This document discusses how analog voice signals are measured, the
units used, and the points of reference used when you measure.
The quality of a transmission system is defined by the difference
between spoken voice at one end and reproduced voice at the other end. Anyone
who uses the telephone experiences both good and bad connections, and can
probably describe the quality of a particular connection in a subjective way.
But how can you define good and bad quality in an objective way?
In transmission, the first step to answer this question is to decide
upon these questions:
This document answers these questions.
There are no specific requirements for this document.
This document is not restricted to specific software and hardware
Refer to the
Technical Tips Conventions for more information on document
Analog is defined as a signal that has a continuously and smoothly
varying amplitude or frequency. Human speech, and everything else you hear, is
in analog form, and early telephone systems were analog as well. Analog signals
are often depicted as smooth sine waves, but voice and other signals are more
complex than that, since they contain many frequencies. The
figure in the Analog
Voice Measurement section shows the typical distribution of energy in
The vertical axis is relative energy and the horizontal axis is
frequency. The figure in the
Analog Voice Measurement section shows that the
voice frequencies that contribute to speech can extend from below 100 hertz to
above 6000. However, most of the energy necessary for intelligible speech is
contained in a band of frequencies between 200 and 4000.
In order to eliminate unwanted signals (noise) that can disturb
conversations or cause errors in control signals, the circuits that carry the
telephone signals are designed to pass only certain frequencies. The ranges of
frequencies that are passed are said to be in the pass band. Zero to 4000 hertz
is the pass band of a telephone system voice channel-a VF channel. (Sometimes
this band is called a message channel.) Bandwidth is the difference between the
upper limit and lower limit of the pass band. Therefore, the bandwidth of the
VF channel is 4000 hertz. However, the transmission of speech does not require
the entire VF channel. The voice pass band is restricted to 300 through 3300
hertz. Hence, any signal carried on the telephone circuit that is within the
range of 300 to 3300 hertz is called an in-band signal. Any signal that is not
within the 300 to 3300 hertz bands, but is within the VF channel, is called an
out-of-band signal. All speech signals are in-band signals. Some signaling
transmissions are in-band and some are out-of-band.
Any waveform can be characterized in terms of frequencies and power.
The quantities commonly used to describe various aspects of transmission
performance are frequency and power. Many performance standards are stated in
terms of power at a particular frequency. The unit used to measure frequency is
the hertz, abbreviated as Hz or seen with the f symbol. Hertz equals one
(0.00000000125) cycle or one oscillation per second and measures the waves or
frequencies of electric changes each second..
As is common in most electrical systems, power is measured in units of
watts, abbreviated W. Since the power encountered in transmission systems is
relatively small (compared to the power of a light bulb), power is usually
expressed in milliwatts, abbreviated mW.
1 mW = 1 W = 0.001W = 10–3W
In transmission, the common interest is in power ratios rather than in
absolute power. In addition, transmission is concerned with an extremely wide
range of absolute power values. For these reasons, a convenient mathematical
expression of relative power, the decibel (dB), is commonly used. In order to
describe relative power in terms of decibels, you must define the reference
point from which you measure. Based upon the transmission parameter that is
measured, you can use different forms of decibel measurement. Each form of
measurement has a specifically defined reference point. When you use the
appropriate units of power related to specific references, you can measure
absolute power, relative power, and power gains and losses.
Since the power in telephone circuits is small, the milliwatt is used
as the basic power measurement unit, just as the foot is used as the basic
measurement of length. Most measurements of absolute power in transmission are
made in milliwatts or in units that are directly related to milliwatts.
The frequencies that are used in testing usually fall within the voice
frequency band. Commonly used pure (sine wave) test tones are 404 Hz, 1004 Hz,
and 2804 Hz. (The 4-Hz offset is not always stated. However, actual test
frequencies should be offset by 4 Hz in order to compensate for effects that
some carrier facilities have on test tones.) A measurement of 1004 Hz is near
the voice-band frequencies that carry much of voice power, 404 Hz is near the
low end of the spectrum, and 2804 Hz is in the range of higher-frequency
components of the voice spectrum that are important to the intelligibility of
In addition to pure test tones, "white noise" within specific frequency
ranges is used for certain tests. White noise test tones are complex waveforms
that have their power evenly distributed over the frequency range of interest.
"White noise" is a signal that contains all the audio frequencies in equal
amounts, but which manifests no recognizable pitches or tones
This figure illustrates, in a very general and simplified way, how a
test-tone transmission is set up and how test tones are generated and measured
(demarc A to demarc B).
The equipment is set up to test the circuit between the demarc at A and
the demarc at B. You are going to measure 1004 Hz loss inherent in the circuit
between A and B.
The bridging clips at both demarcs are removed in order to isolate the
segment of the circuit under test.
