.py included in the ZIP package. The lab reader discusses this but as of May 2026 there have been some changes in the packages.

The two Python packages that need to be suinstalled are:

  • sdr_helper which was originally written for this lab; to install from the terminal in your dsp-comm313 or similar virtual environment use
(dsp-comm313) PS C:\Users\mwick> pip install sdr-helper
  • rtlsdr a package that can directly control the RTL-SDR dongle; to install from the terminal in your dsp-comm313 or similar virtual environment use
(dsp-comm313) PS C:\Users\mwick> pip install pyrtlsdr[lib]
In [1]:
# %pylab inline
from numpy import * # better in import numpy as np
from matplotlib.pyplot import * # better to import matplotlib.pyplot as plt
import sk_dsp_comm.sigsys as ss
# As a result of: pip install pip install sdr-helper
import sk_dsp_comm.sdr_helper as sdr
# This package also requires: pip install pyrtlsdr[lib]
import lab6
import sk_dsp_comm.fir_design_helper as fir_d
import sk_dsp_comm.digitalcom as dc
import scipy.signal as signal
from IPython.display import Audio, display
from IPython.display import Image, SVG

%config InlineBackend.figure_formats=['svg'] # SVG inline viewing

Capture 5 Seconds of I/Q Samples at 88.7 MHz (KCME, Colorado Springs)¶

In [2]:
Tc = 5
# Tc, fo=88700000.0, fs=2400000.0, gain=40, device_index=0
x = lab6.capture(Tc, fo=88700000.0, fs=2400000.0, gain=40)

Archive the capture in a wave file¶

In [3]:
lab6.complex2wav('KCME_IQ.wav',2400000,x)

Saved as binary wav file with (I,Q)<=>(L,R)

Restore the IQ Samples into a Complex Baseband Array¶

If need be, restore an archived capture back into the workspace. The sampling rate will be 2.4 Msps as that is how it was originall captured and save.

In [4]:
fs, x = lab6.wav2complex('KCME_IQ.wav')

Look at the Complex Baseband Spectrum from the RTL-SDR¶

In [11]:
Px, fx = ss.psd(x,2**14,fs=2400,scale_noise=True);
plot(fx,10*log10(Px))
xlim(-1000,1000);
title(r'Rx Spectrum at $f_o = 88.7$ MHz with $f_s = 2.4$ Msps')
xlabel(r'Frequency (kHz)')
ylabel(r'PSD (dB/Hz)')
grid();
No description has been provided for this image

Process Samples Through a Mono FM Demodulator¶

In [7]:
z_bb, z_out = sdr.mono_fm(x,fs=2.4e6,file_name='KCME.wav')

Done!
In [8]:
Audio('KCME.wav')  # From file
Out[8]:
Your browser does not support the audio element.
In [33]:
z_bb, theta, y_lpr, y_lmr, z_out = \
        sdr.stereo_fm(x,fs=2.4e6,file_name='KCME_stereo.wav')

Done!
In [34]:
Audio('KCME_stereo.wav')  # From file
Out[34]:
Your browser does not support the audio element.
In [40]:
P_z_bb, f_x_bb = ss.psd(z_bb,2**14,fs=2400/10.,scale_noise=True)
plot(f_x_bb,10*log10(P_z_bb))
title(r'Baseband Spectrum Showing Sub-Carriers')
text(2.0, 22.0, r'$L+R$',c='green')
text(12.0, 17.0, r'19 kHz pilot',c='green')
text(32.0, 13.0, r'DSB $L-R$',c='green')
text(52.0,1.0, r'RDS BPSK',c='green')
xlabel(r'Frequency (kHz)')
ylabel(r'PSD (dB/Hz)')
grid();
No description has been provided for this image
  • Note the subcarrier label RDS BPSK is a binary phase-shift keyed subcarrier responsible for the radio data service that produces information such as the current sone being played by the radio station; https://pysdr.org/content/rds.html
  • Coherent demodulation from the 57 kHz subcarrier is possible using three times the 19 pilot

FSK Simulation¶

The simulation code shown below can be used to understand the the performance of an FSK receiver using a real RF transmnit signal generated by the Keysight 33600A and then received by the RTLSDR.

In [42]:
Ns = 120; Nbits = 1000; Rb = 1000; Df = 2500.0; fs = Rb*Ns;
data = ss.pn_gen(Nbits,7)
#data = ss.pn_gen(Nbits,7)
waveform = signal.lfilter(ones(Ns),1,ss.upsample(data,Ns))
x, b = ss.nrz_bits2(data,Ns,'rect')
n = arange(Ns*Nbits)
xc = exp(1j*2*pi*Df*x/fs*n);
In [43]:
plot(n[:3000]/fs*1000,x[:3000])
xlabel(r'Time (ms)')
ylabel(r'NRZ Bits $d(t)$')
grid();
No description has been provided for this image
In [46]:
Pxc,fxc = ss.psd(xc,2**12,fs,scale_noise=False);
plot(fxc,10*log10(Pxc))
xlim([-5000,5000])
ylim([-60,0])
title(r'1 kbps FSK with Peak Freq Dev = 2.5 kHz (5 kHz p-p)')
xlabel(r'Frequency (kHz)')
ylabel(r'Power Spectrum (dB)')
grid();
savefig('FSK1kbps2_5.pdf')
No description has been provided for this image

FSK Capture $\Delta f = 2.5$ kHz¶

The details here require first configuring the Keysight 33600A to produce a low power FSK signal at 70 MHz.

