Gain (relationship between encoding law and audio level) | ua\d\u)\d
} \(+- | .08 \(+- | .08 \(+- | .05 \(+- | .05
.TE
.LP
\ua\d\u)\d
Measured at one frequency, approximately 820\ Hz or 1020\ Hz
at a level of 0\ dBm0.
.LP
\fINote\fR
\ \(em\ If a sinusoidal test signal is used, uncertainties in the absolute
level position of the companding law characteristic of a practical
encoder require special interpretation of the error limits specified in
modes\ A\(hyD, A\(hyA and (if the signal passes via an analogue point)\ D\(hyD. In these modes, the figures represent the accuracy with which the \fIenvelope\fR
of the
characteristic can be located, rather than the accuracy of any \fIsingle result\fR
. For further discussion and the theoretical location of the envelope, see
Annex\ A to this Recommendation.
.nr PS 9
.RT
.ad r
\fBTable 3/O.133 [T3.133], p. \fR
.sp 1P
.RT
.ad b
.RT
.sp 1P
.LP
4.2.2
\fIReturn loss\fR | optional)
.sp 9p
.RT
.PP
See Table 4/O.133.
.RT
.ce
\fBH.T. [T4.133]\fR
.ce
TABLE\ 4/O.133
.ps 9
.vs 11
.nr VS 11
.nr PS 9
.TS
center box;
cw(72p) | cw(36p) | cw(24p) sw(24p) sw(24p) sw(24p) , ^ | ^ | c | c | c | c.
Parameter Indicated result Error limits (dB)
A\(hyD D\(hyA A\(hyA D\(hyD
_
.T&
lw(72p) | cw(36p) | cw(24p) | cw(24p) | cw(24p) | cw(24p) , ^ | c | c | c | c | c.
Return loss | ua\d\u)\d \ 0 to 30 dB \(+- | \(+- | \(+- | \(em
30 to 40 dB \(+- | \(+- | \(+- | \(em
.TE
.LP
\ua\d\u)\d
Measured at a level \(>=" \(em10 dBm0.
.nr PS 9
.RT
.ad r
\fBTable 4/O.133 [T4.133], p. \fR
.sp 1P
.RT
.ad b
.RT
.LP
.bp
.sp 1P
.LP
4.2.3
\fILongitudinal conversion loss (LCL)\fR | optional)
.sp 9p
.RT
.PP
See Table 5/O.133.
.RT
.ce
\fBH.T. [T5.133]\fR
.ce
TABLE\ 5/O.133
.ps 9
.vs 11
.nr VS 11
.nr PS 9
.TS
center box;
cw(48p) | cw(36p) | cw(24p) sw(24p) sw(24p) sw(24p) , ^ | ^ | c | c | c | c.
Parameter Indicated result Error limits (dB)
A\(hyD D\(hyA A\(hyA D\(hyD
_
.T&
lw(48p) | cw(36p) | cw(24p) | cw(24p) | cw(24p) | cw(24p) , ^ | c | c | c | c | c.
LCL | ua\d\u)\d \ 5 to 40 dB \(+- | .5 \(em \(+- | .5 \(em
40 to 56 dB \(+- | .5 \(em \(+- | .5 \(em
.TE
.LP
\ua\d\u)\d
Measured at a level \(>=" \(em10 dBm0.
.nr PS 9
.RT
.ad r
\fBTable 5/O.133 [T5.133], p. \fR
.sp 1P
.RT
.sp 1P
.RT
.ad b
.RT
.sp 1P
.LP
4.2.4
\fILongitudinal conversion transfer loss (LCTL)\fR | optional)
.sp 9p
.RT
.PP
See Table 6/O.133.
.RT
.ce
\fBH.T. [T6.133]\fR
.ce
TABLE\ 6/O.133
.ps 9
.vs 11
.nr VS 11
.nr PS 9
.TS
center box;
cw(48p) | cw(36p) | cw(24p) sw(24p) sw(24p) sw(24p) , ^ | ^ | c | c | c | c.
Parameter Indicated result Error limits (dB)
A\(hyD D\(hyA A\(hyA D\(hyD
_
.T&
lw(48p) | cw(36p) | cw(24p) | cw(24p) | cw(24p) | cw(24p) , ^ | c | c | c | c | c.
LCTL | ua\d\u)\d \ 5 to 40 dB \(+- | .5 \(em \(+- | .5 \(em
40 to 56 dB \(+- | .5 \(em \(+- | .5 \(em
.TE
.LP
\ua\d\u)\d
Measured at a level \(>=" \(em10 dBm0.