At A, an oscillator is attached to transmit and receive leads (also
called tip and ring leads). At B, a transmission measuring set (TIMS) is
attached to transmit and receive leads.
The oscillator at A is set to generate a pure test tone with a power of
1 mW at 1004 Hz. At demarc B, the TIMS is set to read power in the range of 1
mW. The power reading at B is 0.5 mW. Therefore, the power lost between A and B
1 mW – 0.5 mW = 0.5 mW
A more useful way to express the loss is in terms of the relative loss,
or the ratio between power out (B) and power in (A):
Relative loss = Power out (B)
Power in (A)
Relative loss = 0.5 x 10-3
1 X 10-3
Relative loss = 0.5
Half the power that the 1004 Hz test-tone introduced at A is lost by the time
it reaches B.
This example repeats the test with the use of less test-tone power. The
oscillator at demarc A is set to generate 1004 Hz tone at a power of 0.1 mW. At
demarc B, the power measurement is 0.05 mW. Then, the absolute power loss is:
0.1 mW – 0.05 mW = 0.05 mW
The relative loss, or the ratio between power out (B) and power in (A),
Relative Loss = Power out(B)
Power in (A)
Relative Loss = 0.05 x 10-3
1 x 10-3
Relative Loss = 0.5
The relative loss, or power ratio between B and A, is the same whether
you use a test signal of 1 mW or 0.1 mw.
Mathematically, the decibel is a logarithmic measure. The logarithm, or
log, of a particular number is the mathematical power to which a base number
must be raised in order to result in the particular number. The base number you
use when you deal with the decibel is 10. For example, what is the logarithm
(log) of 100? Another way to ask this question is 'To what power do you raise
10 to get 100?'. The answer is 2 because 10 x 10 = 100.
log (100)= 2
log (1000)= 3
log (10,000)= 4
and so on.
You can also use logarithms to express fractional quantities. For
example, what is the logarithm of 0.001? Another way to ask this question is
'To what power do you raise 1/10 (0.1) to get 0.001?'. The answer is 3. By
convention, the log of a fractional number is expressed as negative.
log (0.001) = -3
Logarithms of numbers that are not integral powers of 10 can be
calculated when you look them up in a table or when you use a hand calculator.
The decibel uses logarithms to express power ratios. By definition, the
deciBel, or dB, is the logarithmic (base 10) ratio of two powers, P1 and P2
dB = 10 log P2
P2 and P1 are power measurements expressed in consistent units. The
number of decibels is positive if P2 is greater that P1. The number is negative
if P1 is greater that P2 (see the table). It is
important that the two powers be expressed in the same units, such as milliWatt
(mW) or Watt (W). Otherwise, this leads to errors in calculation.
* Approximate dB value.
The power ratio between the power measured at B and the power measured
at A was one-half. Expressed in decibels:
(Loss, A to B) = 10 log (0.5)
(Loss, A to B) = –3 dB
With the use of decibels, you can express the loss or gain of a circuit
or piece of equipment without having to explicitly state the actual values of
the input and output power. In the example, the loss between A and B is always
3 dB, regardless of the absolute amount of power that is transmitted.
Absolute power is expressed in milliwatts and relative power is
expressed in decibels. When you establish a relationship between the decibel
and the milliwatt, you can eliminate the milliwatt as an operational unit of
measure and deal exclusively with the decibel and related units of measure. The
unit of measure that is used to express absolute power in terms of decibels is
dBm = 10 log (Power, measured in mW)
Since a milliwatt is the standard power reference in communications, it
is logical that 0 dBm (the absolute power reference when decibel units are
used) is equal to 1 mW of power. Mathematically:
0 dBm = 10 log Power out
0 dBm = 10 log (1/1)
0 dBm = 10 x 0 = 0
Because the power is an alternating current waveform and impedance can
vary as a function of frequency, it is necessary to state what frequency the 0
dBm standard is based upon. The standard frequency is 1004 Hz.
You must also know the resistance or impedance (load) of the circuit.
The standard impedance is 600 Ohms.
Therefore, the reference of 0 dBm is equal to 1 mW of power imposed
upon an impedance of 600 Ohms of a frequency of 1004 Hz.
Tests are usually performed with the use of test signals that are less
powerful than 1 mW (0 dBm). If you apply a 1004 Hz test tone of –13 dBm at A,
you read –16 dBm on the TIMS at B. The loss is still –3 dB.
In any discussion of the performance of a circuit, it is necessary to
describe the power at a particular point in a circuit with reference to the
power present at other points in the circuit. This power can be signal power,
noise, or test tones.
The description of this power is similar to the description of the
height of a mountain (or the depth of the ocean). In order to measure the
height of a mountain, it is necessary to pick a reference height from which to
measure. The standard reference height is sea level, which is arbitrarily
assigned a height of zero. When you measure all mountains from sea level,
comparisons of their height can be made even though they can be many miles
This figure shows test tone transmission from demarc A to demarc B.
