On Fri, Jun 22, 2018 at 10:38 pm, Ashhar Farhan wrote:

It may be improper level of LO injection. On the transmit side, the mixer has _very_ strong input termination and LO termination. It will be interesting to study this. Mixers can provide endless amount of fascination or frustration depending on which side of the RF probe you are.

 

Yes they are a conundrum in three ports.  However I did explore the path of the TIA amp now working well 
and terminated it with a 6db pad... no improvement at all.  Same for the output to the filter 6db pad and
again no improvement.  In both cases there was a shift in levels of 6db and 12DB (ran both!) and no
difference in spur production.    The only thing that seems to impact the amplitude of the spur is the
port to port isolation more so than termination.  I believe it to be a alias image as it has the same
relationships.  It only shows when the relationship is such that its below filter cutoff.  

I replicated it at 45/73mhz and for experiment I did a 9mhz if, 23mhz LO  and without filtering I could
see the desired 14 and also 5mhz spur!  Its relationship is output frequency 14mhz minus IF and its
about 38db down from the desired 14mhz (two different ZLW-2 connectorized mixers).  I never noticed
a mixer spreadsheet I have from somewhere predicts it.  Likely reason I never noticed it was most
applications there is fairly strong filtering for the desired. 

I checked and it doesn't happen to the same extent with analog multiplier mixers (gilbert  cell) like
sa602 as they are not bidirectional (very high input to output isolation).

Nyquist was right.  Its the sampling behaviour of switching mixers.

Allison

 

 

. . .