These are some comments that I recieved from CR Magnetics related to my application.
"I would kick up the voltage to 100 mV and use a 20 ohm on the burden resistor. I would kick up R2 and R3 to 10K to minimize input voltage differential effects. (50 mV/100 ohms = 0.5 mA error on a 5 mA signal).
I would decrease R5 to 500K for a gain of 50.
Please also note that using the 1000 turns is the ideal transfer ratio, however, no CT is perfect, and corrective factors are given in the spec sheet. Even with this adjustment, you will get variance part to part, and for the most complete and accurate system, an adjustment pot is usually added to calibrate. Without calibration, you will probably get a sensor that is +/- 5% part to part.
Filtering depends on acceptable ripple. Full wave rectifying on 60 Hz results in 120 Hz ripple on a DC offset. Assuming single pole filtering 20db per octave, you will get 50 mV of ripple using a filter with a corner frequency of 12 Hz. (V/100). 1 divided by 12 Hz = time constant = .083 = 1/(2 * pi * f * R * C). Assuming C = 10 uF, R would be around 2K. 4.7 uF and 10K would be a pretty good filter."
Crutschow: What simulation program are you using?