[David Grant]

Will it be possible (or maybe it already is) to digitize a mic level signal
with quality similar to current line level ADC standards? What is the
phenomenon (on a physical level) that prevents converter thresholds from
being set in this manner?

Dave



--
Posted via a free Usenet account from http://www.teranews.com

[Richard Crowley]

"David Grant" wrote ...
Will it be possible (or maybe it already is) to digitize a mic level
signal with quality similar to current line level ADC standards? What is
the phenomenon (on a physical level) that prevents converter thresholds
from being set in this manner?

The main problem to this point has been that mic levels are
way down in the noise of your typical ADC. It is not that
easy to make a good ADC, but to make one that operates
with the kind of performance we would want from good
audio, AND at microphone levels is beyond the capabilities
of current integrated circuit technology.

I've always thought it might be worth investigating using
a condenser mic as part of an RF tuned circuit, and then
using a frequency counter to recover the audio from the
radio frequency. Essentially a sound/capacitance-to-
frequency converter with counter. Voltage to frequency
conversion is the technology used by the really high-end
test&measurement equipment like Audio Precision, etc.

[Scott Dorsey]

David Grant wrote:
Will it be possible (or maybe it already is) to digitize a mic level signal
with quality similar to current line level ADC standards? What is the
phenomenon (on a physical level) that prevents converter thresholds from
being set in this manner?

Well, first you have to define what mike level is....
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Don Pearce]

On Tue, 3 Oct 2006 13:23:41 -0700, "Richard Crowley"
wrote:

"David Grant" wrote ...
Will it be possible (or maybe it already is) to digitize a mic level
signal with quality similar to current line level ADC standards? What is
the phenomenon (on a physical level) that prevents converter thresholds
from being set in this manner?

The main problem to this point has been that mic levels are
way down in the noise of your typical ADC. It is not that
easy to make a good ADC, but to make one that operates
with the kind of performance we would want from good
audio, AND at microphone levels is beyond the capabilities
of current integrated circuit technology.

I've always thought it might be worth investigating using
a condenser mic as part of an RF tuned circuit, and then
using a frequency counter to recover the audio from the
radio frequency. Essentially a sound/capacitance-to-
frequency converter with counter. Voltage to frequency
conversion is the technology used by the really high-end
test&measurement equipment like Audio Precision, etc.


Just go for the MKH series of Sennheiser mics - this is exactly how
they work.

d

--
Pearce Consulting
http://www.pearce.uk.com

[Jay Kadis]

In article ,
(Don Pearce) wrote:

On Tue, 3 Oct 2006 13:23:41 -0700, "Richard Crowley"
wrote:

"David Grant" wrote ...
Will it be possible (or maybe it already is) to digitize a mic level
signal with quality similar to current line level ADC standards? What is
the phenomenon (on a physical level) that prevents converter thresholds
from being set in this manner?

The main problem to this point has been that mic levels are
way down in the noise of your typical ADC. It is not that
easy to make a good ADC, but to make one that operates
with the kind of performance we would want from good
audio, AND at microphone levels is beyond the capabilities
of current integrated circuit technology.

I've always thought it might be worth investigating using
a condenser mic as part of an RF tuned circuit, and then
using a frequency counter to recover the audio from the
radio frequency. Essentially a sound/capacitance-to-
frequency converter with counter. Voltage to frequency
conversion is the technology used by the really high-end
test&measurement equipment like Audio Precision, etc.


Just go for the MKH series of Sennheiser mics - this is exactly how
they work.

d

Do they work by altering the frequency of oscillation or by the
amplitude of the resonant-frequency response? (I really should know
this...)

-Jay

--
x------- Jay Kadis ------- x ---- Jay's Attic Studio ----x
x Lecturer, Audio Engineer x Dexter Records x
x CCRMA, Stanford University x http://www.offbeats.com/ x
x---------- http://ccrma.stanford.edu/~jay/ ------------x

[[email protected]]

Scott Dorsey wrote:
David Grant wrote:
Will it be possible (or maybe it already is) to digitize a mic level signal
with quality similar to current line level ADC standards? What is the
phenomenon (on a physical level) that prevents converter thresholds from
being set in this manner?

