To me the main problem become word clock control and converter changes over time? Much less the following thought.

So you want to change the way it works from converters attached to the workstation to converters built into the mic's? I'm just now getting used to digital tape and digital mixing. You wan't to go digital Mic's too? If we get a digital board, what do we do with our existing mic's and preamps?

Digital is better than analog for tape. But not for mixing and mic's. At least not this year. Will be a cold day in hell to pry all the external reverbs away from people and go completely digital.

I don't think anyone is suggesting that analog gear goes away. There are significant advantages to putting the ADC as close as possible to the sound source, though, and they're outlined in the papers presented in this thread. This isn't just for recording - the advantages in noise immunity and optimized gain structure have serious implications for live sound too! Think of a mic with automatic feedback handling, for instance.

Of course you'll keep your analog gear, but it's because you LIKE the imperfections inherent in that gear, and you use it to your creative advantage.

Mark my words - in 20 years, I predict that the "normal" audio path will be ADC as close to the source as possible, transport over Ethernet and IP between devices (wired and wireless), and DAC conversion back as close as possible to the output devices. You'll still ship audio out to and back from analog devices through DA/AD outboard interfaces, but everywhere else it's gonna be bits in packets.

A digital mic is in part analog.. It's just a mic with a converter in it
it has to have some mechanical device in

Well yes but; the Neumann digital mics go straight from diafragm>impedance converter> AD converter.

So a minimum of analog devices and the AD converter knows exactly what signal to expect.

The "analog gain" adjusts the voltage coming from the diafragm. Lowering this raises the maximum SPL the mic can handle.

Neumann DMI-8 via AES to Lawo stagebox. This box is connected via ravenna to a Lawo MC2 36 for the broadcast mix. (there is a second MC2 36 in the network for the FOH mix)

Huub

You need full mode-2 control to change settings.

But you *can* adjust the level (in mode-1) if you have the AETA 4MinX.

John, if by "level" you mean gain it can be done in the 788T also.

Good, I wasn't sure - some recorders have no level control at all on the digital inputs.

There is control for gain at recording box, but Mode 2 allows separate control for gain at mic, right?

Just for everyone's information:
there's true analog gain, which is the voltage of the diafragm.
Basically only needed in extreme circumstances.
Then there's digital gain in the mic itself (before the internal brickwall limiter)
and there's digital trim after the mic, in the console or DAW.

We have only been using digital trim, because in most circumstances this is all that is needed.
Because of the 2 AD converters there is close to no noise and like I said earlier, the AD converters know what level to expect.

Huub

I am hoping Merging will release a Mode 2 AES42 card for Horus/Hapi. We are lucky to have the Merging marketing manager in Aus next week and I am going to ask him about AES42 support.

With an optimized AD converter and limiter in the mic, i'd think that trim while recording would be rare (since you don't worry about clipping). Maybe in live sound you'd trim up the volume if the source is relatively quiet?

I'd like to see a digital mic with DANTE and ip based remote control. Use POE for phantom, and you've solved your noise/splitting/etc - just route the bits where needed, and your stage boxes are ethernet switches.

Not really. The concept of gain implies amplification. And with digital microphones, that can be a non sequitur because of the lack of amplifiers in the chain.

For example, in the Neumann KM D, there doesn't seem to be an amplifier: "The KM D digital output stage converts the signal from the capsule directly into a digital signal." Yet, they go on to say: "The integrated digital signal processor (DSP) permits remote control of the microphones, including the digital gain..." I think of digital gain (and I could be very wrong) as pushing the digits back and forth in their 16 or 24 bit envelope. I visualize it sorta like a histogram in digital photography, flawed as that analogy might be.

My point is, if a microphone chain doesn't use microphone preamps, the concept of gain needs to be rethought.

In my fuzzy little mind I'm thinking that setting the digital gain at the microphone (via AES 42) is kinda/sorta like setting the trim (input), and twiddling the pots on the front of the recorder is sorta like setting a fader (output). But in an all digital system, both these things are just shoving the bits back and forth in that 16-24 bit wide envelope.

This is going to make people (OK, me) scratch their heads over the concept of gain staging. In a digital system, is gain staging even defined?

There will always be a need for some form of amplification, because of the nature of a capacitor transducer. Even if it is only in the form of impedance transformation between capsule and ADC input, which is practice is a power gain (unity voltage gain, but a power gain because the output has higher current capability).

