Hi Chavdar,

> What I want to add is that in my opinion the 2nd order IMD is much 
> more important in direct conversion (DC) RX than in superhet RX. Not 
> only around half LO frequency, but any two input strong signals with 
> close frequencies will give a spur f1-f2 into the low frequency  
> baseband  of the DC  RX. This same IMD2 is responsible for direct AM 
> detection of strong broadcasting stations which sometimes is 
> difficult to be filtered out with input filters. For example I have 
> very often AM detection from a strong broadcasting station working at 
> 13.600 MHz when my DC RX is on 14 MHz band - very near to be filtered 
> effectively. 
> 
>    The 2nd order spurs should be removed theoretically when a balance 
> mixer is used but practically not more than 30 -40 dB additional 
> attenuation can be achieved. The balance depends very much from the 
> quality of the broadband transformers and parasitic capacitive 
> balance of the practical hardware. My own measurements show IP2 in 
> order of 55-65 dBm (@MDS:  -125 dBm)  for balanced  QSD mixers with 
> 74HC4052 switch.
> If  I have to rate the main problems in SDR with DC RX hardware they 
> are:
> 1. Insufficient broadband image rejection
> 2. IMD2 giving AM broadcast detection.
> Very rarely in normal use I have any IMD3 problems. So it will be 
> nice if we see more often detailed data for 2nd order response of the 
> SDR equipment.

There is another second order problem with these mixers. When the LO 
is set to e.g. 7.050 and the desired signal is at 7.060 (IF=10kHz)
a signal at 7.055 or at 7.045 will produce an IF of 5 kHz and its
second order harmonic would be on top of the desired signal.
I ran into this problem when designing the WSE RX2500 unit. 
The solution was to add feedback to make the current through 
the switch zero. 
http://www.sm5bsz.com/linuxdsp/rxiq/mixer.htm

It is a dramatic improvement for the "audio overtone" case.
I do not see any of f1-f2 coming out on the IF (audio) port, but I 
have no idea whether it would help against f1+f2 (half the
frequency) because the the RX2500 has a good RF filter that 
prevents signals far away from 2.5 MHz to reach the mixer.

73

Leif / SM5BSZ

Reply via email to