My response ended up somewhat lengthy, but hopefully
it will be of help to those who are interested in
quiet microphones.
--- Bruce Wilson <> wrote:
> If the noise comes from the 6800 ohm
> resistor only, then the noise would
> be -119 dbu, just like every other
> phantom supply
0 dBu =3D -2.21 dBV
The 20 kHz bandwidth noise from a 6.8 kOhm resistor is
-116.65 dBV; -2.21 gives 118.86 dBu (working from
memory now, I seem to have mislaid my calculator and
reference books), which is very close to Bruce's
figure.
But this is NOT a limit to the performance of phantom
powered microphone systems. Since such microphones
always have either an amplifier or a transformer at
the output, their output impedances are determined by
the output impedance of the amplifier or transformer,
which is always much lower than 6.8 kOhm.
Effectively, the output impedance of the microphone
forms a voltage divider that divides down the noise
voltage from the phantom power to something much
smaller. In a previous email to Rob I used the output
impedance of the Schoeps microphones as a point of
reference (the schematic is pinned to my wall!).
You can see the spec on output impedance here:
http://www.schoeps.de/E-2004/specs-cmc.html
I see that the output impedance is spec'd at 35 Ohms
(when the phantom power is 48 Volts), so the
attenuation is about .005, or -46 dB. It would take a
lot of noise from a phantom supply to cause a problem
by strictly adding noise to the microphone output.
What might be a problem is if there were ultrasonic
tones in the phantom power, which then might
intermodulate with ultrasonic signals inside the
microphone. That would depend a great deal on the
design of the microphone.
> I noticed he's doing something funny with
> the noise figures he quotes (-157 dbFS).
By referencing dBFS, Len Moskowitz is now referring to
the noise as it is translated into the digital domain.
He doesn't say this, but it is also obvious that he
is referring to spectrum level, and not broadband or
weighted noise levels.
What's the difference? The finer the spectrum is
divided up, the less noise is seen in each spectral
band. For instance, an FFT of length 16384 results in
8192 spectrum bins (the name for each spectrum band in
an FFT). The energy in each bin goes as the square
root of the bandwidth, and thus each bin has 1/90 of
the total energy, which is to say that it is -39.13 dB
lower in level than the total. So if the total level
at the point of measurement was -119 dBV, then the
spectrum level would be -158 dBV, which is very close
to the level that Moskowitz quotes.
There might be some reaction to this as being
confusing, but getting the measureemtn into the
digital domain is an important distinction. If the
microphone preamp, A/D converter, and recording device
are all in one block, then once the signal has passed
into the microphone input, all we care about are the
levels in the digital domain. And those are expresed
in dBFS, or dB relative to digital Full Scale. There
is an AES (Audio Engineering Society) standard which
describes how to do measurement like this. That
standard is AES 17 - AES standard method for digital
audio engineering =97 Measurement of digital audio
equipment.
Eric Benjamin
> I noticed he's doing something funny with the noise
> figures he quotes
> (-157 dbFS). If the noise comes from the 6800 ohm
> resistor only, then
> the noise would be -119 dbu, just like every other
> phantom supply. That
> gives a full-scale reference of +38 dbu, or an
> unlikely 62 v (remember,
> phantom power is only 48 v, +36 dbu). Fortunately
> mics have capacitors
> which (I think) will kill some of the thermal noise.
> Maybe the internal
> resistance of a capacitor adds thermal noise, but
> the resistance can't
> be high, so I doubt the added noise will be either.
>
>
> Bruce Wilson
> http://science.uvsc.edu/wilson
> -----Original Message-----
> From:
> On Behalf
> Of Rob Danielson
> Sent: Tuesday, June 27, 2006 10:16 AM
> To:
> Cc:
> Subject: [Nature Recordists] 2Phant Core Sound
> Portable Phantom (was How
> would you clean up this file?
>
> Hi Aaron--
> Apologies for the subject line change. Good find.
> Being able to see
> the state of the batteries is also a nice addition.
>
> The unit seems of special interest to people with
> low-noise phantom
> powerred mics and non-phantom recorders or recorders
> with poorly
> performing phantom circuits. Some of the later seem
> to have been
> introduced to the market. The first question that
> jumps to my mind is
> how does one know if the $200 cost difference is
> better invested in a
> 2Phant or perhaps towards better mics?
>
> If I assume that the performance of the phantom
> supply in the Sound
> Devices 722 is among the best in terms of noise
> performance, one way
> to start this evaluation is to compare:
>
> NT1-A-> PB224-> 722 (Phantom Off)
>
> and
>
> NT1-A->722 (Phantom On)
>
> which I've done several times with consistent
> results. One of them is
> at the head of this test:
>
>
http://www.uwm.edu/~type/audio-art-tech-gallery/pages/page_38.html
>
> I'm cc-ing this to Eric Benjamin who is testing the
> PB224 along with
> the WL-93 and other mics in case he has comment.
>
> Rob D.
>
> =3D =3D =3D =3D
>
> At 1:20 AM -0700 6/27/06, Aaron Ximm wrote:
> >Interesting new 2x phantom powering module from
> Core. Pricey at $275
> but it
> >has some nice features (the ability to swap
> batteries without
> interrupting
> >use is nice!).
> >
> >Interestingly Len claims the PB224 puts out less
> than 48V, he's got an
> >incomplete section comparing the new box vs.
> various other phatom pow
> >
> > http://core-sound.com/2Phant/1.php
> >
> > aaron
|