Advertisement

Article

Measurement System for the Characterization of Hi-Fi Audio Cables

The transmission of an electrical analogical signal between two devices requires a suitable cable that should allow for transferring it with low distortion and losses. Ideally a cable does not introduce power-losses and the electrical signal applied at the input is transferred integrally to the output, without any modification of its parameters. Unfortunately, in real conditions, there is always an alteration of the transmitted signal, because of reactive and resistive (parasite) parameters of the interconnection system constituted by the cable and the two end connectors.

The behavior of a real interconnection system can be evaluated considering different parameters, such as the signal attenuation and distortion that can become critical with the increasing of the frequency. The analysis of the problem can be carried out taking into consideration a set of electrical parameters, depending on the adopted modeling methodology. The traditional approach refers to the lumped elements: resistance (R), inductance (L) and capacity (C), as shown in the electrical model of Figure 1 . The behavior of a cable can be also described by means of the transfer function. Moreover, it is also important to analyze the electric and magnetic coupling between different lines of the interconnect system, because they can be a source of crosstalk noise.

Of the different application fields that can be mentioned as a basic reference for the problem under consideration, one may refer to the audio signal transmission in high-fidelity (Hi-Fi) applications.

Figure 1:  Lumped parameters model of a Hi-Fi cable.

Hi-Fi audio signal bandwidth spans the entire audible range of frequencies, from the 41 Hz (and below) of bass guitar and synthesizer to the 20 kHz harmonics of keyboards and cymbals. Professional applications demand a wide bandwidth interconnecting system to preserve the signal integrity. In this field, there are two main kinds of applications:

  1. The interconnection between audio systems in which the output system has an high impedance ZL (about 100 kΩ and the input system a low impedance ZS (in the range of 80-800 Ω); in these conditions the transmitted signal has low power (approximately some tens of milliwatt);
  2. The interconnection between amplifiers and acoustic diffusers in which the transmitted analogical signal transfers power to the diffusers, which have typical impedance values in the range of 4-16 Ω.

The general opinion of Hi-Fi end users is that the choice of audio interconnection systems has an audible effect upon the perceived sound. This is the reason for frequently experiments with various types of cables and the motivation to purchase expensive cables which claim to provide a high level of subjective performance.

In recent years, the production of these devices has grown out, even if some manufacturers publicize special cable qualities without any established scientific basis.

Even if, from a technical point of view, it is normal that the properties of a connection system may affect the transmitted signal, the perceived subjective differences between cables can have different explanations.

Most would agree that the ideal connecting system is one which brings us as close to the experience of the original musical event, hearing the music as it was recorded without adding or missing signal components (cable neutrality). This ideality can be obtained only with zero series inductance and zero parallel capacitance, providing an unlimited transmittable bandwidth.

Some cable manufacturers adopt a different philosophy, based on the use of networks, filters and additional elements to improve the audio signal by altering it, producing colorations that destroy the natural, musical reproduction. The result is artificial and contrived, rather than ideal and neutral, even if it can be appreciated by some people more than a neutral system.

According to the IEC standards, the measurement of cable parameters requires the adoption of complex procedures, also because some standards refer to cable for general purposes . The aim of this paper is the development of a measurement procedure for the performance evaluation of the high-quality cables adopted for professional audio applications. The main features of these high-performance cables are: attenuation lower than 0.25 dB in the frequency range of DC-50 kHz, resistance around to 0.1 ohm, and inductance in the range of 6-24 H.

In the paper we will report and discuss the experimental results related to the characterization of a set of Proel Die-hard professional cables.

The Measured Parameters

In an interconnection system, the cable's performance is important, but most important is the connector's performance, where one can suppose the signal degradation occurs, especially in live concert applications, where the extremely dynamic cable movement produced by the artists can generate a connection mechanical instability. For this reason, we linked the systems under test to the measurement instruments using the connectors complementary to those mounted by the manufacturer at the end of the cable, so the measured parameters refer to both the cable and the two connections. The obtained results will be pejorative, compared with those obtained testing only the cable, but they will reflect more realistic applications.

To characterize the interconnection systems, we measured their main parameters, such as the electrical R-L-C, the frequency response (magnitude and phase characteristics) and crosstalk, at different signal frequencies.

