> Seems to me 40dB gain covers most situations. Or am I in left field here?
You're in the ballpark; sometimes the drummer isn't.
I once recorded a classical percussion solo and needed LOSS between mike and recorder.
Mike sensitivities:
Output at 74dB SPL:
1.0 mV to 2.5 mV - German-tradition large condensers
0.2 mV to 0.6 mV - small condensers
0.1 mV to 0.2 mV - US dynamics
vintage ribbons: 5dB to 10dB lower than dynamics, assuming ~150Ω nominal impedance
Few mikes want to be impedance-matched. Some don't mind. Others specify >1K or >2K loading. Some mikes expect some load: some of the BBC 300Ω ribbons wanted 600Ω loading to tame the bass. Many people prefer SM-58 with a heavy load.
Dynamics can drive a dead short without harm. The response will deviate from the spec, but it won't go to hell. Some condensers with class AB output will drive a short, most are rated 100Ω minimum. Condensers with single-ended output device may have trouble driving less than 1KΩ (or long cables). General custom is to give a 2KΩ input; you can always add a loading resistor if you like the sound better. A current fad is "variable impedance" implemented several ways: plain loading, or tapping a transformer. One designer advocates zero impedance inputs, though he has to put a few ohms in so the condenser mikes don't freak. Very high impedance inputs are not favored because when the mike cable is unplugged, they buzz (and maybe hiss) badly, though Valley People sold a lot of 200K mike inputs.
Performance acoustic levels:
Harpsichord in audience: 74dB SPL
String quartet in audience: 94dB SPL
Full orchestra, in audience, rare peaks: 126dB SPL
Trumpet at 4 feet: 118dB SPL; at 1 foot, ~130dB SPL
High power speaker, near-field: 130dB SPL
Using AKG414 (1.25mV at 74dB SPL) around orchestra, I assume that once or twice a season I will have 500mV coming off the mike.
I also do harpsichord in that hall: about 1.25mV off the mike which would need 400 times more gain than the orchestra to peak at zero dBfs. In fact I use only 10 times more gain because the room rumble is not a lot lower than the harpsichord (though a lower frequency range). The high-end noise of mike and room is 15dB SPL, so I can only get 60dB dynamic range from the performance. I don't have to jam this into the top of the 95dB range of 16-bit recording. (Also the applause may be louder than the harpsichord, and I'm too lazy to reach for the knob when the music stops.)
When dynamics ruled, it was customary to spec input iron for 30mV for small acts, 100mV for orchestral work. 1.25mV/0.2mV is about 5:1 which is about the difference between my LD numbers and the old dynamic-mike mixer specs. This is a good benchmark for "normal" performance. Of course since Dick Dale got Leo and Jim to build him a GITAR AMP, and the drum makers caught up, "normal" is abnormal.
Large classic percussion, vigorously played, miked at 8 feet, with a 2.5mV/74dBSPL condenser, gave peaks over 2V. Indeed that seems to be the upper un-padded spec for many inputs of today. If you don't give 2V input overload, someone is going to clip.
I have heard that dynamics in front of some singers or inside a bass drum can make a Volt of signal. That would be 1V/0.2mV= 74dB above 74dBSPL or 148dB SPL. I have a hard time accepting this, but am not prepared to reject it either.
As Paul says, noise of the passive mikes runs 0.2uV, and they give up to 0.2mV at 74dB SPL. Assuming noise floor near 14dB SPL, these mikes need near-zero noise figure in the preamp. (Studios have acoustic noise from below 10dBSPL to over 30dBSPL, so total noise is not always dominated by the mike and amplifier.)
The large condensers give 1mV-2.5mV output because they have internal amps, which make noise, 1uV for my AKG414. These do not need low noise figure amps: a 10dB NF input would be just noticeable. The logic is impeccable: if you adopt the condenser capsule, you are stuck with an amp in the mike. But now that you have an amp, you may as well bring the level up to overcome line noise and allow a low-price console input. The board-amps for the hot condenser mikes can be seen as high-gain line-inputs, rather than low-noise inputs. OTOH, there are condensers with outputs (and output noise) scaled to dynamic levels, so you can't just go by "type".
You can build a low-noise amp, you can build a low-gain amp, you can even build a low-noise low-gain amp. But it gets very hard to build a low-noise amp with gain that varies from high to near-unity. And since hot mikes can easily exceed a volt, and console internal level may be less than a volt, you may need gain of unity or less.
