HackRF SDR Skill

Operate the HackRF One software-defined radio for scanning, capture, analysis, and demodulation.

Prerequisites

• HackRF One connected via USB
• hackrf_sweep and hackrf_transfer CLI tools installed
• Python 3 with numpy, scipy, matplotlib

Verify: hackrf_info should show device info.

Workflow

1. Scan (Frequency Sweep)

Use hackrf_sweep to discover signals in a frequency range:

hackrf_sweep -f \x3Cstart_mhz>:\x3Cend_mhz> -l 40 -g 40 -w \x3Cbin_hz> -N \x3Csweeps>

• -l 40: LNA gain 40 dB (max for weak signals)
• -g 40: VGA gain 40 dB
• -w 1000000: 1 MHz bins for wide scans, 50000 for detailed
• -N 10: Number of sweeps (more = better averaging)

Quick overview: -w 1000000 -N 5. Detailed: -w 50000 -N 20.

2. Waterfall Visualization

Run the waterfall script for a combined spectrum + waterfall plot:

python3 scripts/waterfall.py --start \x3Cstart_mhz> --end \x3Cend_mhz> --bin \x3Chz> --sweeps \x3Cn> --outdir \x3Cdir>

Or from saved sweep data: python3 scripts/waterfall.py --input \x3Cfile> --outdir \x3Cdir>

Outputs a PNG and prints detected signal groups with center/bandwidth/power.

3. IQ Capture

Record raw IQ samples for detailed analysis:

hackrf_transfer -f \x3Cfreq_hz> -s \x3Csps> -l 40 -g 40 -n \x3Csamples> -r /tmp/iq_capture.raw

• Sample rate (-s): 10M for wide, 2M for narrow signals
• Duration: -n = sample_rate x seconds (e.g., 10M x 10s = 100000000)
• Always capture with enough BW to include the signal + margin

Always delete IQ captures after analysis — they are large (~200 MB for 10s at 10 MSPS).

4. Signal Analysis

Analyze captured IQ data for modulation type, bandwidth, SNR:

python3 scripts/analyze.py /tmp/iq_capture.raw --freq \x3Ccenter_hz> --rate \x3Csps> --outdir \x3Cdir>

The script:

• Computes PSD and spectrogram
• Detects peak frequency and SNR
• Measures 3dB and 10dB bandwidth
• Classifies modulation (CW, AM, FM, PSK, QAM, etc.)
• Detects pulsed vs continuous signals
• Finds secondary peaks
• Generates a 6-panel analysis PNG

5. Demodulation

Demodulate audio from IQ captures:

python3 scripts/demod.py /tmp/iq_capture.raw --mode \x3Cfm|am|usb|lsb> --offset \x3Chz> --out \x3Cfile.wav>

• --offset: signal frequency offset from center (from analyze.py output)
• Output: WAV at 48 kHz

6. Cleanup

After analysis, delete IQ capture files:

rm -f /tmp/iq_capture.raw

Tips

• Start with a wide scan (100 MHz), then narrow down to signals of interest
• Use -l 40 -g 40 for maximum sensitivity; reduce if signals saturate
• HackRF int8 format: 2 bytes per sample (I + Q), so 10 MSPS ~ 20 MB/s
• For signals near DC, offset tune by 1-2 MHz to avoid DC spike
• The peak frequency from analyze.py is the offset from center -- add to center freq for absolute freq
• Reference: references/frequency_bands.md for band allocations and modulation classification