Audio Interconnect Performance: Claims Versus Laboratory Measurements

by Robert A. Cooper

Submitted to the Department of Electrical Engineering and Computer Science in Partial Fulfillment of the Requirements for the Degrees of Bachelor of Science in Electrical Science and Engineering and Master of Engineering in Electrical Engineering and Computer Science at the Massachusetts Institute of Technology May 22, 1998 @ 1998 Robert A. Cooper. All rights reserved.
The author hereby grants to M.I.T. permission to reproduce and distribute publicly paper and electronic copies of this thesis and to grant to others the right to do so.

Author

, Dep'tment o('Electrical Engineering and Computer Science May 22, 1998
Certified by_ Byron M. Roscoe Thesis Supervisor Accepted by Arthur C. Smith Chairman, Department Committee on Graduate Theses
by Robert A. Cooper Submitted to the of Electrical Engineering and Computer Science Department May 22, 1998 In Partial Fulfillment of the Requirements for the Degree of Bachelor of Science in Electrical Science and Engineering and Master of Engineering in Electrical Engineering and Computer Science

ABSTRACT

With advancements in high fidelity home audio and theater, many claims, theories, ideologies, and products concerning the improvement of sound in high fidelity audio systems have hit the market, regularly confusing consumers. Often, in the quest for better sounding audio systems, audio enthusiasts succumb to false or unresearched claims of ways to improve sound quality, and in doing so, often spend a lot of money. As an example of an unresearched area, companies exist that charge hundreds to thousands of dollars for a "specialized" power cord that connects an audio component (e.g. amplifier) to the AC-line, and purports to offer improved bass response, increased dynamic range, and greater transient-delivering capability. Another area in the high fidelity audio market that has seen relatively little research is audio component interconnects. These cables are used to connect audio components together in order to transfer the signal from one to another, such as for sending the signal from a CD player to a preamplifier. For many years, simple coaxial cable (RCA/phono connectors on both ends) was used to accomplish this task. However, in recent years, companies have surfaced that have technologies (some of which are patented) which tout improved dynamic range, imaging, clarity, or other audible quality enhancement over the simple coaxial cable. The goal of this thesis, and the research presented in it, is to determine whether these audio interconnects have any measurable qualities which would affect the audible sound quality of an audio system. Specifically, the intent is to show whether interconnect quality/construction has an impact on sound quality and to what extent price and performance are correlated. Thesis Supervisor: Byron M. Roscoe Title: Technical Instructor; Director of Undergraduate Teaching Labs

Acknowledgments

Thanks are due to various people who helped me gain the idea for this thesis and who helped in the completion of it by offering guidance, assistance, equipment or just moral support. First and foremost, I should recognize Ron Roscoe, my thesis advisor, who helped lock down the idea for this thesis and got the project going. In addition, it was he who provided the equipment that I used to conduct my tests and measurements, personal contacts to whom I went for guidance and advice, and the many references that I used for the thesis. Secondly, I must acknowledge Dave Smith, my senior year roommate, whose many trips to the hi-fi audio store (and quest for better sound) planted the idea for this thesis. In particular, it was his desire to buy expensive interconnects and speaker wire that led me to wonder if the short length of interconnects could possibly have any audible effects on the performance of an audio system. I wish to thank Monster Cable Products, Inc., of San Francisco, CA, for lending me all of the Monster Cable products that were tested for this thesis. Without their support, obtaining an adequate number of cables to test for an extended period of time would have been very difficult. Thanks and love go out to my fiance, Sharonda Bridgeforth, for the encouragement she gave me while writing my thesis and while trying to finish my final year at MIT. She is a wonderful blessing, and I look forward to our future together. Finally, I have to thank Bryan Bilyeu and Phillip Rowe, my two roommates who not only helped me proofread this thesis, but who provided often needed distraction from all the stress of academia, and Shahram Tadayyon, who drove me all over the place whenever I asked, especially to the golf course.

