I've attached a magnitude spectrum of an AD9361 negative SSB tone output which could use some explanation. I played IQ for a -30KHz complex tone sampled at 240 KHz, using the TX coefficients of 'GSM.ftr', and an analog bandwidth snapped to the lower limit (1.2MHz). I took two traces, one with 0 dB attenuation and the second with 40 dB attenuation.
What I don't understand is the high level of opposite sign 3x harmonic. In the first trace, it is ~33 dB higher than the negative 3x harmonic. The 2x harmonics at +/- 60 KHz (missing data-tips) are roughly the same level.
When I sweep a tone, I see the opposite sign 3x harmonic not following filter characteristics of my digital filter chain (different spectral "envelope" compared to the main tone response.) This data isn't shown because I don't have a good way to take a video from my specAn.
Additionally, I tried to further characterize this 3x positive harmonic by dropping the power level in the signal chain. With 40 dB attenuation, I see the level of the 3x positive harmonic drop much more than the 40 dB in the variable attenuator (~60 dB drop on the harmonic.)
Are there any recommendations to try to reduce this harmonic content from the analog sections of the transmit chain/voltage adder/Tx amp?