This thread has been created to discuss the design below. The person who started this thread can change this introductory post by clicking on the Edit button above.
The q8 and q9 are placed in the headshell. They must be extremely low noise type (with very small internal Rb' resistance). Unfortunately, such transistor are high-frequency (microwave) types, triggering other problems (such as oscillation). The additional RC parts are necessary to prevent microwave oscillation.
The MM/MI input stage (with 47 kohm input) is not displayed here. It contains a double jFET device (U404), that's why it cannot be tested here (lack of jFET model).
The main amplifier starts with q4 aqnd q5. They also provide power for the input stage via the headshell cable.
The opamps op1 and op3 only for setting up the operating point of the input current mirror (eliminating the current differences produced by the input stage).
The op2 makes the correction.
I would like to approach the 0 dB noise figure this way as much as possible. This design helps me to analyze the possible noise sources.
As you say, the main noise source is the base resistance of the input transistors. A trick that I remember from 30 years ago is to use a power transistor instead. Even better: put a number of (those) transistors in parallel.
To see the improvement you will need to use the real SPICE models as SV apparently does not model base resistance.
Thanks for sharing this! I had to look up the meaning of RIAA.
You can reuse this model in your own design by adding it to "My Favorites" using the RMB.
Thanks Darrel, I will try it!