Table of Contents

1. Introduction.... 1.1 Background -.. 1.2 Previous Work 1.3 Objective.-. ---------------------------------------... 17 17. 35 37

1.4 Thesis Organization.

2. Overview of Equipm ent Used.. 2.1 Audio and Test Equipm ent ------.. 2.2 Audio Interconnects ---.
3. Interconnect Measurem ents... 3.1 Lumped-parameter Model 3.1.1 Matlab Simulation 3.1.2 ICAP/4 Spice Simulation 3.2 Interconnect Impedance Measurements 3.3 Amplitude 3.4 Phase 3.5 Total Harmonic Distortion and Noise (THD+N)... 3.6 Noise 3.7 Interm odulation Distortion (IM D)... 3.8 Crosstalk -- -- - -..38.. ------------------------------------.-----------.----.. ------------------------------------------------------------
------------------------------------------------
4. Sum m ary and Conclusions..
Appendix A: Electronic Industries Association Test Methods.. 42 Appendix B: Audio Precision Portable One Specifications.. Appendix C: Matlab and Spice Simulation Code... References..
List of Figures & Tables
Figure 2-1 Figure 2-2 Figure 3-Audio Component Output Impedances Versus Frequency ------------------------14 Monster Cable M1000i Interconnect Construction Diagram --------------------Lumped-parameter Model of the Interconnect Characteristics, Shown with Source and Load Impedances ---- -----------------------

Table 3-1 Figure 3-2

21 Interconnects' Lumped-parameter Values at 10 kHz ------------------------------Matlab Magnitude and Phase Plots for System Under Ideal Conditions, Plus 20 Hz - 20 kHz Zoomed Plots ------------------------------

Figure 3-3

Matlab Magnitude and Phase Plots for System Using IHF 24 Standard Load Values, Plus 20 Hz - 20 kHz Zoomed Plots ----------------------ICAP/4 Spice Magnitude and Phase Plots for System Under Ideal Conditions ICAP/4 Spice Magnitude and Phase Plots for System Using Zs 1 kQICAP/4 Spice Magnitude and Phase Plots for System Using IHF Standard Load ICAP/4 Spice Zoomed Magnitude and Phase Plots for System Under Ideal Conditions Interconnect Impedances Versus Frequency Interconnect Phase Responses --------------------

Figure 3-4

Figure 3-5

Figure 3-6

Figure 3-7

Figure 3-8 Figure 3-9

------------------------------
Figure 3-10 Total Harmonic Distortion + Noise Measurements Versus Frequency -------------------------------------------Figure 3-11 Interconnect Noise Measurements Versus Frequency --------------
Figure 3-12 Interconnect Crosstalk Measurements Versus Frequency -----------

Chapter 1

Introduction

1.1 Background

In the audio industry, a noteworthy debate exists in the area of audio component interconnects (hereafter referred to as interconnects or cables). These interconnects carry the audio signal from one piece of audio equipment to another, and some assert that they affect the sonic performance of the audio system. Often, when a compact disc player or other piece of audio equipment is purchased, a pair (one for each of right and left channels) of basic interconnects is packaged with it. These interconnects, if purchased separately, are relatively inexpensive - costing two to three dollars. Their construction is simple: typically a one-meter coaxial design with a 20 - 24 gauge, multi-stranded center conductor, a braided or wrapped copper shield/ground, and RCA/phono plugs on each end. Particularly in the last five to ten years, a handful of interconnect manufacturers have made a name for themselves by offering interconnects that are supposedly superior to basic interconnects in construction, design, and performance. The manufacturers patent their cables as the result of a new method of wire winding technology or because their design employs exotic insulation. For example, Monster Cable Products, Inc., which is a
well-known interconnect and speaker cable manufacturer, produces the M550i.
addition to winning awards and top reviews1 , this product boasts "greater overall clarity, extended frequency response, lower noise floor and extended dynamic range, and improved reproduction of imaging, soundstage and depth". The advertised characteristics of the M550i include a 100% copper foil shield, 24k gold contacts, special dielectric insulation, two conductors with "multiple-gauge wire networks", and heavy duty construction that is capable of withstanding years of use. One group of devout audiophiles suggests that cable construction and quality has a large effect on the quality of sound produced by an audio system. On the other hand, another group insists that these supposedly higher-quality interconnects do nothing more than add cost to the audio system; they believe that the cables negligibly affect the signal being passed through them. In other words, they contend that interconnects do not affect the signal enough to alter quality of the sound. While both sides have convincing and compelling points, gray areas still exist in both arguments. The side advocating the

