6. Measurement of Interference and Wow & Flutter
Setups:
o
SNRA_AA.SAC
o
SNRA_AD.SAC
o
SNRA_DA.SAC
o
SNRA_DD.SAC
o
SNRC_AA.SAC
o
SNRC_AD.SAC
o
SNRC_DA.SAC
o
SNRC_DD.SAC
Definitions
and test conditions:
The S/N ratio is the ratio in dB of the nominal output
voltage to the sum of the broadband or weighted measured output voltages with
the source EMF set to zero.
To determine the S/N ratio, the output voltage of the amplifier is measured
under nominal conditions (ie the nominal output voltage V2ref
at full-scale amplitude of the DUT is measured). Then the source EMF is reduced
to zero and the noise voltage V2' is
measured. The result is indicated as noise level V2'
or as S/N ratio 20 lg (V2ref/V2')
db.
Audio Analyzers UPD and UPL provide S/N ratio measurements as automatic test
sequences. S/N ratio measurements
are covered by a variety of test standards and procedures. These differ mainly in:
o
the type of weighting filter used for simulating
hearing sensitivity as a function of frequency,
o
the type of detector used.
For linear audio noise voltage
measurements, the unweighted rms noise voltage is measured in accordance with
DIN 45412 "Noise voltage measurements on sound broadcast receivers and
related equipment". In this measurement, a bandpass filter of 22.4 Hz to
22.4 kHz is used for limiting the measurement bandwidth approximately to the
range of audibility. DIN 45412
further defines a commonly used method of S/N ratio measurement in which an A
filter is used and the rms noise voltage determined. The steep roll-off of the A curve with decreasing
frequencies results in a strong attenuation of hum components, which is
expedient for this measurement as it truly reflects hearing conditions. The standard prescribes the use of an
rms detector, so the average noise power is measured. However, the ear is very
sensitive to sound containing pulses (noise peaks, clicking noise). Therefore,
increasing use is made of a quasi-peak detector to CCIR 468-4 or DIN 45405. The standard DIN 45405 "Noise
voltage measurement in sound engineering" technically coincides with CCIR
Recommendation 468 "Measurement of audio-frequency noise in broadcasting,
in sound-recording systems and on sound program circuits". It defines, for
example, filter curves for weighted and unweighted measurements.
For unweighted noise level
measurements, the same bandpass filter as defined by DIN 45412 is used. Measurement of noise as defined by DIN
IEC 268-3 (amplifiers) provides for measurements using A filters and rms
weighting as well as measurements to CCIR 468-3 (corresp. to DIN 45405).
Notes
on measurements:
Setups SNRA... measure the S/N
ratio as a weighted rms value using an A filter, setups SNRC... use a quasi-peak
detector and a CCIR filter.
It should be noted that the quasi-peak detector requires a settling time of
approx. 3 s to supply valid results. This time is set in the setup.
Apart from the filters used in the above S/N ratio measurements, a wide variety
of other weighting filters is in use in practice. In digital applications, for
example, weighting is performed with a CCIR ARM filter which is also known as
CCIR 2k filter. It differs from the CCIR weighted filter in its reference
frequency of 2 kHz (normally 1 kHz). Moreover, this measurement is rms-weighted.
In the ANALYZER panel, any other weighting filter can be selected in the Filter
line. The automatic S/N test sequence can be switched on or off in the S/N
Sequ line.
A comparison of results of noise voltage measurements is possible only if the test
conditions regarding detectors, weighting filters and measurement bandwidth are
observed. Depending on the type of measurement, deviations of more than 10 dB
may be obtained.
Setups:
o
CRSS_AA.SAC
o
CRSS_AD.SAC
o
CRSS_DA.SAC
o
CRSS_DD.SAC
Definitions
and test conditions:
In accordance with DIN IEC 268-3,
the level difference between the output signal of a fully driven channel and
the output signal of a channel that is not driven is measured. The measurement
is prescribed for both directions, and the results may differ due to
asymmetries of the setup. The measurement is mandatory at the reference
frequency and optional at further frequencies. Often measurements are made over
the entire frequency range and results displayed graphically.
