Hey Loren, nice post -yea the first pass at getting Cf=0.175pF seemed a low.
I am guessing the reason the original calculation for Cf was too low was the unity gain closed loop bandwidth of 1.8GHz was used in Eq. 2 where that should have been the GBP, more like 900MHz - Eq. 1 is implicitly solving to place the feedback pole at Fo. that will give a Q=1 and likely the final Cf shown was slightly increased to empirically reduce the peaking (1.25dB) from that condition.
I touched on these issues a bit in a new blog showing up on PLanet Analog, Feb. 3, 2019.
The latest direct radio frequency (RF)-sampling transceivers provide a number of powerful capabilities that enable advanced system features like multiband and multimode operation, as well as frequency translation and fast frequency hopping.