Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Condenser Mic Moise

46 views
Skip to first unread message

seanbrode...@yahoo.com

unread,
May 14, 2013, 9:42:34 PM5/14/13
to
I'm wondering if anyone knows what the typical dominant source of
noise is in an externally polarized condenser mic utilizing a Jfet for
impedance conversion. Is it the Johnson or 1/f noise of the
polarizing resistor, voltage noise of the Jfet, or some other source?

In this article:

http://www.etymotic.com/publications/erl-0053-1967.pdf

it's claimed that a larger polarization resistor is quieter, but I
tried a 5 Gigohm thick -film resistor in place of a 200 Megohm, and
noticed little difference. I'm already using a Jfet with 1.5 nV/root
Hz voltage noise, is there anywhere to turn except liquid N?

Thanks,

Sean B



Adrian Tuddenham

unread,
May 15, 2013, 9:02:03 AM5/15/13
to
You could try using RF instead of DC bias. Peter Baxandall got results
with germanium transistors in 1963 that stand comparison with some of
the best DC modern designs:

http://www.coalcanal.org/other/RFMic1.pdf
http://www.coalcanal.org/other/RFMic2.pdf

Even a DC-polarised capacitor mic, if it is well designed, will have
electronic noise no higher than the Brownian noise of the air molecules
colliding with the diaphragm, so theoretically perfect electronics would
only give an improvement of about 3dB.

You could try increasing the signal by getting nearer to the sound
source or by placing a pressure-operated mic against a wall to take
advantage of the pressure-doubling effect. If the frequency response is
less important, a parabolic reflector could be used to increase the
signal (a shotgun-type microphone does not increase the wanted signal,
it just cancels ambient acoustic noise).

Using several capsules and combining their outputs is another technique.
Each doubling of the number of units gives 3dB increase in the random
noise and 6dB increase in the coherent signal; thus improving the S/N
ratio by 3dB. If the capsules are stacked vertically, the directional
response in the horizontal plane is unaffected (at the expense of much
worse effects in the vertical plane, so don't record a troupe of musical
acrobats this way).

One of the problems with large numbers of capsules is to make an
arrangement that doesn't require too high a supply voltage which might
cause damage to individual capsules (all in series) and doesn't give
such a low output impedance that it is diffcult to match to a low noise
amplifier (all in parallel). Another source of noise that is sometimes
overlooked by experimentrs is the power supply for the capsule.

The arrangment shown below overcomes both of these difficulties. It
uses two back-to-back arrays of 16 Panasonic WM 55A103 capsules in
vertical 'honeycomb' formations, so as to give omni, cardioid and
bipolar (ribbon) responses..

http://www.coalcanal.org/other/LNUmic.gif


--
~ Adrian Tuddenham ~
(Remove the ".invalid"s and add ".co.uk" to reply)
www.poppyrecords.co.uk

Scott Dorsey

unread,
May 15, 2013, 9:07:05 AM5/15/13
to
seanbrode...@yahoo.com <seanbrode...@yahoo.com> wrote:
>I'm wondering if anyone knows what the typical dominant source of
>noise is in an externally polarized condenser mic utilizing a Jfet for
>impedance conversion. Is it the Johnson or 1/f noise of the
>polarizing resistor, voltage noise of the Jfet, or some other source?

It's the Brownian noise of the air inside the capsule.

>In this article:
>
>http://www.etymotic.com/publications/erl-0053-1967.pdf
>
>it's claimed that a larger polarization resistor is quieter, but I
>tried a 5 Gigohm thick -film resistor in place of a 200 Megohm, and
>noticed little difference. I'm already using a Jfet with 1.5 nV/root
>Hz voltage noise, is there anywhere to turn except liquid N?

Secondary to the Brownian noise of the capsule is 1/f noise in the FET,
with excess noise adding to that. The polarization resistor adds very
little to the noise because the capsule is a big capacitor in parallel with
it, shunting all that noise. Same goes with the leak resistor. You _may_
increase the output level of the capsule by increasing the resistors but
there are diminishing returns on that (depending on the capsule capacitance
the value for diminishing returns may be as low as 500k or as high as 2G
for typical capsules).