Well, first you have to define what mike level is....
--scott

I don't have an exact amplitude in mind... I figure you'd want to
attenuate the signal most of the time to get it to the amplitude
desired by the ADC, as opposed to amplify it. Design the ADC for an
amplitude for which an average of 90% of recorded mic levels are above
said amplitude.

[Scott Dorsey]

Jay Kadis wrote:

Do they work by altering the frequency of oscillation or by the
amplitude of the resonant-frequency response? (I really should know
this...)

It's a tuned circuit with the capsule part of the tuning. Capacitance
changes, the frequency changes, and the output goes into a discriminator.
Get the schematic for the MKH104 which is a lot simpler than the more
modern variants.

If you had an oscillator going into a tuned circuit with the capsule part
of it, you'd do a lot better to check the phase of the output than the
amplitude. The math is in Terman if you want to see it.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Don Pearce]

On Tue, 03 Oct 2006 13:38:28 -0700, Jay Kadis
wrote:

In article ,
(Don Pearce) wrote:

On Tue, 3 Oct 2006 13:23:41 -0700, "Richard Crowley"
wrote:

"David Grant" wrote ...
Will it be possible (or maybe it already is) to digitize a mic level
signal with quality similar to current line level ADC standards? What is
the phenomenon (on a physical level) that prevents converter thresholds
from being set in this manner?

The main problem to this point has been that mic levels are
way down in the noise of your typical ADC. It is not that
easy to make a good ADC, but to make one that operates
with the kind of performance we would want from good
audio, AND at microphone levels is beyond the capabilities
of current integrated circuit technology.

I've always thought it might be worth investigating using
a condenser mic as part of an RF tuned circuit, and then
using a frequency counter to recover the audio from the
radio frequency. Essentially a sound/capacitance-to-
frequency converter with counter. Voltage to frequency
conversion is the technology used by the really high-end
test&measurement equipment like Audio Precision, etc.


Just go for the MKH series of Sennheiser mics - this is exactly how
they work.

d

Do they work by altering the frequency of oscillation or by the
amplitude of the resonant-frequency response? (I really should know
this...)

-Jay

The capacitance of the capsule forms part of the tuned circuit in an
RF oscillator, so the frequency changes as the capacitance does. An FM
discriminator follows this to produce the audio output.

d

--
Pearce Consulting
http://www.pearce.uk.com

[Jay Kadis]

In article ,
(Scott Dorsey) wrote:

Jay Kadis wrote:

Do they work by altering the frequency of oscillation or by the
amplitude of the resonant-frequency response? (I really should know
this...)

It's a tuned circuit with the capsule part of the tuning. Capacitance
changes, the frequency changes, and the output goes into a discriminator.
Get the schematic for the MKH104 which is a lot simpler than the more
modern variants.


I have one of the more modern schematics and you are certainly right.

If you had an oscillator going into a tuned circuit with the capsule part
of it, you'd do a lot better to check the phase of the output than the
amplitude. The math is in Terman if you want to see it.
--scott

Thanks. Terman's library is right down the street.

-Jay

--
x------- Jay Kadis ------- x ---- Jay's Attic Studio ----x
x Lecturer, Audio Engineer x Dexter Records x
x CCRMA, Stanford University x http://www.offbeats.com/ x
x---------- http://ccrma.stanford.edu/~jay/ ------------x

[Scott Dorsey]

wrote:

I don't have an exact amplitude in mind... I figure you'd want to
attenuate the signal most of the time to get it to the amplitude
desired by the ADC, as opposed to amplify it. Design the ADC for an
amplitude for which an average of 90% of recorded mic levels are above
said amplitude.

Your goal in life is to have the highest possible S/N. This means keeping
levels as high as possible and never attenuating anything if you can help
it.

