Previous index Next

Are these Mic Pre's really close!!?

Recordists who wish to capture subtle auditory experiences like amibence in "quiet" natural spaces know such conditions demand very low noise mic preamps and low-noise mics. Its becoming increasingly popular to conduct and/or access on-line equipment comparison tests to determine mic preamp noise performance before making purchases or planning gear combinations.

An audible comparison test is simple in concept and not too hard to make. You need the mics you are going to use, the recorder or external mic pre in question and another recorder whose mic input noise performance is up tp your standards. You set-up the mics in the quiestest place you can access, you set the record gain of all recorders at the maximum or near maximum level and you record at least 10 minutes of the ambient "room tone" with a very quiet sound source several feet away (like clock ticks, above). Take short passages from each recording where the background sounds are the quiestest. Place them into a multi-track sound mixing application and adjust the volume of the softer recordings to match the loudest one. When you play them back, back to back, you can hear the changes in noise "across" the edits between the passages. Very short duratiion cross-fades between the passages can make it easier to hear the changes.

If the "self-noise" rating of the mics you are going to be using is above 18 dB(A), there's not much point in running a comparison test because the noise of the mics is probably going to be loud enough to mask or hide the "input noise" generated by the mic preamps in most recorders. Likewise, if the "input noise" measurement of the recorder's mic pre is less than -127 dBu (A weighted-) -- a smaller, negative number, there's not much point in a comparison test because one can't buy a mic with low enough self-noise to hear the mic preamp noise "behind" the self-noise of the mics. There are some exceptions but these principles are useful to keep in mind and can ward off lengthy discussions about sound qualities that don't seem to be readily audible.

A comparison test of "pro-quallity" recorder and external mic preamp made by Curt Olson is an example of this. The QuickTime movie above is from Curt's test recordings. His test conditions and my analysis follows:

To: Curt Olson <>
From: Rob Danielson <>
Subject: Re: Fostex FR-2LE - MixPre Comparison [lnm]
Cc: "David Michael" <>


Curt & David--
Regarding Curts recent, maximum gain, "clock" test comparing these equipment combinations:

(A) Audio Technica AE5100 mics -> FR-2LE (internal mic pre)
(B) Audio Technica AE5100 mics -> SD MixPre (line out) > FR-2LE (line in)
(C) Audio Technica AE5100 mics -> SD MixPre (tape out > Zoom H2 (line in)

I had to set the range and gain at very high values in the sonograms order to produce any visible differences in the recordings. Here are my thoughts about what we can learn and not learn from the test:

(1) The MixPre has more response under 40 Hz and the H2 has a pronounced drop in response starting around 16K Hz. Neither are huge reasons to choose one over the other unless you're recording bats or material to be feautured on a subwoofer. Everything else is probably attributable to differences in the environment at the times of the recordings.

(2) We can use Table 3 of the Rane Note 148 web publication, or refer to just the chart, to compute the noise floor equivalent of the AT AE5100 mics that Curt used in his comparison test. From Audio Technica's website, we can determine that these mics have a sensitivity of 15.8 mV/Pa and self-noise of11dB(A)]. With these numbers, one can compute the equivalent mic input noise number of -117 dB-SPL, A weighted. I'm using the mic manufacturer's numbers, but if these are correct, these mics would produce the same amount of (self) noise as the input noise of a mic preamp with -117 dBu noise. The Rane chart is designed to convert mic self-noise into pre input noise to help folks determine if the mic pre they are considering using with a particular set of mics is quiet enough. These "EIN" numbers are rarely provided these days for consumer and prosumer gear, but basically, the mic pre number needs to be 7-10 dBu less to be inaudible "behind" the self-noise of the mics used.

(3) Raimund Specht's actual input noise measurement for the FR-2LE is -129dBu (A weighted). The MP2/MixPre is very close to that number.

(4) The noise from the mic pres we are trying to compare audibly is about 12dBu less than the noise produced by the AE5100 mics. That's a large difference. Any number greater than 10 dBu is inaudible. What we are hearing in the above QuickTime movie I made from Curt's test, is the self-noise of the mics and the sounds they are reproducing-- not noise from the mic preamps. If you listen carefully, you can hear that the character of the hiss is exactly the same in all three recordings. So, if we want to hear other detectable differences in the pre performance of the Fostex and MixPre, we need mics with 3 dB(A) self-noise to barely begin to hear them. (Please let me know if you come across such mics!-- unless they are very expensive).

The bottom line is we can audibly evaluate mic pres that approach "pro-quality" noise performance ONLY if we use the quietest of mics like the Rode NT1-A's. As reference, the surprising low mic pre noise of a Sony Hi-MD recorder has been measured at -124 dBu (A weighted). This amount of noise is only partially audible behind the self-noise of NT1-A's. With a self-noise rating of 5.5 dB(A) and a sensitivity value of 25 mV/Pa, the NT1-A computes to -118 dBu (A weighted) using Table 3. The difference between -124 and -118 of "6 dBu" falls within the audible 10 dB range. One might say that "pro-quality" low- noise, mic pre performance effectively starts around -127 dBu (A weighted) because its input noise is quiet enough to hide behind the quietest of mics.

Another factor is the "color" or audio spectrum distribution of the noise being produced. When the noise is comprised mostly high frequencies, like >8 K Hz, the noise is usually less noticeable than when the noise extends low into the spectrum. This characteristic of the pre input noise is discernable through comparison listening and not evident in the published or measured input noise numbers themselves. Rob D.