benefits of cable quality often cites listening tests in which subjects decided which cable(s) made the overall system sound better in comparison to others. However, this group has no measurements or anything else concrete to support the listening test results. Those who do not believe that interconnects affect system performance insist that because of the short length of the cables and the small frequency range of audio, the parasitic elements - series inductance and resistance and shunt capacitance - of the interconnect are of no noticeable consequence.
1The M550i was the winner of the Hi-Fi Grand Prix Award from AudioVideo Magazine International and
was given a "unanimous, unequivocal top rating" by Home Theater Magazine.

1.2 Previous Work

Much of the work done thus far in audio component interconnection has been in the area of speaker cables. Various audio magazines have conducted tests on speaker cables, as has the Audio Engineering Society in more than one issue of their journal 2. These magazine articles, as well as basic engineering knowledge, make it clear that it is best to be cautious when buying speaker cable. Despite the fact that people still disagree as to which type of cable is best, it is clear that the lengths of cables used to connect speakers to amplifiers - often between 15 and 30 feet in today's surround sound systems - are significant. Therefore wire should be chosen that is sized to handle the current involved in driving a speaker, low enough in resistance to avoid wasting a significant amount of power, and low enough in capacitance to avoid attenuating high frequencies or causing power amplifier instability. The main motivation for this thesis came after reading Brent Butterworth and Al Griffin's article, "String 'Em Up", in the August 1997 issue of Home Theater. In this article, the authors present the results of a supposedly single-blind listening test (the listeners did not know which cable was connected to the system). The article states that, "Each panelist ranked the cables in three groups, roughly corresponding to 'OK', 'good', and 'great', and each panelist also picked a favorite." This rating scheme seems
acceptable; however, in the authors' final results, there were numeric quantities for the interconnects' performance, build quality, ergonomics, features, value, and overall rating, leading one to wonder from where the numbers came. The source of the numbers is never explained, and the qualifications of performance are a bit arbitrary. For example,
Reference section for Audio Engineering Society publications concerning speaker cable tests.
in the review of AudioQuest's "Turquoise" product, "edgy" was a word used by one listener to describe the sound, while another said, "The percussion had a nice, full sound, with lots of drive." These descriptions and classifications are not at all scientific and are indeed subjective opinion. It is, therefore, the goal of the work conducted in this thesis to show scientific measurements and test results which determine whether these interconnects have any perceptible effects in audio systems.

1.3 Objective

The objective of this thesis is to evaluate various models of interconnect cables in order to determine the impact of their electrical characteristics and performance on the rest of the audio system. In particular, 11 different models of interconnect were tested and evaluated based upon their lumped parameter values, impedance, amplitude and phase response, and total harmonic distortion, crosstalk, and intermodulation distortion performance. An approximately 15-year-old Tandy Corporation interconnect was used as a reference to which all of the other cables will be compared. In running the aforementioned tests, the following pieces of N.I.S.T. (National Institute for Standards Technology) traceable equipment were used: Hewlett Packard 4192A Low Frequency Impedance Analyzer, Hewlett Packard 34401A Digital Multimeter, Hewlett Packard 33120A Arbitrary Waveform Generator, and the Audio Precision Portable One Plus Analog Domain Audio Analyzer. In addition, various pieces of audio equipment were evaluated for such things as input and output impedances, so that realistic numbers could be obtained for use in frequency and phase response calculations.

1.4 Thesis Organization

This thesis is divided into three major sections. Chapter 2 provides an overview of the different pieces of test equipment, audio equipment, and interconnects that were used or tested in the various experiments. Primarily, this chapter is provided for
background purposes in the event that future experimenters intend to reproduce the results presented in this thesis. Chapter 3 lays out the various electrical tests that were run on the interconnects, and finally, Chapter 4 summarizes all of the experimental findings in a qualitative manner and discusses the importance of interconnects in audio systems.