.nr PS 9
.RT
.ad r
\fBTable 6/O.133 [T6.133], p. \fR
.sp 1P
.RT
.sp 1P
.RT
.ad b
.RT
.sp 1P
.LP
4.2.5
\fIAttenuation/frequency distortion\fR
.sp 9p
.RT
.PP
See Table 7/O.133.
.RT
.ce
\fBH.T. [T7.133]\fR
.ce
TABLE\ 7/O.133
.ps 9
.vs 11
.nr VS 11
.nr PS 9
.TS
center box;
cw(60p) | cw(36p) | cw(24p) sw(24p) sw(24p) sw(24p) , ^ | ^ | c | c | c | c.
\ \(em\ If a sinusoidal test signal is used, uncertainties in the absolute
level position of the companding law characteristic of a practical
encoder require special interpretation of the error limits specified in
modes\ A\(hyD, A\(hyA and (if the signal passes via an analogue point)\ D\(hyD. In these modes, the figures represent the accuracy with which the \fIenvelope\fR
of the
characteristic can be located, rather than the accuracy of any \fIsingle result\fR
. For further discussion and the theoretical location of the envelope, see
Annex\ A to this Recommendation.
.nr PS 9
.RT
.ad r
\fBTable 9/O.133 [T9.133], p. \fR
.sp 1P
.RT
.ad b
.RT
.LP
.rs
.sp 6P
.ad r
BLANC
.ad b
.RT
.LP
.bp
.sp 1P
.LP
4.2.8
\fIVariation of gain with input level\fR
.sp 9p
.RT
.PP
See Table 10/O.133.
.RT
.ce
\fBH.T. [T10.133]\fR
.ce
TABLE\ 10/O.133
.ps 9
.vs 11
.nr VS 11
.nr PS 9
.TS
center box;
cw(84p) | cw(48p) | cw(24p) sw(24p) sw(24p) sw(24p) , ^ | ^ | c | c | c | c.
Parameter Level range Error limits (dB) | ua\d\u)\d
Gain variation (sinusoidal test signal at approx. 420,
820 or 1020\ Hz)
} {
\ +3 to \(em40 dBm0
\(em40 to \(em50 dBm0
\(em50 to \(em55 dBm0
} {
\(+- | .10 | ub\d\u)\d
\(+- | .20
| ub\d\u)\d
\(+- | .25
| ub\d\u)\d
} {
\(+- | .10
\(+- | .15
\(+- | .20
} {
\(+- | .15
\(+- | .20
\(+- | .25
} {
\(+- | .10
\(+- | .15
\(+- | .20
}
.TE
.LP
\ua\d\u)\d
Error referred to measurement of \(em10 dBm0.
.LP
\ub\d\u)\d
Provisional value, to be studied further.
.LP
\fINote\fR
\ \(em\ If a sinusoidal test signal is used, uncertainties in the absolute
level position of the companding law characteristic of a practical
encoder require special interpretation of the error limits specified in
modes\ A\(hyD, A\(hyA and (if the signal passes via an analogue point)\ D\(hyD. In these modes, the figures represent the accuracy with which the \fIenvelope\fR
of the
characteristic can be located, rather than the accuracy of any \fIsingle result\fR
. For further discussion and the theoretical location of the envelope, see
Annex\ A to this Recommendation.
.nr PS 9
.RT
.ad r
\fBTable 10/O.133 [T10.133], p. \fR
.sp 1P
.RT
.ad b
.RT
.sp 1P
.LP
4.2.9
\fICrosstalk measurement\fR
.sp 9p
.RT
.PP
See Table 11/O.133.
.RT
.ce
\fBH.T. [T11.133]\fR
.ce
TABLE\ 11/O.133
.ps 9
.vs 11
.nr VS 11
.nr PS 9
.TS
center box;
cw(36p) | cw(60p) | cw(24p) sw(24p) sw(24p) sw(24p) , ^ | ^ | c | c | c | c.
Parameter Remarks Error limits (dB)
A\(hyD D\(hyA A\(hyA D\(hyD
_
.T&
cw(36p) | lw(60p) | cw(24p) | cw(24p) | cw(24p) | cw(24p) , ^ | l | c | c | c | c.
Crosstalk {
Sinusoidal test signal | ua\d\u)\d
} \(+- | \(+- | \(+- | \ | \(+- |
{
Conventional telephone signal | ub\d\u)\d (optional)
} \(em \(em \(+- | .5 \(em
.TE
.LP
\ua\d\u)\d
Measurement to be performed while injecting an auxiliary signal
in the disturbed channel. Appropriate auxiliary signals are defined in
\(sc\ 3.2.4. Error includes effect of finite rejection of the auxiliary signal
by the measurement filter and of quantizing distortion in the measurement
bandwidth.