And as far as RF capacitor mics are concerned, there needs to be an audio stage of some sort incorporating an FM discriminator to produce the voltage that can be passed on to the ADC. (No-one, it seems, has managed to produce a practical direct frequency counter which would offer the prospect of direct conversion to a digital audio value without going through an audio stage.)

There needs to be some clarifcation between gain adjustment to present the signal from the capsule at a level optimum for the dynamic range of the incident sound and the capability of the ADC (and hence recording medium), and gain adjustment to include that signal at the desired value in the final product. With a digital microphone, the first transfer function is solely under the design control of the microphone designer, and is probably valued for what it excludes rather than what it includes. Yes, the idea of "voltage gain" no longer directly relevant. It is more a transfer or conversion function, though it can be referred back to a voltage output from a reference DAC system when it is returned to audible form. The applicable concept is "sensitivity" - the SPL that corresponds to -x dBfs. Most digital mic manufacturers still seem to quote in x mV/hPa.

Yes, this makes sense. But this impedance transformation hardware is probably not adjustable to the end user. Or is it? I'm thinking this is a design decision by the design engineer of the microphone -- it's basically a parameter of the ADC circuit, to give the ADC what it wants.

I can see that the Sennheiser / Rode RF condensers might be a little more complicated between the capsule and the ADC. But again, would there be any reason to make this transform function available to the end user to manipulate? I'm thinking not, but IDK.

That's what I'm looking for -- a clarification of those two functions. I'm thinking the first, the transfer function from capsule to ADC, it basically a "black box" to the user. So the only thing the end user sees, is the adjustment that gets the signal to the desired value in the final product. And that adjustment isn't related to analog gain, but is moving the bits closer to or farther away from 0 dBfs. So to the end user, it feels like a fader.

How am I doing? I'm hoping I'm getting closer...

Good point on the whole "body with different capsules" scenario... I'd still think that if I were setting the functional specs for such a device I'd have the gain/impedance matching/whatever-you-call-it be a specific value for a specific capsule such that the maximum signal the capsule can physically generate without significant distortion is maybe -1 dBFS, and not user adjustable. Choosing that specific SPL would require careful listening, since some small amount of distortion might be deemed pleasing.

Gone are the need for analog pads - the analog portion of the circuit simply CANNOT be driven beyond its limits unless the mechanical portion of the microphone is being driven beyond its limits. If the mic sounds bad, you know it's a mechanical distortion issue and you have to address it by placement or choosing a new capsule.

Perhaps the interchangeable capsule could have the appropriate analog electronic circuitry in it, so that the main body of the mic is set to expect certain signal strengths period.

Slight digression: the 788T has gain control in the digital domain for AES 42 1 digital mics. If Schoeps would just manufacture the CMD bodies that worked to their satisfaction I am sure they would be working to my satisfaction. To have my current capsules still useable, and they are good, against a background of inky-black silence would keep me happy for a long, long time.

There are several reasons digital mics haven't caught on. The main one is that few problems are actually solved, and a few new ones are introduced. Having led the AES standards process to define AES42, I can report that the issues were known then and haven't changed. Mode 1 works fine but it means that every mic signal will go through a sample rate converter, and most people would like to avoid that although good SRCs are much easier to do now than when the interface was designed. Mode 2, which was designed by Wolf Buchleitner and Stephan Flock, is a much better approach, but is not supported by most equipment. And most of us know how rapidly converter technology evolves. If you hear a digital mic that sounds good to you, by all means use it, but most people like the options they have with the analog interface, no matter how limiting it is.

The reason I spent the time I did on this project was to get the concept in play so that there would be places for true digital mics (of which there still aren't any on the market) once they became practical. Everything is still an analog mic with an a/d converter inside. There are several great ideas for mics (different capsule designs) that aren't practical using analog technology ... and I think it would be better to wait for them. But I haven't seen a business case that they would be that world-changing, so no one seems willing to spend the couple million dollars it would take.

Thanks for the reply.

Yes, I am waiting for a similar solution as well, in our case I'd prefer Ravenna but yes.

We're using a bit of trimming because we're making a mix, not just recording.

Yes - and a lot more.

Well, Sennheiser digital mics do not have the "wire to AD" solution, they have a traditional mic with the AD section straight after the analog circuitry.

Also sounds very good and detailed and clean.
But not the purist Neumann approach.

But the question is how? Is it changing the sensitivity (analog) or just level shifting post conversion (which can just as effectively be done at the receiving end)?

