A bipolar gain device, with resistive load with 12 volt VDD, can only provide 12/0.026 == 480X gain.
Thus you need two gain stages, plus buffer stage into power amp.
For good signal-noise-ratio (if 80 dB is adequate), you need 0.2 microvolts total random noise RTI referred to input.
In 10,000Hz bandwidth, need about 2 nanoVolt ( 0.002 uv) noise density;
this requires a total input_referred Rnoise of 90 ohms. A bit of a challenge.
Interestingly, RIAA preamplifiers are easier to design, because their bandwidth is a mere 50Hz, with that 50Hz rolloff continuing cleanly out past 10,000 Hz.
With 20X slower bandwidth, the Rnoise can be 20X higher, what with
- Vnoise = sqrt( 4 * K * T * Bandwidth * Resistance)
Note the tradeoff between Bandwidth and Resistance.
ahhhh and the power supply rejection of standard resistor-load-gain-stages is approximately ZERO. Thus a very clean LDO is needed.
a) use GAIN=30DB, RIAA_ROLLOFF, GAIN=30/50DB, OUTPUT_BUFFER to have the vinyl energy at 20,000 Hz be strongly attenuated; the diyAudio preamp uses 30 volt supplies and NJFET devices
b) planning the PCB to avoid echoes, caused by delays through the 50Hz primary pole with output_to_input feedback in the PCB, requires a LONG aspect_ratio PCB; to adequately reduce out_in crosstalk; each square of foil in the ground will give about 12dB attenuation; when the circuit has 60 to 80 dB gain, the physical layout must give attention to out_in crosstalk within the ground system; without echo_suppression to -120 dB (or at least to -100 dB), the music gets muddy what with inter_tonal time crosstalk.
c) diyAudio "simplistic NJFET RIAA" thread gives much attention to shunt regulator with just a few NANOVOLTS of noise; granted the Vdd design uses numerous large capacitors to handle high frequency trash.