Hi Stefan,

You wrote:
> ...

I'm not interested in frequencies
above 600 kHz.
...

Why not trying OPs like the LT1056? Noise? The current noise levels
of those devices are at some fA/sqrt(Hz) and voltage noise does not
play a big role at E-Field probes.

In contrast to simple Fet or Dual-Gate MOSFet input stages I never
observed any kind of IM here where "KIL" (353 kHz) is my neighbor.

Furthermore using a balanced input stage like a vertical dipole,
2 x 30cm, feeding an INA111 there are much less interferences
from local origin.

All problems I observed using those devices are due to input limiting
diodes, bad decoupling from power devices and high common mode
capacitance of decoupling transformers at both ends of the feeder line.

Another solution could be using 2 or 4 Fets in parallel. But doing such
feeding from batteries may be problematic. I remember soldering 4 x BF981
piggybacked improved noise AND big signal handling at low frequencies.

Peter, df3lp