A subsequent version running at 40G (2x 20G in series, with the midpoint bootstrapped by a matching lower-Z string) got the effective shunt capacitance down to about 14 nF, so we ran at a voltage gain of about 70.
Fun.
(In most cases this sort of elaboration is a net SNR lose, but up in the gigohms it turned out to be a win. Pity about the company--they got killed off by a big customer not paying on time, to put it mildly.)
Cheers
Phil Hobbs
Cheers
Phil Hobbs