- collect: set up a test transmission at each available frequency, then
then record the signal seen by the USRP2
- norm: extract parameters necessary for normalization
- viz: visualize the spectrum with gnuplot
- row: generate rows of spectra with "viz"
- fft.c (usage, main): added optional averaging parameter for -d (dump)
- fft.c (do_fft): split the sample into several parts, FFT each of them
individually, and average the FFT results
- fft.c (window, fft_complex): added Hamming window function
- fft.c (do_fft): shift dump output such that center frequency is in the
middle, not at the edges
Examples:
doall 'run/*' # no arguments
doall -o foo.png 'run/*' # arguments for plscan
doall -a 1 -- -o foo.png 'run/*' # arguments for evscan and plscan
- usrp/doall: added usage()
- usrp/doall: arguments before -- are passed to evscan
We now have a much wider dynamic range of the signal (with better SNR).
The outliers are mainly WLAN interferences.
- usrp/evscan: changed threshold from -50 dB to -70 dB
- usrp/evscan: increase outlier tolerance from 2 dB to 10 dB
- usrp/plscan: changed display range from [-35:-25] to [-40:-5]
The pre- and post-processing of FFT data was tweaked based on results
obtained from a mis-configured measurement setup, in an only partially
successful attempt to right the wrong data.
- usrp/fft.c (fft_complex): square the magnitude on output but don't
square the scaling
- usrp/fft.c (fft_real): square the magnitude on input (does this make
sense ?)
- usrp/fft.c (fft_real): scale the output with the square root of the
FFT size (dubious)
- usrp/doall: new script to directly evaluate and plot data sets
- usrp/fft.c (fft_complex): don't square the magnitude
- usrp/plscan: if an argument name has the form title=name, use "name" as
the file name and "title" as the title
- usrp/evscan (usage, main): treat all arguments but the last as options
and pass them to "fft"
- usrp/fft.c (fft_complex, do_fft): moved actual FFT calculation to
separate function
- usrp/fft.c (fft_real, do_fft): added real-valued FFT (DCT) as alternative
- usrp/fft.c (main): undocumented option -a to set the FFT algorithm
- usrp/fscan: perform a frequency scan in my test setup
- usrp/evscan: determine the frequency-dependent signal strength from a
frequency scan
- usrp/plscan: plot one or more data sets produced by evscan
- usrp/fft.c: utility to read a set of samples obtained with
usrp2_rx_cfile.py and to perform a Fourier transform on them
- usrp/Makefile: use LDLIBS for libraries, not LDFLAGS
- usrp/Makefile: added "fft"
- usrp/p.c (find_peak, main): moved data processing to separate function
- usrp/p.c (find_peak, usage, main): new option -s skip to set the number
of samples to skip (default: 0)
- usrp/p.c (find_peak, usage, main): new option -d to dump the histogram
instead of picking the peak
- usrp/p.c (find_peak, usage, main): the percentile can now be set on the
command line (default: 0.9)
- TODO: changed reference to atspi-txrx to atrf-txrx
- TODO: removed renaming from the list of pending issues
- usrp/step: atspi-txrx is now called atrf-txrx
- usrp/step: the default number of runs (10) can now be overridden on the
command line
- usrp/step: usrp2_rx_cfile.py likes to busy loop from time to time. Add
a CPU time limit to prevent this from getting out of hand.
- usrp/d.c: reduce the number of data points in a file produced by
usrp2_rx_cfile.py and print them as text suitable for gnuplot
- usrp/p.c: determine the peak amplitude in a series of transmissions,
filtering noise and artefacts at the beginning of the data file
- usrp/step: step through all channels and measure TX power (for antenna
tuning)
- usrp/Makefile: built "p" and "d"
