Working with sufficiently-powerful Cuda-based GPUs, the DDC option allows on-the-fly baseband signal extraction, and is offered as an alternative to FPGA-based DDC and its associated programming.
“The GPU, which can have thousands of cores working in parallel, then allows processing software to be created using C/C++,” according to Spectrum. “This makes for a much easier DDC implementation as customisation can be made with normal programming skills.”
Spectrum Instrumentation M5i series digitiser streaming at 12.8Gbyte/s with on-the-fly GPU down-conversion – The down converted (green) signal almost 10dB of signal to noise improvement. Display is though Spectrum’s SBench 6 software.
In a demonstration, a M5i.3337-x16 digitiser card (above photo) sampled a 702MHz signal at 6.4Gsample/s, and the result streamed into a 6,144 core Nvidia RTX A4000 GPU over PCIe at 12.8Gbyte/s.
The GPU mixed the data with a digitally-synthesised complex sinusoid, applied a moving average, decimated the data by 512x, FIR filter it and then re-scaled the result for storage in PC memory. “The conversion dramatically reduces the resultant data set while also improving signal quality and measurement accuracy”, claimed Spectrum.
The DDC feature is now part of the Spectrum’s SCAPP (Spectrum’s Cuda access for parallel processing) software.
Spectrum Instrumentation was founded in 1989 and it based near Hamburg. It makes its digitisers in Germany.