--- In [email protected], "Dave Wade" <[EMAIL PROTECTED]> wrote: > > Having read these I was wondering why these folks use the "divide/4" approach when there are now DDS chips available very cheaply, > that provide I/Q outputs like the AD9851. >
The problem with using a DDS with I/Q outputs (the 9854, NOT the 9851) is that, while the DDS chip actually produces outputs at exactly 90 degrees, after the DDS you have to place anti-alias filters and amplifiers, all still in the analog domain. Those filters and amplifiers cannot keep the 90 degrees relationship exact. Component value tolerances and aging cause that value to change. Remember that to get 80 dB image rejection, you need a pretty impressive phase/amplitude match: approximately arctan(0.0001) (0.005 degrees), and for 40 dB you need about 0.6 degree. So that's why a solution based on a Johnson counter (the divide by 4) is usually better, as it does not rely on any matching of analog components. 73 Alberto I2PHD
