Rx flatness

#1

Hello,
The RBB filter is bypassed and the input signal is swept from LO to (FADC/decimation/2 i.e., the effective first Nyquist zone).
We noticed that the gain drop by almost 7-8 dB (signal strongest closest to the LO).
Is this to be expected? (some sort of ADC sinc response?)
Putting numbers to the settings:

  1. CGEN = 200MHz
  2. FADC = 50MHz
  3. Decimation = 2^1 (Hence effective sampling rate is 25MSps)
  4. First Nyquist Zone = 0 to 12.5MHz.
    The input signal was swept from f_LO to f_LO + 12.5MHz.
    Thank you for looking into this case,
    Regards,
#2

Hi @EnthuMan,

Because of decimation filter, your signal RF bandwidth will be calculated as:

Fp = x * Fclk / K, where Fclk is ADC sampling frequency, K is decimation factor and x is:

x <= 0.27 for >= 108dB decimation alias suppression,
x <= 0.30 for >= 75dB decimation alias suppression,
x <= 0.32 for >= 60dB decimation alias suppression.

Check “Interpolation” and “Decimation” section in datasheet for more information.

In your case the numbers will be as follows:

Fp = 0.27 * 50MHz / 2 = 6.75MHz

Hence if you sweep the signal up to Nyquist frequency, you start to see decimation filter (HB1) response.

#3

Hello @Zack,
Thank you for the quick and clear answer.
Does this have an impact on modulation EVM for bandwidths more than Fp?
If yes, then can it be compensated by one of the FIR filters?
Regards,

#4

Hello @Zack,
We did a frequency sweep and a “max hold” in our daq software.
The settings are FADC = 100MHz, followed by decimation of 2, everything else is bypassed. Overall filter BW is set to 20MHz.
A 1024 pt FFT is being performed as an input single tone is sweeped in frequency (from LO to first Nyquist frequency). We can see along with ripples a drop in the signal amplitude by almost 15dB from dc to first Nyquist. Please guide us on how to interpret this observation.
IMG_20190823_122115%20(Small)