.LP
\ub\d\u)\d
Measurement error includes tolerances of the weighting filter
given in Recommendation\ O.41.
.nr PS 9
.RT
.ad r
\fBTable 11/O.133 [T11.133], p. \fR
.sp 1P
.RT
.ad b
.RT
.sp 2P
.LP
\fB5\fR \fBOperating environment\fR
.sp 1P
.RT
.PP
The electrical performance requirements shall be met when operating at
the climatic conditions as specified in Recommendation\ O.3,
\(sc\ 2.1.
.RT
.ce 1000
ANNEX\ A
.ce 0
.ce 1000
(to Recommendation\ O.133)
.sp 9p
.RT
.ce 0
.ce 1000
\fBIntrinsic \fR \fBerrors in the PCM encoding process\fR
.sp 1P
.RT
.ce 0
.ce 1000
\fBwhich may affect the interpretation of measured results\fR
.ce 0
.LP
A.1
\fIIntroduction\fR
.sp 1P
.RT
.PP
Pulse Code Modulation (PCM) has some inherent limitations which affect
measurements on PCM encoders. This pertains especially to the
measurement of the variation of gain with input level and of the quantizing
distortion ratio. Due to the limited number of quantizing steps available
for encoding an analogue signal, the output signal of a PCM decoder is
not a
replica of the input signal to the encoder. Depending on the actual amplitude
of the signal samples to be encoded, as compared with the quantizing
thresholds, the output values at the decoder are sometimes greater and
.PP
sometimes smaller than would occur in a linear system. The differences are
called quantizing errors, and exist even for an ideal PCM encoder/decoder
pair conforming to a practical encoding law. A test signal will experience
the
average effect of the quantizing errors in all its samples, which depends on
the amplitude distribution of the signal. For Gaussian noise, the errors
tend to average out, and no measurement problems arise. However, this is
not the
case for sinusoidal signals, and measurement results for gain linearity and
quantizing distortion ratio must be interpreted with care.
.RT
.sp 1P
.LP
A.2
\fIMeasurement of gain and variation of gain with input level\fR
.sp 9p
.RT
.PP
As mentioned in the introduction, the signal at the output of a PCM decoder
may differ from what would occur at the output of a linear system. This
means that a PCM channel may appear to have unexpected gain when measured
with a sinusoidal signal. This \*Qquantizing gain\*U is sometimes positive
and sometimes negative and varies with input level. In the case of linear
encoding, the more quantizing steps available for encoding the analogue
input signal, the smaller the quantizing errors and hence the gain variations.
With a truly logarithmic encoding characteristic the quantizing error would
be independent of the input level.
.PP
The encoding laws used in practice (A\(hy and \(*m\(hylaw) approximate the
logarithmic characteristic by a segmented curve. For the A\(hylaw, this
results in a gain variation which follows the same rules for the segments
No.\ 7 to No.\ 2 and which increases with decreasing input level for segment
No.\ 1. Because the values at the segment end points of the \(*m\(hylaw
characteristic are not multiples of\ 2 (as with the A\(hylaw), the gain
variations for the corresponding segment
portions are similar but not identical.
.PP
Figures A\(hy1/O.133 to A\(hy4/O.133 show the (calculated) variation of
gain with input level when measuring a PCM channel with an asynchronous
sinusoidal signal. Because the gain variation in the upper segments is
always between
+0.043\ dB and \(em0.048\ dB, only the level range below \(em30\ dBm0 is
shown. The gain has a sharp minimum each time the peak of the sinusoid
passes through a
decision value. As the input amplitude is increased, the gain rises quickly
to a maximum before falling again. In the vicinity of the minima, the gain
can
vary substantially when the input level is varied only by small amounts.
With the A\(hylaw, for example, the gain changes by approximately 0.8\
dB (selective
measurements) when the input level is varied between \(em57.00\ dB and
\(em57.066\ dB.
.PP
In this case the ratio of level\(hyto\(hygain variation is 1:11.8. For
greater input levels and for the \(*m\(hylaw, the variation of gain with
input level is smaller but still not negligible.
.PP
For signal levels above \(em60 dBm0, the maximum excursions are within
a range of approximately \(em1.3 to +0.65\ dB (\(em1.0 to +0.9\ dB) for
the A\(hylaw, and
approximately\ 0.5 to 0.3\ dB (\(em0.45 to 0.35\ dB) for the \(*m\(hylaw
depending on the measurement mode selective or (wideband).