If we could get some semiconductor technology to operate at decent voltages,
we'd be a lot better off. Give me a monolithic DAC that will run on +/-48V
rails and it'll have a lot higher S/N than a +/-5V unit with the same amount
of device noise.

There are ways to reduce the noise of a particular semiconductor device by
changing the geometry, but there is always a compromise between operating
voltage, noise, linearity, output drive, input impedance, etc. With
monolithic stuff, the designer is very limited in the types of devices he
can put together on one chip (and of course things like resistors become
difficult and expensive on some fab lines and impossible on others, while
capacitors are mostly impossible for all of them). This is part of the reason
why discrete technology is still so popular in most higher grade audio gear
today.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Ron Capik]

Scott Dorsey wrote:

....snip..

Well, first you have to define what mike level is....
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Doesn't that have something to do with equal
drool from th... umm, ooops. Never mind.
[ Sorry, couldn't resist. ]


Later...

Ron Capik cynic-in-training
--

[Mark]

If you had an oscillator going into a tuned circuit with the capsule part
of it, you'd do a lot better to check the phase of the output than the
amplitude. The math is in Terman if you want to see it.

this is true if the swing is small....

if the low frequency swing is large however, there will be a problem
because most phase detectors are not linear beyond +/- 360 deg.

Mark

[Eeyore]

David Grant wrote:

Will it be possible (or maybe it already is) to digitize a mic level signal
with quality similar to current line level ADC standards? What is the
phenomenon (on a physical level) that prevents converter thresholds from
being set in this manner?

Cross-posted to s.e.d

Graham

[Mike Rivers]

David Grant wrote:
Will it be possible (or maybe it already is) to digitize a mic level signal
with quality similar to current line level ADC standards? What is the
phenomenon (on a physical level) that prevents converter thresholds from
being set in this manner?

Neumann is already doing it with their Solution D series. The thing is
that a microphone signal has a huge dynamic range and they have to play
tricks with A/D conversion in order to accommodate the lowest and
highest level signals. It's not cheap, but it works quite well. You can
go to the Neumann web site and find a paper on how the Solution D A/D
conversion works.

[David Grant]

Neumann is already doing it with their Solution D series. The thing is
that a microphone signal has a huge dynamic range

To my understanding, no more so than the resultant line level signal comming
from a mic preamp and feeding a regular ADC converter (unless a compressor
is in the chain). You set your preamp trim to get your line level signal's
magnitude to fit within the highest ADC threshold. Don't see why you
couldn't do something similar to a mic level signal using a bit of gain or
attenuation.

Someone mentioned that the problem with attenuation is that it yeilds a
lower S/N ratio but the intruiging thing to me is to hear a microphone's
output without the effects of any class A, B, or A/B active circuitry in a
preamp.




--
Posted via a free Usenet account from http://www.teranews.com

[PeteS]

Eeyore wrote:

David Grant wrote:


Will it be possible (or maybe it already is) to digitize a mic level signal
with quality similar to current line level ADC standards? What is the
phenomenon (on a physical level) that prevents converter thresholds from
being set in this manner?


Cross-posted to s.e.d

Graham


What type of mic? An electret type has an output of a few 10s of
millivolts RMS, but has a fairly high output impedance.

On the basic question, such levels can be successsfully digitised, but
the issues become PSRR, CMRR (for differential signals) and system and
thermal noise at very low level signals.

As mics typically have relatively high impedance outputs, they are not
suitable for direct connection (even through caps) to most A-Ds (which
have fairly low impedance input networks), so buffers are usually
required. That said, if you have a buffer, it may as well be an
amplifier to give a full swing of a volt or two, which is easily digitised.

I would be interested as to why you asked.

Cheers

PeteS

[David Grant]

"PeteS" wrote in message
...
Eeyore wrote:

David Grant wrote:


Will it be possible (or maybe it already is) to digitize a mic level
signal
with quality similar to current line level ADC standards? What is the
phenomenon (on a physical level) that prevents converter thresholds from
being set in this manner?


