[ProAudio] Microphones question

Chris Caudle 6807.chris at pop.powweb.com
Wed Jun 23 09:21:16 PDT 2021

On Mon, June 14, 2021 2:49 pm, Dan Mills via ProAudio wrote:
> The interesting takeaway from that calculation is that actually given a
> reasonable ADC (-97dBu self noise or so), and a reasonable mic (-120dBu
> self noise or so) we only need about 30dB of front end gain going into a
> modern ADC to make the mic self noise completely dominant

That is exactly the point made by Paul Frindle in one of my favorite
preprints, pp4126 from the fall 1995 AES convention.   Paul presented on
the design considerations for the conversion system designed for the Sony
Oxford large format digital console, and section 4.2 makes exactly the
point you brought up:

"As the input stage for the ADC converter, the microphone amplifier is
required only to make up the shortfall in the dynamic range of the
converter and provide impedance matching to produce the appropriate input
referred noise specification in the digital domain.
Beyond this, extra gain to increase the levels of signals is best
performed in the digital domain. Therefore if a converter had a total
dynamic range in excess of 128dB and input matched to 150R referred to
0dBu, no microphone amplifier would be required at all.

"Similarly, for a converter with an input noise of 112dBu (with 0dBfs
referred to 0dBu), the input noise specification could theoretically be
achieved with gain of around 16dB providing no noise was added (i.e 128 -
112dB). However an amplifier with gain of 16dB and output noise below
-112dBu is a difficult specification to achieve due to the relatively high
impedances required within the amplifier circuit and the practical limit
of power dissipation for the circuit."

That entire section 4 of the preprint is good and covers lots of the
practical considerations.  I have not ever had access to one of those
conversion systems, but from the description it seems that it is made for
microphones or instruments only, it does not appear that it was designed
to handle typical professional line levels (i.e. +20 dBu to +24 dBu range)
without a pad.  The input for full scale was +12 dBu, but I am not sure if
the system used a 12 dB pad for line level, had a separate front end for
line level, or just did not accommodate traditional line level signals.

> All of a sudden those preamps with 60dB of gain range on the pot start to
> seem rather pointless, just do a fixed +30dB and hit the ADC

You probably still want some adjustment range to optimize the gain
structure, but you won't need as much gain.

That is all assuming you only care about getting into the digital domain
as quickly as possible.  If you have interoperability requirements with
other equipment you might still need to be able to drive traditional line
levels, but yes it does seem kind of pointless to take lots of pains to
get large amounts of gain, to then throw it away when you have to
attenuate the signal by a large amount to get into a converter running
from +5V.

> On the subject of Ein, does nobody discuss Iin, particularly for a bipolar
> input stage it is often a more significant noise source then Johnson once
> the source gets above a few k.

I think that would primarily apply to instrument inputs where you are
amplifying the output of a coil with a lot of windings, or a piezoelectric
pickup.  A microphone should only have output impedance in the hundreds of
ohms, and a line level output would be in tens of ohms, so input current
noise should not be dominant for those cases.  From what I have seen the
typical solution for instrument inputs is to put a JFET buffer in front of
your mic amp circuit.

Chris Caudle

More information about the ProAudio mailing list