.PP
When measuring the gain variation of a PCM channel with a sinusoidal stimulus,
the theoretical considerations described above must be taken into
account. Because the relative level at the encoder input need only be set
within a limit of \(+- | .3\ dB (Recommendation\ G.713\ [10]), and because the
analogue signal generator used for the measurement has some uncertainty
in the send level setting, it is not possible to exactly predict the actual
position on the encoding characteristic or even to avoid the minima. For
this
reason, any single measurement result must be treated as relative to the
envelope of the gain variation characteristic. Additionally, it has to be
.PP
considered that Figures A\(hy1/O.133 to A\(hy4/O.133 represent theoretical
values
with ideal encoders having no quantizing threshold errors. In practice,
deviations from the ideal characteristics due to encoder threshold offset
must be expected.
.bp
.PP
This limitation also applies to measurements of gain, although at high
levels the error is small\ \(em\ of the order of \(+- | .04\ dB.
.PP
To simplify the interpretation of measurement results,
Tables\ A\(hy1/O.133 to A\(hy4/O.133 list the extreme values of the gain
variation
with input level for the A\(hy and \(*m\(hylaw for selective and wideband
measurements. The tables have 64\ lines (multiple of\ 16), so one line
contains the values of corresponding segment portions. For the A\(hylaw
the corresponding gain values in the first three columns are identical.
.RT
.sp 1P
.LP
A.3
\fIQuantizing distortion measurements\fR
.sp 9p
.RT
.PP
The quantizing error results in quantizing distortion which varies as function
of input level.
Figures\ A\(hy5/O.133 and A\(hy6/O.133 illustrate the (calculated) quantizing
distortion characteristics for the A\(hy and \(*m\(hylaw when
measuring a PCM\(hychannel with a sinusoidal stimulus. As with gain measurements,
the quantizing distortion ratio can vary substantially as a result of small
variations of the input signal. The variation ratio reaches its maximum
at the segment end points.
.PP
For the same reason as described above, one can again only refer to
the envelope of the variation of the quantizing distortion ratio when
interpreting individual measurement results. The warning with respect to
quantizing threshold errors in a non\(hyideal encoder applies to quantizing
distortion ratio measurements as well.
.PP
Tables A\(hy5/O.133 and A\(hy6/O.133 contain the extreme values of the
quantizing distortion ratio of an ideal encoder when measured with a sinusoidal
signal. In the tables, \*Qlevel\*U is the input level; \fIS\fR /\fIQ\fR
is the ratio of the corresponding level (at the output) of the stimulus,
measured selectively, to the quantizing noise, measured flat and with a
fixed correction to normalize
the noise bandwidth to 3.1\ kHz.
.PP
\fINote\fR \ \(em\ Tables A\(hy5/O.133 and A\(hy6/O.133 and their accompanying
graphs are mainly indicative, since:
.RT
.LP
1)
the calculations (flat \fIS\fR /\fIQ\fR ) do not compare with the weighted
ratio (
\fIS\fR +\fIQ\fR )/\fIQ\fR result of the method of
Recommendation\ O.132. They are more similar to the use of a tone stimulus
with the filters of Recommendation\ O.131;
.LP
2)
the correction to the 3.1\ kHz bandwidth assumes the quantizing noise
spectrum is flat, whereas it is non\(hyflat and level\(hydependent (so
that no fixed correction will compensate for the lost bandwidth of the
stimulus
rejection filter).
.sp 1P
.LP
A.4
\fIGeneral notes to tables and graphs\fR
.sp 9p
.RT
.PP
The input levels are stated based on values of \fIT\fR\d\fIm\fR\\d\fIa\fR\\d\fIx\fR\uof
exactly 3.14\ dBm0 for the A\(hylaw and 3.17\ dBm0 for the \(*m\(hylaw.
(On this
basis, the selective levels of 1\ kHz sequences of Recommendation\ G.711\
[1] are \(em0.0016\ dBm0 for a the A\(hylaw and \(em0.0024\ dBm0 for the
\(*m\(hylaw.)
.PP
The envelope of a characteristic is a pair of smooth curves tangential
to the characteristic at or near all its extreme values.
.RT
.LP
.rs
.sp 14P
.LP
.bp
.ce
\fBH.T. [T12.133]\fR
.ce
TABLE\ A\(hy1/O.133
.ce
\fBVariation of gain with input level, A\(hylaw.\fR
.ce
.ce
\fBGain calculation based on a selective measurement of the stimulus\fR