by Rupert Neve
Noise can be specified in several ways:
With Gain at Unity Better than -100 dBu
With Gain at 66 dB Better than -62 dBu
Equivalent Input Noise Better than -128 dBu
Or: Noise Factor Not exceeding 1 dB
Or: Equivalent Noise Resistance 50 ohms
Or: Short Circuit Input Noise -136 dBu.
Decibels are used to express relationships and to make comparisons. One could state that amplifier ‘a’ has 3 dB less noise than amplifier ‘b’ but unless you have a point of reference for one of the two amplifiers, you would not know how good they were by comparison with other amplifiers. ‘a’ and ‘b’ might both be pretty bad amplifiers but you would only know that ‘a’ was less bad than ‘b’ by 3 dB.
The term “dBu” is derived from the original telephone standard, dBm, indicating 1 mW, (0.001 watts) of power in a 600 ohm circuit. This works out at 0.775 Volts.
Today we use many different impedances in Pro Audio. The term dBu indicates the same 0.775 Volts RMS as a voltage reference level, not a power level.
There are several types of electrical noise but here we are mainly concerned with noise from the devices used in amplifiers plus the noise from circuit resistance, known as Thermal or Johnson noise. Noise is heard, ideally, at an extremely low level, as a “hiss” accompanying the music signal. (Or in some cases, as a “hum” or ‘buzz” – but these are the result of interference, not circuit noise). Thermal or Johnson noise results from the Brownian motion of ionized molecules within a resistance. Thermal noise is entirely fundamental and cannot be eliminated.
Present day microphones are designed to have a source resistance (more correctly, impedance) of between 150 and 300 ohms. So even if there is no sound input to the microphone, its own source resistance will be generating noise that depends on its resistance.
The amplifier to which the microphone is connected also generates noise from its amplifying devices and circuit resistances. When these two noises – the external and the internal noises are added together, we have an “Equivalent Input Noise”.
A microphone signal is often at a very low voltage level and needs a considerable amount of amplification. Assume that we provide 66 dB of gain: Noise from the input, measured at the output of the amplifier is now, obviously 66 dB greater than at the input. This output noise is relatively easy to measure but, as we’ve seen, it consists of the microphone source resistance noise plus the amplifier noise.
Taking gain into account, this is the Equivalent Input Noise (or E.I.N.). But it does not really tell us exactly how good the amplifier is by itself. Fortunately, if we know all the factors that govern thermal noise, such as temperature, bandwidth and the value of the resistance itself, we can calculate thermal noise very accurately. So with this figure calculated, any difference between our measured noise and the thermal noise must be amplifier noise.
Most designers of good microphone amplifiers just quote the E.I.N. because this figure is often close enough to the calculated resistor noise that amplifier noise should be very small indeed.
There are several ways to provide more information about the noise performance of a microphone amplifier:
For example, a graph can be drawn showing Noise against Resistance. Thermal, or pure resistor noise is a straight line whilst amplifier noise is a curve that depends upon the optimum source resistance for that particular amplifier. The difference between the two noise measurements in dB is the “Noise Factor”. (“N.F.” Familiar to R.F. and communications designers.)
So if we know the source resistance of the microphone, we can state that the Noise Factor is, for example, 1 dB. Such a graph can be very informative if we substitute impedances of an actual (acoustically silent) microphone, because microphone source impedances are not the same at all frequencies – but that subject is for another day!
Another method that reveals more than just the E.I.N. combined figure is to measure the combined amplifier and resistor noise over a range of resistances right down to zero resistance. With a zero value resistance connected to the input (in other words when the input is short-circuited), there is, theoretically, NO resistor noise so any residual measured noise must ALL be due to the amplifier. We then calculate the value of a resistor that would produce noise equal to the residual measured amplifier noise.
So we can state the amplifier noise performance by equivalence to a resistor having the same noise. This is a good way because it’s easy to measure and takes account of any resistance in the amplifier circuit such as printed circuit tracks, connector resistance and length of cable between the amplifier proper and the point at which you have applied an input short circuit. (Little point in having a superb amplifier noise performance, equivalent to a very low resistance if we then add a microphone cable or snake that worsens the noise by several dB!)
Traditionally, high quality microphones such as ribbons had very low source impedances – as low as 30 ohms at the output of a ribbon matching transformer. Moving coil microphones were higher but had not been standardized as they are today. Condenser microphones, before the days of semiconductors, used tube head amplifiers that were coupled to the outgoing line with a transformer. Microphone amplifiers, such as those in a mixing console, also used tubes and these typically have high input impedance.
Microphones are voltage generators, not power generators. It is essential to deliver the maximum possible signal voltage into the amplifier. It was traditional to provide an amplifier input impedance of about 1,000 or 1,200 ohms (i.e. about 5 or 6 times the source impedance of the microphone). This provided relatively low loading on the microphone – whatever its type – and went a long way to avoid voltage loss.
In the early 1960′s when the Pop music scene was exploding and sound levels in the studio became very high, there was concern that the head amplifiers in condenser microphones would overload if the console input impedance was too low. In those early days of consoles, when 48-volt phantom powering was first introduced, I was asked to provide a higher input impedance than the normal 1,000 or 1,200 ohms in order to lessen the load on the built-in microphone head amplifier. This of course, resulted in less “step-up” in the console input transformer and there were fears that with other types of microphone, for example, dynamics, we would lose out on noise at the other end of the scale. The fact that microphones were to be less heavily loaded allowed an increased microphone signal voltage. The reduced loading also resulted in less deviation of frequency response due to variation of microphone impedance and less distortion at high levels.
The Rupert Neve Designs Portico 5012 and 5032 microphone amplifier provides an input impedance of 10,000 ohms. This high input impedance means that variations in microphone source impedance versus frequency have less effect on the sonic quality; it benefits microphone loading, allows greater maximum output, reduces microphone distortion and gives a noticeable improvement in transparency.