April 2012 Speaker Cable Listening Test
The following question has vexed audiophiles for decades: Do speaker cables make a difference? Are expensive cables better than cheap cables? The Audio Society of Minnesota tackled these questions with an experiment involving over 50 members and guests as our test group. As much as was possible under the circumstances, the experiment was set up as a blind test of four sets of cables ( A, B, C, & D) ranging from cheap to expensive and auditioned in pairs in the order shown below.
Test 1: A & B Test 5: B & C
Test 2: C & D Test 6: D & A
Test 3: B & D Test 7: A & C
Test 4: C & A
Note: Test 7 is a repeat of test 4 except that the order of cables was reversed.
Cable “A” was a 7ft pair of common 16ga zipcord terminated with hardware store copper spades. Approximate cost: less than $3.
Cable “B” was 6ft pair hollow tube design wrapped with an exotic copper winding and terminated with copper spades. Retail price in the $8000 range.
Cable “C” was a 15ft heavy gauge twisted pair in rubber plenum terminated with gold plated copper spades. Price unknown but expensive.
Cable “D” was a 6ft prototype silver cable in a Teflon jacket, 18ga twisted. Bare wire ends coated with solder. No price quoted.
Blind tests were performed. The audience did not know ahead of time how many cables were involved in testing. Cables were kept hidden behind the speakers and amplifiers. Preamp and amplifier levels were not adjusted during the tests or between tests (more on that later). Seven separate listening tests pitted one unknown cable pair against another. The same 1 minute 50 second listening track consisting of various music selections was used in every test. To assure proper connection between the amp and speakers, the listening began with a phase test, an “out of phase” male voice followed by an “in phase” voice. These short few seconds were taken from a Stereophile Test CD. The 5 second phase test was immediately followed by the opening 38 seconds of the “Firebird Suite” from the Reference Recordings Sampler with the Minnesota Orchestra with a short fade out. This was followed by 34 seconds of the beginning of Mark Knopfler’s “Shangri-La” from the same titled album, again with a short fade out. This was followed immediately by 33 seconds of Tracy Kent’s rendition of “You’re the Top”. This mix of music features demanding peaks from the Firebird, a well recorded male vocal and background, and a well recorded female vocal with Jazz trio and piano solo. Track mix was performed using Pro Tools and burned to disc in various formats. After private disc listening comparisons, the decision was made that the DVD-A 24/96 format was most faithful to the original track selections and was used for the tests.
For each of the seven comparison tests, the music track was played using the first pair, a pause for cable switching (20 to 35 seconds), followed by the same track using the second pair, followed by cable switching, followed by music track using the first pair a second time, cable switching, then the second pair a second time. At that point, test #1 was concluded and the audience was asked to mark sheets with votes indicating the listener’s preferred cable They were asked to rate the cables on bass, midrange, treble, and overall preference. An option was also given to indicate a “no preference” vote in each of the categories. The audience was informed that cable “winners” would only be based on the “overall” preference votes. The remaining six tests were conducted in the same fashion. There was a 10min stretch break following Test 4.
Of the approximately 55 people in attendance, 40 submitted voting sheets. In addition to voting on listening tests, the audience was asked to answer a short questionnaire on the same sheet.
After test Seven was complete, voting sheets were collected and the audience was then shown the four different sets of cables involved in all of the tests. While votes were being tallied capacitance measurements and frequency response measurements (not presented here) were conducted on each of the cables under test.
The room used for this event is in the ASM’s usual meeting venue, the Pavek Museum of Broadcasting. It is approximately 20ft wide, 38ft deep, and 10ft high, rectangular in shape with an open L-shaped area at the rear on the left channel side. It features a tight woven commercial carpet floor, an acoustic panel ceiling, and hard plaster walls. Windows along the right channel wall are covered in aluminum blinds. Speakers were positioned 8ft apart at their center and 5ft from the back wall with ribbon tweeters on the outside, toed in to converge about 12ft into the room. An equipment rack was centered between speakers. Two “cable changers” were placed behind the speakers and equipment. The first row of listeners was approximately 7ft from speakers.
The Audio Society of Minnesota/Pavek Museum house system consists of a modified/upgraded Oppo BDP-83 multi format player with analog stereo output into an ARC SP-17 preamplifier feeding two ARC M-300 mono power amplifiers. (“Murphy’s Law” full disclosure: One of the two M-300 MkII amps blew a fuse early during test #3 and rather than attempting to determine the cause of the failure it was decided to switch to a back-up amplifier, an ARC D-115MkII. This created a 15 minute downtime period followed by several quick attempts to adjust levels and balance to match that of prior tests. The second half of test #3 through all remaining tests were done using this second amplifier. There were several comments to the effect that the sonics of the second amplifier were slightly preferred. Hopefully this did not significantly influence the results of test #3 and remaining tests.). Interconnects were Anti-Cables while the speakers were Magnepan 3.6. Connections on both speakers and amps are stock screw down block terminals. Because of this, we preferred to use cables of spade terminations for easiest installation and changing without the use of any adaptors on terminations. Cables were kept hidden from view at all times.