But really, the Brownian noise is the big offender today. You can reduce
that by making the capsule larger or doing some tensioning trickery (a
la MKH-20), but all those things have side effects.
--scott

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

Paul

unread,
May 15, 2013, 11:08:17 AM5/15/13
to
On 5/15/2013 6:02 AM, Adrian Tuddenham wrote:
> seanbrode...@yahoo.com <seanbrode...@yahoo.com> wrote:
>
>> I'm wondering if anyone knows what the typical dominant source of
>> noise is in an externally polarized condenser mic utilizing a Jfet for
>> impedance conversion. Is it the Johnson or 1/f noise of the
>> polarizing resistor, voltage noise of the Jfet, or some other source?
>>
>> In this article:
>>
>> http://www.etymotic.com/publications/erl-0053-1967.pdf
>>
>> it's claimed that a larger polarization resistor is quieter, but I
>> tried a 5 Gigohm thick -film resistor in place of a 200 Megohm, and
>> noticed little difference. I'm already using a Jfet with 1.5 nV/root
>> Hz voltage noise, is there anywhere to turn except liquid N?
>
> You could try using RF instead of DC bias. Peter Baxandall got results
> with germanium transistors in 1963 that stand comparison with some of
> the best DC modern designs:
>
> http://www.coalcanal.org/other/RFMic1.pdf
> http://www.coalcanal.org/other/RFMic2.pdf
>

It's interesting that in the broadcast RF world,
FM is cleaner than AM, where noise affects the amplitude.

So I would assume part of the reason the AM is cleaner in
the microphone situation is because of the Faraday cage, or
metal shielding of the mic case.


> Even a DC-polarised capacitor mic, if it is well designed, will have
> electronic noise no higher than the Brownian noise of the air molecules
> colliding with the diaphragm, so theoretically perfect electronics would
> only give an improvement of about 3dB.
>

That's great. I doubt it, but I wonder if you could ever get a
measurable difference by lowering the temperature of the air.

But then your musicians would freeze! haha!

John Williamson

unread,
May 15, 2013, 11:19:41 AM5/15/13
to
Paul wrote:
> On 5/15/2013 6:02 AM, Adrian Tuddenham wrote:
>> Even a DC-polarised capacitor mic, if it is well designed, will have
>> electronic noise no higher than the Brownian noise of the air molecules
>> colliding with the diaphragm, so theoretically perfect electronics would
>> only give an improvement of about 3dB.
>>
>
> That's great. I doubt it, but I wonder if you could ever get a
> measurable difference by lowering the temperature of the air.
>
> But then your musicians would freeze! haha!
>
So, no real downside, then. ;-)

Sorry, I'm listening to a track for the umpteenth time and it still
doesn't sound *great*. Time for a break, I think.

--
Tciao for Now!

John.

Paul

unread,
May 15, 2013, 11:37:06 AM5/15/13
to
On 5/15/2013 8:19 AM, John Williamson wrote:
> Paul wrote:
>> On 5/15/2013 6:02 AM, Adrian Tuddenham wrote:
>>> Even a DC-polarised capacitor mic, if it is well designed, will have
>>> electronic noise no higher than the Brownian noise of the air molecules
>>> colliding with the diaphragm, so theoretically perfect electronics would
>>> only give an improvement of about 3dB.
>>>
>>
>> That's great. I doubt it, but I wonder if you could ever get a
>> measurable difference by lowering the temperature of the air.
>>
>> But then your musicians would freeze! haha!
>>
> So, no real downside, then. ;-)
>

"Sorry if you guys are shivering a bit, but we've gotta
keep the Brownian motion of the air molecules down...."

:)


> Sorry, I'm listening to a track for the umpteenth time and it still
> doesn't sound *great*. Time for a break, I think.
>

How about the 1000th time or so? Keep listening to it, and
eventually it will grow on you. You'll learn to love all the
mistakes!

:/


Adrian Tuddenham

unread,
May 15, 2013, 12:46:44 PM5/15/13
to
Paul <Quill...@gmail.com> wrote:

> On 5/15/2013 6:02 AM, Adrian Tuddenham wrote:
> > seanbrode...@yahoo.com <seanbrode...@yahoo.com> wrote:
> >
[...]
> > Peter Baxandall got results
> > with germanium transistors in 1963 that stand comparison with some of
> > the best DC modern designs:
> >
> > http://www.coalcanal.org/other/RFMic1.pdf
> > http://www.coalcanal.org/other/RFMic2.pdf
> >
>
> It's interesting that in the broadcast RF world,
> FM is cleaner than AM, where noise affects the amplitude.

