What SDR to Buy for SIGINT - and the One Spec That Mattered Most

After a few weeks of rooftop SIGINT with a single cheap dongle, I knew exactly where it fell short. This is the buyer's guide I wish I had started with - grounded in real limitations I hit, not spec-sheet worship.

01The one spec that mattered: ADC bit depth¶

My RTL-SDR is 8-bit. That single fact caused two of the biggest problems in this whole series:

• The faint LoRa that would not decode - 8 bits of dynamic range cannot pull a weak signal out from next to a strong one.
• The L-band spur forest in the spectrum sweep - 8 bits plus a +41 dB DVB-T monster in band produces intermodulation artefacts that masquerade as signals.

Dynamic range, not frequency range, is what separates "I can see there is something there" from "I can decode it." For weak-signal work, bit depth beats bandwidth. That reframed my whole shopping list.

02The three radios¶

RTL-SDR stays useful. It is cheap enough to dedicate to one permanent job - mine sits parked on the meter band as a 24/7 telemetry node. Do not throw it out; specialise it.

SDRplay RSP1B is the sensitivity upgrade. 14 bits is the answer to the weak-LoRa problem. Pair it with a band filter and a low-noise amplifier at the antenna and those marginal signals climb over the decode threshold. RX only.

HackRF Pro is the explorer. 100 kHz to 6 GHz reaches everything the others cannot - 2.4 GHz, drones, the upper ISM bands - and it transmits. The new Pro adds a built-in TCXO (frequency stability, which matters for LoRa and any precise work) and a 16-bit precision mode that finally addresses the original HackRF's weak-signal weakness.

03Do not buy three of the same thing¶

The mistake is buying three general-purpose radios. The win is specialisation:

• RTL-SDR - permanent meter/telemetry node, parked on one band, cheap and disposable.
• RSP1B - the precision receiver for anything weak or narrow, with an LNA and band filter.
• HackRF Pro - wideband reconnaissance across 6 GHz, plus transmit for antenna testing and your own device work.

Three lenses on the spectrum, not three identical eyes.

04The front end matters more than the radio¶

For weak signals, a band-specific SAW filter plus LNA at the antenna often buys you more than upgrading the radio. The filter rejects the strong out-of-band signals (that DVB-T transmitter) that desensitise a wideband front end; the LNA sets the noise figure. Spend here before you spend on a fourth SDR.

And if you are going to build or tune antennas, get a NanoVNA (around 350 PLN). A HackRF can measure radiation patterns and compare antennas in the air, but it is a poor reflectometer. The NanoVNA measures SWR, impedance, and resonance directly - it is the right tool for actually building an antenna, and the two complement each other perfectly.

05On transmitting¶

The HackRF transmits, which opens antenna testing, your own device work, and radar experiments. It also opens a fast path to breaking the law. Transmit power in the 868 ISM band is capped (typically 25 mW, 1% duty cycle); outside ISM you need a licence. A power amplifier will exceed those limits trivially. The rule is simple: transmit into a dummy load or a shielded setup, stay in ISM at legal power, or hold an amateur licence. Receiving is free; transmitting is a responsibility.

06The shortlist¶

If I were starting over: keep the RTL-SDR for grunt work, add an RSP1B plus an 868 SAW+LNA for real decoding, and a HackRF Pro for the 6 GHz frontier and transmit experiments. That trio, plus a NanoVNA, is a complete SIGINT and antenna bench for well under the price of one professional radio.

This wraps the hardware thread. The series so far has been about receiving and understanding what is in the air. The next chapter is bigger: turning the firehose of captured signals into a machine-learning pipeline that classifies, detects anomalies, and learns the RF environment on its own.

Receiving is legal in the EU. Transmitting is regulated - stay in ISM at legal power or get licensed.