A question we often hear in Technical Support goes something like this, “I’ve cranked up the gain of my [pick a Sound Devices product] microphone input and it sounds noisy. What gives?” The short answer is, “that is a meaningless ‘measurement’”. The long answer is far more interesting for those who want a better understanding of microphone preamplifiers. Performing a valid, meaningful measurement of microphone preamplifier input noise requires more than “cranking open” an input and listening in headphones.
Before discussing how to measure noise, it is important to know the primary source of noise. All electronic components generate noise. Resistors, a key component in preamplifier circuitry, have several noise-generating mechanisms. Johnson noise (or thermal noise) is the primary component of a resistor’s noise.1 Here is the equation to calculate thermal noise:
V^2 = kB x T x R
kB is Boltzmann’s constant in joules per kelvin, T is the resistor’s absolute temperature in kelvin, and R is the resistor value in ohms (Ω). More info here: https://en.wikipedia.org/wiki/Johnson%E2%80%93Nyquist_noise
Looking at the equation above, noise is primarily affected by the value of the resistor and by temperature. Higher temperatures and higher resistance result in higher noise. If a resistor could be cooled to absolute zero, voila, it would be noiseless. Bottom line, resistors make noise; resistors are a fundamental component in all electronics like microphone preamplifiers, and the higher the resistance, the higher the noise.
Input Impedance Noise
The input impedance (resistance) of a typical microphone preamplifier is between 1k ohms and 10k ohms. This input resistance is intentionally much higher than the source impedance of a microphone (50-200 ohms) so that the preamplifier won’t load down the tiny signal from the microphone. A typical microphone with a source impedance of 150 ohms generates 0.22 uV of noise (or stated another way, -133 dBV).
The noise generated by a 10k ohm resistor (based on the thermal noise formula above) is around 1.8 uV (-114 dBV). This is almost ten times higher than the noise of the microphone.
Crank up the Preamp?
With an understanding of resistance and input impedance noise, let’s now examine why turning up the gain control and listening to its noise is misleading.
- An unterminated input (with no microphone or load connected) is a high impedance (resistance).
This is a common mistake when trying to determine the noise of a microphone preamplifier. The problem here is that with no connection at the input you are listening to the noise of the input impedance alone. It is going to be much higher, and would never be the case with any kind of microphone attached. The instant a microphone is plugged in, this noise drops significantly. This is also a good reason to turn down, or mute, unused inputs.
- Different microphone preamplifiers (and subsequent gain stages like the fader, headphone amp, etc.) have different amounts of available gain.
This sounds obvious, but it is needs to be stated, “if a product being measured has more maximum gain (a good thing!) than another product, it will sound noisier when both are at maximum.” For instance, all other things being equal, let’s say a Sound Devices mixer has 100 dB of gain from microphone input all the way to a listener’s headphones. Comparing this to a product which has only 80 dB of gain, it will sound like the Sound Devices product has 20 dB more noise, which is a lot. But in reality, the actual noise at the microphone preamplifier is not higher – it is just being amplified more.
The Right Way to Compare Preamplifier Noise
A valid noise measurement between several microphone preamplifiers requires a true “apples-to-apples” measurement. The steps below are a quick way to accurately compare preamplifiers noise performance.
- Calibrate the gain between units under test using a tone generator (or other source with a known output level) plugged into a mic input. Adjust the gain for the same level out of the headphones. Doing this makes sure you have the same amount of total gain through both preamps.
- Plug in a 150 ohm resistor across pin-2 and pin-3 of the XLR connector. This emulates a microphone, but without the acoustic output of a microphone. Alternatively, a dynamic microphone with an on/off switch will accomplish the same.
- Now listen to the difference between the units.
What is the Equivalent Input Noise (EIN) Specification?
Many manufacturers, including Sound Devices, specify preamplifiers using EIN. This specification is helpful to evaluate different mic preamps. As discussed above, the noise heard in headphones is dependent of how much gain is applied by the preamplifier. EIN is simply the noise at the output, less (minus) the gain of the preamp.
EIN is helpful as it removes gain from the equation and makes apples-to-apples comparisons easier. EIN is expressed in dBV or dBu. The lower the number, the better the EIN. This number is properly measured using 150 ohms as an input terminator. The very best EIN that can be achieved is -133 dBV, since this is noise purely from a 150 ohm resistor.
Due to the reasons listed above, measuring microphone preamplifier noise solely by cranking up the gain on with an open XLR connector is not the way to go. Properly terminating inputs and using a known amount of measured gain allows one to make meaningful comparisons between products. The 8-Series and MixPre Series have negligible amounts of additional noise above theoretical limits, allowing for very transparent recordings.
1There are several other noise mechanisms which contribute to preamplifier noise. These noises include bulk noise in resistors due to bias currents, excess noise due to the actual resistor type (metal film vs carbon etc.), input noise current vs. input noise voltage from amplifiers. These other noises contribute much less to preamp noise than Johnson noise.
Also, humans can hear a bandwidth of 20-20 kHz, more or less. Some test gear, however, can measure noise well beyond human audibility, sometimes up to 1 MHz. Since noise is broadband and doesn’t just magically go to zero above 20 kHz, a meter with higher bandwidth will show higher amounts of noise. The important thing to keep in mind is that different pieces of test gear have different bandwidths.