The system engineering problem is the matching of the capsule transfer characterisitics to the conversion process. Our basic requirements are to make the capsule noise dominant at the low end, and accommodate as high an SPL as possible at the top. Let's assume that capsule equivalent noise is single figures dB SPL. While for minimal influence the convertor noise should be 10dB lower, let's set our spec at hearing threshold (0 dB SPL). Now what is a practical upper threshold for general use (wighout using an attenuator for special purpose applications - jet engines, kick drum, et al)? Something in excess of 120dB SPL might be a good starting point.

Current convertor technology can deliver about 20- to 21-bit equivalent performance (dynamic range 120dB +/- 3dB), so some compromise is currently necessary. The early approach of a hybrid two-stage design allowed a figure apparently in excess of this at the expense of high complexity and precision in manufacture. Single chip designs can deliver the 120dB SINAD - how to use it? The Schoeps compromise was to set the convertor noise equal to the capsule noise (around 10 dB SPL equivalent) and set -1dBfs at 128dB SPL. With the signal fitting in this window, there is no particular need for gainshifting in the microphone system, except in the situation where the signal is to be directly delivered to a mix, for keeping faders in a preferred working range. (As a direct broadcast engineer, personally this is a valuable feature.)

So, I still remain convinced of the value of digital mic systems, if they deliver optimal sound to digital conversion. Avoiding further analog degradation in transmission from my revered MK21 capsules to final destination is the most valuable benefit. So how about it, Schoeps?

The Neumann KM-D has a 28-bit ADC and uses two 24-bit converters in a patented method.

The capsules are matched to the converter.

The gain is in the digital domain and is really only needed so you can monitor.

A solution of its time, to prove a point, it seems to me. I consider it similar to the BBC's development in the ' 70s of NICAM for Sound-in-Syncs TV audio, where 14-bit ADCs were available but only 10 bits per channel was available for transmission, and there was a similar 4-bit (24dB) effective gain ranging. The near-instantaneous approach (over 32 samples) to companding avoided the more obvious noise modulation of the instantaneous companding digital systems (A-law, µ-law). The Neumann solution was sophisticated and effective, and of course with the noise floor 120dB down instead of 60dB, the noise modulation problem was several orders of magnitude less.

Current ADC chip technology is capable of delivering satisfactory performance in all but the most extreme environments. The biggest problem is the market, which is not yet ready to shift its paradigms to allow the full embrace of the technology.

What do you mean with "wire to AD"? Any digital microphone is analogue with a FET stage and at least two stages of analog gain, although Schoeps in its CMD2u only used one fixed gain, and this microphone was noisy for low signals. Using two or more gain levels is called gain staging, and that is how the microphone does not need to be adjusted. For loud signals the low amp stage is used, for soft signals the high gain stage is used, each having their own AD channel. Clever processing switches or cross-fades between those modes. Prism and Stage Tec introduced these techniques for their AD converters already in the 90s. This way it is possible to create an AD stage with an dynamic range of 28 bits, but not a signal to noise ratio of 28 bits.

I mean: the neumann mics have diaphragm>impedance converter>AD
Senneheiser mics have the whole "traditional inner electronics of the microphone">AD

I don't think the electronics of a digital microphone differ so much from the conventional path. Basically it is the conventional path + AD converter. It is not that simple as you state above.

Digital microphone:

diaphragm>impedance converter>gain staging>output stage feeding AD>AES out.

Conventional microphone:

diaphragm>impedance converter>output stage capable of feeding long wires

I did not see less electronics in the CMD2u from Schoeps, I saw actually more electronics.

No, I am correct here
Please ask the manufacturers.

There is no analog gain staging in the neumann, analog gain is adjusting the diaphragm voltage.

Here you can see the patent of Otmar Kern.

Patent US6707402 - Method for converting an analog signal to a digital signal - Google Patente

http://patentimages.storage.googleap...316-D00002.png

The principle of the Neumann converter is gain staging, only that it differs from the "common gain staging theory" that the second converter never clips if the signal becomes high. This because as you mention, indeed the signal from the capsule is adjusted, but it is swithing anyway as any gain staging set-up, Neumann says only less critical in their system, but I have never heard Stage Tec or Prism switch. Neither there is a "wire" of gain. The first set-up gain is 30 dB minus the 6 dB of the second converter substracted, gives a total of 24 dB.