Ultra low-noise mic pre?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Martin B. Kantola

Well-known member
Joined
Jun 24, 2005
Messages
209
For experimental use I will need a microphone preamplifier, must have standard +48V phantom feed and typical input impedance (2k) because I'll use it to test typical condenser microphones. Well, maybe not so typical...

The only important is to get the noise as low as possible, even if it means sacrificing other qualities. Frequency response, distortion, output impedance or offset etc. are not critical. About 60dB of gain should do, but another 10dB would be great.

My question is which path or combination to choose for best results:

- dedicated chip such as THAT 1502
- low noise OP such as LT1115
- transformer input (higher than 1:4 ratio?)
- discrete bipolar input pair
- some kind of cooling arrangement?
- parallel devices

And ideas appreciated. The more creative (crazy) the better...

Martin
 
Could you be a bit more specific? The 'lowest noise' solution varies a lot depending on several parameters.

Pretty much all condenser microphones have an integrated amp between the capsule and the XLR. For most reasonably low-noise mic pres, this amp's noise will dominate the system noise figure.

Most audio preamps have significant 1/f noise. You say frequency response is not critical, does that mean you're allowed to, say, include a 1kHz high pass filter to get rid of the 1/f noise?

For 60dB gain, a simple THAT1512 has a 1.3dB noise figure when terminated with a microphone with a resistive impedance of 150 Ohm. Is that good enough?

What are you trying to do ?

JD 'insufficient data' B.
 
Good points, thanks!

I'll try to explain my needs a little better. What I want to do is to plug in a typical condenser mike (Z probably below 250 Ohms) and get an idea of the noise amount and type of the mike itself. As you wrote, that's usually the dominating source.

As for the 1/f noise issue, it would be very useful to include all audio band frequencies 20-20k, so a high pass is not a good solution. But the system could be limited to the audio band only. Noise we can hear.

Guess what I'm asking is how close to the thermal noise I can get and how. 1.3dB is a great noise figure (think it's @ 200 Ohms for the THAT) but does it stay that small all the way down to 50 or even 25 Ohms? These might be more likely actual source impedances.

Martin
 
For a condenser mike, the head amp and not the output impedance will determine the output noise.

For example, the RØDE NT1-A, one of the lowest noise (6dBA) commercially available condenser microphones, has a ~ -117dBu noise floor on its output terminals. A vanilla THAT1512 mic pre has a ~ -128dBu noise floor @60dB amplification measured in 20-20kHz, thus producing a <0.4dB increase in noise compared to an ideal noiseless mic pre.

Have a look at http://www.rane.com/note148.html for more useful stuff.

JDB.
 
Thanks again for the patient reply, I'm learning all the time :)

Regarding the mike impedance I was thinking that the theoretical minimum noise is lower with a lower source impedance, or?

The Rane document is great, actually found and printed it a couple of weeks ago!

So, if the mike I wish to test has an output impedance of 50 Ohms, the perfect noiseless pre-amp should have a noise floor lower than -140dB. But as you pointed out, it's the contribution to the total noise that matters.

Another reason I brought this up is the noise difference I seem to be getting with two of my usual pre-amps, the Millennia HV-3D and the Portico 5012. Don't have a noise figure for the HV, but EIN is -133dB if I remember correctly. Isn't the Rode a stunning 5 dB(A) BTW?

Martin
 
Remember as well that you can subtract out the noise of your preamp. As mentioned it should be small anyway, but measure it with a dummy load equal to the mic output impedance. Square the total noise, and subtract the square of the preamp noise with the dummy load. Take the square root---that will be the mic noise alone. If you like, correct for the thermal noise of the dummy termination knowing the resistance and the bandwidth.

As JDB points out, the capsule amplifier and other noise sources in the front end will be far larger than thermal noise in the output impedance of the line driver. Actually "theoretical minimum noise" for a mic with a mythical noiseless capsule amp doesn't have to be thermal noise of the magnitude of the impedance either, for that matter---one can contrive circuits to look like a given Z over a useful frequency range without having full thermal noise of the equivalent resistor*. However, it would be a wonderful world if such considerations were relevant for microphones.



*I got into this behind a few glasses of wine at AES once, when Jürgen Wahl was presenting the Neumann Solution D, and asserted that regular mics were limited by the thermal noise of their impedance magnitude. My comments were not well-received, possibly owing to the presence of some Neumann execs in the audience :razz:
 
To realize low NF at 25 or 50 ohms source you might want to check out something like 2SB737/ 2SD786 series. They were designed for MC phono preamps so low noise at low source impedance. They are obsolete now but available in small quantities.

