[hpsdr] Odyessey-Siren Rev B.

[Graham Haddock]

 >>Graham Haddock wrote:
 >> 1.) In reading the schematics for the Siren Rev. B board, and 
comparisons
 >> to the Rev A design, I note that the ISD (QSD) is a 180 degree switch as
 >> compared to the 90 degree switch used in the Rev. A board.
 >>
 >> As I understand the operation of these switching detector/mixers,
 >> this will result in an extra 2 dB decrease in gain/NF or so in the
 >> conversion process.


 > Robert McGwier wrote:
 > I don't believe this is correct.  I believe the two switches are 
operated
 > in quadrature and they are inverted to achieve the opposite polarity.
 >
 > The I line is turned on at both 0 and 180 and the input to the Op Amps
 > is inverted by the switches at these points.   The Q line is turned on
 > at 90 and 270 and the input to the Q line Op Amp is inverted by its 
switch. 
 > Did I get this wrong?


Hi Bob:
Your comments above are correct, and the Rev B circuit will work.

BUT...

Rev A. used an FST3253 dual 4:1 mux, which with quadrature clock input
should do something like

Positive input routed to            Negative input routed to
I-Pos   0 to 90 degrees            I-Neg  0 to 90 degrees
Q-Pos  90 to 180 degrees           Q-Neq  90 to 180 degrees
I-Neg  180 to 270 degrees          I-Pos  180 to 270 degrees
Q-Neg  270 to 360 degrees          Q-Pos  270 to 360 degrees

At any given time, a single OpAmp-integrator is being driven for just
90 degrees of clock phase.


Rev. B  uses a pair of 74LVC2G53  SPDT switches. One driven by I clock,
one by Q clock so input routed to
I-Pos   0 to 180 degrees
I-Neg   180 to 360 degrees
simultaneously the input is split and also routed to 
Q-Pos   90 to 270 degrees
Q-Neg  270 to 90 degrees

By integrating across 180 degrees in Rev B, versus just 90 degrees in
Rev.A you are incurring an additional 2 dB reduction in gain and increase
in NF in the Rev B. mixer, assuming the implementation losses of both
circuits are the same.

If you wanted to simplify the receive circuit, using just one 4:1 mux
in a single balanced circuit would probably work just as well, with
possibly lower implementation losses.

Best regards, --- Graham / KE9H

==


 1166583075.0