In [333]:
Tc = 20 # or 10s
# Tc, fo=88700000.0, fs=2400000.0, gain=40, device_index=0
x_FSK2_5 = sdr.capture(Tc, fo=70000000.0, fs=2400000.0, gain=40)
In [ ]:
psd(x_FSK2_5,2**14,2400);
xlim(-50,50);
xlabel('Frequency (kHz)');

Save and Restore IQ Data¶

In [113]:
sdr.complex2wav('FSK_70M_IQ2_5.wav',2400000,x_FSK2_5)

Saved as binary wav file with (I,Q)<=>(L,R)
In [22]:
fs, x_FSK2_5 = sdr.wav2complex('FSK_70M_IQ2_5.wav')

Processing the Capture¶

$N_2=5$ Final Decimation Stage Lowpass Filter¶

In [27]:
b_FSK = fir_d.fir_remez_lpf(1000,4000,.1,60,2.4e6/10)

Remez filter taps = 205.
In [ ]:
# FSK demod code

Bit Synch to Received Data Bits¶

In [31]:
data_hat, clk, track = sdr.sccs_bit_sync(data_hat_FSK2_5,48)
In [63]:
plot(track)
title(r'Bit Synchronization Relative to Ns = 48 Samples/Bit')
ylabel(r'Start of Bit Mod Ns')
xlabel(r'Bits Processed')
grid();
No description has been provided for this image

Prepare an Equal Length Stream Transmitted PN7 Bits¶

In [56]:
def PN7_33600A(Nbits):
    """
    Create an Nbits stream equivalent of the Keysight 33600A PN7 PBRS
    
    Mark Wickert April 2019
    """
    m = 7
    PN7_oct_taps = [0o211,0o217,0o235,0o367,0o277,0o325,0o203,0o313,0o345]
    taps = np.array([1, 0, 0, 0, 0, 0, 1, 1])
    sr = np.ones(m)
    # M-squence length is:
    Q = 2**m - 1
    c = np.zeros(Q)
    for n in range(Q):
        tap_xor = 0
        c[n] = sr[-1]
        for k in range(1,m):
            if taps[k] == 1:
                tap_xor = np.bitwise_xor(tap_xor,np.bitwise_xor(int(sr[-1]),int(sr[m-1-k])))
        sr[1:] = sr[:-1]
        sr[0] = tap_xor
    PN7 = c[::-1] # reverse the sequence
    PN7_stream = zeros(Nbits)
    for n in range(Nbits):
        PN7_stream[n] = PN7[mod(n,127)]
    return PN7_stream

Generate Tx Reference Bits and Error Detect¶

In [ ]:
tx_bits = PN7_33600A(len(data_hat))
N_bits, N_errors = dc.bit_errors(tx_bits,int16((data_hat[5:]+1)/2))
print('N_bits = %d, N_errors = %d, BEP = %1.2e' % (N_bits,N_errors, N_errors/N_bits))

Other Support Code¶

Using strips() for Plotting a Long Waveform¶

In [57]:
# data = ss.pn_gen(127*5,7)
# Make 5 periods of the 33600A PN7
data = PN7_33600A(5*127) # create 5 periods of a 127 bit sequence
x, b = ss.nrz_bits2(data,48,'rect')
dc.strips(x,48*127);
title(r'127-Bit PN Sequence Repeated 5 Times')
savefig('strips_example.svg')
No description has been provided for this image

Sinusoidal FM Transmitter¶

In [50]:
def bb_fm_tx(m,fd,fs):
    '''
    A simple complex baseband FM modulator
    
    Mark Wickert April 2019
    '''
    return exp(1j*2*pi*fd/float(fs)*cumsum(m))
In [75]:
# Here m is a 1kHz tone with fs = 240kHz, fd = 25kHz
fm = 1000.; fs = 240e3; fd = 25e3;
n = arange(0,100000)
x_tx = bb_fm_tx(cos(2*pi*fm/fs*n),fd,fs)
subplot(211)
P_x_tx, f_x_tx = ss.psd(x_tx,2**14,fs,scale_noise=False)
plot(f_x_tx,10*log10(P_x_tx))
title(r'Unshifted: $f_m = 1000$, $\Delta f = 25$ kHz')
ylabel(r'PSD (dB)')
xlabel(r'Frequency (Hz)')
xlim([-40e3, 40e3])
ylim([-80, 0])
grid();
subplot(212)
# Include a tuning frequency error of +2000 Hz
x_tx *= exp(1j*2*pi*2000./fs*n)
P_x_tx, f_x_tx = ss.psd(x_tx,2**14,fs, scale_noise=False)
plot(f_x_tx,10*log10(P_x_tx))
title(r'Shifted by 2000 Hz: $f_m = 1000$, $\Delta f = 25$ kHz')
ylabel(r'PSD (dB)')
xlabel(r'Frequency (Hz)')
xlim([-40e3, 40e3])
ylim([-80, 0])
grid();
tight_layout()
# savefig('FM_tx_Python.svg')
No description has been provided for this image
In [ ]: