A Simple GnuRadio and RTL-SDR Based NBFM Deviation Meter
This is a complete rewrite of my original deviation meter to eliminate many earlier limitations caused by the use of Gnu Radio's built in NBFM Demodulator block. While that block works very well for building receivers, it does not expose some parameters needed for wideband use and accurate measurements, so I decided to use lower level blocks to create my own demodulator.
The only prerequisites are that you have a recent version of GnuRadio installed that includes the OsmoSDR support for SDR radios, and that you have a properly installed library for you RTL-SDR.com version 3 or 4 radio. On some platforms you need to install the OsmoSDR support as a separate step.
Previous versions allowed you to set audio bandwidths greater than 20 KHz but didn't handle them properly. An option has been provided to allow bandwidths up to 43 KHz has now been added, but to handle these bandwidths, the sampling rate is increased and CPU usage will rise significantly.
The wide bandwidth functionality may not work well on Pi 4 or Pi 400 computers. As this capability is not required for most uses, it was made optional.
Note also that limitations in my test equipment make it impossible to confirm that operation beyond 30Khz audio bandwidth is completely correct. I'm seeing some signs that the filters may be aliasing, but I cannot tell for certain. Use this capability with caution.
To enable 96000 Hz sampling, the command line to the generated flow graph:
"python3 deviation.py --sampleRate 96000"
can be used, or it can be made the new default by changing the value in the "Parameter Block" in the upper left hand corner of the flow diagram to "96000" instead of "48000". These are the only two values allowed.
There is a newer Gnu Radio Control called the "QT GUI Digital Number Control" that can be used to set the receiver frequency. If you'd like to try it, and if it shows up in your version of Gnu Radio and the block is not labeled as "missing", you can enable it and disable the "QT Range" control in the lower left corner of the flow diagram labeled "tuning". Both do the job, but the method of tuning is completely different in each of them.
Hovering over the scope display will show the Y value of the peak deviation at that point.
While hovering over the scope display, the mouse wheel can be used to zoom the Y axis gain up or down. Of course, if you zoom out, what you see will not display much beyond the max_deviation and current bandwidth settings.
You can also select a rectangular portion of the display with the left button. When you release the button, the display will zoom to the selection.
Clicking with the right button will restore the original zoom settings.
The oscilloscope display runs free instead of being triggered. If you have a 3 button mouse, the middle button can pop up a menu of scope display parameters which can be used to modified this and other behaviours. You can also save the scope display as a jpeg file.
Wayland based systems may output lengthy warnings to standard output that X windows features used by this popup menu may not be supported in the future. These are annoying, but they may safely be ignored and don't affect operation at this time.
This sets the full scale value on the display and also sets the RF bandwidth needed to pass the relevant FM sidebands for that deviation.
This sets the baseband filtering bandwidth. This is the highest received audio frequency that can be displayed. On startup, it defaults to 5 KHz which is a good setting for most NBFM communication use at 5 KHz deviation.
Note that for NBFM, this wider bandwidth will often result in a lot of "jagged" or "raspy" appearing signals because that is what the transmitter is sending. They're normally outside the passband of NBFM receivers. If you want to simulate a typical transceivers's receiver instead of measuring the total transmitted deviation, you can reduce this setting to 3500 Hz.
Wireless microphones and broadcast measurements require a much wider setting. The value you select will depend on the highest modulating frequency of interest. Many wideband devices send additional subcarrier information like pilot tones, ultrasonic squelch tones, stereo difference information, supplementary subcarrier broadcasts, and additional digital channels above the basic transmission.
If you want to measure the total deviation of these, you must use a high enough bandwidth to pass all of them; if you are only interested in observing the behavior of the basic analog channel, you can use a lower value to filter them out.
For most US FM broadcasts, the maximum deviation should be set to 75 KHz. You can set the bandwidth to 16 KHz to see just the basic Mono signal. To see the full composite signal including pilot tone, L-R, SCA and Digital subchannels, you might need to use a bandwidth setting well over 40 KHz.
Note that this program now limits setting the bandwidth to slightly less than 1/2 the audio sampling rate in use. That defaults to 48000 but can be increased to 96000 samples per second as described above.
There is a selector that allows Deemphasis to be applied to the audio output. This is provided as a convenience for listening and roughly monitoring loudness. This filter only affects the monitor -- not the displayed deviation.