Cross-posted to s.e.d

Graham


What type of mic? An electret type has an output of a few 10s of
millivolts RMS, but has a fairly high output impedance.

I had typical studio quality condenser microphones in mind

On the basic question, such levels can be successsfully digitised, but the
issues become PSRR, CMRR (for differential signals) and system and thermal
noise at very low level signals.

Not sure about PSRR but if you digitized both sides of the balanced signal
outputted from one of these mics you could perform CMR in DSP, couldn't you?
I'd expect it to be fairly effective and easy to implement.


As mics typically have relatively high impedance outputs, they are not
suitable for direct connection (even through caps) to most A-Ds (which
have fairly low impedance input networks), so buffers are usually
required. That said, if you have a buffer, it may as well be an amplifier
to give a full swing of a volt or two, which is easily digitised.

I would be interested as to why you asked.

As different classes of amplifiers and different amplifier topologies each
have their own sonic "signature" I'm curious to hear a microphone's output
without the effects of a class A, B, A/B microphone preamplifier (granted
there may be a different sonic signature associated with the electronics
necessary to digitize such a low amplitude signal but my curiousity
remains).

Also, if the preamp can be cut from the chain and the cost of doing so in
quality fashion isn't exorbitant, this could means pretty big cost savings
(for someone who WANTS their signal in the digital domain, of course).


Cheers

PeteS



--
Posted via a free Usenet account from http://www.teranews.com

[Eeyore]

PeteS wrote:

Eeyore wrote:
David Grant wrote:

Will it be possible (or maybe it already is) to digitize a mic level signal
with quality similar to current line level ADC standards? What is the
phenomenon (on a physical level) that prevents converter thresholds from
being set in this manner?

Cross-posted to s.e.d

Graham

What type of mic? An electret type has an output of a few 10s of
millivolts RMS, but has a fairly high output impedance.

You buffer them with a jfet and the output impedance drops to a few hundred ohms.



On the basic question, such levels can be successsfully digitised, but
the issues become PSRR, CMRR (for differential signals) and system and
thermal noise at very low level signals.

As mics typically have relatively high impedance outputs

Dynamic mics are just a few hundred ohms actually.


, they are not
suitable for direct connection (even through caps) to most A-Ds (which
have fairly low impedance input networks), so buffers are usually
required. That said, if you have a buffer, it may as well be an
amplifier to give a full swing of a volt or two, which is easily digitised.

Mics can deliver several *volts* of output when presented with a high level
input.

Graham

[hank alrich]

David Grant wrote:

Someone mentioned that the problem with attenuation is that it yeilds a
lower S/N ratio but the intruiging thing to me is to hear a microphone's
output without the effects of any class A, B, or A/B active circuitry in a
preamp.

You need _something_ to buffer the output of the capsule for impedance
matching. What'll you have?

--
ha

[David Grant]

You need _something_ to buffer the output of the capsule for impedance
matching. What'll you have?


Hadn't thought of that... Would a transformer do the trick? Note that I'm
not thinking about cutting out the gain stage that is in the mic body - I'm
toying with the idea of replacing outboard preamps.




--
Posted via a free Usenet account from http://www.teranews.com

[Mike Rivers]

David Grant wrote:
Neumann is already doing it with their Solution D series. The thing is
that a microphone signal has a huge dynamic range

To my understanding, no more so than the resultant line level signal comming
from a mic preamp and feeding a regular ADC converter (unless a compressor
is in the chain). You set your preamp trim to get your line level signal's
magnitude to fit within the highest ADC threshold. Don't see why you
couldn't do something similar to a mic level signal using a bit of gain or
attenuation.

Thing is that if you put the A/D right on the capsule, you don't have a
trim control to get it within range. That's what makes the Solution D
so special

[Paul Stamler]

"David Grant" wrote in message
.. .


You need _something_ to buffer the output of the capsule for impedance
matching. What'll you have?


Hadn't thought of that... Would a transformer do the trick? Note that I'm
not thinking about cutting out the gain stage that is in the mic body -
I'm
toying with the idea of replacing outboard preamps.

