Post Go back to editing


Category: Datasheet/Specs
Product Number: AD9957


We need test setup used to get the spectrum graphs in 'TYPICAL PERFORMANCE CHARACTERISTICS' of datasheet for AD9957.

  • Was there any 500MHz Low Pass Filter or any other filter in the test setup?
  • What is the "Lower Sideband Suppression" that are mentioned in the right hand side spectrum shots in the datasheet?
  • Is there any spur beyond 500MHz that needs attention while designing our custom board filtering circuit (1GHz sampling clock) ?

There is a 400MHz LPF used in the EVAL-AD9957 as shown below,

  • Can you share the -3dB level frequency, roll-off region and the rejection performance ?

Thanks & Regards,

Arunkumar P

  • The details of the test setup are not available. However, the Typical Performance plots generally make use of the Evaluation Board (EVB) in conjunction with a bench test setup. That said, the filter on the EVB schematic would be the filter used for making those measurements.

    I am unable to find any records regarding the filter design. You would need to model the filter using a mathematical tool (e.g., Matlab, MathCAD) based on the schematic and determine the filter response by simulation. Be aware the input impedance driving the filter is not equal to the load impedance at the output of the filter. Although I am near certain the filter design was done with an expectation of matched 50 ohm impedance, the EVB has an intrinsic impedance mismatch with regard to the filter. That is, the filter load is 50 ohm (the measurement instrument), but the input to the filter is 99.8 ohm (the impedance as seen by the filter input looking back toward the transformer).

    The "sideband" plots are intended to demonstrate the ability of the digital quadrature modulator to produce a single sideband output signal. The modulator input includes a baseband quadrature sinusoid combined with a "local quadrature oscillator" (which is the DDS in this case). An ideal quadrature modulator will produce a single sideband with no carrier or "alternate sideband" feedthrough. However, a real quadrature modulator will always exhibit some feedthrough. For example, the spectrum shown in Figure 8 uses a baseband frequency of fB=16.625kHz and a carrier frequency of fC=102MHz. The unsuppressed sideband (i.e., the desired output) appears at fC+fB=117.625kHz with a level of ~0dB. The carrier (at 102kHz) is suppressed by approximately 80dB and the alternate sideband (fC-fB=86.375MHz) is buried in the noise floor (>95dB suppression).

    Because the filter is lowpass by design, spurious content beyond the filter cutoff frequency (~400MHz) is a function of the filter components and PCB layout.

    The filter topology suggests the filter response is of the Chebyshev variety. That said, it should exhibit some passband ripple before starting to roll off at the cutoff frequency to follow a constant attenuation slope of 60dB/decade, Again, the actual stopband attenuation performance will depend on the filter components (self resonance, parasitic inductance, parasitic capacitance) as well as the physical characteristics of the PCB layout.

  • Regarding the EVB filter design... I came across the attached set of plots, which may prove useful.