Chapter 2

Overview of Equipment Used
2.1 Audio and Test Equipment
The following audio and test equipment was used in conducting the experiments for research presented in this paper. A brief description 3 of each, with pertinent
performance specifications, has been provided for background information in case one desires to reproduce the tests/results that are discussed.
Hewlett Packard 34401A Digital Multimeter: S/N US36047341, Calibrated (NIST traceable) 2/5/98 by HP * 6/2 digit, 15 parts-per-million basic DC accuracy * AC Voltage Measurement Accuracy: (0.06% of reading + 0.03% of range) for 10 Hz to 20 kHz. Hewlett Packard 33120A Arbitrary Waveform Generator: SN US36018116, Calibrated (NIST traceable) 2/10/98 by HP * Features sine, triangle, square, ramp, and noise waveforms, a 12-bit, 40 Mega Samples/second arbitrary waveform generator, with sweep and modulation capabilities. * 6/2 digit, 15 MHz synthesized function generator * Amplitude (into 50 Q): 50 mVp., to 10 Vp_, * Accuracy (at 1 kHz): 1% of specified output * Flatness (sine wave relative to 1 kHz): 1% (0.1 dB) for frequencies less than 100 kHz.

years old, with basic nickel- or zinc-plated connectors that were somewhat dirty and oxidized. Likewise, the wire was of a basic coaxial design - the center conductor was a small gauge and the return was a braided or wrapped copper shield return with no EMI foil shielding (foil shielding is found in most of the interconnects tested). The top of the line product that was tested was the Monster Cable M1000i, one of Monster's elite audiophile cables, selling for $129.95 per two-meter pair. The M1000i was constructed of two identical, twisted-pair conductors (for signal and return), a 100% foil shield, 95% braided copper shield, 24k gold contacts for better conductivity, and, among other features, specially wound wire networks (see numbers 11 and 12 in Figure 2-2) which separately carry low, middle and high frequencies. See Figure 2-2 for a
4 diagram of the M1000i's construction. All of Monster Cable's other interconnects
included some of these features, although only the Interlink 800, M850i and M1000i included both the copper and the foil shielding.
Interconnect Construcuon viagram Figure 2-2: Monster Cable Mi UUUo

4 Monster

Cable M 1000i information taken, and diagram reproduced, from its retail packaging.
The following items list the cables tested and their physical characteristics. 5 All cables tested were two meters (6.6 feet) in length, unless otherwise noted. Prices are shown strictly for comparison purposes. 6 * Tandy Corporation "Reference" Cable (1.82 meter length) - model number and price unknown: Coaxial design with multistranded center conductor, braided shield/return, and nickel or zinc RCA connectors.
* Radio Shack Cat. #42-2605 (1.82 meter length) - $10.95: Coaxial design with multistranded center conductor, braided copper shield/return, and gold plated RCA connectors * Monster Cable Interlink 100 - $12.95: Approximately the same construction as the Radio Shack product. * Monster Cable Interlink 250 - $23.95: 24k gold plated split-tip center pin, heavy duty RCA connector, 100% foil shield that is grounded at the source, and two balanced twisted-pair conductors.

* Monster Cable Interlink 400 Mk II - $41.95.: In addition to the I250's characteristics, two different types of special dielectric insulation material, dual solid core center conductors, and separately wound low and high frequency wire networks. These networks supposedly bundle two separate wire windings into one larger wire (for example, see numbers 11 and 12 in Figure 2-2). Monster Cable claims that these separate networks better carry different frequency bands for "precise imaging and natural tonal reproduction." * Monster Cable Interlink 800 - $83.95: In addition to the 1400 Mk II's characteristics, silver content solder joints, 100% foil and 95% braided copper shield. * Monster Cable Interlink CD - $30.95: Same as 1400 Mk II's characteristics, but an older design. * Monster Cable M350i - $49.95: Same as the 1400 Mk II's characteristics, in a newer design, with seemingly heavier gauge conductors, minus one of the special dielectric insulating materials.
5 Characteristics taken from the respective interconnects' retail packaging. The technologies used in Monster's interconnects were not explained in technical detail on the packaging, and are thus not discussed in this paper.
prices shown for most of the Monster Cable products were obtained from retail store chain Tweeter,
Etc.'s Monster Cable displays. Prices for the 1250, 1400 Mk II, 1800, and ICD were approximated from the price listings on Monster Cable's website, www.monstercable.com.
* Monster Cable M550i - $69.95: Same as the 1800's characteristics, minus the braided copper shield. Possible heavier gauge conductors. * Monster Cable M850i - $99.95: Same as the M550i's characteristics, plus braided copper shield. * Monster Cable M1000i - $129.95: Same as the M850i's characteristics, but with three multiple-gauge separately wound networks (instead of two), for low, middle, and high frequencies, and an improved RCA connector.