Not really; you need something with a very high input impedance, higher than
is practical for a transformer. The only things anyone's tried in my
experience are FETs and vacuum tubes. (Well, and perhaps an opamp with a FET
at its input.)

Peace,
Paul

[Anahata]

David Grant wrote:

I'm
not thinking about cutting out the gain stage that is in the mic body - I'm
toying with the idea of replacing outboard preamps.

I don't think there's any future in that.
What do you hope to achieve by doing it that way?

All the problems of noise and distortion that are present in a mic
preamp are still there in whatever technology you use in the A-D
converter. However you do the conversion there's at least one analogue
comparator in there and that has the same noise issues as any amplifier,
in fact liable to be worse since its performance is optimized
differently. And anything producing a digital output tends to produce
noise itself that you have to keep away from the input.

Better to design the A-D to operate at whatever level gives it best
dynamic range and linearity, then boost the mic output to that same
range with an amplifier designed solely to do that one job as well as
possible.

Anahata

[David Grant]

All the problems of noise and distortion that are present in a mic preamp
are still there in whatever technology you use in the A-D converter.
However you do the conversion there's at least one analogue comparator in
there and that has the same noise issues as any amplifier, in fact liable
to be worse since its performance is optimized differently. And anything
producing a digital output tends to produce noise itself that you have to
keep away from the input.

Supposing you have A-D that IS optimized however (which is what my original
post was inquiring about). You're always going to have some noise associated
with A-D - I understand that - but if you can somehow (someday, perhaps with
some future technology) get A-D to perform as well on these low-amplitude
signals as current A-D performs on higher amplitude signals then overall
noise added through the system could be less (think of it as effectively
yanking the preamp - a noise source - out of the chain). Could there not be
some decent gains to be had?

Better to design the A-D to operate at whatever level gives it best
dynamic range and linearity, then boost the mic output to that same range
with an amplifier designed solely to do that one job as well as possible.

I understand the design mentality... it's sort of like RISC - design the
processor to perform fundemental functions really well rather than get
mediocre performance of complex (combined fundemental) functions. I
understand that the reason things are done the way they are right now is
because it's what works best. I'm talking about years down the road however.

I'd be surprised if with all the advances in semiconductor technology we
won't one day be able to very accurately digitally measure small electrical
signals without first having to condition them so drastically - at a cost
that isn't prohibitive. Although, I wonder if such advances wouldn't lead to
even greater noise reductions in existing topologies. I don't really know
enough about it, as you can probably tell from my questions, but that's why
I'm asking them.



--
Posted via a free Usenet account from http://www.teranews.com

[Paul Stamler]

"David Grant" wrote in message
.. .

I'd be surprised if with all the advances in semiconductor technology we
won't one day be able to very accurately digitally measure small
electrical
signals without first having to condition them so drastically - at a cost
that isn't prohibitive. Although, I wonder if such advances wouldn't lead
to
even greater noise reductions in existing topologies. I don't really know
enough about it, as you can probably tell from my questions, but that's
why
I'm asking them.

It's certainly possible, and carries the advantage of being able to do cable
runs without interference problems, which could be very useful in
electrically noisy environments. (But, as Anahata pointed out, you then have
digital circuits sitting next to low-level microphone signals inside the mic
body, which has its own problems.)

Thing is, in terms of noise, existing preamp designs can get very quiet,
down within a decibel or two of a dynamic mic's inherent self-noise. That's
pretty close to perfect from a noise point of view, at least until you start
to deal with ribbons.

Peace,
Paul

[Kevin Aylward]

David Grant wrote:
"PeteS" wrote in message
...
Eeyore wrote:

David Grant wrote:


Will it be possible (or maybe it already is) to digitize a mic
level signal
with quality similar to current line level ADC standards? What is
the phenomenon (on a physical level) that prevents converter
thresholds from being set in this manner?


Cross-posted to s.e.d

Graham


What type of mic? An electret type has an output of a few 10s of
millivolts RMS, but has a fairly high output impedance.

