Handbook for Sound Engineers

1226 Chapter 33

is a range of possible values that the waveform can take
on. The possibilities are infinite (within a device’s
dynamic range) for an analog component and finite for
digital components, since digital signals are made up of
discrete samples that must be quantized to fixed steps.
Fig. 33-6 shows a 1 kHz sine wave driving a component
to just below clipping. The level difference between
clipping and the noise floor describes the component’s
dynamic range.

An example is in order. Let us consider a line level
audio signal processor. We can pick any rail voltage that
we like, since no universally recognized standard exists.
A rail voltage of +17.5 Vdc (and –17.5 Vdc for the
negative rail) will allow a peak voltage of 17.5 V to be
realized at the device output. It is customary to express
this voltage in terms of the rms value of the largest sine
wave that the device (peak 3 dB) can produce with an
acceptable amount of distortion. An oscilloscope allows
observation of the wave and any deformation due to
clipping. This rms voltage becomes the maximum
output voltage, and when expressed in decibels becomes
the component’s maximum output level. This level is
expressed in dBm (dB ref. 0.001 W) for impedance
matched interfaces (note that knowledge of the circuit
impedance is required) or dBV (dB ref. 1 V) or dBu
(dB ref. 0.775 V) for constant voltage interfaces
(assuming the bridged impedance condition is
maintained).

Assume that the thermal noise measured at the

device output is about 200μVrms, as measured using an

rms broadband voltmeter. Expressed as a level in dBV,

the thermal noise floor becomes:

This audio component thus has a dynamic range on

the order of 100 dB, a very good figure, and one that is

typical for a well-designed piece of audio equipment,

whether analog or digital.

With the dynamic range established, it is still neces-

sary to use it effectively. If a weak signal is fed to the

component, it may fall far short of the clipping point

established by the power supply voltages, placing it

unnecessarily close to the component’s noise floor. This

will produce a poor signal-to-noise ratio, SNR, even in

a component that has a wide dynamic range, Fig. 33-7.

If the input level control is increased beyond unity,

the thermal noise will likely increase with the signal

voltage, and no increase in SNR is realized. Increasing

the drive (source) voltage will improve the SNR,

assuming that the sending device has a noise floor lower

than the driven device. In some cases, an additional gain

stage may be required, as with microphones and phono-

graph cartridges.

If too strong a signal is fed to the component, the

highest amplitude parts of the waveform may not fit

within the constraints of the power supply voltages and

may drive the component into a nonlinear mode of

operation (clipping). This may yield an excellent

signal-to-noise performance, but a distorted output

Figure 33-6. Dynamic range.

Lout= 20 log 17.5 3–

= 21.8 dB dBV

20

17.5

0.775

= log©¹§·------------- 3–

= 24 dBu

Figure 33-7. Poor SNR.

Lnoise= 20 log 0.0002

–= 74 dBV

20

0.0002

0.775

= log©¹§·----------------

–= 71.8 dBu