That is a completely different situation, with a weak signal and
potentially high electrical noise level.
>
> So I would assume part of the reason the AM is cleaner in
> the microphone situation is because of the Faraday cage, or
> metal shielding of the mic case.

Baxandall's circuit was based on Blumlein's capacitance bridge. It is
neither AM nor FM, it phase-detects the off-balance voltage of a bridge
energised with RF. Later RF-energised mics seem to have abandoned that
principle (to avoid patents, perhaps?) and gone for frequency detection
and other methods.

The second part of that article deals with the sources of noise, which
are mainly transistor noise in the oscillator and rectifiers. It can be
made relatively insignificant if the circuit is correctly balanced and
the switching transistors are swung through their noisy half-on state as
rapidly as possible. If that is what can be achieved by a competent
engineer in a prototype using relatively noisy low gain germainium
devices, think what could be achieved nowadays.

Les Cargill

unread,
May 15, 2013, 1:43:41 PM5/15/13
to
Paul wrote:
> On 5/15/2013 6:02 AM, Adrian Tuddenham wrote:
>> seanbrode...@yahoo.com <seanbrode...@yahoo.com> wrote:
>>
>>> I'm wondering if anyone knows what the typical dominant source of
>>> noise is in an externally polarized condenser mic utilizing a Jfet for
>>> impedance conversion. Is it the Johnson or 1/f noise of the
>>> polarizing resistor, voltage noise of the Jfet, or some other source?
>>>
>>> In this article:
>>>
>>> http://www.etymotic.com/publications/erl-0053-1967.pdf
>>>
>>> it's claimed that a larger polarization resistor is quieter, but I
>>> tried a 5 Gigohm thick -film resistor in place of a 200 Megohm, and
>>> noticed little difference. I'm already using a Jfet with 1.5 nV/root
>>> Hz voltage noise, is there anywhere to turn except liquid N?
>>
>> You could try using RF instead of DC bias. Peter Baxandall got results
>> with germanium transistors in 1963 that stand comparison with some of
>> the best DC modern designs:
>>
>> http://www.coalcanal.org/other/RFMic1.pdf
>> http://www.coalcanal.org/other/RFMic2.pdf
>>
>
> It's interesting that in the broadcast RF world,
> FM is cleaner than AM, where noise affects the amplitude.
>
> So I would assume part of the reason the AM is cleaner in
> the microphone situation is because of the Faraday cage, or
> metal shielding of the mic case.
>


No. This is apples and oranges. The bias is out of band with the actual
signal.

<snip>

--
Les Cargill

Mark

unread,
May 15, 2013, 1:47:58 PM5/15/13
to

>
> >>> it's claimed that a larger polarization resistor is quieter, but I
> >>> tried a 5 Gigohm thick -film resistor in place of a 200 Megohm, and
> >>> noticed little difference.  I'm already using a Jfet with 1.5 nV/root
> >>> Hz voltage noise, is there anywhere to turn except liquid N?
>
>


Re the FET contribution to the electrical noise, both the Voltage
noise and the CURRENT noise performance of the FET need to be
considered especially when the source Z is high.

http://intrator.us/2012feb6/jfet/JFET_noise_Siliconix_appnote_70599.pdf

Mark

Scott Dorsey

unread,
May 15, 2013, 6:33:16 PM5/15/13
to
Adrian Tuddenham <adr...@poppyrecords.invalid.invalid> wrote:
>
>Baxandall's circuit was based on Blumlein's capacitance bridge. It is
>neither AM nor FM, it phase-detects the off-balance voltage of a bridge
>energised with RF. Later RF-energised mics seem to have abandoned that
>principle (to avoid patents, perhaps?) and gone for frequency detection
>and other methods.

Well, frequency and phase detection are really two halves of the same thing
with a dot on top of one of them. Honestly, if you look inside an MKH102
you'll see a conventional frequency discriminator, but really it's the same
thing. Same thing goes with the MKH-20 which has a more linear but
effectively identical discriminator.