Both broadband and selective measurements are possible. Since with high-quality
DUTs, the crosstalk level is in the vicinity of noise, only selective
measurements are expedient in this case.
For crosstalk, measured values will always be < 1 (negative dB value)
since the measured voltage is referred to the nominal output voltage.
For crosstalk attenuation, which is likewise specified, values > 1 (positive
dB values) will be obtained since
the reference used is reversed.
Crosstalk is measured at a level of -20 dBFS in accordance with AES 17.
Notes
on measurements:
Crosstalk measurement is not a separate measurement
function on Audio Analyzers UPD and UPL, it is performed as a level
measurement, with results being referred to the level values obtained for the
other channel in each case.
The setups determine crosstalk in both directions. Measurements are made
selectively with the aid of the sweep function from 10 Hz to 20 kHz. The
following additional settings are to be made:
o
In the basic setting of the setup, the test
signal is output to channel 1 only. The level for the "analog" setup
is at 1 V and must be set to maximum level of the DUT.
For all other setups, the level is to be set to -20 dBFS.
o
The sweep is started and crosstalk coupled into
channel 2 displayed graphically. In the measurement, the level of channel 2 is
determined and continuously referred to the level measured for channel 1
(setting: Reference MEAS CH 1).
o
Now set TRACE A to HOLD in the
DISPLAY panel and activate TRACE B to display the results of channel 1.
o
Set the reference for the results of channel 1
to MEAS CH 2 in the Reference line.
o
Set the Channel(s) line in the GENERATOR
panel from 1 to 2 so that the other channel will be driven.
o
When the sweep is restarted, the crosstalk from
channel 2 to channel 1 will be displayed in the diagram.
Since settings in various panels
have to be modified when changing from one channel to the other, the use of the
status panel is expedient in this case. In this panel the key command lines for
an application can be combined and settings made in the status panel. In the
setups, this feature has been taken into account, see Fig. 10.
Setups:
o
SEPS_AA.SAC
o
SEPS_AD.SAC
o
SEPS_DA.SAC
o
SEPS_DD.SAC
Definitions
and test conditions:
Measurement of the stereo
separation is very similar to crosstalk measurement. In the latter, one channel
is driven and the levels measured for the two channels and correlated to one
another. In stereo separation measurement, on the other hand, only one channel
is measured and the test signal switched between the two channels. If, as is
usually the case, the input levels for the two channels are equal, identical
results will be obtained for stereo separation and crosstalk measurements.
Crosstalk measurements can however easier be integrated into a sweep and are
therefore performed almost exclusively today. It has become common practice
however to use the term "stereo separation" instead of
"crosstalk".
For digital applications to AES 17, stereo separation measurements are
performed the same as crosstalk measurements. For stereo separation, positive
dB values are obtained.
Notes
on measurements:
For the measurement procedure, the information given
under 6.2 "Crosstalk" applies analogously.
6.4. Wow & Flutter
Setups:
o
WFI_AA.SAC
o
WFN_AA.SAC
o
WFJ_AA.SAC
Definitions
and test conditions:
When storing analog sound signals
on moving media, the sound quality depends on the mechanical precision of the
transport mechanism used. Short-term variations in speed will result in
frequency fluctuations of the sound signal.
To measure the frequency fluctuations, which are referred to as wow &
flutter, a sinusoidal tone is played, FM-demodulated and the signal measured.
Since hearing sensitivity is greatest at modulation frequencies of 4 Hz, wow
& flutter measurements are frequently performed by means of a 4-Hz
weighting filter.
Wow & flutter measurements are covered by various standards differing in
the test signal and the detector used:
o
DIN 45507 / IEC 386 / CCIR 409-2
Reference frequency: 3.15 kHz Evaluation:
quasi-peak detector
o
NAB Recommendation
Reference frequency: 3 kHz Evaluation: average
detector
o
Japan Industry Standard
Reference frequency: 3 kHz Evaluation: rms detector
Notes
on measurements:
Three setups matched to standards
DIN/IEC, NAB and JIS are available for wow & flutter measurements.
Not only the analyzer but also the generator is set, for instance for the
recording of signals on test tapes.