On 12/02/2014 11:29 AM, Michael Black wrote:
On Mon, 1 Dec 2014, dave wrote:

Direct conversion is all the rage these days. Or Direct conversion to
I/Q quadrature directly into a DSP. A crystal radio is direct
conversion so I guess we are back where we started (except now we have
the DSP).

No, a crystal radio is not direct conversion, not in the sense that we
came to know it.

WIth a crystal radio, the front end selectivity (and there usually isn't
much of it) is what determines selectivity. In reality you could have two
or more local signals, all about the same level, and they'd ride through
that front end filter, and land at audio. There is nothing you can do
then to get rid of the unwanted signals, except improve the front end
selectivity. The carriers of each station are mixing with the sidebands
and putting signals into the "baseband", ie audio. Since they exist in
the same audio space, they can't be eliminated.

A direct conversion radio actually uses a local oscillator to convert a
signal down to audio. That local oscillator will be much stronger than
any of the incoming signals, so it can dominate. That means that the
front end selecitivty doesnt' matter so much. You have a station on
1000KHz, and set the BFO there. The BFO in effect "captures" things, so
that signal on 1000KHz will dominate. If there's a signal at 1010KHz
that's also strong (not likely since adjacent channels aren't allocated
locally), the BFO at 1000KHz will convert the adjacent signal to "audio"
too, except since it's 10KHz away from the wanted signal, it will start
at 10KHz rather than 0KHz. Above audio range for many people, if it's a
bother one can put a low pass filter after the mixer so the unwanted
signal will be well attenuated. Same thing happens on the audio image,
the unwanted signal at 990KHz will be out of audio range once converted
to audio. (Note that this works since AM broadcast channels are 10KHz
apart, if the signal was closer, it would translate to audible audio and
be a nuisance, and you'd never be able to get rid of it with a simple
mixer and audio filter).

And it goes on, up and down from 1000KHz. The front end selectivity has
little effect, audio selectivity can knock out unwanted signals. There
was that wave of "build your own AM receiver from a PLL" articles forty
years ago, and they all had little or no front end selectivity, yet were
selective because of this. A crystal radio would receive multiple
signals with little or no front end selectivity.

Michael

If you have an LO on freq how much gain can there be in the front end?
I have heard these on YouTube and they compete with Regenerative as the
most annoying radios ever built. Here's a chip that uses one analog
mixer at 70 mHz. There are others for TV tuners that span a GHz or more.
You can directly inject RF voltage of virtually any frequency into a
chip like this and get intelligence out, with a minimal parts count. Are
you saying the tiny radios today use monodyne architecture?