DDS differential to single ended conversion

[SArepairman]

http://www.analog.com/static/imported-files/data_sheets/AD9106.pdf

I am using a 150 MHz clock to generate a audio signal.

My route was

filter each rail individually (ferrite bead - resistor - grounded capacitor) -> common capacitor between rails -> differential op amp -> band pass filter op amp (~10KHz)

The datasheet only says to use a transformer, I would rather use a amplifier because of higher CMRR and the ability to gain (i'd prefer not to gain in my band pass filter).
In another datasheet that recommended this method there were pullup resistors (right before the signal entered the passive filter) on each rail, going to AVDD. I don't know if I should include that.

I also don't know what to set my passive filter too, the idea to filter the signal of HF crap that will get rectified in the op-amp and couple through it and whatnot. I was thinking to put the passive filter at like 50MHz cutoff (the ferrite/inductor resistor capacitor part) or so... but should I go higher? I wanted to get rid of the stuff the op-amp would have problems dealing with essentially. Or do I need more? 100MHz cutoff, 10MHz cutoff, then my active band pass?



so, what frequency should the passive filter be at and do I need to pull up the outputs to AVDD? I'd rather not use a transformer.

I would use one of these: http://www.digikey.com/product-search/en/filters/emi-rfi-filters-lc-rc-networks/3408034?k=low%20pass%20filter for each output of the DDS to make the passive filter... But I don't know what is the best cutoff frequency to choose for 150MHz clock.

[tonyarkles]

Maybe I'm misunderstanding something here, but that seems like serious overkill for generating audio.

[SArepairman]

Maybe I'm misunderstanding something here, but that seems like serious overkill for generating audio.

well im pretty sure the DAC count put out some HF crap that can get through the amplifier, harmonics of 170MHz (main oscillator, square wave) and high frequencies in general can have real world effects on circuits

for instance if you have this happen to the circuit you connect your DDS to it could cause a problem!
http://www.analog.com/static/imported-files/tutorials/MT-096.pdf

i want a nice circuit I can trust to be as good as it appears to be on paper without hidden demons.

[tszaboo]

I would not use opamp. CMRR will be reduced because of the resistor matching and they do funny things with signals higher than the bandwidth. Use an instrumentation amplifier (something like INA331). Those are specified, setting the gain is easy, and HF stuff is described to some megahertz, so you only need to filter above.
Than have fun listening to 12 bit sine waves.

[SArepairman]

I would not use opamp. CMRR will be reduced because of the resistor matching and they do funny things with signals higher than the bandwidth. Use an instrumentation amplifier (something like INA331). Those are specified, setting the gain is easy, and HF stuff is described to some megahertz, so you only need to filter above.
Than have fun listening to 12 bit sine waves.

what I meant by differential amplifier is a single resistor gain programmable chip, or an instrumentation amplifier chip like you said (still not commited to anything specific)
ina331 looks good
but I still thought it would be a good idea to put some passives before the op amp.