It would seem logical to build an input with <0.2uV noise and 100mV overload, and supply a 5:1 pad for use with hot condensers. It is hard to build a 5:1 14dB pad with appropriate in and out impedances; if hot-condenser is the only goal, you can let the input impedance be less than 1K. For general use, you wish the input to be over 2K and output under 200Ω, which suggests at least 20dB pad, but that puts a hot-condenser's self-noise below the thermal noise of a 200Ω input. We live with imperfection.
> Seems to me 40dB gain covers most situations. Or am I in left field here?
Take my 500mV orchestral peaks and amplify 40dB. We have 50V signal! I could engage the 10dB pad on the mike, but the lines run past dimmered lamps so I'd rather not. I could pad 10dB-15db at the amp, but then why did I pay for 40dB gain and throw-away a lot of it?
What I use is 20dB gain, pot, 20dB gain, 2Vfs recorder. For the largest orchestra works, I run 28dB loss in the pot. I can bring 5mV or 86dBSPL signals up to 0dBfs. I don't try to bring 74dBSPL harpsichord up to 0dBFS; let it peak at -12dB, it aint supposed to be loud, and the digital dither garbage is hidden under room and mike noise.
Note that the 20dB gain in the first stage is nearly all I can run with +/-15V supplies. 500mV times 20dB is 5V, just 6dB-8dB below clipping. Actually I would be OK with 26dB gain: 126dB SPL peaks are so rare and so strained that touching clipping would not ruin the recording. It just happened that the commercial boxes I used were easily set to 20dB, fussy to set to 26dB.
Note that, with passive pot, I need gain after the pot. I need 34db-40dB of gain-range, for the same mikes in the same room for different acts. If I took all the gain before the pot, then to get 2V at the recorder with 34dB attenuation for the big orchestra, I'd need 100V level out of the preamp! Yeah, we could do it with tubes or high-dope transistors.
If I used dynamics, I would need 14 to 20dB more gain. Say 40dB, pot, 20dB.
Ribbons could need 5dB to 10dB more, even if they are nominal ~200Ω impedance. Say 46dB, pot, 20dB.
If I worked at +4dBu nominal levels, I'd need 12dB more gain. For hot condenser: 20dB, pot, 32dB. For ribbon: 46dB, pot, 32dB. This IS 78dB maximum gain, and only pulls 86dB SPL up to recorder max input.
So there are a lot of issues. Mike sensitivity varies from 0.05mV to 2.5mV, 24dB. Acts range from 74dB SPL to >130dB SPL. Nominal electronic level can be -10dBV or +4dBu (or +8dBm), 2.8V peak or 17V peak, or some other level (or no level discipline at all).
Mikes: 24dB range
Acts: 56dB range
Level: 12dB range
So we could need 92dB range of gain.
In fact 74dB SPL acts don't need to be peaked; with modern recorders a 90dB SPL level is reasonable, so only 76dB range of gain.
The 12dB difference between Tascam and Pro levels is usually covered by a switch or added jack (or no-support): 64dB range. Many sound-folk know to use a pad on bass-drum: 44dB range.
The situation is very different from phono-amps. The noise and max signal on a disk is fairly well fixed: ultra-virgin vinyl may have 10dB lower noise, some 12" singles may have 10dB bigger wiggles, but nothing like the 50dB range of acts found on stage. Phono needles have converged on two basic types: 5mV and 0.25mV, and you don't change pickups often. Mikes ran 0.05mV to 2.5mV and we do swap them frequently to find "the sound".
> I get over 72dB SNR with my all tube phono stage
You can only claim the last 12dB of that. The LP standard aimed at 60dB from surface noise to nominal recording level. The 5mV 47K pickup standard evolved because the surface noise would be slightly higher than the thermal noise of 10K-30K impedance at resonance and the grid noise of a semi-selected tube. (An MC pickup with a transformer works about the same as an MM pickup, though a little less trouble from parasitics.) Improved vinyl and cutters have got noise down a bit, your skilled tube-work has improved on early 6SC7 phono inputs to keep up. And I suspect that you reach over-kill: if you listen with the needle up, then drop it in a blank groove, doesn't the hiss increase? You "don't have to be that good" (though I agree that pure hiss is different from surface noise, less interesting, should be kept far back).