To maintain dynamic range its necessary that the diode ring mixer (and
surely FET mixers that switch like a diode ring) be terminated broad
band. Otherwise the undesired signals and output products (always at
least the sum and difference of the input spectrum and the LO) may be
reflected (especially from a narrow band IF filter) in such phase to
raise the voltage of some unwanted signal a the mixer or to decrease the
desired output. That destroys dynamic range on the mixer, sometimes
increasing conversion loss by 10 dB and dropping the top end intercept
as much as 40 dB. Hence the broadband push pull grounded gate stage
after the mixer. That stage makes a better mixer terminator (without
introducing noise or loosing signal) than a 10 dB attenuator which would
provide the same quality of mixer termination.
Yaesu's FT-767 VHF modules did the mixer with broad band termination and
they have a 10 dB better MDS and 30 dB better 3rd order intercept than
their FT-736 which came later. The VHF modules in the 736 connect the
diode ring mixer directly to the crystal filter.
Last winter I did a study for my own enlightenment of mixers and
especially mixers with lowered conversion loss. The properly terminated
diode ring runs 5 to 7 dB conversion loss. At VHF and up, if that
conversion loss (essentially also its NF) could be reduced to 4 dB, the
required RF stage gain for the RF stage to control the receiver NF is
significantly smaller, thus the maximum signal seen at the mixer is less
making the receiver intermod intercept greater for the same mixer
intermod limit.
I found literature from some decades ago concerning a mixer that had
image terminations that carefully reflected the unwanted mixer output
back to the diodes and that did reduce the conversion loss. The author
promised further articles and none appeared. Later articles and a book
on mixers that I did find showed that while that particular termination
did decrease mixer loss to 4 dB, it increased the mixer NF several dB
above the conversion loss and reduced the 3rd order intercept nearly 40
dB. So the concept has been abandoned.
I don't know what the mixer termination requirements for a simple source
injected FET or push pull FETs might be. Few circuits seem to take the
care for matching or reflecting the image outputs and they may be
trading circuit simplicity for performance. I suspect that some rely on
the relative unilateral drain output not letting an output image
reflection getting back to the gate and so don't get remixed like they
do in the multilateral diode ring mixer.
For quiet band work, the Corsair (II) has a failing. There's no image
reject filter between the RF stage and the mixer. So the RF stage noise
at the image frequency gets converted to the IF. At HF that's not so
much of an issue, antenna noise tending to be so much greater. There
also is little shielding to prevent odd signal paths (and TVI creation
to the shack TV from harmonics of the mixed VFO) coupling. The Corsair
uses the Anzac transformer fed back bipolar stage as RF stage and first
IF stage. The receiver design books written by Ulrich Rohde (recommended
reading!) specify that circuit combination as the best possible. There's
a patent liability though and I suspect the multiple grounded gate (the
large gate TI transistor specified in European designs of that era not
being made any more) JFET does quite well for NF and broad band gain
with very good handling of large signals. As far as I can tell only the
Corsair and Corsair II use that transformer feedback circuit. Everything
since uses the grounded gate circuit. The grounded gate circuit is less
expensive (no complex three winding transformer to create and wire by
hand) in parts and assembly.
73, Jerry, K0CQ
--
Entire content copyright Dr. Gerald N. Johnson, electrical engineer.
Reproduction by permission only.
|