Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[klaus]

So I get this email from a gent who suspects that his KM84 Neumann is hissy noisy "when I increase gain".

Here is what I wrote him back:

... before you go to the next step and involve professional help, I recommend a simple self-diagnostic.

First, you need to put the hissing you perceive into perspective by defining a baseline that allows you to compare the noise of your mic to that of another. That will determine:

*is this just the normal noise floor of a KM84 (they are not the quietest) you are hearing, or
*does your KM84 objectively have too much noise, compared to its specs? 

As you may not have a second KM84 to compare noise types and floors with, I recommend to record with your KM84 a sine wave or, preferably white noise at a reasonable distance and level from the speaker that has been fed the test sound. Then turn the tone/noise generator off, but continue to record for a few more seconds (i.e. quite room).

Now, on a second channel, record with a different professional condenser mic-hopefully a healthy one-the same white noise, after adjusting the level on the recording meter to exactly the same as for the KM84 recording. Continue recording a bit more without sound input (quiet room).

After you have equalized playback levels between channels during playback of the test tones, crank up the listening level and listen for the self-noise level during the quiet section at the end. THAT will give you the information whether your KM84 is unusually noisy, or still within a 3dB range* of the other mic's noise floor, therefore "within spec".

I have used the testing method I described for many years, and never had anyone tell me that the results were not at least roughly in sync with either their hearing or more sophisticated methods of noise analysis.

But i am open to modify or abandon how I comparatively test noise, if any of you deep thinkers see a fatal flaw in my methodology. To emphasize, for this kind of fault finding I am not chasing after the last dB. That's the realm of fine tuning at a different level for a different purpose.

*Yes, I find that the hissy kind of noise level of professional mics still varies by as much as 3dB from one specimen to the next.

[RuudNL]

IMHO this is a very reliable and objective method to check the self noise of a microphone.
Anyway, this is the method I also use myself.
An other way of checking for noise (if you have a second microphone of the same type), is to listen to the two microphones with headphones.
One in the left channel and the other one in the right channel.
Carefully adjust for the same level when speaking in both microphones close together and after that listen to the noise in a quiet room. If there is any significant difference in self noise, you will hear the noise 'out of center'.

[Jim Williams]

The scientific way without subjective guess work is to use a graphical analyzer to "view" the noise vs frequency plots from 20~20k hz. That will show whether the noise is hi frequency hiss, mid range grunge or lower frequency hum components, or all three.

Before that is done the mic preamp needs to be tested in the same way. Some designs like Audio Precision allow one to extrapolate the mic noise from the measured preamp noise to eliminate that preamp noise contribution as all noise sources are accumulative.

Then one would have a proven, scientific base line of the specs of those models in which to compare to others. AP will allow you to nest those sweeps into a group to show the baseline and any variants. Then you can create a bell curve to show all the permutations as well as the base line at the top of the bell.

This sort of advanced testing does require an initial investment in rather expensive test equipment which is not commonly found in the smaller pro audio companies.

[klaus]

The intention here was to self-help someone who does not have access to sophisticated and expensive test equipment. So, Jim, will the procedure I recommended work in this specific case?

[Kai]

The methods you describe perfectly works, preamp self noise should be of no concern in this case because the microphone's self noise is way higher than any quality preamp's.
Analyzing the noise by ear is OK too, because what you hear is what is annoying in the noise.
The method could be simplified if you have a stereo preamp (I assume that both channels are of the same quality):
Both microphones could be recorded on separate tracks AT THE SAME TIME, mics located as close as possible side by side, with any kind of signal (except pure tones) as reference, e.g. speach or white or pink noise out of a smartphones speaker.
After level matching the signal in the DAW the noise level can be compared either by ear or with any kind of meters.
The only problem that might show up is: you need a very very quiet room, especially if you want to measure the noise with a meter.
When I do noise measurements on microphones I put them inside a damped box in my very very quiet recording room.
So record the reference signal, then close the box while continue recording the selfnoise, without changing any setting inbetween, would be the way to go.

[Jim Williams]

The intention here was to self-help someone who does not have access to sophisticated and expensive test equipment. So, Jim, will the procedure I recommended work in this specific case?

Probably if the noise is high enough over the base line. Without a comparison one wouldn't know if that noise is at a normal level or whether it's an outlier. Without a normal baseline unit to compare to one wouldn't now if it's in the normal region or not.
There are low cost graphical analyzers available in software, I would try to use those as they also have the ability to store the results for later comparisons. 

[gtoledo3]

Seems like the only possible issue with the process in the original post, if I understand correctly, is that it introduces the variables of another preamp channel and cabling. But those could be verified first by testing each with a single mic and repeatable sound source.

[klaus]

To eliminate these variables, I always use the same input, phantom power source, and cabling.

[Kai]

Without a comparison one wouldn't know if that noise is at a normal level or whether it's an outlier. Without a normal baseline unit to compare to one wouldn't now if it's in the normal region or not.
There are low cost graphical analyzers available in software...

Klaus method does not use an absolute reference like a 94dBSL Sound Level Calibrator.
To use a Calibrator and Analyser needs a certain amount of professional technical knowledge about accoustics to do the measurement right and interpret an meter and analyser result.
For an amateur Klaus' comparison to a "known good sample" is much simpler and gives enough information for the decision "good or bad".

[Kai]

Without a normal baseline unit to compare to one wouldn't now if it's in the normal region or not.

In the last consequence his would mean doing a measurement to check if the unit meets factory specs.
As mentioned this is not easy and needs a certain amount of knowledge and equipment.

Seems some really want to know, so here is how to do it:
First and most important you need to create a sound field using a calibrated Sound Level Meter:
- 1kHz sine wave,
- 1m distant to the loudspeaker,
- homogenous within the size of the tested specimen,
- univectorial (all soundwaves propagate into the same direction at the measurement position),
- with a certain sound level, preferable 94dBSL (1Pa).

The precision of the Sound Level Meter and sound field creation directly affects the precision of the measurement.

This can only be done in an unechoic or very low reverberant environment, else you would get reflected sound waves crossing the original one, which destroy the precision of the measurement.
The reason for this high demand here is, the reference sound field is measured with an omni-directional microphone, but the tested specimen e.g. Neumann KM84, is usually uni-directional. So any crossing soundwave would be interpreted different between the two microphone-types.

Then you would replace the Sound Level Meter by the Device Under Test and record this reference soundwave. Next put the DUT in an isolation chamber and record it's self noise, without changing the preamp gain setting.
Now, using a meter with proper weighting filters (e.g. A- or CCIR468-4) and rectifiers (e.g. RMS, DIN quasi-peak) that correspond to the ones mentioned in the factory specs you can compare the readings between reference sound and noise to determine the absolute noise level, referenced to 0dBSL.

Even if this all is done very accurately, don't expect it to be of better precision or repeatability then +/- 1dB at best.

One uncertainty arises from the fact that the DUT changes the sound field due to its physical size.
That's why I use a corresponding sized measurement microphone for sound field creation, 1 inch for LDCs and half inch for SDCs.

 So, after all, Klaus' method is much easier to handle.