[hpsdr] Signal Paths for QSD

[Jim Miller]

I've been giving the QSD signal path some more thought. I understand that 
motivation for active integrators (with or without a twist) is the effect of 
a passive integrator's impedence on input bandpass filters.

I would think that solution to that would be to follow a bandpass filter 
with a suitable, e.g. pushpull Norton amp, followed by a pad. The pad would 
limit impedence variations to the amp and thus to the BPF. It would also 
limit impedence variations to the QSD filter.

The combination of the pad, amp and BPF would also limit radiation from the 
VFO to the antenna.

The output capacitors of the QSD would then be followed by suitable high 
impedence butters with the modest remaining gain necessary to match to the 
input range and impedence of the differential A/D converter. The requirement 
for fancy slewrates or gainbandwidth would then be eliminated for the 
buffers making their selection considerably simpler. I suspect the critical 
component count would actually be a wash or even lower for this 
configuration.

Since a pushpull Norton can have a significant gain without compromising IP3 
or IP2 the pad could be pretty lossy perhaps as much as 10db for a return 
loss of 20db. That could still leave a 6db net gain for the combination of 
amp and pad. Wouldn't that be enough?

What have I missed? (again...)

73

jim ab3cv