Topic: Schoeps CMD-42

[Colin Liston]

This looks pretty cool.  Sorry I couldn't find this link on youtube only on instagram.

https://www.instagram.com/reel/C53Q3T2iDrs/?igsh=Y2Z0ZTFqa3EzMjQ3

[ero3030]

Old tech getting a new push

[DSatz]

There's actually quite a lot of new stuff in these amplifiers, not just "old tech". It's unlikely to interest most folks here directly, though, given the lack of consumer or semi-pro recorders with AES42 digital microphone inputs.

If there's a more general interest in digital microphones and/or these mikes, we can definitely talk about them, but I don't want to push.

[breakonthru]

If there's a more general interest in digital microphones and/or these mikes, we can definitely talk about them, but I don't want to push.

I wouldnt anticipate any resistance, please feel free!

[Sebastian]

Yeah, I'm definately interested in all the technical details.

[DSatz]

OK, I'll post some notes here in a few days if that's all right; I'm still getting over jet lag from spending nearly all of May over there, plus some bad personal news when I got back. I'm not up for much writing at the moment, but soon.

[DSatz]

Sorry for the delay. Let me divide this into two topics. This message will be about digital mikes generally; later I'll talk about the new Schoeps product.

-- I think folks get the general idea that a digital microphone is mostly a refactoring or rearrangement of the same parts that everyone uses today--the capsule is analog, the first stage of the amplifier is analog, then there's an a/d converter and some (digital) output circuitry. What goes down the mike cable is basically an AES3 signal--two channels, 24 bits, sample rate to be discussed. The first obvious benefit is that interference in the mike cable becomes next to impossible.

The second benefit is less obvious--in fact it took me a long time to catch on; occasionally someone who had switched from analog mikes of one manufacturer to comparable digital mikes from the same manufacturer would say, wow--these sure are quiet! And I couldn't see how that could be so, unless the person had previously been using a setup that was far from optimal. I saw some claims from one manufacturer that I thought severely abused the whole concept of headroom in order to claim some huge dynamic range improvement, and I left it at that, until finally I heard a straight explanation that really got my attention. This was discussed in the Team Schoeps thread over in the team boards section, but I'll summarize it here.

Basically, if you draw a block diagram of an analog mike setup (feeding a digital recorder) and a digital mike setup (also feeding a digital recorder), they're largely the same except that more of the components are now inside the microphone. But one place where there's a significant difference is what goes between the first stage of the microphone amplifier (the FET) and the a/d converter. With an analog mike, you've got the output circuit of the mike, then the cable, then some kind of preamp that takes in the mike-level signals and amplifies them to line level to drive the a/d converter's inputs. With a digital mike, the output of the FET stage (possibly with some signal conditioning that wouldn't necessarily involve much if any voltage amplification) could drive the input of the a/d converter more or less directly.

Now, everyone understands that mike-to-line preamps always add some amount of noise to the signals that they amplify. And the amount of noise they add is very much a function of their gain setting--the more you turn up the gain, the more noise you'll hear. What a lot of people apparently don't realize, though, is that in a well-designed preamp, the noise increase at higher gain levels will be less than the increase in gain. As a result, for typical analog preamps, the quietest setting overall is the highest gain setting that doesn't risk driving the preamp into overload. That seems counter-intuitive to some people (they think: the less gain I ask for, the less noise will be added), but it's verifiably true for the vast majority of preamps that you can improve your signal-to-noise ratio by using higher rather than lower gain levels in the preamp. (If this comes as news to you or if it seems wrong, I suggest that you take a pause here and absorb the information before proceeding, since it's not wrong, and obviously has implications for recording with analog mikes and preamps/mixers/recorders.)