Note that the Tau applied if "WBFM" is chosen is the standard FCC value of 75 microseconds. Outside of North America, 50 microseconds is normally used. There is a variable called TAU in the flow graph where you can set this to any value desired.
The use of "NBFM" deemphasis is not specified by the FCC and varies widely in the industry. This filter applies a commonly used 750us value.
This may not be what you want as manufacturers use a variety of other demodulator response curves depending on their technology and desired speaker response! There is a variable called NBFM_TAU which can be used to adjust this value.
There is a checkbox that allows the application of a low pass filter to eliminate the DC component of the incoming signal. This DC offset is an indication of the difference in frequency between the Deviation meter and the incoming signal.
Being able to see the DC component is useful for detecting whether the transmitter is roughly on frequency or "chirping" during transmission, but in some cases it is desirable to reset the baseline to 0 for easier measurements. Note that this will also remove low frequencies below about 50 Hz. For maximum accuracy when measuring continuous tone squelch deviation, this option should NOT be selected.
A popup box in the lower right corner of the display allows you to select audio output from either an adjustable sine wave oscillator or from the monitor receiver. You may need to add a capacitor and/or transformer to your sound card output to get proper input to your transmitter's microphone circuit, as they often have DC bias voltage on them.
In the upper Left corner of the flow diagram, you will find a variable named "CALIBRATION". In theory, the correct setting for this should be 1.0 since all the demodulation is done digitally based on calculated ratios and should be accurate. If you find that checking against another deviation meter, or Bessel nulls disagrees with this, you can trim the reading by using this variable.
Within the limits of the Low Pass filtering, the measured single sine wave deviation should be accurate within 3% of the measured value. Note, however that for wideband measurements, the scope display has some limitations above 12 Khz and some aliasing may be seen.
The bandwidth filters specify the frequency at which their response is significantly reduced, so the response typically begins to drop at about 90% of the set value. This is the standard way to describe bandwidth.
The RMS trace displayed is for use as a loudness reference only and is highly waveform dependent.
There's a disabled "QT GUI sink" block included in the flow diagram used for debugging. For troubleshooting, it can be connected to inspect the signal anywhere. It takes a lot of screen space and CPU.
When connected to the output of the RTL-SDR source, it has sufficient resolution to observe Bessel Nulls for calibration if appropriately zoomed. The first two carrier nulls are useful because they occur at modulation indices of 2.4048 and 5.5200.
The Raspberry Pi operating system's audio system is sometimes highly unstable, particularly for a few months after the release of a new upgrade. This program requires that ALSA audio be operating properly, in particular that the ALSA default output device be working.
You can test if your ALSA system and it's default output device are working properly using the built in linux tool called "speaker-test". Typing this command from a terminal window should report if the device is working right and begin producing sound. If it does not, you should correct this problem before expecting Gnu Radio (or many other programs that produce sound) to work properly.
A workaround for some OS bugs might be to open the Audio Sink block in deviance.grc and explicitly select your ALSA device by putting its name (e.g. "plughw:CARD=vc4hdmi1,DEV=0") into it, rather than leaving it blank to selects the default ALSA device. You can find your device name by running "aplay -L".
FMDeviance may work with other radios that have Gnu Radio support, but I've only tested it with the official RTL-SDR.com version 3 and 4 dongles.
I've only seen one problem in moving FMDeviance between different platforms and versions of Gnu Radio. The various Low and High Pass Filter modules may report that they can't find DSP windowing options like "hamming" because over time the suffixes for the windowing option names in the popups have changed.
To fix this, open each complaining module and pop up the selector for the windowing option, then reselect the windowing option. This is easier than it sounds - the correct names are obvious variations of the incorrect ones.
I designed this experimental NBFM Deviation meter as an aid to operation and experimentation. It is not designed to be highly accurate or foolproof, or to be a high performance receiver. It works well when used as intended.
Be careful not to destroy your SDR receiver input by having the Radio and SDR antennas close together! HTs can produce quite high rf voltages in nearby equipment.
In some cases, you may need to use a power attenuator and/or sampling tap connecting the transmitter output and the SDR dongle to safely obtain a signal that is not leaking from earlier transmitter stages.
This software was made possible thanks to the excellent work done by the Gnu Radio team and the rest of the community!
Mario Vano AE0GL
