[hpsdr] Differential reconstruction filter help

[Howard Long]

Folks

I am trying to come up with a low chip count high performance differential
reconstruction filter for a Cirrus CS42448. The data sheet comes up with the
following example which presents a single ended result:

http://www.g6lvb.com/images/cs42448DacRecon.JPG 

The point of this seemingly rather esoteric design is to present a constant
in-band impedance. Now, I could just replace the single-ended opamp with a
differential one, and indeed I've tried just that, but perceived wisdom
seems to be that the feedback arms of a differential opamp should be
symmetrical to reduce second harmonic distortion.

Now what I could do is simply take this circuit as-is and buffer with a
unity gain differential op amp, but then this increases my chip count and
thus current consumption (I am trying to keep power down to a minimum and
have limited board space). In addition (although a minor point for this
application) I loose the benefit of common mode rejection by passing through
a single-ended interim stage.

Another option is to simply buffer the DAC outputs first, thus presenting a
constant impedance to the DAC, and then use a simpler differnetial
reconstruction filter design. Again though, I'm doubling the chip count.

My other option is to simply stop thinking so hard, put a much simpler
single differential opamp stage LPF and correct for any impedance artifacts
in software. The nature of the application is that in-band spurii in the
region of 40dB down will probably be acceptable.

Are there any opamp design experts out there who can suggest a fully
differential single (8 pin) design here?

Many thanks, Howard G6LVB


 1198753243.0