Chapter 3

Interconnect Measurements
3.1 Lumped-parameter Model
The lumped-parameter model for the cables' resistance, capacitance, and inductance characteristics was used to calculate the theoretical magnitude and phase response of the interconnects, with specific load and source impedances. It should be noted, however, that this lumped-parameter model, which is shown in Figure 3-1, is only approximate and therefore has been verified via other tests. See sections 3.2, 3.3, and 3.4 for these other tests, which include impedance, magnitude, and phase response (respectively). Output Impedance

of Component

Input Impedance
Interconnect Equivalent Circuit of Component
Figure 3-1: Lumped-parameter Model of the Interconnect Characteristics,
Shown with Source and Load Impedances An explanation of the component variables follows. Zs is the output impedance of the audio component from which the signal is being sent. R1 and L 1 are the resistance
and inductance of the center conductor of the interconnect, while R2 and L2 are the resistance and inductance of the ground (return) wire. Technically, R2 and L2 exist in the ground connection between the load impedance, Zin, and the capacitor. However, after running simulations on both circuits, it was found that over the frequency range 10 Hz - 50 kHz, the choice of placement for R2 and L 2 made no difference in magnitude or phase response. However, with higher frequencies, R2 and L2 placed in the ground led to responses that were monotonically decreasing, rather than exhibiting spikes near 10 MHz (see Chapter 3, section 3.1.1 and 3.1.2), since the inductance L2 has no effect on the circuit and thus no L2-C resonance occurs. C is the lumped-capacitance of the interconnect, and is thus the most inaccurate value of measurements, since the capacitance is distributed along the length of the cable, while the measurement is taken across the center conductor and shield at one end of the cable. Finally, Zin is the input impedance of the audio equipment to which the interconnect is delivering the signal. The Hewlett Packard 4192A Low Frequency Impedance Analyzer was used to obtain the values for R 1, R2, L 1, L2, and C. In order to measure C, the interconnect was connected to the analyzer, with the center conductor connected to the "high" side and signal return to the "low" side, and the other end of the interconnect left open-circuited so as to quantify the center conductor-to-signal return capacitance. For the inductance and resistance measurements, the end of the interconnect not connected to the analyzer was shorted - the center conductor was connected directly to the return so that the center conductor and return wire resistance and inductance would be determined in one measurement. The resulting inductance and resistance readings given by the analyzer were, therefore, the values for the entire length of the cable twice - down the center

Cable Reference Radio Shack Int 100 Int 250 Int 400 Int 800 Int CD M350i M550i M850i M1000i
Resistance R1 + R2 (i) 0.210 0.234 0.256 0.378 0.137 0.122 0.418 0.266 0.116 0.112 0.073
Inductance L, + L2 (jIH) 0.68 0.85 1.23 1.98 1.34 1.18 1.19 1.68 1.37 1.35 1.36
Capacitance C (nF) 0.1966 0.1371 0.1598 0.1822 0.3328 0.2853 0.3328 0.2413 0.3411 0.3386 0.3655
Table 3-1: Interconnects' Lumped Parameter Values at 10 kHz
This table gives all of the lumped parameter measurements taken on all of the interconnects that were used in the tests. The listings and description of each
interconnect is given in Chapter 2. The Hewlett Packard 4192A was used to take each of the above measurements for R, L, and C at 1 kHz, 10 kHz, 25 kHz, and 50 kHz. Values at 10 kHz were used. See Chapter 2 for a description of the HP4192A. The values for R 1, R2 , L 1, and L2 were worst case values from actual interconnect measurements (see shaded boxes in Table 3-1 above) - R1 = R2 = 0.209 Q and L 1 = L2 = 0.99 pH. The value chosen for the capacitance, C = 0.3386 nF, was nearly the highest value of all the cable measured, and it was the worst-case measurement at the time that these simulations were run. The actual worst-case capacitance measured was 0.3655 nF, which would presumably add 7.4% error over using C = 0.3386 nF (i.e. a 7.4% lower break point frequency). The following magnitude and phase plots in Figure 3-2 are the result of the Matlab simulations with Zin equal to a purely resistive 100 ko load and Zs set to an ideal value of zero ohms. Zoomed plots that cover only the audible frequency range have been provided since it is the range of interest in dealing with audio system performance. As a
note: Matlab wraps the phase; therefore, any discontinuities in the phase response are places where the phase falls too low to fit into the -180' to 1800 y-axis values.

