Microphones in parallel to reduce noise
in [Design]
|
From: Eeyore on 17 Dec 2009 05:10 Michael A. Terrell wrote: > krw wrote: >> On Thu, 17 Dec 2009 11:29:21 +1100, "Phil Allison" <phil_a(a)tpg.com.au> >> wrote: >> >>> "krw" >>>> Some "dynamics" are really electrets in drag. >>> >>> ** Totally stupid bullshit. >>> >> In fact, you are. You and DimBulb; a perfect pair. > > > Just don't try to seperate them. You know what happens when you try > to divide zero by zero. So, you're as thick as 'krw' are you ? Graham
From: Jan Panteltje on 17 Dec 2009 07:18 On a sunny day (Thu, 17 Dec 2009 02:14:30 +0000) it happened Eeyore <rabbitsfriendsandrelations(a)removethishotmail.com> wrote in <016e08e7$0$15175$c3e8da3(a)news.astraweb.com>: >Phil Allison wrote: >> <osr(a)uakron.edu> >>> My boss laid a unit on my desk today. Its got ten electret mics fed >>> into one low noise opamp. Any ideas how they did this without a ton >>> of noise. All I can see from the board layout is it is NOT the classic >>> op amp summer. >>> >>> Any suggestions for papers on low noise signal combining? >> >> ** Likely all the mics are simply wired in series and supplied with power >> from a DC current source and the signal taken from the top most mic. The s/n >> advantage comes from *correlation*. >> >> The 10 noise signals ( from the internal JFET pre-amps) are un-correlated >> while the sound arriving at each co-sited mic will be heavily correlated. >> So the output signal will be 10 times that of one mic, while the self noise >> will be only 3.16 times - so a 10 dB improvement is possible. >> >> I suspect the mics are arranged to create a directional pick up pattern too. > >Very likely. > > >> Dynamic mics have much less self noise than typical electrets - cos there is >> no internal pre-amp. Means it is easy to get a 10dB or more improvement, >> just by using one instead. > >But dynamic capsules cost more ! Dynamic mikes have a heavier diafragm, and because of that are less sensitive. So you may gain at one point, but have less signal to process, still more noise. This is *my* view, IIRC Phil has a different idea about that. >Graham >
From: whit3rd on 17 Dec 2009 15:23 On Dec 16, 7:46 pm, George Herold <ggher...(a)gmail.com> wrote: > o...(a)uakron.edu wrote: > > My boss laid a unit on my desk today. Its got ten electret mics fed > > into one low noise opamp. > > Any suggestions for papers on low noise signal combining? > Crazy idea, Is there anyway to add them in series? Sure, if you transformer-couple the outputs; use 9 transformers to put ten in series. The individual units need filtered DC on the power terminal to bias 'em, though, and there's DC on the outputs. I'm not happy paralleling the (FET) outputs; seems like the DC levels would interact poorly. One could, however, use a low-Z input, high-Z output amplifier on each unit (grounded base transistor) and just tie all the outputs (collectors) together, with suitable bias. More troubling, the multiple small microphones are a phased array of elements; you'd want to form the array with respect to the desired microphone directional gain (which will be frequency-dependent).
From: Michael A. Terrell on 17 Dec 2009 17:40 Eeyore wrote: > > Michael A. Terrell wrote: > > krw wrote: > >> On Thu, 17 Dec 2009 11:29:21 +1100, "Phil Allison" <phil_a(a)tpg.com.au> > >> wrote: > >> > >>> "krw" > >>>> Some "dynamics" are really electrets in drag. > >>> > >>> ** Totally stupid bullshit. > >>> > >> In fact, you are. You and DimBulb; a perfect pair. > > > > > > Just don't try to seperate them. You know what happens when you try > > to divide zero by zero. > > So, you're as thick as 'krw' are you ? > > Graham That's it. Change the subject when tou can't do anything else. Typical loser attack. -- Offworld checks no longer accepted!
From: George Herold on 17 Dec 2009 17:41
On Dec 17, 3:23 pm, whit3rd <whit...(a)gmail.com> wrote: > On Dec 16, 7:46 pm, George Herold <ggher...(a)gmail.com> wrote: > > > o...(a)uakron.edu wrote: > > > My boss laid a unit on my desk today. Its got ten electret mics fed > > > into one low noise opamp. > > > Any suggestions for papers on low noise signal combining? > > Crazy idea, Is there anyway to add them in series? > > Sure, if you transformer-couple the outputs; use 9 transformers > to put ten in series. The individual units need filtered DC on the > power terminal to bias 'em, though, and there's DC on the outputs. > > I'm not happy paralleling the (FET) outputs; seems like the DC > levels would interact poorly. One could, however, use a low-Z > input, high-Z output amplifier on each unit (grounded base transistor) > and just tie all the outputs (collectors) together, with suitable > bias. > > More troubling, the multiple small microphones are a phased array > of elements; you'd want to form the array with respect to the desired > microphone directional gain (which will be frequency-dependent). "> I'm not happy paralleling the (FET) outputs; seems like the DC > levels would interact poorly." Hmm, can't you just AC couple them into the opamp? George H. |