The analog interface (ADC and gain control) is now completely verified, and a bandpass polyphase FIR filter, with decimation factor 8, was added to the project to reduce the sampling rate downstream and filter out-of-band noise. As the ADC is used with a sampling rate of 768 kHz, this produces an output rate of 96 kHz, with the Nyquist frequency comfortably above the minimum for the symbol frequencies used. The filter has passband 19-29 kHz, producing a 16-bit output from the ADC data. We used the buffering module (a BRAM with some I/O timing control) to see parts of the internal waveforms on the PC, and the entire chain up to the filter output is working correctly, both with direct function generator input and with the on-board analog filter.

We also started looking at the test data from last week, but the results (in Simulink) have been poor. This may be because of the channel coherence time, which seems to be very large. We will investigate this issue further next week.