The R-L-C parameters have been measured with the Wayne Kerr 4265 impedance meter at a frequency up to 100 kHz, with the connections shown in Figure 2 and Figure 3 .

The transfer function has been measured by means of the Stanford DS345 function generator, the Keithley 2001 voltmeter, the HP 5335A counter and the LeCroy LC584AXL digital oscilloscope (Figure 4 ). Sinusoidal signals at different frequencies (from 10 Hz to 100 kHz) have been applied at the cable input, terminated with the generator characteristic impedance (50 Ω). The V1 and V2 signal amplitudes and the phase differences have been measured by means of the voltmeter and counter; visualizing the signals on the oscilloscope. The crosstalk has been measured according to , using a Stanford DS345 function generator and a Stanford SR770 FFT Signal Analyzer. A 2 Vrms sinusoidal signal (Vg) in the frequency range 10 Hz — 100 kHz has been applied to the driven line, terminated with the generator characteristic impedance (50 Ω). The coupled signal, induced on the quiet line (Vi) is measured with the spectrum analyzer.

Figure 2:  Resistance and inductance measurement

Figure 3:  Capacitance measurement

Figure 4:  Frequency response measurement

The quiet line is terminated with the analyzer characteristic impedance (50 Ω for the single-ended input and 100 Ω for the differential input). We measured two parameters:

    1. the NEXT (near end crosstalk ratio), the ratio of the signal amplitude measured at the line in proximity to the generator to the generated signal amplitude Vnext / Vg (Figure 5 ).
  • the FEXT (far end crosstalk ratio), the ratio of the signal amplitude measured at the far end of the quiet line to the generated signal amplitude Vfext / Vg (Figure 6 ).
The Measurement System

In order to guarantee the repeatability of lumped parameters and crosstalk ratio measurements, an automatic testing system has been adopted. The system architecture for the crosstalk measurement is shown in Figure 7 .

A PC hosts an IEEE 488 board for the communication with:

  • The Stanford DS 345 that generates the sinusoidal input signals at the different frequency values;
  • The Stanford 770SR FFT Network Spectrum Analyzer that performs the amplitude spectrum evaluation and measures the crosstalk signal amplitudes;
  • The Keythley 2001 multimeter that performs a high accuracy measurement of the input voltage amplitude to control the stability of the signals at the different frequency values.

The control and measurement software has been developed in the National Instruments LabVIEW environment. It has been developed to perform the following steps:

  1. the reset of all the apparatuses;
  2. the setting of the function generator for the sinusoidal input signal generation;
  3. the monitoring of the input signal amplitude;
  4. the setting of the frequency, input configuration and measurement mode for the analyzer;
  5. the synchronized running of the different instruments;
  6. the acquisition of measured values from the spectrum analyzer and digital multimeter.

The crosstalk ratio measurements have been performed setting the function generator output at 2.000 Vrms. For the accurate measurement of the crosstalk voltage amplitude, the spectrum analyzer has been configured using suitable values of frequency span range, a differential input stage and a peak hold mode. The adopted center frequency and span values are tabulated in Table 1 .

 

Center Frequency Span 10 97.5 Hz 20 390 Hz 50 780 Hz 100 780 Hz 200 780 Hz 500 780 Hz 1,000 780 Hz 2,000 780 Hz 5,000 780 Hz 10,000 780 Hz 20,000 780 Hz 30,000 780 Hz 40,000 780 Hz 50,000 780 Hz 100,000 1.56 kHz

Table 1:  Frequency and span for crosstalk measurements.

The multimeter has been configured to operate in high resolution mode, averaging 100 measures. We carried out the measurements on sale-ready cables, with length ranging from 5 to 10 meters. This choice has involved some problems due to their very low values of the lumped parameters. Specifically, a very critical aspect is the contact resistance compensation, because of the reduced resistances that the tested cables have shown. To obtain satisfactorily results, we performed the instrument calibration procedure at each measurement frequency.

The crosstalk voltage measurement is critical; specifically, the measured rms voltage values are in the range 20 µV – 100 mV. The adoption of the differential voltage measurement helps to reduce the noise effect; however, for each frequency value we also performed the voltage measurement without the input signals, to evaluate the noise level. The characterization of the multipolar cables requires the crosstalk measurements on a high number of conductors with the respective shield.

