Double post. Mods, please delete :-)

[ACS]

Oops 

[deafbutpicky]

I'd rescale the lowpass to 47k and 15n at first glance and 0.1µ for the shape seems a bit big.
The Vb Rs would be better off with 10k -100k as you feed the OPA with 1M already providing
a more stable bias too and a gain of 100 in the makeup stage is maybe too much, driving
IC2 into overdrive itself, which might not sound too pleasing, but the rest looks legit.

What's your problem with LTspice output?
With what source do you feed the simulated input?