Turn a WiFi 7 Router into a 320Msps Software Defined Radio Receiver

Original Article ↗ Hacker News Discussion ↗

A new hack repurposes a sub‑$100 Wi‑Fi 7 router into a wideband software-defined radio (SDR) receiver, delivering 14‑bit IQ samples at up to 320 mega‑samples per second around 5 GHz — performance previously reserved for far more expensive lab gear. Commenters highlight the potential for affordable amateur radio, satellite, and RF analysis work above 2 GHz, while comparing it to existing RTL‑SDR and HackRF options. Much of the debate centers on the role of large language models in reverse‑engineering the hardware, with some praising the dramatically lowered entry barrier and others questioning documentation quality, understanding of the firmware, and future vendor lock‑downs.

Technical achievement

  • Router’s Wi‑Fi 7 chipset is repurposed as a wideband SDR receiver: ~14‑bit IQ, up to 320 Msps, around 5 GHz, data streamed over 10 GbE to a host for live spectrum/waterfall.
  • This is done by abusing the modem’s factory/test mode to read out raw IQ at full rate.
  • Biggest engineering challenges (per thread): keeping testmode running continuously without gaps, and extracting the “firehose” of data reliably; also fuzzing undocumented configurations (e.g., higher bandwidth on 2.4 GHz than vendor tables allow).

Hardware significance & cost

  • Several commenters call it a major jump: similar performance previously required 4–5‑figure lab gear; now sub‑$100 hardware does 320 Msps at >2 GHz.
  • Noted that high‑speed standalone ADCs with comparable specs are very expensive; Wi‑Fi chips deliver multiple such ADCs cheaply due to volume production.
  • Some see it as linear progress rather than a revolution, but most agree it’s a big deal for 13 cm / 5 cm amateur bands and higher‑frequency SDR work.

Role of AI / reverse engineering

  • Project heavily leveraged a large language model to understand firmware, registers, and bring‑up sequences.
  • One side sees this as an “existence proof” that LLM‑assisted reverse engineering dramatically lowers the barrier that once required deep RF/embedded expertise and lots of time.
  • Critics argue parts of the README read like keyword “word salad,” suggesting the author doesn’t fully understand the generated text, likening it to 1990s script‑kiddie behavior.
  • Others counter that this is a reverse‑engineered proof‑of‑concept, not vendor‑grade documentation, and that expecting polished chip docs is unrealistic.

Comparison to other SDRs and related hacks

  • Placed in lineage with RTL‑SDR (cheap DVB‑T dongles) and ESP32‑based SDRs; those opened low‑cost SDR at lower bandwidths, this opens much higher‑frequency/high‑rate territory.
  • Discussion of ESP‑SDR variants (including 5 GHz C5 chips) and hints at TX being technically possible but often intentionally disabled to avoid spectrum abuse/regulatory issues.
  • Some note this router platform is more niche than ESP32 or common USB Wi‑Fi dongles; broad adoption may depend on similar hacks for widely available chips.

Capabilities, limitations, and open questions

  • Currently RX‑only; several people speculate about future TX and phased‑array / beamforming uses, while others jokingly “shush” this due to regulatory concerns.
  • Questions raised about SNR, dynamic range, IQ‑imbalance calibration, and multi‑channel/MIMO support; these details are not yet documented in depth and remain unclear.
  • One commenter claims its usable bandwidth is too narrow for “most SDR use cases”; others strongly disagree, pointing out it already functions as a general‑purpose SDR for many applications.
  • Suggested uses include ham radio on higher bands, satellite links (with external front‑ends), DIY service monitors up to 5 GHz, and general signal analysis.

Wider SDR usage & education themes

  • Multiple side threads describe using cheap SDRs for:
    • ADS‑B/aircraft tracking, weather satellites, pager decoding, consumer sensor monitoring (433 MHz, 900 MHz), and trunked radio systems.
  • Several people report using LLMs as “professors” and assistants for SDR learning, signal classification, pipeline scripting, and dashboarding, emphasizing how much this accelerates entry into RF.