Topic: Digital Microphones and Their Advantages: Is This the Future?

[H₂O]

Hhb has had a memory card recording microphone for 3 years or so called flashmic

[MIQ]

Sure I can see that being a product already.  The really inventive part, that I'd look forward to, is that all the mics in use are wirelessly sync'ed and you are left with capsule choice, placement, and what you want to do in post.

-MIQ

[rastasean]

Hhb has had a memory card recording microphone for 3 years or so called flashmic

Yeah, I remember seeing that in the b&h catalogue/website and thought it was rather expensive but I don't recall the actual price.

[kleiner Rainer]

hi all,

there are two objections to use WLAN technology to link digital mics and recorders:

- security
- capacity

Security: a venue operator could shut down stealth taping with a simple jammer. Ever thought of that?

Capacity: there are limited numbers of WLAN channels available, with a limited channel capacity. Imagine arriving late at a venue only to find out that all "channel capacity" is used up by fellow tapers using 24/96. What now?

This cannot happen with a wire connection. As long as there is still space to squeeze in a cable, you have your own secure channel with nearly unlimited capacity.

Think about it.

Greetings,

Rainer

[Gutbucket]

Capacity: there are limited numbers of WLAN channels available, with a limited channel capacity. Imagine arriving late at a venue only to find out that all "channel capacity" is used up by fellow tapers using 24/96. What now?

Wireless bandwith for sync & control only is minimal if the high resolution audio data is stored locally.  This is a taper specific solution though, since the live feed is unavailable.

Since I'm digital dreaming and Jon has mentioned DSP, another thing I've invisioned in my George Jetson recording future is independant and selectable convolution for each mic capsule-  Used to match less than perfect capsules and adjust them to the desired response.  Use a little management app to select a basic set of mic convolutions that correct and match each of a set of capsules, and combine those with secondary sets of convolutions that adjust for specific applications: degrees of free field vs diffuse EQ, boundary mounting, adjusting for mounting my miniature omnis in small spheres (or big head sized ones), and odd user-specific setups like adjusting response for body mounted configurations.  Select the desired convolutions, combine them for each microphone and load the results into the DSP.

[John Willett]

The one thing that is being forgotten is latency.

An AES42 digital microphone has a latency so short that it's about the equivalent of, I think, about 13" of the speed of sound in air.

This means that it is not noticeable and the microphones can be used in a live situation.

As soon as the latency gets above about 2.5 to 3ms then it becomes noticeable.

[SmokinJoe]

I'm still interested what people think of the following... let's say we have a digital mic...  Your diaphragm/capsule can tolerate some loud SPL before it's sonic accuracy really goes to hell.  For the sake of argument lets say it's 150db, and we balance things so our A/D is 1111 1111 1111 1111 1111 1111 (24 bit digital max) at that point.  If you are recording audio at 75db, you are utilizing only about 1/5000th of your A/D's bandwidth.  Is that a big deal? I don't know.  I would think people would be freaking about that.

Here are a couple analogies:
- If someone on here ran Analog condenser Mics > phantom power supply > high end A/D (with no preamp) and added 75db of gain in post everyone would tell him he's nuts.
- If someone had a high-end digital camera with a bazillion pixels, and only used a 28mm lens, when things were far away you just digital zoom and crop in post.

How is this digital mic situation different from my analogies?

[Hypnocracy]

Smokin...I find I prefer the results of running my analog Preamp HOT and bumping along with peaks on the digital side close to cliping too...doing so it is almost like compression...

The days of running everything at low 12db peak because it is 24 bit...is somewhat hollow to me...I'd almost rather run closer to the edge and have 16bit...

for example...I had ran a string band that needed some boosting to just be in the 12db 24bit range...then the next band came on stage and rocked a sound check...I was distracted at first and came over to check the levels and saw I would be turning it down for the set...

Once I got home and heard the results....IT BLEW ME AWAY...I'm almost sure I ran right up against the V2's clip point...and I know where to set the FR2LE to match the max V2 signal with a level that approaches the digital clip point...

http://soundcloud.com/hypnocracy/preston-frank-with-donna-the

[SmokinJoe]

Hypno, that is probably largely a change in flavor. I'm talking about resolution, and quantization noise.

To make the math cleaner I'll use 60db instead of 75. suppose full scale on ADC happens at 140 and quiet studio recording at 80db. With 60db below max the voltage is 1/1000th of max. 10bits is a factor of 1024, which is close enough to 1000. This means we are only using about 14 bits out of 24 and the rest is zeros. I recorded good sounding tapes at 16 bit, and I do record at -12 at times because I'd rather let it ride than clean up my gain adjustments in post, so I can't say that 14bit resolution is crap, but I think the whole thing is worthy of consideration.