So what did we find? Do cables make a difference and are expensive cables better than cheap ones? Looking at the proportion of votes for the cable pairs does indicate that some cables were preferred over others.
Cable Cable No Preference
Test #1: A & B: B: 56% A: 39%. 5%
Test #2: C & D: C: 54% D: 32%. 15%
Test #3: B & D: D: 59% B: 15%. 27%
Test #4: C & A: C: 41% A: 29%. 29%
Test #5: B & C: C: 49% B: 32%. 19%
Test #6: D & A: D: 41% A: 27%. 32%
(At this point, all 4 cables have been tested against one another – each being involved in three tests.)
Test #7: A & C: C: 49% A: 41%. 11%
As noted earlier, test #7 is a repeat of test #4, except order of cables is reversed. It is interesting to note that even though cable order was reversed, results came out nearly the same.
Here’s another way of looking at the results:
Cable C: 3 wins 0 losses
Cable D: 2 wins 1 loss
Cable B: 1 win 2 losses
Cable A: 0 wins 3 losses
Note: The win loss records represent a clear ranking in that no cable defeated a cable ranked higher than itself.
Of all Votes that noted a cable preference (all first 6 tests combined) results are:
Bass: Cable A: 26% B: 23% C: 24% D: 27%
Mids: Cable A: 18% B: 24% C: 32% D: 27%
Treble: Cable A: 22% B: 22% C: 29% D: 27%
Overall: Cable A: 20% B: 22% C: 30% D: 28%
This shows that listener votes of “preferred cable” align more with Mids and Highs preference than with Bass response preference.
55% of voting audience uses bulk wire (not branded) on their home systems.
Of those that use branded cables, 72% cost $500 or less per set in retail dollars.
Therefore, 13% of audience use “expensive cables”, 33% use branded cables $500 or less.
Before this test 69% of our test subjects thought that speaker cables could make a significant difference. After this test, 86% believe that speaker cables can make significant differences.
Since no statistical tests of significance have been applied to the results it is not possible to demonstrate that one set of cables was found to be superior to any others in an objective sense. Subjectively, however, it was shown that cable C received more preference votes than the others with cable D, the prototype silver/Teflon, running in second place. Cable B fell more or less in the middle of the rankings. Cable A, the zipcord, was not preferred on any of the three comparison tests in which it was included, suggesting that, yes, these cheap and commonly used cables do not sound as good as cables designed for the audiophile market. Based on the results of this informal but honest attempt to address the questions posed earlier, it does appear that in this case at least, there was a preference for the expensive cables and a definite non-preference for the cheap zipcord. So cables do make a difference.
After the listening session, Frank Van Alstine measured the cable capacitance per foot (picofarads):
A: (zipcord) 29.84pF C: (heavy gauge) 97.3
B: (garden hose) 93.6pf D: (Teflon silver) 38.3
Ron Ennenga then performed a measurement on the cables. These measurements represent the transfer function of the cables only. The reference probe was attached to the amplifier output and the measurement probe was attached at the speaker input. The cables were measured with an MLS signal using Speaker Workshop with a Roland UA-1EX sound interface at a 48kHz sample rate. A signal length was chosen to give ~1.5Hz resolution. Some of the measurements had artifacts at 60 or 120Hz, these were removed for this graph, and the measurements were 1/6 octave smoothed.
From Top to bottom in this graph are cables B, C, A, D. Peak-to-peak variations were very small and highest (as expected) in the cable with the smallest wire.
The Impedance of the Magnepan 3.6R was measured and found to be largely resistive with a value of ~4.6 ohms and small crossover related ripples in the midrange. Of note is the extremely flat impedance phase. Note that the response of the cables reflects the impedance of the speaker, so speakers with higher min-max impedance magnitude swings need larger gauge wire to minimize ripples, all other things being equal.
In hindsight it would have been more interesting, and perhaps easier to correlate with the listening tests if we had been able to measure the transfer function of the whole system with the cables in circuit - essentially comparing the output of the source with the voltage at the speaker terminals to capture any amplifier/load interactions.
Data Mining/Synopsis/Interesting observations
See PDF file at the bottom of the page, there are some very interesting (unexpected even) observations about our listener preference data.