In my last post, I mentioned a technique for measuring the frequency response of an RF circuit. The method is, you apply a white noise signal and measure the response. Any deviations from white are due to the circuit.
In practice, we found it more convincing to compare the measured spectrum to another with the circuit unplugged or disabled. Here's the tool chain we've been using.
The spectra are measured using rtl_power, a command line spectrum analyzer, available as part of the rtl_sdr package.
An rtl_power command looks like this:
The -f specifies the frequency start:stop:step. The -i specifies the time to to take per sweep. The -e gives the total time to run (so setting -i and -e equal means one sweep). The last argument is the output file.
Phil then stole a pair of python programs to remove some of the formatting of the output csv and then plot the spectrum.
We then wrote a bash script that collects two spectra (prompting the user to alter the circuit between the collections), then flattens them and plots them together.
Here's an example output:
Green - No Filter Blue - Low-Pass Filter[/caption]