[sdiy] Weaver frequency shifter

[Eric Brombaugh]

On 4/27/22 17:41, Brian Willoughby wrote:
> Do these tuner stages at the beginning and end of the Weaver signal chain end up shifting the frequency band of the content?

Yes. The first tuner stage is used to create an analytic signal by doing 
a quadrature shift of the center of the audio band, resulting in a 
two-sided signal centered on DC. The quadrature LPF removes the unwanted 
image that's created in the process.

The tuner at the output shifts the analytic signal back up from DC the 
difference between the input shift and output shift is the total shift 
you end up hearing.

> Would this help keep all signals within the sample rate? Are these questions moot because of the potentials of under sampling?

This depends on how you plan out the bandwidths and tuner frequencies. 
In DSP systems the initial shift is often chosen to be Fs/4 which 
simplifies a lot of the math (the quadrature LO becomes a sequence of 
[1+j, -1+j, -1-j, 1-j] which is very easy to do). That also allows the 
quadrature LPFs to be implemented with half-band filters that can also 
simplify the math. Regardless, you have to keep the bandwidths within 
that allowed by Nyquist or suffer aliasing when the final real signal is 
created.