OK. So, again following the logical implications of all this: Any given capsule and first-stage amplifier (FET or tube) has some given noise floor, and if you want to add the least possible noise to that, you'll set your preamp gain as high as you can safely get away with. When you do that, what is the maximum sound pressure level (SPL) that you can handle? It will depend on one of two things, whichever limit is reached first: either the maximum SPL of the microphone, or the maximum output voltage of your preamp. If your preamp is set to, say, 60 dB gain because that adds the least noise to the microphone's noise floor, and if it's a more or less typical 10 mV/Pa microphone with (say) a 125 dB SPL maximum, then the microphone can put out about 350 mV--which, when amplified by 60 dB, is 350 Volts! Clearly the preamp would be setting the limit on the overall dynamic range of that setup. Just as clearly, if you reduce the gain of the preamp to accommodate high sound pressure levels, then you will incur more than the minimum of noise at the preamp input. It's a forced choice--you can either have the lowest input noise in the preamp OR you can have the full headroom of the microphone OR you can find some point in between that you hope will be satisfactory, BUT it can't possibly be optimal.

(Note that having a 32-bit recorder doesn't help AT ALL with this situation because all the 32-bit stuff comes AFTER the a/d converter, which can only have about a 21-bit dynamic range at best. The relationship between the levels at the analog mike inputs and the signals driving the converters needs to be optimized first; then you can record 24 or 32 or 69 bits if you like, it won't make any difference to the signal-to-noise ratio of your recording. But that's a whole other topic.)

The thing about a digital microphone amplifier is that it's always putting out the same amount of power regardless of the SPL at the moment. 1s and 0s are changing, but the signal itself still has the same overall voltage feeding the same impedance, thus the current is constant as well, and power is voltage times current, so that stays constant. There's no risk of overloading the amplifier's output circuit, so the dynamic range of the microphone depends entirely on its input circuit (except if you use a control signal to set the amplifier to something other than its default gain). So in a well-designed digital microphone amplifier you have the best of both: the lowest possible input noise AND the greatest possible headroom, without having to raise the gain for low-level recording or reduce it for extremely loud sounds (e.g. sound effects recording). You can, of course, accommodate the softest and loudest sounds by adjusting the gain on your analog preamp--but for situations in which either or both extremes might reasonably occur and you can't (or don't want to) "ride gain" in real time, the digital approach wins over analog by quite a substantial margin.

No surprise, film and video sound people are the main fans of digital microphones. Live recording would probably be the second-favorite application IF there were more recorders with AES42 digital microphone inputs--especially affordable ones--or standalone interfaces that would support multiple AES42 microphones, which unfortunately don't currently exist (some used to exist in the $1,000-and-up category). I would really like to see simple, compact two- and four-channel interfaces that could power the mikes and support basic Mode 2 operation; to my knowledge there's no particular reason why such things would have to be very expensive. Unfortunately, though, there's a real chicken-and-egg problem at least for the time being. [Edited later to add: Another thing that affects widespread practical acceptance is the powering issue. Digital microphones need entire Watts of powering, so battery operation isn't nearly as convenient--or as concealable, for those whom that may concern--as with analog.]

[Chanher]

what an interesting read, it really does seem like a contradiction to everything that we've learned so far, but im open minded.

i wonder how these work in pairs or even 4 like you mention; is there any word clock involved?

[DSatz]

With digital audio there are always word clocks, and each digital microphone that follows the AES42 standard generates its own. However, there are two possible modes of operation (if the given microphones implement them both), and in Mode 1, each microphone's clock is entirely free-running and independent; thus no two are ever exactly the same. To combine the signals from any two or more microphones running in Mode 1, the receiving interface must implement some form of sampling rate conversion (SRC) which may either be used on all input channels--or, as an alternative, one microphone may be chosen to provide the master clock for all the others (in which case, that one microphone's signal wouldn't go through SRC, but any and all others have to). If I'm not mistaken, at least one model of Sound Devices recorder handles or handled things that way.