)1 0.0.

, 0.005

-0.005

(. -0.005

10e 10[ 10e2

Figure 3-2: Matlab Magnitude and Phase Plots for System Under Ideal Conditions, Plus 20 Hz - 20 kHz Zoomed Plots
Figure 3-2 shows that the system response is flat over the audible frequency range. Furthermore, neither the magnitude nor phase deviates from flat until frequencies above 1 MHz. However, these simulations were performed using the ideal conditions of zero source impedance and very high input impedance of 100 kQ. A more relevant simulation would be one that embraces a more worst-case scenario. One such simulation that better approximates real-world situations is to use a source impedance that is closer to that found in a piece of audio equipment. A source impedance, Zs, equal to 1 kQ is a

more realistic value since, during output impedance measurements, the Teac EQ-220A Graphic Equalizer exhibited approximately this value over most of the audible range Therefore, this value for Zs was added to the circuit model. impedance Zin was changed to mirror a worst-case load. The Electronics Industries Association (E.I.A.) has published guidelines on standard loads to be used in audio component tests 9. One guideline is that audio

In addition, the input

components should be loaded with a 1 kQ resistor in parallel with 1000 pF capacitor during testing, called the IHF (for the Institute of High Fidelity, which created the testing guideline) standard load. Therefore, one more simulation was conducted with Zn set equal to this standard load of 1 kQ in parallel with 1000 pF. This can be considered a worst-case simulation for the system, since the E.I.A.'s recommendation for input impedances is 100 kC2 minimum, and none of the audio equipment input impedances tested for this thesis even approached 1 kQ over most of the audible range. Figure 3-3
shows the 0 - 100 MHz and 20 Hz - 20 kHz results of this IHF standard load simulation with Zs set equal to 1 kQ2, as discussed in the previous paragraph.
The actual impedance was approximately over most of the range, going as high as 1178.at 20 Hz. See Figure 2-1 for a plot of output impedance versus frequency. 9 E.I.A. Standard RS-490: Standard Test Methods of Measurement for Audio Amplifiers, Electronic Industries Association, November, 1981. See Appendix A for a listing of pertinent standards.

3.-4 -5 Frequency [Hz]

Figure 3-3: Matlab Magnitude and Phase Plots for System Using IHF Standard Load Values, Plus 20 Hz - 20 kHz Zoomed Plots
In this worst-case scenario, there are a few points of interest. According to these simulations, the magnitude of the system is approximately 6 dB down, compared to the previous simulation shown in Figure 3-2, right from the beginning. However, this -6 dB level is due to the output impedance, Zs, of 1 kQ forming a voltage divider with the 1 kQ input resistance of the IHF standard load values. This divider relationship caused the 6 dB of loss. Therefore, the other -0.022 dB (see zoomed magnitude plot) is due to the 0.418 Q of cable resistance. This very small loss is constant over most of the audible range, so it would unnoticeable to the human ear since there are no magnitude variations. Near 10 - 20 kHz, though, the response falls another -0.03 dB. This change is due to

resulted in Figure 3-3, except that Figure 3-7 was taken over the range from 10 Hz to 50 kHz, rather than 20 Hz to 20 kHz as in the Matlab simulations. The simulations were run for comparison purposes with the Matlab results.
-IMegritude of 1OMEG 1MEO 100 1K [Hz] Frequency Function from Vin to the Transfer 1
-50 -Vs 1K Phase 1OOK 1 K [Hz] Frequency of Vini 1 ME IOMEG
Figure 3-6: ICAP/4 Magnitude and Phase Plots for System Using IHF Standard Load

