> Robert Sands <k7vora...@gmail.com> wrote: > > Is the processor in the K4 fast or big enough to handle the demand of the > anticipated predistortion algorithm?
Yes. Modulation and demodulation is handled by a dedicated, very fast floating-point DSP with various on-chip accelerators. The panadapters run from separate hardware, so the DSP has plenty of cycles to burn, especially in transmit mode. > Is this predistortion algorithm a proprietary matter or is it constructed de > novo from existing knowledge? There are three parts to this. First you need the appropriate hardware paths, which we already have in place: RF sampling of the reference waveform as well as the internal or external transmit signal. Second, the DSP has to apply complex curve fitting at baseband. This is typically done with several terms of a Volterra series, etc. Finally you need an algorithm for adaptive convergence on the ideal coefficients. This can range from LMS (least-mean-squares) to back-propagation (neural net). This is the "fun" (time consuming) part of the R&D, with tradeoffs in performance vs. processing time vs. staff availability. Given the multitude of distractions I can't say exactly how long this will take. It's likely that a simpler version of the algorithm will be available initially, allowing for early evaluation. 73, Wayne N6KR ______________________________________________________________ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html Message delivered to arch...@mail-archive.com
