Why does the output spectrum get corrupted when using the NCO to place the IF with AD9122 using single tone vectors generated with the DPG2 pattern generator? Is there a work around?
When using the AD9122 or AD9125 DACs and the NCO is used to place the IF, the output spectrum can become corrupted when the amplitude of single tone vectors have a digital backoff. This happens when using unsigned binary data format and can be avoided by switching to 2's compliment data format on both the DAC SPI register setting and the vector format chosen in the DPG2 panel.
Here is an exapmle of what can happen
The Data sample rate is 250MSPS
Using 4x interpolation with NCO
Screen shots are taken at the QDAC output on the AD9122 evaluation board
When the single tone is selected with 0dB back off you get the propper output shown in Figure 1
When you change the digital back off, in this case -3dB with unsigned binary data format the output becomes
corrupt when NCO is used to place the output, shown in figure 2. If you change the vector and AD9122
both for 2's compliment data you will get the correct vector with digital backoff. However if you mistakenly have
the vector in Binary and AD9122 in 2's compliment the out put will look corrupt as shown in Figure 3.
Or if you select 2's compliment for the vector in the DPG2 pannel and leave the AD9122 in unsigned Binary
you also get a similar corrupt output shown in Figure 4.
Retrieving data ...