You didn't specify with what you measured the output signal with. If it was a computer soundcard you may have been dealing with 10k impedance. Together with the 1 uF output cap that would explain your -3dB at 20 Hz.
I used a conventional, "analogue" measurement scenario, as follows:
1. I used a conventional sig-gen, an oscilloscope and Jim Hagerman's reverse RIAA unit.
2. I fed the sig-gen into the reverse RIAA unit. One of its outputs reduces by 40dB (1/100th out) so I adjusted the output of the sig-gen to 500mV, thus getting 5mV - the nominal output of an MM cartridge - going into the phono stage.
3. Using the Hagerman reverse RIAA, I should get an even response on the scope, from the phono stage. IE. the output at any frequency should be the same as the output at 1KHz (or within +/-0.2dB, anyway).
Can you see anything wrong with the above method?
Regards,
Andy