RadioBanter - View Single Post

[email protected] · September 8th 06, 12:58 AM posted to rec.radio.amateur.homebrew

On Thu, 07 Sep 2006 21:11:03 -0000, (Dave Platt)
wrote:

In article ,
i3hev, mario held wrote:

Last, but not least, a simple DCR is inherently a DSB receiver, lest you
implement it with rather complex mixers - which add much to the circuit
total complexity.

As I understand it, in order to do SSB via direct conversion, you must
use a modern version of the old IQ phasing technique. This requires
doing the zero-IF mixdown twice, on two different versions of the
signal (with a uniform 90-degree phase shift being applied to one copy
of the RF or LO signal). The resulting two baseband outputs are then
processed (with a further 90-degree phase shift being applied to one
of them) and carefully mixed. This results in reinforcement of the
desired sideband and suppression of the other.

The circuitry needed to apply the necessary phase shifts is not
trivial (if you want enough accuracy to deliver acceptable
opposite-sideband rejection), and is not necessarily simpler or less
expensive than the filtering and extra stage of mixing done in a
traditional IF-based SSB receiver or transmitter.

It is trivial, three stages of all pass using opamps for both the I
and Q paths and sum or difference the result. Using 1% parts and
decent opamps opposite sideband suppression of 45db is easy
and 50db or better attainable. Read Breeds and KK7Bs works.

However to build a good Image reject DCR and a good superhet
is not trivial when you consider shielding and mechanical issues
to get the best results.

These days, of course, you can apply the phase shift by converting the
two baseband signals to digital format, and implementing the final
90-degree phase shift via a digital FIR all-pass filter. This of
course requires your design to have a pair of high-linearity ADCs, a
DSP, and a DAC to reproduce the final (mixed) signal.

This is more reproduceable and works very well.

The "holy Grail" these days seems to be a direct-from-RF system, in
which the RF signal is _directly_ sampled (at a ferociously-high
sampling rate), and all of the phase shifts and downconversion and
mixing are done digitally. RF-grade ADCs with the necessary linearity
and speed aren't particularly inexpensive, especially if you need your
system to deliver a very high dynamic range which can work properly
even in the presence of strong in-band or near-band interferers.

Usually this technique is the "tailend" of a more conventional
Superhet so the RF detection is done at some low IF rather than
baseband. Typically 12-50khz for that IF. Problem is thes end up
being triple (or more) conversion reciever with the need for better
than average shielding or the birdies are prolific.

Allison