AnsweredAssumed Answered

AD9364 noise spectrum peaking at 61.44Msps

Question asked by pete_harbour on Jul 26, 2018
Latest reply on Aug 16, 2018 by travisfcollins

AD9364 noise spectrum peaking at 61.44Msps

 

 

I am running the AD9364 at 61.44Msps at IQ data port.
I have set the TRX clock chain up as follows.
RX {983040000, 491520000, 245760000, 122880000, 122880000, 61440000}
TX {983040000, 245760000, 122880000, 61440000, 61440000, 61440000}

 

AD9361_RXFIRConfig rx_fir_config = {
3, // rx channels programmed 3=0b11=CH2&CH1;
-6, // rx_gain;
2, // rx_dec;

{121,377,447,98,-284,-116,293,149,
-348,-198,421,269,-515,-359,627,474,
-766,-620,940,811,-1167,-1068,1475,1433,
-1930,-2005,2688,3055,-4264,-5719,9927,29364,
29364,9927,-5719,-4264,3055,2688,-2005,-1930,
1433,1475,-1068,-1167,811,940,-620,-766,
474,627,-359,-515,269,421,-198,-348,
149,293,-116,-284,98,447,377,121},

64 // rx_coef_size
};

 

AD9361_TXFIRConfig tx_fir_config = { // BPF PASSBAND 3/20 fs to 1/4 fs
3, // tx;
-6, // tx_gain;
1, // tx_int;
{21, -65, -160, -167, -20, 235, 423, 342,
-68, -606, -878, -549, 360, 1346, 1649, 748,
-1151, -2986, -3268, -891, 4098, 10281, 15391, 17369,
15391, 10281, 4098, -891, -3268, -2986, -1151, 748,
1649, 1346, 360, -549, -878, -606, -68, 342,
423, 235, -20, -167, -160, -65, 21, 0},

48 // tx_coef_size
};

 

ad9361_set_trx_fir_en_dis(ad9361_phy, 1);


Operating in this mode produces substantial noise skirts near the outer edge of the
baseband spectrum (upper real frequencies in I & Q), the skirt peaking is roughly 8dB.

 

 

Modifying the TRX chain to run the FIR filters in bypass (off) also shows gradual
noise slope with increasing baseband frequency, 6dB overall spectral density increase
across the band. I wish for the baseband noise spectrum to be flat versus frequency.
What changes to the TRX clock chain setup should I implement to realize a flat noise
spectrum across the baseband spectrum?

Outcomes