The mutual position and the distance between the driven line and the quiet line can introduce many relevant effects in the measurement crosstalk parameters. We measured the NEXT and FEXT swapping the driven and quiet lines, to evaluate the crosstalk symmetry.

We tested a set of conductor couples in different mutual positions from the centre to the border of the cable.

The experimental results

The described measurement procedures have been adopted for the characterization of 20 PROEL cables, constructed using high purity oxygen-free copper (OFC): i) professional flexible musical instrument cables (with and without connectors for evaluating the effect of the connectors on the signal transmission); ii) professional flexible passive speaker cables OFC red copper conductors; iii) professional flexible microphone cables OFC red copper conductors.

As an example of the obtained results, the diagrams in Figure 7 through Figure 17 show the lumped elements, frequency response (amplitude and phase) and crosstalk characteristics of three cables: the Die-hard DH340LU5, the HPC640BK (4 x 2.50 mm2 ) and the CMN20.

Figure 7:  Resistance from the beginning to the end of the white signal conductor [DH340LU5]

Figure 8:  Inductance from the beginning to the end of the white signal conductor [DH340LU5]

Figure 9:  Frequency response: amplitude characteristic measured from the conductor (white) and the conductor (blue) [HPC640BK]

Figure 10:  Frequency response: phase characteristic measured from the conductor (white) and the conductor (blue) [HPC640BK]

Figure 11:  NEXT Crosstalk characteristic: stimulus circuit (blue) (red), measure circuit (white) (brown) [HPC640BK]

Figure 12:  Frequency response: amplitude characteristic measured [DH340LU5]

Figure 13:  Frequency response: amplitude and phase characteristics [DH340LU5]

Figure 14:  NEXT Crosstalk characteristic: stimulus circuit (1+) (1-), measure circuit (2+) (2-) [DH340LU5]

Figure 15:  FEXT Crosstalk characteristic: stimulus circuit (1+) (1-), measure circuit (2+) (2-) [DH340LU5]

Figure 16:  Frequency response: amplitude characteristic measured between a couple 1 [CMN20]

Figure 17:  NEXT Crosstalk characteristic: stimulus circuit (couple 1), measure circuit (couple 13) [CMN20]

Figure 18:  The DH340LU5 speaker cable

The DH340LU5 is a cable for the passive speaker, with a 12.5 mm overall diameter, a 5 m length and two Speakon 4 pole connectors (Figure 18 ). The cable shows a flat attenuation in the entire audio band (around -0.1 dB variation), a linear phase variation (only -6 degree) and a cutoff frequency greater than 2.1 MHz.

The cable without connectors is the HPC640BK, a flexible 4-conductor twisted loudspeaker cable with an 11.16 m length. Also the cable HPC640BK shows a flat attenuation in the audio band (around -0.1 dB variation).

The CMN20 is a multipair cable that consists of 20 couples of conductors with 0.14 mm2 of section, individually insulated with 100% foil shield and 9.80 m length.

The cables DH340LU5 and HPC640BK show a similar behavior in terms of frequency response. The DH340LU5 has not significant difference between NEXT and FEXT for each pair of pole; a small difference of behavior has been measured in the crosstalk of the two pole pairs.

The CMN20 cable has a different behavior in comparison with the other ones due to the smaller section of conductors.

The results have been analyzed with reference to the standard IEC 61938 that defines some characteristics for the cables used to transfer analogue signals between audio systems:

  1. speaker cables
    1. the usual value for the conductor resistance should be less than 1/100 ZL ;
    2. the usual value for the inductive reactance at the maximum frequency of interest should not exceed 1/3 ZL ; this condition guarantees less than 1 dB loss at the maximum working frequency;
  2. cables for interconnection between audio systems
    1. the usual value for the conductor resistance should be less than 1/10 ZL ;
    2. the usual value for the minimum capacitive reactance for conductors carrying different audio signals should be greater than 1000 ZL at the highest frequency of interest; this ensure that the relative crosstalk level is approximately of -60 dB;
    3. the usual value for the minimum capacitive reactance for conductors to screen should be greater than 3 ZS (impedance of the signal source); this condition ensures less than 1 dB loss at the highest frequency of interest.