The fancier way to do it is to operate the microphones in Mode 2 rather than to use SRC. In that mode, the receiving interface sends brief control signals to each microphone which "nudge" it into alignment with the interface's own clock. The clock signal for each microphone is still generated within that microphone--no word clock is sent up the cable to the microphone--but the receiving interface extracts the clock signal from each incoming data stream and compares its frequency and phase to its own word clock. If/when a difference is detected, the interface sends a packet of data indicating which direction (up or down in frequency) the microphone's clock oscillator needs to move, and this process repeats itself until convergence is reached and sync is established across all the microphones. Since there are thousands of data frames per second no matter what sampling frequency is chosen, the process is so quick that no audible delay or distortion occurs.

Back when the AES42 standard was new, it was difficult to implement Mode 2, but newer chips make it quite straightforward. It is technically a little nicer than the SRC required by Mode 1, which always adds some increment of noise, however small.

P.S.: If I understand what it is that might seem "like a contradiction to everything that we've learned so far", it's probably what I said about preamp input noise. That discussion probably belongs in another thread in another section, but people might note that the spec sheets for nearly all preamps cite the noise figure achieved at the maximum gain setting, because that's usually when the input circuit of the preamp adds the least noise to whatever the theoretical minimum might be for the given level of gain.

However, that doesn't have to be true in all cases. Maybe 10 - 15 years ago I posted a report here about a project in which I measured the noise of all my preamps under conditions that I felt were representative of the way I actually use preamps: driven by the source impedance of an actual condenser microphone amplifier (with phantom power applied, but a capacitive test head in place of a capsule), and set to a gain level around 30 - 35 dB rather than the 60 dB or so at which preamps are typically spec'ed. At that gain level, one of the preamps that had measured as relatively noisy (at full throttle) turned out to be the second quietest of any that I owned. Its designers had stated in the manual that they'd taken an unusual approach and had focused on optimizing moderate gain levels, and the results bore that out. Unfortunately it wasn't a type that could easily run on batteries. Another preamp that I know of, but don't own, was specifically designed (ca. 1980, no longer made) for a uniform noise figure at all the gain levels that it was capable of. For practical reasons this required a gain switch which controlled multiple components at various points in the circuit, rather than a continuously variable gain knob that was a simple, variable resistor (a/k/a potentiometer).

The point is that you can't predict the noise performance of any preamp at gain level X from knowing its noise performance at some other gain level Y. Input noise depends also on the source impedance--listening or measuring with a short or an open circuit at the input can be very misleading. The issue in any event is almost always the input stage of the preamp, since the output stage is usually mostly a current amplifier (buffer / line driver stage), and it's not too hard to make those as quiet as you'd want.

[kuba e]

Thanks for the very interesting posts. I have only vague ideas about electronics. I would like to ask if I can simplify it to the following idea. Classic microphones have an analog signal output. And in order to transmit an analog signal over a long distance, it is necessary to increase its impedance (and make it symmetrical). This is what happens in the body of our classic microphone. And our classic analog preamplifiers are designed precisely for an input signal with a higher impedance. And this causes the disadvantage of noise at low loads and and limited range. Digital microphones, on the other hand, conduct the analog signal over a very short distance. So the preamplifier and a/d converter in the body of the microphone works with a very low impedance signal. And this gives them the advantage that they can process a large-scale signal and with constant low noise.

If my idea is not completely wrong, I have a question. Can our classic preamplifiers be designed similarly? They would first convert the high-impedance input signal to a low-impedance one and then process it just like a digital microphone, so they would reap its benefits.

I remember when I started recording I didn't understand why everything was so complicated - microphones, xlr cables, preamplifier, a/d converter, level settings. I learned it. But now I can see that my original amateur idea of ​​plugging the mics into the box and hitting record is becoming real. That's great. My last question is, does a digital Schoeps microphone cost more than an analog one?

[Sebastian]

If my idea is not completely wrong, I have a question. Can our classic preamplifiers be designed similarly? They would first convert the high-impedance input signal to a low-impedance one and then process it just like a digital microphone, so they would reap its benefits.

This is exactly what the FET does already. A FET has a high input impedance (in the megaohms range) and a low output impedance. If I understand DSatz correctly, the main benefit in digital mics is the short distance between the FET stage and the ADC. The shorter the distance, the less likely it is that any kind of noise is induced into the signal path. And being able to eliminate the pre-amp altogether means that it also can't add any additional noise. Also, because the output cables transmit digital signals, analog noise is a problem of the past.