CL -800

fMaignltucde of the Transfer
10K 1K [Hz) Frequency Function from

Frequency[Hzl of Vin

Figure 3-7: ICAP/4 Zoomed Magnitude and Phase Plots for System Using IHF Standard Load
These figures closely agree with Matlab.
For example, the magnitude is down
about 0.02 dB at 20 kHz, and the phase is down approximately four degrees. Again, the roll-offs in the responses are due to the 1000 pF capacitive component of Zin, and are not attributable to the response of the cable itself.
3.2 Interconnect Impedance Measurements
Impedance measurements using the Audio Precision Portable One were taken so as to obtain plots of interconnect impedance verses frequency. These measurements were very useful since they were a reality check for the calculations and simulations presented in the previous two sections. Figure 3-8 shows the plots for all of the various
interconnects that were available for testing.
Cable Impedances vs. Frequency

-- Reference Cable

0 4000-

0 0ttt-

- MCInterlink 100 -MC Interlihnk 250 --- MC Interlhnk MKII 400 MCInterhnk 800

-4MC Interhnk CD

04000-

H Hm i mH Hm WH

-MC M350i -*-MC M550i -- MC M850i --- MC MI001 + Radio Shack

0 1000

Q40' NQ,. - ro ( t
Figure 3-8: Interconnect Impedances versus the Logarithm of Frequency
As can be seen, half of the interconnects' impedances generally fall below the "reference", and half above. It is interesting to note that the half that are below the reference actually climb above the reference at approximately 25 kHz. This is due to the reference cable having the lowest inductance of all the cables tested; inductive impedance becomes significant at higher frequencies since ZL = j2ifL. Regardless, all of the
impedances remain below seven-tenths of an Ohm over the frequency range 10 Hz - 50 kHz. Since the output impedance of audio equipment is to be no greater than 10 kQ and

3.4 Phase

The phase response tests produced as little useful information as the amplitude tests. A phase plot is provided in Figure 3-9 to show how close the phase measurements were to baseline reference. The response for only two cables has been provided because any more cable phase plots would clutter the graph and would yield no additional helpful information since all of the cable phase responses were nearly identical.
Interconnect Phase Measurements vs. Frequency
---- Baseline Test ---- Reference Cable S--A-MC Interlink CD
Figure 3-9: Interconnect Phase Responses
The phase tests were run on the Audio Precision Portable One from 10 Hz to 50 kHz, using 1/3 octave measurement intervals. It turns out, as seen in Figure 3-9, that the phase response tests yielded interconnect performance of -0.0'/+0. 1' from baseline over the entire frequency range.
3.5 Total Harmonic Distortion and Noise (THD+N)
The total harmonic distortion plus noise (THD+N) tests yielded somewhat more interesting results. In this case, THD+N was measured against frequency (10 Hz - 50 kHz), with the reference (baseline) test being performed with the XLR-to-phono connectors and RCA-RCA plug as in the previous two tests discussed in sections 3.3 and 3.4. Then, the RCA-RCA plug was removed and the interconnect under test was put in its place. The plot in Figure 3-10 on the following page shows the results of all the THD+N tests. The Audio Precision instrument was set for medium integration speed, rather than fast, in order to yield more accurate results. The bottom (lowest level) plot is the
reference measurement, labeled baseline test. All of the interconnect measurements fall at, or above, this reference level, with the one exception of the M1000i at 3125 Hz. This discrepancy is likely due to the selected integration speed of the Portable One and the randomness of white noise in the environment. For instance, at the period of time when the Audio Precision instrument was measuring the THD+N performance of the M1000i at 3125 Hz, some of the noise samples may have been slightly lower in magnitude than when the reference measurement was taken. 12 If the measurement falls on the same point
For a discussion of white noise, refer to a textbook dealing with signals and systems or discrete time

signal processing.

as the reference (e.g. the Radio Shack product, MC 1100, and MC M1000i at some frequencies), then the cable is adding negligible THD+N to the system at that particular frequency because the cable and reference measurements have the same value.
Interconnect THD+N Measurements

-90 -91 -92 -93