Post Go back to editing

The AD9834 generates both sine and square waves

Category: Hardware
Product Number: AD9834

Can the AD9834 generate both sine and square waves? Differential generation of sine waves from IOUT and IOUTB. It is followed by an instrumentation amplifier. SQUARE WAVE FROM SIGNBITOUT.

  • Hello Lijie,

    yes: in principle your setup should work. But it may be not really optimal.

    You take the sine-feedback to the comparator (Vin-pin) from the current output Iout. The low pass filter for filtering Vin will load Iout. The load on Iout (200 Ohm in parallel to load pass filter) will be different to the load on Ioutb (200 Ohm). For best perfomance and spectral purity of the sine, your instrumentation amplifier will most probably want to see a symmetric source impedance - which is not the case for your setup.

    Moreover the voltage swing at R5 will depend on the frequency of the sine wave. Your low pass filter R7/C15 has a corner frequency of 5,5 kHz. For sine frequencies well below 5,5 kHz the voltage swing at R5 will be 3mA*200Ohm=600mV. For sine frequencies well above 5,5 kHz the voltage swing at R5 will be 3mA*(200Ohm || 300 Ohm) = 360 mV.

    This might be acceptle, if you only want to generate a singe frequency. But if the frequency may vary, you will most probably not want the sine amplitude to vary as described above.

    This problem should disappear if you do not use Iout as feedback to the comparator, but if you use the output of the instrumentation amplifer to drive a low pass filter and Vin. That output is low impedant and should not have a problem to drive the low pass filter at any frequency with a stable amplitude. 

    Depending on the gain of the instrumentation amplifier you may have to scale down that voltage before its applied to Vin: the comparator accepts signals in the range 100 mVpp to 1 Vpp.

    best regards

    Achim

  • Ups, there is one aspect I hadn't noticed before: as the comparator is AC-coupled, it requires a minimum input frequency of 3 MHz - your setup will not work properly for lower frequencies.

    At frequencies above 3 MHz, C15 will act almost as "short" compared to R7 (as 3 MHz is far above the  the corner frequency of the low pass). If you work only in that frequency range you can restore the symmetry of Iout and Ioutb by simply connecting 300 Ohm in parallel to R6 too.

    best regards

    Achim