How to Calibrate SDR Signal-Level Measurements

[Pavel]

Short answer: Calibrate dBFS readings against a traceable or known signal source at each relevant frequency, gain, bandwidth and RF path. The result is a correction table with uncertainty; an uncalibrated waterfall level is not automatically dBm.

Before you start​

• Define authorized bands, observation period, minimum signal duration and the measurement you need before selecting hardware.
• Use shielding or conducted RF paths for active security tests whenever over-the-air transmission is unnecessary.

Step-by-step method​

1. Step 1: Warm the SDR and source, connect through known attenuation and keep the input safely inside the receiver's linear range. Record cable and adapter loss.
2. Step 2: Inject several known levels at each target frequency and record the SDR power measurement with fixed gain, FFT bandwidth and detector settings.
3. Step 3: Plot measured dBFS versus input dBm, identify the linear region and derive correction or regression. Exclude points near noise floor and compression.
4. Step 4: Repeat after reconnecting and at another temperature, then document combined uncertainty from source, attenuators, mismatch and receiver repeatability.

Concrete example​

If -60 dBm at the connector reads -42 dBFS, one point suggests an 18 dB offset, but multiple levels are required to prove linearity. A -30 dBm input reading only 5 dB higher may indicate compression.

How to judge the result​

Calibration is useful when known check levels are recovered within the stated uncertainty and the exact gain/frequency configuration is preserved with measurements.

What to record​

• Calibrate amplitude with a known source and record frequency-dependent correction; an SDR display is not automatically a traceable power meter.
• Store enough metadata to reproduce each observation: receiver, clock, antenna, gain, bandwidth, location precision and software configuration.
• For occupancy, define threshold, time resolution and treatment of noise before calculating percentages.

Common mistakes​

• Changing several hardware, software or RF variables at once, which removes the controlled comparison needed to identify the cause.
• Treating one autoscaled screenshot or one unusually good result as proof without recording the settings and repeating the test.
• Applying one correction from 100 MHz to several GHz ignores frequency-dependent front-end and cable response.

Final check​

Run a known controlled signal through the monitoring method and verify detection, calibration, timestamps and gaps. Keep authorization and data-handling records with the measurement.