From the obtained results, for all the examined cables, we can highlight the following considerations:

  1. speaker cables
    1. all the cables have a cutoff frequency greater than 2 MHz, a value very greater than the bandwidth of audio signals;
    2. the resistance of the cable (and connectors) is of 0.15 – 0.17 Ω, with 0.03 Ω˜ /m; to have the required R < 1/100 ZL it is only necessary to have ZL > 100 R = 15-17 Ω;
    3. the inductance, an important parameter for this kind of cables, is always lower than 4 µH with around 0.7 µH/m. The condition XL,20kHz < 1/3 ZL requires that ZL > 3 XL,20kHz = 2.3 – 2.7 Ω;
    4. the attenuation at the maximum frequency is lower than 1/10 of that required (1 dB).
  2. cables for interconnection between audio systems
    1. all the cables have a cutoff frequency greater than 2 MHz, a value very greater than the bandwidth of audio signals;
    2. the attenuation at the maximum frequency is lower than that required by the standard (1 dB);
    3. the resistance of the cable (and connectors) is of 0.26 – 1.04 Ω; to have R < 1/10 ZL it is only necessary to have ZL > 10 R = 2.6-10.4 Ω;
    4. to satisfy the condition XC, 20kHz > 1000 ZL it is only necessary that ZL < XC, 20kHz / 1000, for which ZL < 16.5 Ω for example for the cable DH200LU5 ;
    5. the condition for the minimum capacitive reactance for conductor to screen at 20 kHz greater than 3 ZS (impedance of the signal source), requires ZS < 0.33 XC, 20kHz = 0.33*1.66E+4= 5478 Ω, for example, for the cable DH240LU5 .

As a general comment to these results and to correctly analyze the problem related to the cable's impedance and to its variation in the audio band, we must also consider that the impedance of the audio devices connected at the input (source) and output (load) can have a frequency variation.

For example the speaker impedance is usually largely dependent on frequency. Real impedance of a nominal 8 ohm speaker, for example, can vary from 6 ohm to 10 ohm (variation can be much higher on some speakers). The nature of the speaker impedance can also change with the frequency from purely resistive to inductive or capacitive.

Starting from these considerations, the problem of the cable parameter variation with the frequency becomes of secondary importance for most applications.

Conclusions

In this paper, we have implemented some characterization tests for Hi-Fi cables in the real conditions of use. In order to guarantee the repeatability of the lumped parameters and crosstalk ratio measurements, an automatic testing system has been adopted.

A wide set of high performance Hi-Fi cables, with and without connectors, have been characterized, evidencing their behavior differences. The measurement of lumped elements, attenuation, and phase-difference has been carried out for different frequency values. For the multipolar cables, the crosstalk characteristics have been measured.

The experimental results confirm the overall quality of the tested cables, with reference to IEC standard requirements.

Our aim in the next work is to develop other kinds of characterization procedures for high performance Hi-Fi cables. Specifically, the tests on power cables for loudspeakers will be carried out with variable both frequency and load conditions, according to the non linear behavior of real Hi-Fi apparatuses like loudspeakers.

About the Authors
Giovanni Bucci received a degree in electrical engineering in 1985 from the University of L'Aquila in Italy. He is now an Associate Professor in Electrical Measurement at the University of L'Aquila. His current research interests include ADC and wireless device testing, multiprocessor-based measuring systems, digital algorithms for real-time measuring instruments, and power measurements. He has authored more than 80 scientific papers.
Edoardo Fiorucci received a degree in electrical engineering in 2000 from the University of L'Aquila in Italy. In June 2004, he received a Ph.D. in Electrical and Information Engineering from the same university. He is now a postdoctoral fellow of the Department of Electrical Engineering at the University of L'Aquila. His current research interests include wireless device testing, power quality measurement, virtual instrumentation, smart sensors, and power measurements. He has authored more than 20 scientific papers.
Fabio Di Nicola received a degree in electrical engineering in 2003 from the University of L'Aquila in Italy. He is now a Ph.D. student in Electrical Engineering at the university “La Sapienza” of Roma in Italy. His current research interests include analysis and optimization of intervals between periodic calibration for measurement instruments, virtual instrumentation and performance evaluation of Hi-Fi systems.

0 comments on “Measurement System for the Characterization of Hi-Fi Audio Cables

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.