I have been trying to perform a pulse-width modulation gating the output of my AD9951 DDS. The problem is that some of the pulses have a worse turning on than others without any aparent reason (probably my ignorance of the internal architecture has something to do here). The pulses look like this:
The problematic one is that which can be seen in the middle of the figure. All pulses should have the same amplitude. The schematic is practically the same as in the AD9953 evaluation board with an amplifier after the filter. These purple measures are the output of that amplifier.
Detailed view of that pulse.
It looks like that measured at the pin of the DDS.
Detail of the problematic pulse.
Does anybody have any idea why is that and how to solve it? The first 3-4 pulses also seem to be a little weird but I could live with that. Only, if somebody has any idea about that also it would be fine.
The signal is always 80 MHz.
Thank you very much in advance.