Understanding Microphone Sensitivity Specifications
In the course of building microphones and evaluating microphones I have found a lot of confusion around the meaning of what is called "sensitivity". Part of the confusion is due to changes in the what was popular as "standard" measurement units over the years for audio and sound pressure.
Since we actually measure our mics, whether they be modded or built by us, we wanted to compare how we were doing compared to the great mics of the past and the mics of today.
This article will try to help you understand the gobblety-gook that you find with your microphones sensitivityspec and allow you to compare microphones one to the other.
Here is a sample of what we are talking about:
From the u67 manual 1963
From the Original m251 Manual
English Translation:
NO-load Impedance 200 ohm +/- 10%
By Directional Characteristics .......
Cardiod 0.9mV/uBar, Omni 1.2mV/uBar, Eight 0.9mV.uBar
From APEX 580 web page
Sensitivity: -34dB +/- 2dB (0dB=1v/pa@1kHz)
If you know what this all means then read no further but if any of it seems confusing read on!
Overview
A microphone is a part of a family of devices called "transducers". A transducer's function is to convert one form of energy into another form of energy.
A microphone therefore converts sound energy into electrical energy. In the simplest terms "sensitivity" is a measure of how effectively a microphone does the energy conversion. In reality due to the history of acoustic measurements and audio measurements there are many different units of measurement used for both sound and electrical output, that many times are not well understood and this makes understanding what should be a simple specification something of a mystery to many people.
"Standards are great... Everybody should have one!"
Charles Moore, Inventor of Forth
Brief History of Audio Measurements
Not unlike the history of measurement of anything, (distance, weight etc.) audio measurements have a rather inconvenient history of changing units over time.
In the simplest terms audio is just a form of alternating current (AC) electricity. So the outlet in our homes is AC, why don't we just use volts? It turns out the range of possible voltages with audio goes from micro-volts (millionths of a volt) to tens of volts. This is a huge range. And to make things worse humans don't hear audio (sound) in a linear fashion. So if we change the voltage applied to a loudspeaker from 1 volt to 10 volts (ten times increase) humans don't hear the sound as being 10 times louder but more like twice as loud.
This non-linear hearing that we possess gave rise to using the decibel scale to measure audio and therein lies some of the confusion.
And what can also be confusing is that a decibel scale requires a reference point to have any meaning...
Should we use voltages or decibels for the electrical measurement? We'll come back to this question.
Brief History of Acoustic Measurements
Sound is the compression and rarefraction of air. So acoustic measurements in their simplest terms are really a measurement of air pressure. Early pressure measurement systems used milli-meters of mercury or milli-meters of water in a glass tube.
Between English Empire measurements (pounds per square inch) and the metric system the measurement of air pressure has had an interesting history as well. Quick what is a "Bar" versus a "Pascal" compared to a "PSI"? And how do they compare to one "atmosphere"? All are just measurements of pressure.
If we look at early microphone specs we see measurements in micro-Bars (µb) or millionths of a BAR. Later on in specification listings we see references to dB Pa. This is a decibel scale of sound that is referenced to the Pascal. The link to the Pascal has a detailed chart that lets you compare how much pressure each of these units represents in each other's terms but it might not help you either!
And we sometimes also see references to dB SPL, which is a decibel scale that references units of sound pressure that begins at '0' dB as the threshold of human hearing. The chart below shows how that scale works. This is the commonly used scale when people talk about sound "loudness" in the real world.
A Real Example
Here is a sensitivity specification from the Neumann u47 & u48 in Cardiod (Niere) published in 1959:
Translation for the U 47 line:
"Sensitivity 2.5 milli-volts per micro-bar (Cardiod)"
This means that when the microphone diagphragm is deflected by 1µb (micro-bar) of pressure there will be 2.5 mV of output from the microphone electronics.
Hmmm... µb? How loud is that? My mics seem to show a sound pressure level of 94 dB SPL. How many micro-bars is that?
From the Handbook of Sound Engineers (page 516) we get this convenient set of conversions:
| Pressure( µb) | Pascals (Pa) | Dynes/cm2 | dB SPL |
|---|---|---|---|
| 1 | .1 | 1 | 74 |
| 10 | 1 | 10 | 94 |
So that's easy. Volts and µb are linear scales so:
If 1µb generates 2.5mV
10 µb will generate 25mV
So another way to write the U 47 spec is: 25mV/94 dB SPL
But why 94 db SPL?. Why use that volume to test?
