Search results

FFTDSP is a PC program which can detect weak radio signals in real time. Uses the PC's soundcard and advanced signal processing techniques, extracts and displays weak signals from the receiver audio.

PA3FWM's software defined radio (SDR) page documents his extensive hardware and software development efforts between 2004 and 2009. Initial experiments utilized a direct conversion receiver with 90-degree phase difference, feeding a PC soundcard at 48 kHz sample rate, covering 24 kHz of spectrum around a 7080.5 kHz local oscillator. This setup, similar to AC50G's QEX 2002 article, allowed for basic I/Q signal processing to distinguish signals above and below the LO frequency. Limitations included fixed crystal frequencies, 16-bit dynamic range, and narrow bandwidth. Subsequent hardware iterations aimed for enhanced performance, incorporating external 24-bit ADCs with 192 kHz sample rates, connected via 10 Mbit/s Ethernet. A **MC145170-based PLL** and programmable octave divider provided a 58 kHz to 30 MHz tuning range. The **Tayloe mixer** was employed, with differential outputs feeding a PCM1804 ADC. An ATmega32 microcontroller handled serial data conversion to Ethernet frames, though without CRC calculation due to processing constraints. Later designs integrated AD7760 2.5 Msamples/second ADCs and a Xilinx Spartan-3 FPGA, enabling direct reception of 0-1 MHz spectrum and eventually 2.5 MHz bandwidth across the shortwave spectrum. Software was refactored to use an initial 8192 non-windowed FFT for efficient high-bandwidth processing. The project culminated in a two-way QSO on 21 MHz using the developed hardware and software, demonstrating transmit capabilities with a D/A converter. The system exhibited a 2.5 MHz wide spectrum display and a zoomed 19 kHz display, capturing signals like ionospheric chirp sounders and RTTY contest activity. Challenges included noise leakage from digital circuitry and cooling for high-power dissipation components.