I had typical studio quality condenser microphones in mind

On the basic question, such levels can be successsfully digitised,
but the issues become PSRR, CMRR (for differential signals) and
system and thermal noise at very low level signals.

Not sure about PSRR but if you digitized both sides of the balanced
signal outputted from one of these mics you could perform CMR in DSP,
couldn't you? I'd expect it to be fairly effective and easy to
implement.

Not a good idea to to CMR subtraction in the digital domain. Its one of
the few areas where analogue processing will **** all over the digital
approach. A well designed differential pair input could actually achieve
140db+ CMRR.

Its pretty much a given, that the best way to digitise a signal is get
it up to several volts in the analogue domain, say 5-10V, and then a/d
it. Trying to directly digitise 10mv at even 16 bits is like 0.16uv for
an LSB. It aint gonna happen mate, well not unless some new physics
device becomes available.



As mics typically have relatively high impedance outputs, they are
not suitable for direct connection (even through caps) to most A-Ds
(which have fairly low impedance input networks), so buffers are
usually required. That said, if you have a buffer, it may as well be
an amplifier to give a full swing of a volt or two, which is easily
digitised. I would be interested as to why you asked.

As different classes of amplifiers and different amplifier topologies
each have their own sonic "signature" I'm curious to hear a
microphone's output without the effects of a class A, B, A/B
microphone preamplifier (granted there may be a different sonic
signature associated with the electronics necessary to digitize such
a low amplitude signal but my curiousity remains).

This is really a bit of an old wives tale. Modern high quality, decently
designed amplifiers all sound same. The mic and speaker frequency
response and non-linear distortions vastly overshadow any amplifier
blemishes.


Kevin Aylward B.Sc.

http://www.anasoft.co.uk
SuperSpice, a very affordable Mixed-Mode
Windows Simulator with Schematic Capture,
Waveform Display, FFT's and Filter Design.

"There are none more ignorant and useless,than they that seek answers
on their knees, with their eyes closed"

[Chris Hornbeck]

On Thu, 05 Oct 2006 16:16:52 GMT, "Paul Stamler"
wrote:

Thing is, in terms of noise, existing preamp designs can get very quiet,
down within a decibel or two of a dynamic mic's inherent self-noise. That's
pretty close to perfect from a noise point of view, at least until you start
to deal with ribbons.

Much thanks, as always,

Yeah, noise is so often counter-intuitive that all kinds
of things get conflated, and the idea of noise figure is
often missed.

Thermal noise is caused by the very same mechanism in
amplifying stages and in A/D convertor ladders, and in
wires, and in...

Just leads back to Hank's comment, "You need _something_
to buffer the output of the capsule for impedance matching.
What'll you have?"

The OP would probably enjoy a search on the terms "noise
figure" and "thermal (/or) Johnson noise", sometimes lately
also called Nyquist noise. This is the fundamental limit
in all circuits, and everything else was classically called
"excess noise".

Lotsa fun stuff, and a few Wikipedia entries sampled seemed,
strangely, pretty much right on.

Always good fortune,

Chris Hornbeck

[David Grant]

The OP would probably enjoy a search on the terms "noise
figure" and "thermal (/or) Johnson noise", sometimes lately
also called Nyquist noise. This is the fundamental limit
in all circuits, and everything else was classically called
"excess noise".


haha! I remember we covered that in 3rd year ECE. Sadly that was the year I
did next to zero work and squeezed by with a marginal pass. I'll re-learn
it, but this time not on two hours sleep nor with a looming exam 45 minutes
away.



--
Posted via a free Usenet account from http://www.teranews.com

[Chris Hornbeck]

On Fri, 06 Oct 2006 03:21:44 GMT, Chris Hornbeck
wrote:

Lotsa fun stuff, and a few Wikipedia entries sampled seemed,
strangely, pretty much right on.

I take that back. The http://en.wikipedia.org/wiki/Johnson_noise
topic called "Noise in decibels" is AMAZINGLY wrong.

Sorry,

Chris Hornbeck