[DSatz]

Hmm. kuba e, yes, the amplifier of a traditional, analog condenser microphone (a) polarizes the capsule, (b) transforms its output impedance from very high to low, and (c) balances the signal. Since a capsule is a capacitor, its output impedance varies in inverse proportion to signal frequency, so it's highest at the lowest frequencies and vice versa. For a typical capsule capacitance of ca. 40 pF, the impedance at 1 kHz is about 4 MΩ; at 20 Hz it's about 200 MΩ! If you try to feed the input of a preamp, mixer or recorder with a signal like that, the (typical) 1 or 2 kΩ input impedance will cause huge signal losses, which furthermore will be frequency-selective, i.e. you'll have built a very lossy low-pass filter with a 6 dB/octave rolloff throughout the whole audio range. Possible interference from other signal sources, equipment and cables would be a big problem, too. The solution to both problems is to find an amplifying device or circuit with an even higher (preferably by an order of magnitude or more) input impedance but a low output impedance, and place its input as close to the capsule as possible, then use that device's (or circuit's) output to drive the cable. Technical terms for this functionality include "impedance transformer" and "current amplifier".

The idea of a capacitive microphone was already known in the 1880s, but the problem of its output impedance was unsolved until the development of the vacuum tube--specifically the improved, high-vacuum triode version which Bell Labs developed on the basis of the patent they'd acquired from Lee DeForest. (That history is complex and somewhat in dispute; I'm simplifying here.) The vacuum tube made "repeaters" (intermediate signal amplifiers) possible, which were needed for coast-to-coast telephone hookups across the U.S., which hadn't existed before. The same research program led to the first practical condenser microphone, which was patented by E.C. Wente (AT&T / Western Electric) in 1917. It was invented mainly to serve as a reliable instrument that was uniformly sensitive across a wide range of audio frequencies, especially the voice range, for measuring sound pressure levels; this was needed for systematically evaluating the performance of telephone systems, especially long-distance networks. Greatly improved versions were introduced within a few years, the design was licensed to RCA and Westinghouse among others (apparently including AEG and/or Siemens in Germany although I haven't pinned that down yet), and by the mid-1920s (i.e. almost exactly 100 years ago, yay) condenser microphones were the basis on which the Western Electric sound system made "talking motion pictures" possible along with electrical (i.e. microphone-based rather than horn-based) recording of phonograph records. Radio broadcasting and public address applications came along in the early-to-mid 1920s as well.

These early condenser microphones were all pressure transducers with large diaphragms by today's standards, so they were omnidirectional at low and low-mid frequencies with a narrowing of the pickup pattern starting in the voice range, and tight focusing at high frequencies. That was a very useful pattern, and engineers soon learned to take the best advantage of it. When stereo isn't an issue, it's not important for a microphone to have a textbook directional pattern that's uniform at all frequencies--though if there's going to be variation of sensitivity with different angles of sound incidence, the best option is for the off-axis response to be "smoothly shaded" at high frequencies rather than being "peaky".

But I'm digressing. The direct answer to your question is that physical placement of the first amplifier stage as close as possible to the capsule is for the sake of avoiding signal losses (which could well be frequency-selective) and interference both. Having very low-impedance, balanced outputs for professional microphones likewise serves both purposes. There's nothing wrong in principle with the traditional analog arrangement with microphones and preamps, including the preamps built in to mixers and recorders. But the way things have developed over the past century, and especially with the designs and components of the past 50 years or so (technologically, the 1970s was a really interesting time to get involved in recording), the dynamic range of a good condenser microphone can very well be 120+ dB. And a preamp circuit that only adds (say) 1 or 2 dB to the microphone's noise floor when amplifying that signal to line level requires a sensitivity that can't be carried through at the highest signal levels that the same microphone can put out, since the outputs of the preamp would be driven way beyond practical studio levels. Typical line-level studio devices might clip at around +24 dBu, or somewhat higher if they're ambitious. But professional condenser microphones today can often reach or exceed 0 dBu voltages around their maximum SPL, and if a preamp's gain was set to (say) 60 dB for lowest noise, that would require output levels of +60 dBu, or around a kilovolt, which would be insane. Preamps aren't supposed to be 100 W power amplifiers, but that's what they'd have to be, assuming a typical 10 kΩ line input on the tape recorder. And no one designs analog tape recorders to take that kind of voltage on their line inputs, either.