As Brad mentioned to measure the noise of the mic itself you need to first characterize the noise of the preamp separately.

You can measure the preamp with input shorted to get just the input noise voltage term, then measuring it with the input terminated by a resistor to characterize the noise current and termination's johnson noise. Perhaps measurements with a few termination impedances will help parse out the noise current from the johnson noise.

When terminating with your test microphones, you can reverse out the preamp noise voltage, and what's left is the mic's noise and source impedance, with source impedance times noise current.

JR
 
As the others have pointed out, there is absolutely no need for a low noise preamp when using condenser microphones (especially at high gains). For any current low-noise mic (10 dBA or less) the dominant noise source is Brownian motion of air--no way to get around that other than recording in a cooled room.

Just from a theoretical point of view you want a FET input amplifier for lowest noise figure as you'll get around current noise that way. To bring voltage noise down you'll either need to parallel a bunch or choose a low noise part to start with. I'm currently working on a IF3602 based measurement preamp which would be suitabel.

Samuel
 
How about SSM2019? Check the datasheet if it is low noise enough for your application. It requires only a few extra components and if I remember right Analog devices provides samples too.
 
[quote author="audiox"]How about SSM2019?[/quote]
It appears to be as noisy as or noisier than the THAT1512 which was mentioned earlier.

JD 'according to a brief look at both datasheets' B.
 
Many thanks for the input guys, that's a lot to digest, will take me a little while. Having some trouble with combining all the teachings...

As for subtracting the noise of the pre-amp it's a good idea, except that I was hoping to be able to analyze the noise (listening + looking at the spectrum) in real time. As an example, there's this microphone in my collection which has a good dB(A) figure, but it has a very annoying hiss compared to others.

Maybe it's the simple truth that I don't need a better pre, but can't help feeling I'm pushing the ones I have to the limit, especially with the low noise mikes. Just a simple sound engineer talking here, so it's only a gut feeling.

Great idea to look at MC-preamps. Samuel, what kind of noise figure are you hoping to get with your measurement pre-amp? Might give the 1512 a try just because it's so easy to hook up. Thinking that if it's optimized for a 200 Ohm source it could be interesting to also use a matching transformer if the source has lower impedance source. Will that even work?

Martin
 
Samuel, what kind of noise figure are you hoping to get with your measurement pre-amp?
Math indicates an EIN of -141.9 dBu with shorted input (with essentially zero current noise), which is just the noise of a 12 Ohm resistor. But at this level, there is no way around just building it and look how good it really is.

Samuel
 
Been thinking about this micpre noise issue a bit more. Recorded a clip with a Sennheiser MKH-40 cardioid microphone in my studio, using the Portico 5012 and Millenia HV-3. Both set to 60dB gain.

The MKH-40 is suppose to have a self-noise of 12dB(A) so it's only fairly quiet. Here's a spectrum of the noise as well as a couple of mp3's for listening to the difference (soundfiles have +40dB of added digital gain). Blue trace is the Millennia.

pres.gif

http://www.nordicaudiolabs.com/samples/MKH40-5012.mp3
http://www.nordicaudiolabs.com/samples/MKH40-HV3.mp3


Might not be a huge difference with this mike, but I sure would like to have a setup where I can ignore the pre-amp noise and focus on the microphone noise. But even with this setup there are pretty big differences in the noise spectrums of different microphones, so ambient and Brownian noise still seems to be a bit lower.

Martin
 
Terminated with 150 Ohms (according to MKH-40 specs):

pres2.gif


It is interesting to see how little noise the pre-amps have without a real microphone connected. And the difference (especially around 4k) is gone.

Martin
 
What I expected. The differences you observed with the mics are likely changes in ambiance noise (at the low end), gain mismatch (midband) or interference (at the high end, e.g. cell phones).

Samuel
 
That is a LOT of assumptions if you don't mind me saying so...

As for the low rumble (<30hz) it's of course changing with the conditions (weather and distant passing cars), but this is the best I can do in my studio. Haven't finished building a proper isolated "soundtank" to take care of the low end ambience.

Martin
 
Samuel,

appreciate your input, but the fact remains that the mike pre noise does contaminate my microphone tests to some degree, even with microphones as noisy as 12dB(A). That's why I was hoping to find a more transparent solution. Maybe there isn't one.

Please note that I'm testing in the studio, not on a lab test bench, and there is a reason for that. Want to know how microphones behave in a typical environment.

Martin
 
Back
Top