Yes, it won't work. You are mixing down to DC, and you have lost the ability to sense phase. For example, if you read an amplitude of 0 on 'B', you don't know whether this is due to a large attenuation between Sig Out and B, or whether it's due to a 90 degree phase shift between Sig Out and B. Try a quadrature downconverter:
.. To PC .. * .. | .. ------- .. ------ |FT245BM| .. ch A ----- | | | | .. o-+--|SA602|---| 2kHz |------. |USB I/O| .. | |mixer| | LPF | | ------- .. | ----- ------ | | .. | | LO | | .. | | ------ ------- .. | '----. |24bit | | AVR | .. | | | ADC |---| | .. | | |stereo| | uC | .. | | ------ ------- .. | | ------ | | .. | ----- | | | | | .. '--|SA602|-----| 2kHz |----' | .. |mixer| | | LPF | | .. ----- | ------ | .. | | | .. | | | .. | | ------ ------ | .. | f1| | | |Quad. | | .. | '--| 20MHz|---| DDS |-----o .. | SIN | LPF | | | | .. | ------ ------ | .. | | | .. | ------ | .. | | | | .. '-------| 20MHz|-----' .. COS | LPF | .. ------ .
You will need one of these for chan A, one for chan B. You can get by with one if you only want to do 'through' measurements.
I recommend using a single DDS for both stimulus generation and downconversion LO. This avoids problems with phase shifts between the two DDSs in your first solution.
Regards, Allan