Hi. Klaus invited me a few months ago to post here on this topic, but I'm unsure about the degree of interest, so I'll keep it as brief as I can.
Some years ago when Neumann still had an on-line discussion forum, a fairly prominent UK-based engineer posted that his digital Neumann microphones were much quieter than his comparable, analog Neumann microphones. I replied by pointing out that the noise specs were actually about the same, and that digital microphones simply contain a rearrangement of the same components normally used in any digital recording. I told him that if he really was hearing so much less noise with his digital microphones, then something in his previous setup must not have been optimal.
He was taken aback by that, discussed it with high-level technical people at Sennheiser and Neumann, and in the end, I don't think he was convinced. On the other hand I couldn't offer any clear explanation of where the extra noise was supposedly coming from in his previous setup, when his microphones presumably were working properly, so we left it there. -- I would say today that what I posted wasn't wrong, but in the years since then, I came across a presentation that helped me to understand an important aspect of the situation that I hadn't really considered before. I hesitate to link to the presentation since it's from Schoeps, and I have a 50-year relationship with that company and with many people who work or have worked there, including some consulting work for pay, plus an unusual degree of access for historical research. I don't want to impose my biases here. Also, the presentation is in fairly rapid German. But this doesn't have to be about any particular manufacturer's products; I'd just like to follow up on the general situation as I've come to understand it. It turns out that the perception of much lower noise in digital microphones wasn't a one-time thing by that one engineer, but comes up repeatedly, and is truly valid in certain situations.
--To back up a step: I used to be quite active in several on-line discussion groups of amateur recordists. I discovered a prevalent belief among them that's wrong, but in an interesting way: that recordings will be the quietest if you choose input settings on your recorder that allow the "rec level" knob to be set as low as possible. For example, if the levels coming from your microphones allow you to use either "mic in" (without risk of clipping--say, with a built-in pad enabled) or "line in", you should always choose "line in" because less "variable gain" = less noise. This belief was based on observable fact: As you turn up the gain knob on a recorder with no signal coming in, more noise is heard at the output.
But for most preamps, or the preamp stages of recorders and mixers, the added noise due to amplification _decreases_ as gain increases. The noise that they heard when turning up the "rec level" control wasn't rising as much as the amount they were turning up the control, so the signal-to-noise ratio was actually improving. Unfortunately with that particular test, what one hears is deceiving; to find out what's really going on takes either much more careful, controlled listening comparisons, or else measurements and calculations, and many amateurs aren't equipped or inclined to make either one. Still, the fact is that for best signal-to-noise performance, preamp levels should generally be set so that the early stages of the preamp have the _highest_ gain possible short of clipping--not the lowest. Most professionals understand this from experience without working out the millivolts and decibels numerically.
Where this intersects with digital vs. analog microphones is that sometimes we need to record sounds that have extremely wide dynamic range, but for whatever reasons, it's not appropriate to "ride gain" on the preamp--we have to choose one setting and keep it throughout the whole recording. When that happens, we generally set the preamp gain to the highest setting that allows the highest peak voltage from the microphone to be handled without overload.
If the sound approaches the maximum SPL of a professional condenser microphone, the microphone may put out a Volt or more--some go as high as 5 to 10 Volts these days--and that might limit the gain that you can set on the preamp to (say) 20 or 25 dB. At such high sound pressure levels you won't hear analog preamp noise, but then if the sound gets very soft later on, the preamp won't be anywhere near its quietest setting for those weaker signals. This situation is frequently encountered by people who record classical music, or some sound effects or even dialog for film (if it includes, say, yelling and/or gunshots), but much less so by people who record rock music, for example.
And that's where digital microphones have a real advantage that I hadn't quite recognized before: Unlike analog microphones, the voltage they put out is constant regardless of the sound pressure level. As a result, their internal (analog) gain prior to conversion can be set for the lowest noise levels--but they still won't clip at the highest sound pressure levels that the microphone is capable of. This is simply not true for the combination of a highly sensitive analog condenser microphone and a mike preamp.
What I imagine today is that the engineer from the Neumann "Pinboard"--who records classical music among other genres--was in that situation. He was forced by the occasional very high SPLs to set his analog preamps to rather low gain levels; as a result, he heard noise during the quietest parts of the music. But when he switched over to digital microphones, he could leave them at their default gain settings, where digital "full scale" is just above the maximum SPL limit of the microphone. Then the noise contribution from the preamp in the microphone would be extremely low--only a dB or two in very good digital microphones such as Neumann's were (they have all since been discontinued, unfortunately).
In other words his observations were likely valid, and digital microphones (or digital microphone amplifiers for systems with interchangeable capsules) really do have something to offer for applications in which both the highest and lowest sound pressure levels need to be handled without intervening changes in gain. If gain riding is acceptable, though, or with certain specially-designed combination preamps and a/d converters (where clipping doesn't occur because the preamps don't have to be driven to any high output voltages), then that closes the gap considerably.
--best regards
