[Top] [All Lists]

Noise measurment and figures

Subject: Noise measurment and figures
From: Klas Strandberg <>
Date: Sat, 27 Apr 2002 12:19:25 +0200
Walt, all - just theory:

Telinga stereo DATmic design uses 4 x 10-mm omnis in parallel per channel.
Each has a noise level of 22 db(A), which is some 3-6 db lower than average
for a 10 mm capsule.

1/ For each time you "parallel" transistors, you get 3 db less transistor
noise, because noise will randomly get anti-phase. Four elements therefore
means -6db, hence; 22 - 6=3D 18 dbA.

2/ An average amplification of PZM is usually estimated to some 6 db.
Therefore we're down to 10db inherent noise of the stereo DATmic, without
the dish.

Theory is not true, however. If the stereo DATmic was perfectly "flat",
theory would be more relevant. But it isn't flat. It has (like all mikes
with acoustical amplification like PZM, SASS, CVX(?) etc) - a boost
somewhere on the frequency curve. This "error" is generally described in
positive terms like "a nice presence" or "emphasizing the depth of a human

Some noise is more audible to the human ear than other noise. "Natural
noise" - like wind in the trees - seem to be "neglected" by the human ear /
brain. A high-freq. "sssss" however might be more interesting - a dangerous
snake or something to eat.
Or, - for the non-biologists - we're so used to the natural noise around us
that we don't care about it.

If microphone noise resembles the ambient noise, it gets "masked". You can'=
hear it much. High-freq. noise, however, will be notified a lot, even if it
measures "low".

db(A) and CCIR:
(Therefore) one has developed two different standards for inherent noise:
db(A) and CCIR.
As you note, I neglect Asian-developed methods like S/N, as these figures
are not comparable.

Simple description of the method:
You use a sound generator. You expose the mic to a sound pressure of 1
Pascal. You connect a standardized noise/voltage meter to the output of the
mic and adjust the instrument so that it shows 96 db.

Now you switch off the sound generator, and measure the remaining noise of
the mic only, through a filter defined by DIN =3D Deutche Industrie Normen.
There are two such standards: db(A) and CCIR. My instrument (Sennheiser
"Universal Pegelmessger=E4t UPM 550-1) has filters normalized to both.

Both curves raise by approx 10 db/octave, but the db(A) curve flattens out
at 10kHz, while the CCIR flattens out at 100 kHz.

This means that the CCIR curve is more "critical". Most FET-transistors, in
most mic's, have noise over 10kHz, which will be amplified by the CCIR
filter, while the db(A) filter will flatten out the amplification.

If the instrument shows "0" it means that the mic doesn't generate any nois=
at all. If is shows 10 db, it means that the mic makes 10 db - measured
through the filter used.

If you measure the noise through the CCIR filter, it may show that the
inherent noise generated by the mic is 14 db(CCIR) - which mainly means tha=
4 extra db of the noise is over 10kHz, which is not amplified by the db(A)

Ideal: The same CCIR as a low db(A)!
Far from ideal: a low db(A) and a high CCIR!

And of course: Every time we talk about a scientifically approved method of
measuring something, there is also a relevant discussion on how "good" it
is, - how "accurate" it measures what you really need to know.