BUT, you should note that the Sennheiser microphones not only use the RF
circuit but they ALSO do some tensioning trickery, where the microphone
capsule is designed to have reduced Brownian noise at the expense of the
internal resonance being much narrower and larger, and then the electronics
have equalization to compensate for that. This tensioning trick has more
to do with the outrageously low noise floor of these microphones than the
RF electronics.

>The second part of that article deals with the sources of noise, which
>are mainly transistor noise in the oscillator and rectifiers. It can be
>made relatively insignificant if the circuit is correctly balanced and
>the switching transistors are swung through their noisy half-on state as
>rapidly as possible. If that is what can be achieved by a competent
>engineer in a prototype using relatively noisy low gain germainium
>devices, think what could be achieved nowadays.

Yes, but as you reduce the electronics noise, the capsule very quickly
becomes the limiting factor.

LAB

unread,
May 15, 2013, 7:09:00 PM5/15/13
to
- RF Condenser Microphone

Look at this:
http://www.beis.de/Elektronik/HF-Mic/HF-Mic.html

--
Gianluca

Paul

unread,
May 15, 2013, 8:32:56 PM5/15/13
to
On 5/15/2013 4:09 PM, LAB wrote:
> - RF Condenser Microphone
>
> Look at this:
> http://www.beis.de/Elektronik/HF-Mic/HF-Mic.html
>

Fascinating.

But what does it sound like?

Does it really have a lower self noise than a DC bias
topology? How is the dynamic range affected?

And would it be more susceptible to heterodyning
with a Ham radio operator nearby, who happens to use a
carrier close to 7.68MHz?

And if you use the capsule to replace a variable capacitance
diode in a Colpitts oscillator, and you spread the spectrum
by the index of modulation, by the Bessel functions, I assume
the wider bandwidth is still managable, even with 96kHz
sampling rates.



Les Cargill

unread,
May 15, 2013, 11:16:31 PM5/15/13
to
Scott Dorsey wrote:
> Adrian Tuddenham <adr...@poppyrecords.invalid.invalid> wrote:
<snip>
>
> Yes, but as you reduce the electronics noise, the capsule very quickly
> becomes the limiting factor.
> --scott
>


Noises may not add linearly, either.

--
Les Cargill

Adrian Tuddenham

unread,
May 16, 2013, 4:11:25 AM5/16/13
to
Scott Dorsey <klu...@panix.com> wrote:

> Adrian Tuddenham <adr...@poppyrecords.invalid.invalid> wrote:
> >
> >Baxandall's circuit was based on Blumlein's capacitance bridge. It is
> >neither AM nor FM, it phase-detects the off-balance voltage of a bridge
> >energised with RF. Later RF-energised mics seem to have abandoned that
> >principle (to avoid patents, perhaps?) and gone for frequency detection
> >and other methods.
>
> Well, frequency and phase detection are really two halves of the same thing
> with a dot on top of one of them.

My use of the tem "phase-detects" was incorrect, I should have said
"synchronously-detects".

The signal coming from the bridge varies in amplitude with the amount of
bridge off-balance, but it is also necessary to know its polarity with
respect to the oscillator in order to reconstitute the signal. The
detection system uses synchronous rectifiers (which can also be used for
phase detection, but not in this particular instance).

Frank

unread,
May 16, 2013, 6:45:57 AM5/16/13
to
On Wed, 15 May 2013 17:32:56 -0700, in 'rec.audio.pro',
in article <Re: Condenser Mic Moise>,
Paul <Quill...@gmail.com> wrote:

>
> And would it be more susceptible to heterodyning
>with a Ham radio operator nearby, who happens to use a
>carrier close to 7.68MHz?
>

At least here in the U.S., the two nearest amateur radio (ham) bands
would be the 40 meter (metre) band, which extends from 7.00 MHz to
7.30 MHz, and the 30 meter band, which extends from 10.10 MHz to 10.15
MHz.

--
Frank, Independent Consultant, New York, NY
[Please remove 'nojunkmail.' from address to reply via e-mail.]
Read Frank's thoughts on HDV at http://www.humanvalues.net/hdv/
[also covers AVCHD (including AVCCAM & NXCAM) and XDCAM EX].