We all deal with those limitations whether we're conscious of it or not. Of course not very many of us have to record at the maximum SPL of modern professional microphones (typically 130+ dB SPL), and I'm glad about that; hearing loss is real, folks. But the main way we deal with it is by setting our recording levels so that the absolute peaks stay somewhere just below 0 (digital full scale), and the way to get the best signal-to-noise ratio in the recording is to maximize the gain early on (use the most sensitive setting that avoids overload) and if there are controls available for this purpose, doing so at each stage where you have a choice. Use the full available dynamic range of each stage along the way that you can control, in other words--and the earlier the stage is, the more important it is because noise and/or overload in that stage can't be undone in later stages. Sorry if that goes against the hype for 32-bit recorders with analog mike inputs, but if you don't optimize the gain structure before the a/d conversion, you're recording a nice 32-bit version of a signal that almost certainly has less dynamic range than you could be getting with more careful level settings. Expressing your salary in tenths of a cent instead of dollars doesn't increase your actual income. But if you're using AES42 digital microphones and your recordings are destined for a 32-bit post-production workflow, then a 32-bit recorder with AES42 inputs makes excellent sense.

[rocksuitcase]

Fascinating and illuminating discussion. This brought me back to my university days thinking of all those calculations of SPL and frequency graphing.

THANKS DSatz!!!

[Gutbucket]

The best Monday morning read at TS I've come across in a while!

+T

[kuba e]

This is exactly what the FET does already. A FET has a high input impedance (in the megaohms range) and a low output impedance. If I understand DSatz correctly, the main benefit in digital mics is the short distance between the FET stage and the ADC. The shorter the distance, the less likely it is that any kind of noise is induced into the signal path. And being able to eliminate the pre-amp altogether means that it also can't add any additional noise. Also, because the output cables transmit digital signals, analog noise is a problem of the past.

Sebastian, thank you. I was confused about the impedance. I re-read the basics of electricity, it's clearer to me now.

The thing about a digital microphone amplifier is that it's always putting out the same amount of power regardless of the SPL at the moment. 1s and 0s are changing, but the signal itself still has the same overall voltage feeding the same impedance, thus the current is constant as well, and power is voltage times current, so that stays constant. There's no risk of overloading the amplifier's output circuit, so the dynamic range of the microphone depends entirely on its input circuit (except if you use a control signal to set the amplifier to something other than its default gain). So in a well-designed digital microphone amplifier you have the best of both: the lowest possible input noise AND the greatest possible headroom, without having to raise the gain for low-level recording or reduce it for extremely loud sounds (e.g. sound effects recording). You can, of course, accommodate the softest and loudest sounds by adjusting the gain on your analog preamp--but for situations in which either or both extremes might reasonably occur and you can't (or don't want to) "ride gain" in real time, the digital approach wins over analog by quite a substantial margin.

Thanks you very much too Dsatz for the great explanation. I confused the impedances, I had the wrong idea. My question was about "the second benefit of digital microphones". Most of us record with analog microphones into digital recorders that have - preamplifier and a/d converter. What prevents these digital recorder from being designed with only a a/d converter as in digital microphone? They would thus obtain "the second advantage". I'll try to guess again. Each a/d converter of a digital microphone must be designed specifically for the type of microphone (range, sensitivity, impedance) and unfortunately a/d converter without preamplifier is not enough for several types of microphones. I will be glad if you correct me.