In a similar manner, power, at specified points in a circuit, can be
described in terms of the power at a standard reference point.
This point, which is analogous to sea level, is called the zero
transmission level point, or 0 TLP.
Any other TLP can be referenced to the 0 TLP by algebraically summing
the 1004 Hz gains and losses from the 0 TLP to the point of measurement.
The power present at a particular point in a circuit depends on the
power at the signal source, on where the source is applied, and on the loss or
gain between the two points in question.
With the use of the 0 TLP concept, the power in a circuit is described
by stating what the power would be if it were accurately measured at the 0 TLP.
The standard notation is dBm0, which means power referenced to the 0 TLP.
For example, the term –13 dBm0 means that the power at the 0 TLP is –13
dBm. A TIMS that is properly set up measures –13 dBm at the 0 TLP. An example
of a –13 dBm0 signal.
Once the power at the 0 TLP is found, the power at any other point in
the circuit can easily be determined. For example, if the signal is –13 dBm
when measured at the 0 TLP, it is l3 dB below the numeric value of any TLP on
the circuit when measured at that TLP.
If the signal is –13 dBm at the 0 TLP (makes it a –13-dBm0 signal),
then the power at the +5 TLP can be calculated as this output shows:
(TLP) + (Power at the 0 TLP) = Power at the +5 TLP)
(+5)+(–13 dBm0) = –8 dBm
If the –13-dBm0 signal is properly measured at the +5 TLP, the meter
reads –8 dBm.
In a similar manner, if a –13-dBm0 signal is measured at the –3 TLP,
the meter reads –16 dBm:
(TLP) + (Power at the 0 TLP) = (Power at the –3 TLP)
(–3)+(–13 dBm0) = –16 dBm
In order to determine the expected power at any given TLP, it is
sufficient to know the power present at some other TLP in the circuit. And,
just as the mountain does not have to be near the sea in order to determine its
height, the 0 TLP does not have to actually exist on the circuit.
This figure illustrates a circuit between
two demarcs. A –29-dBm test-tone signal is applied at the –16 TLP. What should
you expect to measure at the +7 TLP?
Even though the 0 TLP does not exist on the circuit, you can describe
the power you see at the 0 TLP if it did exist:
TLP)+(Power at 0 TLP) = (Power at the –16 TLP)
(–16)+(Power at 0 TLP) = –29 dBm
(Power at 0 TLP) = –13 dBm
Using the relationship again, you can determine the power at the + 7
(TLP)+ (Power at 0 TLP) = (Power at + 7 TLP)
(+7)+(–13 dBm0) = –6 dBm
Use of the 0 TLP reference permits transmission objectives and measured
results to be stated independently of any specific TLP, and without the
specification of what the test-tone levels are to be or where the test tone is
to be applied.
This figure shows a test tone transmission from demarc A to demarc B.
In addition to the description of test-tone power at various points in
a circuit, decibel-related units of measure can be used to describe noise
present in a circuit.
In order to describe power in a circuit, the term dBm is used, meaning
"power referenced to 1 mW." Since noise typically contains much less than 1 mW
of power, it is convenient to use a reference power that is much smaller than 1
mW. The reference power used in the description of noise is –90 dBm. The
notation used to describe noise in terms of reference noise is dBrn. If you
know the noise level in dBm, you can easily measure the noise in dBrn:
dBrn = dBm + 90 dB
For example, a noise measurement of 30 dBrn indicates a power level of
–60 dBm (30 dB above the –90 dBm reference noise level). This table shows the
relationship between dBm0 and dBrn.
Noise contains numerous irregular waveforms that have a wide range of
frequencies and powers. Although any noise superimposed upon a conversation has
an interfering effect, experiments have shown that the interfering effect is
greatest in the midrange of the voice frequency band.
In order to obtain a useful measure of the interfering effect of noise,
the various frequencies that contribute to the overall noise are weighted based
on their relative interfering effect. This weighting is accomplished through
the use of weighting networks, or filters, within TIMS.
Noise measurements through a C-message weighting network are expressed
in units of dBrnC (noise above reference noise, C-message weighting).
As with test-tone power, noise power can be referenced to the 0 TLP.
For example, if the noise objective for the circuit is 31 dBrnC0, what
is the noise measurement at the +7 TLP?
TLP) + (Noise at the 0 TLP) = (Noise at TLP)
(+7) + (31 dBrnC0) = 38 dBrnC
The noise measurement at the +7 TLP is 38 dBrnC.
What is the noise measurement at the –16 TLP?
(TLP) + (Noise at the 0 TLP) = (Noise at TLP)
(–16) + (31 dBrnC0) = 15 dBrnC
The noise measurement at the –16 TLP is 15 dBrnC.