 The trade off is that this quantization noise is pretty much the only noise anywhere in the system, no chance of noise from ground loops, preamps, cell phones, or any of that stuff.  Probably the overall system is improved, but I'd really want to understand the tradeoffs before I dropped big bucks on a digital mic rig.

[SmokinJoe]

I'm sure you're right Jon, you have more experience than I do with that kind of thing in audio.

I work with A/D's on industrial equipment, much of which is 12bit.   If you have a 0-10VDC = 0-4095 A/D and if we have a sensor that only puts out a tiny portion of full scale, you get digital readings that are clearly visible in discrete steps and they are coarse enough to matter.  The solution is to match the sensor's output span to the A/D span, more so than buying a new A/D with greater bit depth.  Anyway I take that experience and apply it to audio A/D's, and I think "if you are using less than 1% of the A/D input range, that's got to act the same way."  Maybe our eyes are a lot more fussy than our ears.

[John Willett]

The Neumann ADC in an AES42 mic. is 28-bit (although the final output is 24-bit).

With the ADC right by the capsule you have done away with all the analogue circuitry.  The line driver amplifier in the mic. has gone, the analogue mic. cable has gone, Rf interference has gone, etc..  You have lost a lot of circuitry that all adds noise and problems and you can hear the difference.

All I would say is just try it for yourself and make up your own mind.

I find AES42 mics great and they are obviously quieter compared to analogue.

[dream]

I find AES42 mics great and they are obviously quieter compared to analogue.

I really like my (Neumann) AES42 mics ... rarely use something else now. But I would like to add some experiences concerning noise. Neumann asked me to do a
comparison between an analog TLM103 and the digital TLM103D (I own one). For this test I recorded a very quiet watch. The analog 103 went into a Grace Design Lunatec V3 and the
digital 103 was used with a DMI-2p (as if that matters). On the V3 I added full 70dB gain. I level matched the two recordings and the noise floor and the sound was practically the same.
But generally I would agree that digital microphones are quieter in most applications. In this special case the analog 103 and the V3 are very low noise devices anyway.

[DSatz]

SmokinJoe, you're right that microphone capsules can generally handle sound pressure levels far greater than what we find in live music recording, such as 150 dB. But the electronics of most microphones can't handle such high levels without some form of technical intervention--either the use of a capacitive attenuator, or decreasing the polarization voltage on the capsule. In general modern condenser microphones tend to have something more like 130 dB SPL as the upper limit of their dynamic range. In most live situations that seems to work out well enough, especially since that's well into the range where hearing damage becomes a serious issue withing a relatively short time of exposure. And the range from there down to the noise level of the quietest microphones available corresponds to 20 bits and change, not quite 21 bits. So that is really not a problem for a 24-bit AES/EBU data channel.

Of course the quality of the analog electronics and A/D converters in microphones is just as much an issue as it is with traditional analog microphones and outboard A/D converters. Just because those stages are built in to the microphones doesn't earn them automatic exemption from criticism.

The situation that you describe with industrial A/D converters is NOT applicable to audio recording. In audio applications the analog signal is always dithered prior to (or as part of) conversion--and when properly done, this completely eliminates the stairstepping that you refer to. That problem was solved back in the 1980s! Were that not the case, the THD of a tone would increase sharply if you lowered its recorded level to a point just above the noise floor. But that does not occur, as both measurements and listening tests make clear.

--best regards

[John Willett]

I find AES42 mics great and they are obviously quieter compared to analogue.

I really like my (Neumann) AES42 mics ... rarely use something else now. But I would like to add some experiences concerning noise. Neumann asked me to do a
comparison between an analog TLM103 and the digital TLM103D (I own one). For this test I recorded a very quiet watch. The analog 103 went into a Grace Design Lunatec V3 and the
digital 103 was used with a DMI-2p (as if that matters). On the V3 I added full 70dB gain. I level matched the two recordings and the noise floor and the sound was practically the same.
But generally I would agree that digital microphones are quieter in most applications. In this special case the analog 103 and the V3 are very low noise devices anyway.

Good test and is what I would expect - a minimalistic approach with a single mic., short cables, high quality mic. pre. going into a single line in; the noise difference would be extremely small.

The advantages in AES42 are in eliminating the analogue stage at the earliest possible moment - multiple mics, long cables, analogue mixers, etc. all add their little bit of noise and in a big sound system where the sound remains fully digital from the mics right up until the ADC which outputs directly to the power amplifier can be very large.