Scott Dorsey

unread,
May 16, 2013, 8:28:42 AM5/16/13
to
Yes. You might as well do it synchronously since you already have the
reference right there. It's more stable for one thing.

Adrian Tuddenham

unread,
May 16, 2013, 10:04:08 AM5/16/13
to
Scott Dorsey <klu...@panix.com> wrote:

> Adrian Tuddenham <adr...@poppyrecords.invalid.invalid> wrote:
> >Scott Dorsey <klu...@panix.com> wrote:
> >
> >> Adrian Tuddenham <adr...@poppyrecords.invalid.invalid> wrote:
> >> >
> >> >Baxandall's circuit was based on Blumlein's capacitance bridge. It is
> >> >neither AM nor FM, it phase-detects the off-balance voltage of a bridge
> >> >energised with RF. Later RF-energised mics seem to have abandoned that
> >> >principle (to avoid patents, perhaps?) and gone for frequency detection
> >> >and other methods.
> >>
> >> Well, frequency and phase detection are really two halves of the same thing
> >> with a dot on top of one of them.
> >
> >My use of the tem "phase-detects" was incorrect, I should have said
> >"synchronously-detects".
> >
> >The signal coming from the bridge varies in amplitude with the amount of
> >bridge off-balance, but it is also necessary to know its polarity with
> >respect to the oscillator in order to reconstitute the signal. The
> >detection system uses synchronous rectifiers (which can also be used for
> >phase detection, but not in this particular instance).
>
> Yes. You might as well do it synchronously since you already have the
> reference right there. It's more stable for one thing.

I did a similar thing when I needed to know both the magnitude and the
direction of tilt of a radiogoniometer:

http://www.christerhamp.se/phono/tuddenham-manual.pdf
[See pp 40-41]

The fun part was keeping the phase of the signal and reference accurate
at 45 Kc/s whilst using bog-standard audio op-amps.

Scott Dorsey

unread,
May 17, 2013, 6:59:55 PM5/17/13
to
In article <kn1990$hen$1...@dont-email.me>, Paul <Quill...@gmail.com> wrote:
>On 5/15/2013 4:09 PM, LAB wrote:
>> - RF Condenser Microphone
>>
>> Look at this:
>> http://www.beis.de/Elektronik/HF-Mic/HF-Mic.html
>>
> Fascinating.
>
> But what does it sound like?

It's very transparent if the discriminator is done well. The MKH-20 and that
family do have some odd lower midrange weirdness, but it's not due to the
RF arrangement.

> Does it really have a lower self noise than a DC bias
>topology? How is the dynamic range affected?

It does not. When the original Stevens mike was made, the real advantage
was that the large and bulky electronics could be separated from the capsule
with conventional coax; the Stevens mike was the first low profile condenser
design.

When transistors came along, the design became that much more popular, because
you could implement it with ordinary bipolar transistors with no need to use
a tube front end.

Now that we have low noise J-Fets, the real need for the circuit is somewhat
reduced. It's not really any quieter than a FET front end under good
conditions. However, the lack of all the crazy high-Z stuff means it is
not sensitive to leakage issues from dampness.

The dynamic range is still pretty wide; the capsule is still apt to bottom
out before the discriminator becomes appreciably nonlinear on the MKH-20
mikes.

> And would it be more susceptible to heterodyning
>with a Ham radio operator nearby, who happens to use a
>carrier close to 7.68MHz?

If someone were that far out of the 40M band, they would have more to worry
about with a pink ticket from the FCC. But indeed, external RF interference
was a worry on some models.

> And if you use the capsule to replace a variable capacitance
>diode in a Colpitts oscillator, and you spread the spectrum
>by the index of modulation, by the Bessel functions, I assume
>the wider bandwidth is still managable, even with 96kHz
>sampling rates.

Since you do the discriminator on the analogue side, it doesn't matter.
What comes out of the discriminator is what you care about.

It's actually smarter to use a fixed oscillator (often crystallized) and
use the capsule as part of a phase modulator stage. Since you have the
reference signal and the modulated signal to compare, the demodulation
becomes much simpler to do.

Paul

unread,
May 17, 2013, 7:15:46 PM5/17/13
to
Yeah, the crystal fixed oscillator (as in the circuit LAB posted),
would have a lower phase noise, so wouldn't be surprising if it were
quieter.





0 new messages