Scribd
Upload
63 views

Sound Quality Can Be Measured: Converter

1) The document discusses measurements of sound quality that can quantify parameters like frequency response, harmonic distortion, noise, and phase and group delay accuracy of STUDER audio equipment. 2) Testing showed very low harmonic distortion below 0.01% even at lower frequencies, and noise levels that increased only at higher signal levels. 3) Measurements of non-linearity, phase and group delay accuracy, and square wave response all demonstrated excellent performance within accepted standards, indicating very high quality audio reproduction from the STUDER system.

Uploaded by

Emi Cabrera
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
63 views

Sound Quality Can Be Measured: Converter

1) The document discusses measurements of sound quality that can quantify parameters like frequency response, harmonic distortion, noise, and phase and group delay accuracy of STUDER audio equipment. 2) Testing showed very low harmonic distortion below 0.01% even at lower frequencies, and noise levels that increased only at higher signal levels. 3) Measurements of non-linearity, phase and group delay accuracy, and square wave response all demonstrated excellent performance within accepted standards, indicating very high quality audio reproduction from the STUDER system.

Uploaded by

Emi Cabrera
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

SWISS SOUND 38

Sound quality can be measured


Behind the excellent sound of the STUDER equipment there are, of course, technical reasons. The corresponding parameters are measurable and can be weighted. The following report gives some examples of the quality control procedures to which our equipment is subjected. Converter Although the introduction of the sigma-delta converters resulted in a significant improvement over previous designs, the conversion of analog signals to the digital level and vice versa is still a potential weak spot in the observed system. ponents of the equipment (including input and output transformers) and therefore corresponds to the situation encountered in practice.
STUDER MicVALVE (incl. Super-ADC) and MultiDAC
Level THD+N -1 -6 -11 -16 -21 -26 d B r A -36 -41 -46
-80

@1kHz -80 -83 -86 -89 -92 -95 -98 -101 -104 -107 -110 -113 -116 d B r A

Karl Otto Bder

-31

STUDER

MicVALVE (incl. Super-ADC) and MultiDAC


Level THD+N @-1dBFS

-0.8

-51 -56 -61 -61

-0.9

-82

-1

-84

-56

-51

-46

-41

-36

-31 dBr

-26

-21

-16

-11

-6

-1

-1.1

-86

-1.2 d B r A -1.4

-88 d B r A -92

-1.3

-90

Silvio Gehri
-1.5 -94 -1.6 -96

-1.7

-98

-1.8

20

50

100

200

500 Hz

1k

2k

5k

10k

20k

-100

Measurement 1 shows the frequency response (left-hand ordinate) as well as the THD+N (righthand ordinate) at a level 1 dB below full scale. In the THD+N measurement the stimulus is filtered out. Striking is that despite the transformers the harmonic distortion does not increase at lower frequencies. Even at approx. 6 kHz the harmonic distortion is still less than 0.01 %. As shown in measurement 2 this distortion is caused by an analog circuit (same measurement with a stimulus of -30 dB). Here the distortion components are no longer recognizable, the THD+N value consists purely of noise.

d B r A -0.2 +0.4

The measurements show the results of a chain of STUDER equipment of the D19 family connected in series, that is, the MicVALVE with 22-bit A/D conversion and the MultiDAC with 23-bit D/A conversion. This chain covers not only the complete conversion and reconversion in the measurement path, but also all other comSTUDER MicVALVE (incl. Super-ADC) and MultiDAC
Level THD+N -29.8 @-30dBFS -80

+0.2

-0

-0.4 -121

-111

-101

-91

-81

-71

-61 dBr

-51

-41

-31

-21

-11

-1

-29.9

-84

-30

-88

-30.1

-92

-30.2 d B r A -30.4

-96 d B r A -104

-30.3

-100

Whereas the two first measurements were conducted with a sliding frequency but fixed level, measurement 3 shows a recording with a fixed frequency (1 kHz) but rising level. The left-hand ordinate again relates to the level, the righthand ordinate to the THD+N value. At lower levels there is a constant background noise of approx. 111 - 112 dB below full scale which increases only at higher levels. Of major interest is measurement 4 that investigates the non-linearity errors of the converters. The measurement was conducted with a bandpass. Throughout the investigated spectrum the maximum deviation remained below 0.2 dB 1

-30.5

-108

-30.6

-112

-30.7

-116

-30.8 20 50 100 200 500 Hz 1k 2k 5k 10k 20k

-120

SWISS SOUND 38 where the values below approx. 105 dB are caused by inaccuracies of the measurement system.
STUDER MicVALVE (incl. Super-ADC) and MultiDAC
16k FFT Spectrum (BH) -60 @1kHz, -80dBFS

-70

-80

pensation. Due to these high values the system must change over to floating point mode. The continuous mode switching results in the uniform staircase of the measurement. Accurate qualitative statements on the characteristics of the test object cannot be made; certain is only that they are significantly better than those of the measuring equipment.
STUDER MicVALVE (incl. Super-ADC) and MultiDAC
Group Delay +3

-90

-100 d B r A -120

+2.7
-110

+2.4

+2.1
-130

+1.8
-140

m s
-150

+1.5

+1.2
-160 20 50 100 200 500 Hz 1k 2k 5k 10k 20k

+0.9

+0.6

Measurement 5 illustrates an FFT analysis of the noise and distortion spectrum with a measurement tone of 1 kHz at -80 dB. At 16 k FFT points and using a Blackmann/Harris window a constant bin of 2.7 Hz over the measurement range is observed. As this value remains constant the noise does no longer increase with the root of f as it would be typical for white noise.
STUDER
6
+10

+0.3

+0 10 2.009k 4.008k 6.007k 8.006k 10.005k 12.004k 14.003k 16.002k 18.001k 20k Hz

MicVALVE (incl. Super-ADC) and MultiDAC


Phase Nonlinearity

+8

+6

+4

The group delay error was computed from measurement 6 and depicted in fig. 7. It also shows the superposition of the effects from the test object and the measuring equipment. It is apparent, however, that the threshold of 0.4 ms referred to in the literature (Blauert/Schlichthrle) is never exceeded. In the tested STUDER equipment no group delay effects are noticeable; only at extremely low frequencies (below 50 Hz) can a rise be observed.
STUDER MicVALVE (incl. Super-ADC) and MultiDAC
Square Wave Response @2kHz 3 2.5

+2 d e g

-0

-2
2

-4

1.5 1 500m

-6

-8
V 0 -500m -1 -1.5 -2 -2.5 -3 0 100u 200u 300u 400u 500u sec 600u 700u 800u .9m 1m

-10

10

2.009k

4.008k

6.007k

8.006k

10.005k 12.004k 14.003k 16.002k 18.001k 20k Hz

The measurement shows that hardly any harmonic components occur; only at 2 kHz can a very small value be detected. Also discrete low frequency components in the noise cannot be detected. Measurement 6 is a very tricky measurement because for measuring the non-linearity of the phase the basic delay of the measurement path must be compensated. In the measurement system (SYSTEM ONE from Audio Precision) very high phase values occur before the com2

In view of these excellent results in the phase error measurement it comes as no surprise that also the square-wave behavior is very good. Measurement 8 shows a highly symmetric pattern of the overshoots with a 2 kHz signal.

You might also like