Вот что мне ответили AD:
Thanks for pointing out this strange looking plot.
I talked to the AD8599 product engineer and the applications engineer and they believe that it is noise in the test setup.
After 80 kHz, things settle down.
The main information from the graph is that the phase margin at the zero dB crossing at 10 MHz, so things look o.k.
When they have some time, they will look at the part again and see if they can generate a better looking graph.
Kind regards
Chris
что тут скажешь... гнать индусов к чертовой матери