At 22:52 2002-04-26 -0400, you wrote:
>Simon Squire wrote:
>> Noise measurements in the anacroic chamber showed the Panasonic element =
>> 1% THD when it was out of the CVX. As soon as we put the same element in=
>> CVX housing the THD dropped to 0.1%. Why this happens has not been
>> investigated yet, due to limited chamber time, and money etc.
>The location in the housing may be providing some form of pressure
>"preloading" to the capsule diaphragm.
>I suppose it's tradition to use a different mic noise criteria for every
>brand of mic, but lets try and stick with the ones common for mics used
>in nature recording. I don't believe I've seen THD being used for
>discussing noise floor on a mic in the group before.
>I may be all fuzzy in my thinking tonight, but here goes:
>Panasonic's reporting for their capsule:
>S/N ratio More than 62 dB
>Sennheiser's reporting for the MKH-20:
>Equivalent noise level
>A-weighted (DIN IEC 651) 10 dB
>Panasonic reports more than 62 dBA. Though not stated, this is usually
>against 96dBA, making the equivalent to the MKH-20 of 34 dB A-weighted I
>think. Now assuming the housing on the CVX adds some gain before the
>mic. To be generous, say 10 dB. That would lower the noise of the combo
>to 24 dBA. Adding 10 dB gain to the MKH-20 by using a very low noise
>preamp would raise it's noise floor as it occurs after the mic. Which
>would make it 20 dB at the extra gain.
>In any case, the noise will place the upper limit on how far away
>something can be recorded. I've carted my sound meter around measuring
>frogs some. Highest I've measured, at about 1' distance is about 90 dB.
>Working on sound falling off by about 6dB per each doubling of distance
>that works out your mic might pick up that frog (barely) at a half mile
>or so before it would fall into the noise floor. The MKH-20 plus pre
>would do better by 4 dBA or more depending on the actual gain from the
>CVX housing. Which should translate into greater reach.
>I don't have the noise specs for the Telinga, so can't do the same
>comparison there. I think it has about 10 dB gain before the mic. There
>is also a reporting of the relative shortening of the "distance"
>compared to a bare mic to 1/10 of actual. That all indicates that unless
>the Telinga capsules are noisier than the CVX capsule that the Telinga
>is likely to have greater maximum reach.
>It should also be noted that the Telinga stereo mic element is a PZM
>system. I don't know if this provides any of the gain associated with
>PZM or not.
>I'm not sure how the relationship of sensitivity would be resolved.
>Panasonic reports:
>Sensitivity =9635=B14dB (0db =3D 1V/pa, 1kHz)
>Sennheiser MKH-20 reports:
>(free field, no load) (1 kHz) 25 mV/Pa =B1 1 dB
>One could say that the MKH-20 plus a low noise pre is going to be more
>costly than the CVX, but the mics themselves are in a fairly similar
>price group. MKH like prices create that kind of expectation. The
>Telinga is also of the same sort of price, though it's not at all a omni m=
>> You will be interested in our stereo version which is currently in
>> development. The first prototype is working very well, with less noise a=
>> very good stereo separation. No numbers yet though. This will not be as
>> rugged as the original CVX, but will be very light, and stereo, with sim=
>> performance. Also we have been in contact with Cornell, because they wan=
t to
>> compare the CVX with current mics. They do not have one yet though, so D=
>> can be the "comparator" for now if he wants.
>I'll be interested in following the saga. I do particularly like the
>weather resistance of the design. This, of course, makes it fairly ideal
>for a froglogger.
>Speaking of which, what are raindrops hitting it like for sound? Lets'
>say one of our typical summer thunderstorms. I'd kind of expect the
>housing may act as a drum.
>> I will guarantee you have never heard anything like the CVX, or your mon=
>> back. ( I'm starting to sound like a salesman, yuk, I'm a tech, give me =
>> soldering iron...)
>So far it sounds ok. What appears to be the mic's noise floor is clearly
>noticeable in the recordings I've listened to. Though I certainly don't
>know if it's that or some other part of the process. I'm not surprised
>at this in view of the capsule used. Kind of like the things with the
>SASS mic, it would be nice to see some comparisons between the CVX as
>designed and say a "MKH-20" modification.
>Do you know if any frequency response graphs are available? Seems to me
>like the design may partially act to favor some frequencies over others.
>To unsubscribe from this group, send an email to:
>Your use of Yahoo! Groups is subject to

This outgoing e-mail is scanned for viruses with Norton 2002

Telinga Microphones, Botarbo,
S-748 96 Tobo, Sweden.
Phone & fax int + 295 310 01
org. no SE440130067001


Your use of Yahoo! Groups is subject to

<Prev in Thread] Current Thread [Next in Thread>

The University of NSW School of Computer and Engineering takes no responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the naturerecordists mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU