5. Nonlinear Distortion measurements
5.1. Total Harmonic Distortion (THD)
Setups:
· THD_AA.SAC
· THD_AD.SAC
· THD_DA.SAC
· THD_DD.SAC
Definitions and test conditions:
Distortion is defined by DIN IEC 268-2. To measure distortion, an amplifier is driven with a sinusoidal signal under standard test conditions. To determine total harmonic distortion, the amplitudes of the harmonics at the output of the DUT are measured, their rms values added and a ratio is formed to the total signal. The result is indicated as distortion in % or as total harmonic distortion in dB. Total harmonic distortion as a function of amplitude or frequency is measured analogously. Measurement of nth-order distortion is performed in the same way except that in this case it is not the rms value of all harmonics that is determined but only individual harmonics are determined or combinations of specific harmonics used for calculating distortion. An example of such a measurement is the 3rd harmonic specified for tape recorders. Harmonic distortion or THD is a measure of quality mainly in the lower and middle frequency ranges. For a fundamental frequency of 8 kHz, for example, the 2nd harmonic of 16 kHz is already at the limit of hearing. The 3rd harmonic of 24 kHz is outside the audio transmission range. Harmonic distortion is therefore not suitable for describing nonlinear characteristics at higher frequencies.
Graphic display:
Total harmonic distortion can be indicated by means of a single measured value. With UPD/UPL, however, the spectral distribution of intermodulation products can be displayed, see Fig. 6. Distortion as a function of frequency, for example, will be shown as a graph same as for frequency response.
Notes on measurements:
In practice, DUTs frequently have a quadratic or cubic characteristic. This means that even-numbered or
odd-numbered distortion products
are predominant in the spectrum. This allows conclusions to be drawn
as to the cause of harmonic distortion:
· a quadratic characteristic is obtained with unsymmetric distortion. Example: different gain for positive and negative halfwaves of a push-pull stage
· a cubic characteristic is obtained with symmetric distortion; this is typical of any type of overdriving.
Examples: saturation with tape recorders, max. deflection of loudspeaker coils. Audio Analyzers UPD and UPL allow distortion measurement up to the 9th harmonic as shown in the setups presented here. If a single harmonic is to be taken into account, this is selected in the Measurement Mode line. In the spectral display, the selected harmonics are shown as wide bars, the remaining harmonics as narrow bars. The components used for measurement are also indicated in the measured-value display.
Setups:
o
THDN_AA.SAC
o
THDN_AD.SAC
o
THDN_DA.SAC
o
THDN_DD.SAC
Definitions
and test conditions:
Same as THD measurements, THD+N measurements use a sinusoidal signal to drive the DUT. However, in THD+N measurements, all spurious signals are taken into account in the result. This means that, in addition to harmonic distortion and noise, other signal components such as mixture products formed with the clock frequency in digital signal processing are taken into account in the result. To evaluate such spurious signals, spectral analysis must be performed in addition to THD+N measurement. When comparing measurements the bandwidth must be taken into account. In accordance with AES 17, THD+N are to be performed at a level of -1 dBFS and -20 dBFS. The measurement bandwidth is limited to half the sampling frequency and must not exceed 20 kHz. In the setups described here a measurement bandwidth of 100 Hz to 20 kHz has been selected, the analog output level is 1 V, the digital level -1 dBFS.
Graphic
display:
The THD+N value can be indicated by means of a single measured value. With UPD/UPL, however, the spectral distribution of output products can be displayed using the post-FFT function and harmonics marked automatically as shown in Fig. 7. This enables nonharmonic signal components to be detected very easily.
Notes
on measurements:
This parameter too can be measured as a function of
frequency or level.
With setups THDNS_AA.SA THDNS_DD.SAC
a linear frequency sweep from 20 Hz to 20 kHz is
performed and the THD+N value displayed versus
frequency.
Setups:
o
MOD_AA.SAC
o
MOD_DD.SAC
Definitions
and test conditions:
Instead of a single sinusoidal signal, a signal
composed of two frequencies, f 1 and f2,
is used, which yields not only the harmonics mfl and
nf2 described above but also combination signals with the frequencies
(mf1 ± nf2).
The occurrence of these signals is referred to as intermodulation. To determine the modulation distortion
in accordance with DIN IEC 268-3, an amplifier is operated under standard test
conditions and driven with a two-tone signal. The frequencies of the two
sinusoidal input signals should be such that f1 is
0.5 to 1.5 octaves above the lower limit and f2 0.5
to 1.5 octaves below the upper limit of the transmission range. The level ratio
is 4:1. To calculate modulation distortion, the squares of the four mixture
products formed by the 2nd-order intermodulation distortion (f 2 +
f1 and f2 - f1)
and the 3rd-order intermodulation distortion (f2 +
2f1 and f2 - 2f1)
are added up and the result referred to the level of signal f2 with
the higher frequency. The result
is indicated in % or in dB.
Graphic
display:
As in the case of distortion measurements, the spectral distribution of the components can be displayed in addition to the measured value.
Notes
on measurements:
In the setups, a level ratio of 4:1 of the sinusoidal
signals is selected. Setup MOD_AA.SAC uses 60 Hz and 8 kHz with -10 dB level.
Setup MOD_DD.SAC uses 41 Hz and 7993 Hz with full-scale amplitude of the
DUT, the latter in compliance with AES 17. If other test signals are to be used, the relevant lines in
the GENERATOR panel are to be changed accordingly. No modifications are
required in the ANALYZER panel; the analyzer automatically adjusts to the test
signal.
5.4. Difference Frequency
Distortion (DFD)
Setups:
o
DFD_AA.SAC
o
DFD_DD.SAC
Definitions
and test conditions:
The difference frequency distortion is determined in a
similar way as modulation distortion but using a test signal composed of two
sinusoidal frequencies fl and f2 of equal amplitude. The difference between the
two frequencies is smaller than the lower frequency value. The voltage of the
difference frequency f 2 - f1 is measured whose position in the spectrum
does not change as long as the frequency difference remains constant (2nd-order
DFD). The 3rd-order DFD is determined from mixture products 2f 1 - f2 and
2f2 - f1. Because of the small frequency
differences used here, great demands are made on the selectivity of the
instrument regarding the measurement of 3rd-order DFD, especially when bandpass
filters are used. Modern
audio analyzers employ FFT analysis for this measurement, results are
calculated automatically in line with standards. Measurement of difference frequency distortion is defined by
various standards that differ as follows:
o
For measurements on amplifiers, DIN IEC 268-3
defines the test signals on the basis of a fixed frequency spacing (mainly 80
Hz) and the arithmetic mean frequency.
Results are referred to twice the output voltage of f2,
the absolute values of the two components 2f1 -
f2 and 2f2 -
f1 being added for determination of the 3rd-order
DFD.
o
IEC 118 defines the DFD for measurements on
hearing aids. Here, the upper frequency and the difference frequency are
specified.
Results are referred to output voltage f2,
and the 3rd-order DFD is determined by means of component 2f1 -
f2 only.
The results obtained with the two standards thus
differ by 6 dB for d2 and are equal for d3 provided
the levels of 2f1 - f2 and
2f2 - f1 do not
substantially differ from each other.
o
For measurements on digital components,
differential frequency distortion measurement is defined by AES 17, the
measurement being in this case referred to as intermodulation measurement. The standard defines as test
frequencies the upper limit frequency based on the selected sampling rate as
well as the frequency 2 kHz below the limit frequency. The peak value of the
total signal is to be adjusted such that it is equal to the peak value of a sinusoidal
signal at full-scale amplitude. As with IEC 268, results are referred to the
total output signal of the DUT.
Graphic
display:
Same as modulation distortion analysis.
Notes
on measurements:
The setup for measurements on purely analog components is in line with IEC 268-3, the center frequency is 10 kHz, the difference frequency 80 Hz. If other frequencies are to be used, the relevant lines in the GENERATOR panel are to be changed. No modifications are required in the ANALYZER panel; the analyzer automatically adjusts to the test signal. For measurements to IEC 118, the relevant lines in the GENERATOR and ANALYZER panels are to be changed. The setups for measurements on digital components give a test signal with a 20-kHz and an 18-kHz tone at full-scale amplitude.
5.5. Dynamic Intermodulation
(DIM)
This measurement function is implemented in Audio Analyzer UPD only. In addition, Option UPD-B1 (Low Distortion Generator) is required for measurements on analog interfaces.
! NOT
SUPPORTED ON THE UPL !
Setups:
o
FFT_AA.SAC
o
FFT_AD.SAC
o
FFT_DA.SAC
o
FFT_DD.SAC
FFT analysis is used where the spectral composition of
a signal is to be examined. Audio Analyzers UPD and UPL provide a highly
efficient tool for this purpose.
The setups described here generate a 1-kHz or 997-Hz test signal with a
level of 1 V or -20 dBFS which can be applied to the DUT. The FFT analysis
itself requires a minimum off settings. FFT Size defines the number of
samples on which calculation is based. Higher FFT Size values give
higher frequency resolution but at the same time entail longer measurement
times. The setups were generated for 8k FFT; due to the high measurement speed
of UPD/UPL it will only very rarely be necessary to select a lower number of
points and thus obtain an even faster measurement. Various windows are available to accommodate for a
wide variety of applications. The setups use the Rife-Vincent window, which is
characterized by steep slope of the bell lobe and excellent far-off
interference suppression. With
noisy signals, spectral averaging may be useful sometimes. This can be
performed using the Average Mode function; the type and number of
averaging measurements can be entered.
Very closely spaced frequency components can be analyzed by means of the
Zooming function. In contrast to the zooming function of the graphic
display, the FFT zooming function actually yields higher measurement resolution
since the signal is preprocessed in the time domain before the FFT calculation
takes place. By entering Center and Span, the center frequency
and the spread range for the zoom FFT are defined.