I have not followed the posts regarding a "high bandwidth" ModEEG to which
you refer to but here are some more comments:
1. Remember that there is another LPF pole on the digital board (R104/C106)
which you also need to take into consideration. The best is really to do a
Spice simulation. The given component values are 10 to 1 already
rounded-off values. If you move to E48 resistors you might find more
optimal values. High order filter design is a complex subject and a
trade-off between a number of parameters such as the phase response,
pass-band flatness, etc.
2. For EEG-only applications a higher bandwidth is not required and most
commercial 2-channel amps are using a sampling rate of 256sps.
3. If your application is other than EEG, you will most likely have to
reduce the overall gain.
4. I have mentioned it often enough already, but be aware that changing the
sampling rate from the standard 256sps will make your hardware incompatible
5. Increasing the frequency bandwidth will increase the noise level