We have implemented AD9910 for generating a single tone signal in the range from 75 to 100 MHz. After adapting register values and PLL filter, we get an output signal with the following spectrum
When observing this signal with a scope and analog persistence, we can clearly see that there some amplitude modulation on the output signal. By comparison, we have measured the frequency spectrum on the evaluation board and it looks better. The same components and register values have been populated on the evaluation board.
Note that resolution BW was 10 kHz on the first spectrum and 1 kHz on the second.
Both layouts are similar. On our own board, we referred to the layout of the evaluation board.
However, there's one potential issue that we have discovered. at page 47, it is said that pin 6 can be grouped to together with DVDD 1.8V. We implemented this and isolated this pin with a ferrite bead. By comparing with the evaluation board, we have seen that pin 6 is connected to AVDD 1.8V. Any comment? We tried to patch our layout but the spectrum did not improve.
Another point that I'd like to elucidate: we thought that we should isolate all the analog pins by creating separated AGND plan (like done for evaluation board). In my understanding, the baluun transformer shall be used for splitting AGND and GND, which means that the output connector shall be routed to GND plan and not AGND plan (this is what we have done). In fact, in the evaluation board, it looks like you have isolated DVDD and DGND instead and connected AGND pins on the general GND plan. Any comment?
We have populated the following components for the PLL filter: C15 =47p, C13=1.5n, R37 =1k
Stability is good on both evaluation board and custom board but we still don't understand why we get such a difference on the output frequency spectrum. Can you help me?