3.5mV is fine, and should give excellent performance. Today I would estimate that the sweet spot for Eros is 1.5 to 5mV, and for Seduction 3 to 10mV - but tomorrow I may say something different.
Even though I wrote the white paper (posted under "Community") on signal to noise ratio, the whole subject still frustrates me. Now that we've been working on an MC stepup device to replace transformers, I am diving into the question again. There are quite a lot of data and even more opinions - and that's just in the peer-reviewed academic papers. The papers are not very consistent with each other, and even less consistent with our experience. Internet claims naturally cover an even wide range.
Here's the basic example: our first phono product was the Seduction, still in production after 9+ years. Good solid product, with 40dB midband gain when the C4S is installed. The math works really well - a 3.5mV cartridge makes a nominal 350mV, very close to the unbalanced consumer level of -10dBv or 315mV. Assuming the normal headroom of 14dB for tape and vinyl, that's a peak output of 1.75v, which is very close to the CD standard of 2.0v peak. So you would expect vinyl and CD to be equally loud. But almost everyone who has a Seduction has noticed that the phono is significantly quieter that CDs and they have to turn the gain up, usually by 6 to 9dB (2 or 3 clicks of the Foreplay level control). Clearly, either the levels or the headroom assumptions are not typical of the recordings that people are listening to! Incidentally, this is the reason Eros has more gain - when theory does not match our customer's reality, a good engineer accepts the reality. A good scientist of course looks for a better theory, which is what I am also trying to do.
It gets worse. Shure published a lot of work on peak levels back in the day, and claimed that peaks can be more than 20dB greater than the nominal (maximum VU). For the Eros, that corresponds to a 10v peak output with a 3.5mV cartridge. Eros can handle it, it runs out of steam somewhere above 20v. But if that were really happening then the downstream amplifier would be clipping severely unless the preamp is turned way, way down - which nobody does in practice because you can't hear the music then. I can only conclude that such high recorded levels are extremely rare and unusual.
That's where it stands today, to the best of my knowledge.
