This presentation by Hans provides a retrospective on his 25 years in Software Defined Radio (SDR) development, highlighting his contributions to both receiver and transmitter technologies. He discusses the evolution of his designs, from early direct conversion sampling receivers to the standalone ADT 200R shortwave transceiver, renowned for its superior audio. Key to this "better sound" were three innovations: optimized, "roundish" audio filters for intelligibility, a novel noise blanker and automatic gain control (AGC) system employing a moving trigger level and a unique feed-forward, preemptive approach for handling impulse noise and strong signal jumps. Hans then details an optimized SDR frontend design, emphasizing preselection and antialiasing filters to manage noise and dynamic range. For transmitters, he introduces a power enhancement algorithm that linearly scales audio signals between zero crossings to increase perceived power without significant distortion, and a new real-time polar linearization (predistortion) technique implemented entirely in digital FPGA logic, compensating for both amplitude and phase distortions, including memory effects, achieving over 80 dBc suppression in simulations.
Summarizing the Future of SDR: Hans's Innovations
0:01:47 25 Years of SDR Development: Hans reflects on 25 years since his first article on "software radio," predating the widespread use of the SDR term.
0:02:44 Early Direct Conversion Receiver: In 2002, he presented his first direct conversion sampling receiver, a simple but foundational step.
0:03:06 Standalone Shortwave Transceiver (ADT 200R): Later, he developed a standalone shortwave transceiver (ADT 200R), producing around 100 units sold worldwide, noted for its superior sound quality.
0:04:05 Secrets to Superior Sound (ADT 200R): Hans reveals three key measures for the ADT 200R's audio performance:
0:04:19 Optimized Audio Filters: Filters are designed with a "roundish" response, not "brickwall," to avoid metallic sound, and feature a specific product of lower and upper cutoff frequencies (e.g., 300Hz lower, 3000Hz upper) for best intelligibility.
0:05:10 Innovative Noise Blanker: An impulse shaper algorithm defines a moving trigger level that adjusts to peak signal levels and controls the signal's increase/decrease slope.
0:05:18 Novel Automatic Gain Control (AGC): Features a feed-forward control algorithm with a "preemptive" part that anticipates strong signal jumps (e.g., 100dB) to prevent overshoots and maintain a quiet, smooth response.
0:12:18 Optimized Frontend Design for SDR: Key components for an optimal SDR frontend include:
0:12:42 Narrow Preselector: To prevent overload of the sensitive Analog-to-Digital Converter (ADC).
0:12:58 Preamplifier: A specific, low-noise preamplifier (e.g., Mini-Circuits PGA-103+) with low noise figure and strong IP3 is crucial.
0:14:11 Antialiasing Filter: Essential for cutting signals beyond the Nyquist window (e.g., 30MHz) and preventing higher-frequency noise from folding back into the first Nyquist window, which would increase overall noise.
0:16:35 Frontend Performance: With this optimized frontend, the system achieves excellent performance with an IP3 of +4.5/+5.5 dBm, a noise figure of 13.7/23.7 dB, and a dynamic range of 130 dB.
0:19:33 All-Diode Switched Preselector Filters: To manage the large number of narrowband filters (approx. 150 from 1.8-30 MHz), Hans developed a board with fully diode-switched capacitors and varactor diodes for fine-tuning, with compensation to maintain good intermodulation performance (IP3).
0:21:10 Power Enhancer: A small algorithm shifts the audio signal through a circular buffer. Signals exceeding a certain threshold between zero crossings are scaled down linearly. This method differs from conventional clippers by producing minimal distortion, particularly around zero crossings where there's less energy, resulting in a perceived increase of up to 6dB in signal strength without significant out-of-band distortion.
0:24:32 Polar Linearization (Predistortion): A new, real-time digital predistortion technique implemented in an FPGA. It feeds back the transmitted signal, uses a phase and envelope detector, and applies adaptive amplitude and phase compensation to the transmit path.
0:25:15 Low Latency: Critical for real-time operation, requiring delays in the order of microseconds.
0:27:54 Memory Effect Compensation: The system also compensates for the memory effects of the power amplifier.
0:28:24 All-Digital Implementation: Performed entirely in the digital section of an FPGA, avoiding analog complexities.
0:29:41 Simulation Results: Shows significant suppression of intermodulation products, with simulations demonstrating over 80 dBc suppression (compared to 33 dBc without linearization).
0:30:40 Implementation using Xilinx System Generator: The complex digital predistortion was implemented using Xilinx System Generator, which allows block-level simulation and automatic VHDL code generation, reducing manual coding.
0:33:32 Encouragement for Future SDR Development: Hans concludes by encouraging continued innovation in SDR technology.
0:34:27 Q&A - Filter Shapes and Perception: Discusses how "roundish" filters, despite potentially having higher overshoot in step response, contribute to better intelligibility than brickwall filters for human perception.
0:35:41 Q&A - DPD Robustness: Confirms the digital predistortion (DPD) approach is robust, based on theory and experimental work.
0:36:55 Q&A - Memory Effect Model: Explains that the memory effect model for the power amplifier considers thermal effects due to the power dissipated.
0:37:40 Q&A - Power Enhancer User Experience: A user of the ADT 200R praises the power enhancer, noting that receiving stations perceive a significant increase in power (like switching on an amplifier) when it's activated, even at 6dB.
0:38:29 Q&A - Power Enhancer vs. CESB: Hans clarifies that his power enhancement method is different from Controlled Envelope SSB (CESB), which primarily aims to increase amplifier efficiency. His method limits the audio signal in a way that minimizes in-band intermodulation, unlike conventional compression or limiting followed by filtering.