Well... there is another standard measurement of pressure called the Pascal (Pa). One Pascal ( 1 Pa) of Sound pressure is equal to exactly 94 dB SPL.
I can hear the entire world say Ahhh!
So 1 Pa = 94 dB SPL.
And just because they can, some very smart person created a decibel scale that uses 1 Pa as it's '0' dB. This is called 0 dB Pa
... which is the same as 1 Pascal which is the same as 94 dB SPL
So we could also write the spec as: 25mV/ Pa
Confused? Just wait. I'm not done yet!
Milli-volts vs dB
But some peoples children can't leave well enough alone. Oh no. They just got to add their twist to things. So if a U 47 creates 25 mV how many dB is that? Well now that just depends on what kinda dB you are talkin' about.
Here is where our little shop at F.A.R. is rather traditional. We came from a world where professional audio levels were always measured in dBm. That is 1 milli-watt of power and typically, but not always, that mW was going through 600 ohms of "termination" impedance. The voltage measured this way was 0.775 volts and this was the '0' reference for our audio levels in the studio. Period. (well except at one place where everything was wired for 150 ohms, but that's a longer story!)
If we couldn't know that the impedance was 600 ohms then we still used 0.775 V as the '0' dB reference but we wrote down that we measured "dBu". This means the measurement was "un-terminated" but we still used .775 volts as our reference. Simple.
Ah but then the home audio industry got into the mix. They didn't have a history with terminated audio lines and so everything was just voltage to them so why not just use 1 volt as '0' dB. And so they did and this is written as dBV. (re-read the quote from Charles Moore above)
So look at the APEX 580 spec we showed earlier:
Sensitivity: -34dB +/- 2dB (0dB=1v/pa@1kHz)
Look at the part in brackets. First they didn't use proper capitals and this should read:
(0dB = 1V/Pa@1kHz)
What they are saying here is that for their reference they use:
'0' dB is 1 Volt of output, for 1 Pascal of pressure
And therefore the microphone output is 34 dB below that reference with 1 Pa of pressure which can be written as -34dBV. (and only with a 1KHz sound source)
Clear? I thought so.
My Preamp doesn't use dBV
Since pro audio gear uses dBu/dBm not dBV, it means if you look at the specs for your preamp input level this will not quite help you.
1 dBV is about +2 dBu (1 volt vs 0.775 volts)
So at Fox Audio Research if we measured your APEX 580 we would spec it as:
Sensitivity: -32 dBu/Pa
This says the reference is .775 volts, the test pressure is 1 Pa (94 dB SPL and mic is 32dB below the reference.
Why should I care?
As it turns out in with the current state of the art preamps it has become much less important to worry about microphone sensitivity. Most good preamps have enough quiet gain to boost up even very low sensitivity microphones to usable levels.
But if your favourite preamp is vintage and has less than 65 dB of gain and you record delicate acoustic music, you want to understand this stuff.
For comparison we have put together a table with sensitivities normalized to one way of measurement. We converted mV to dBu or dBV to dBu and uBars to Pa as needed.
It's Cool now that you know what all that means eh?
We included some common favourite mics here so you can see the variances as "apples to apples" comparisons.
Microphone (Cardiod) |
Test SPL | Output |
|---|---|---|
U 47 |
94 dB |
-30 dBu |
c414 XLS |
94 dB |
-31 dBu |
U 67 |
94 dB |
-32 dBu |
U 87ai |
94 dB |
-29 dBu |
U 87 (original) |
94 dB |
-40 dBu |
Shure SM58 |
94 dB |
-52 dBu |
AKG c12 |
94 dB |
-38 dBu |
FAR 12 SE |
94 dB |
-38 dBu |
ELA m251 |
94 dB |
-36 dBu |
Royer R-121 |
94 dB |
-48 dBu |
TLM103 |
94 dB |
-30.5 dBu |
Last Word:
Just to keep Charles Moore accurate, some of the measurements in the table are made with a 1,000 ohm resistive load across the microphone output when the tests are made and other explicitly say the microphone must be "open circuit" (no load, very high impedance).
So I gave up and ignored that difference!