Analog Discovery 2 - OFDM application

[Ergest]

Hi,

I'm using an Analog Discovery 2 board with Python to transmit and record signals for an OFDM system. However, the output signal, when checked with a spectrum analyzer, shows peaks instead of a flat frequency response. I've tried using the AnalogOut_Custom.py script but it's not working as expected.

Can anyone suggest how to get a better frequency response with this setup, or any similar project that can show me the path? Also, how does the 'cSamples' variable affect the OFDM symbol samples?

Edit: This is what the spectrum analyzer shows: 

It is completely off, since I am using a center frequency of 5 kHz and a bandwidth of 3 kHz, 1024 subcarriers.

Thanks for any help!

Edited November 16, 2023 by Ergest

[attila]

Hi @Ergest

I would recommend to use the WaveForms application where it easier to setup such tests.
The 3rd device configuration for AD2 provides more buffer for the AWG.

[Ergest]

What are the limitations of implementing this in Python, and are there any exemplary guides for accomplishing this in the Waveform application? Should I script the generation of OFDM symbols or simply import them from a .csv file? Solely reading the file appears insufficient.

The OFDM consists of more than 62k samples. It's uncertain whether the buffer can manage this volume, but surprisingly, when I examine the time domain of the generated signal, it seems to encompass the entire signal, although the frequency spectrum is incorrect.

[attila]

Hi @Ergest

The WF app uses the same API but I was referring that it is easier to control the device with the app, adjust parameters...

In the first (default) configuration of AD2 the AWG has 4Ki and in the 3rd 16Ki samples of device buffer allocated.

For lower sample rates (<1MHz) you can also use the Play/streaming mode for 'unlimited' length. See Play in WF app/Wavegen and example Python scripts: AnalogOut_Play.py AnalogOutIn_PlayRecord.py AnalogOutIn_PlayRecordStereo.py AnalogOutIn_RecordPlay.py

 

[Ergest]

I still can't get this working. Is the device capable of producing a flat frequency response for, let's say, a 5 kHz bandwidth

[attila]

Hi @Ergest

The AWG -3dB BW is 9-12MHz
https://digilent.com/reference/test-and-measurement/analog-discovery-2/hardware-design-guide#awg_spectral_characteristics

Have you tried the Play in WaveForms application Wavegen ?
Please provide more information so I can help you, like the file samples, script, workspace...

[Ergest]

Hi @attila, 

Thank you for all your help in this matter. 

I use the following scriot to generate white Gaussian noise and bandpass it from 3k-5kHz:

 

from scipy.signal import butter, filtfilt
import numpy as np
import matplotlib.pyplot as plt
import scipy.signal as sg
# Define the sample rate and the number of samples
fs = 100e3 
num_samples = 4096  # Number of samples

def bandpass_filter(signal, lowcut, highcut, f, order=5):
    nyq = 0.5 * f
    low = lowcut / nyq
    high = highcut / nyq
    b, a = butter(order, [low, high], btype='band')
    y = filtfilt(b, a, signal)
    return y

# Generate white noise
np.random.seed(0)
white_noise = np.random.normal(1, 2, num_samples)

# Define the bandwidth of the signal (1 kHz)
lower_cutoff = 3000  # Hz
upper_cutoff = 8000  # Hz

# Apply bandpass filter
band_limited_signal_adjusted = bandpass_filter(white_noise, lower_cutoff, upper_cutoff, fs)

plt.figure(figsize=(10, 4))
plt.plot(band_limited_signal_adjusted)
plt.title('Adjusted Band-limited Signal')
plt.xlabel('Time (seconds)')
plt.ylabel('Amplitude')
plt.grid(True)
plt.show()


plt.figure()
f, P = sg.welch(band_limited_signal_adjusted, fs)
plt.plot(f / 1e3, 10 * np.log10(P))
plt.xlabel('Frequency [kHz]')
plt.ylabel('PSD [dBm/Hz]')

# Frequency shift
fc = 50e3  # 50 kHz
# Time array
t = np.arange(num_samples) / fs

# Complex exponential for frequency shifting
complex_exponential = np.exp(1j * 2 * np.pi * fc * t)

# Shift the signal
pass_band_band_limited_signal = np.real(band_limited_signal_adjusted * complex_exponential)

plt.figure()
f, P = sg.welch(pass_band_band_limited_signal, fs)
plt.plot(f / 1e3, 10 * np.log10(P))
plt.xlabel('Frequency [kHz]')
plt.ylabel('PSD [dBm/Hz]')

 

Then I followed the examples from AnalogOut_Play.py / AnalogOutIn_PlayRecord.py to write the following code:
 

from ctypes import *
import time
from dwfconstants import *
import sys

if sys.platform.startswith("win"):
    dwf = cdll.dwf
elif sys.platform.startswith("darwin"):
    dwf = cdll.LoadLibrary("/Library/Frameworks/dwf.framework/dwf")
    
else:
    dwf = cdll.LoadLibrary("libdwf.so")

cSamples = len(pass_band_band_limited_signal)

hdwf = c_int()

# Convert the numpy array to ctypes array
rgdSamples = (c_double * cSamples)(*pass_band_band_limited_signal)

channel = c_int(0)

version = create_string_buffer(16)
dwf.FDwfGetVersion(version)
print("DWF Version: "+str(version.value))

# prevent temperature drift
dwf.FDwfParamSet(DwfParamOnClose, c_int(0)) # 0 = run, 1 = stop, 2 = shutdown

#open device
print("Opening first device...")
dwf.FDwfDeviceOpen(c_int(-1), byref(hdwf))


if hdwf.value == hdwfNone.value:
    print("failed to open device")


# 0 = the device will be configured only when calling FDwf###Configure
dwf.FDwfDeviceAutoConfigureSet(hdwf, c_int(0))

dwf.FDwfAnalogOutNodeEnableSet(hdwf, channel, AnalogOutNodeCarrier, c_int(1))

dwf.FDwfAnalogOutNodeFunctionSet(hdwf, channel, 0, funcCustom) 
dwf.FDwfAnalogOutNodeDataSet(hdwf, channel, 0, rgdSamples, c_int(cSamples))
#dwf.FDwfAnalogOutNodeFrequencySet(hdwf, channel, 0, c_double(fs))
dwf.FDwfAnalogOutFrequencySet(hdwf, channel, c_double(fs)) 
dwf.FDwfAnalogOutNodeAmplitudeSet(hdwf, channel, 0, c_double(4))
dwf.FDwfAnalogOutNodeOffsetSet(hdwf, channel, 0, c_double(0))
dwf.FDwfAnalogOutRepeatSet(hdwf, channel, c_int(10000000)) # repeat 

dwf.FDwfAnalogOutConfigure(hdwf, channel, c_int(1))
dwf.FDwfAnalogOutReset(hdwf, channel)
dwf.FDwfDeviceClose(hdwf)

I have connected the BNC cable from AD2 to my picoscope and here is what I get in the spectrum analyzer.




I also saved the 4096 samples in a file and upload to the waveform app, but it didn't solve anything. However when I generate a sweep from the Waveform app, I see the expected frequency spectrum.

Best, 
Ergest

 

Edited November 20, 2023 by Ergest

[attila]

Hi @Ergest

FDwfAnalogOutNodeFrequencySet for standard and custom signals expects waveform frequency and not sample rate, you probably want c_double(fs/cSamples)
FDwfAnalogOutRepeatSet for existing devices it supports value up to 32768 or use 0 for infinite repeat