On Fri, 12 Jan 2007 05:09:07 GMT, "numeric" wrote:


My objective is to homebrew a 10khz to 60Mhz software defined radio; one
that uses the power of the PC for all modulation and demodulation functions.
There are several references for a direct conversion receiver that uses a
sample and hold detector or Tayloe detector which seems to provide
remarkable performance. If possible I would like to use this detector; it
embraces the KISS approach ;-). The problem I find however, is that the
detector responds to odd harmonics of the receiving frequency, for a doubly
balanced configuration. ;-(

clip

After
all, the switch function is not linear. A Fourier analysis of the switch
pulse closely agrees with the detected result; the third harmonic is about
10db lower then fundamental. Does anyone have experience with this type of
detector?

The odd harmonic response is typical for switching mixers.

If you want to avoid the odd harmonic response, you should either
filter out before the mixer (e.g. octave filters as used in many
receivers) or use a true linear four quadrant multiplier and feed it
with a pure sinuous local oscillator signal.

However, the noise and distortion performance for a practical linear
four quadrant multiplier may be worse than a switching filter with
octave preselector filtering.

Anyway, some preselector filtering would anyway help the mixer handle
the input signal power and use different attenuators/amplifiers for
different bands. For VLF/LF/MF and lower HF you might need some
attenuation, while for the upper-HF and lower VHF, you might need a
preamplifier.

If direct conversion is used, quite a lot of AF amplification may be
needed, so for the upper bands, it would make sense to use some
RF-amplification ahead of the mixer to distribute the gain between RF
and AF frequencies.

I would suggest using some switchable band filters and a programmable
attenuator if you really want to make the receiver usable and not just
demonstrate that it can tune the specified frequency range :-).

Paul OH3LWR