You guys said that the design makes no sense as far as the filters are concerned?
I probably missed something, but this should get you started:
1) Performance of first stage is highly dependent on unknown source.
2) The first two filters were described a 'high pass tuned to 120kHz'. This is untrue. They are bandpass.
3) The filters are very sharp. Nice in theory, a nightmare in production.
4) First and second stages not decoupled. Low output impedance of first stage swamps performance of second stage's network.
5) Supply voltage on gates not shown. Inferred at 10V but not confirmed.
6) 470k to 5V at output of second stage doesn't do anything.
7) Unusual (but not wrong) to use capacitive gain setting on third stage.
8) Third and fourth stages not decoupled. Feedback around fourth stage swamped by third stage output.
9) Output stage is biased in the linear (uncertain, logic 'maybe', etc.) region. With no signal, its output is unknown.
With signal, its output is '1', which may or may not be different from idle state.
10) If point D is given to a micro processor, its 0-10V swing may destroy one or both of them.
You really really need to post more of the circuit. If 220VAC is applied to point A then it will all explode.
[edit] I must correct myself - the 10M bias resistor at point A is irrelevant, since this point is dominated by the DC feedback from the gate output. This makes the 5V bias point unimportant, so the supply voltage can't be inferred. (points 5 and 10). [/edit].