Sunday 14 June 2015

MEASUREMENTS: Speaker Cables / Wires (Canare, Kimber, and the "Frankenstein" Zip Cord)


Awhile back - in October 2014 to be exact, I spoke a bit about speaker cables, posted some electrical data, and made myself a pair of 4' Canare 4S11 bi-wireable lengths with locking banana plugs to use at home. These are the black cables you see above to the right. I've been happy with them since and as I noted in that previous post, I heard no discernible difference compared to standard 12G copper zip cord nor did I feel bi-wiring made a difference so I reinserted the speaker jumpers to effectively make this a single 11G cable (no risk losing the jumpers!).

However, I presented no measurements at that time. As I mentioned last week, since I've been doing some room measurements lately (which I will talk more about in a future post), I figured it was time to put my impressions to the test and see if I can detect a difference objectively with speaker wires. Today, I present a few measurements in 2 parts - Part I is with a very "poor" zip cord (the above image, left cable), and Part II with a well known and "audiophile approved" commercial cable.



Part I: Canare vs. "Poor" Zip Cord

As in my previous cable tests, I usually perform measurements comparing what I would consider a "good cable" to what I judge would be "bad construction"; that is, something I would never use in real life and which I hope no commercial product ever descends down to! Therefore, I'm going to compare that 4' Canare cable to this poor quality "beauty" - let's just call it Frankenstein's Speaker Cable:

I trust it is self-evident why nobody should be using such a thing :-). I basically took 3 shorter lengths of 12G zip cord copper cable, stripped about 3 cm of insulation off the ends and twisted the bare copper connectors together to make a cable in total ~11' in length. One of the 3 short lengths consisted of a 15-year old zip cord I had from my previous house bought at Home Depot. The other 2 short lengths were bought about 2 years ago from MonoPrice, OFC copper. Although hard to see in the picture above, this middle segment is looking a little tarnished at the ends and under the insulation given the age. For terminations, I just used some simple but good gold-plated Acoustic Research brand screw-on banana plugs (non-locking), also of 15-year-old vintage and no longer being sold as far as I am aware. You can see the terminations below as well as the thick Paradigm Signature S8 jumpers removed when bi-wiring to those speakers (made of brass?).


Using my multimeter, I was able to measure the Canare wire having a resistance of <0.05-ohms and the Frankenstein's Cable had a higher resistance of 0.15 ohms likely on account of the poor twisted metal contacts.



As you see above, this is how the set-up looks using the "poor" cable between my Emotiva XPA-1L mono amp to the Paradigm Signature S8 V3 speaker. Like last week, I placed my calibrated Behringer ECM8000 measurement microphone on axis 65" in front of my right speaker at midrange driver height. Measurements done using Room EQ Wizard, DAC was my TEAC UD-501 played off the Emotiva XSP-1 pre-amp, ADC was the E-MU 0404USB connected to my Intel Pentium G3220-based home theater PC. Test signal amplitude calibrated to 70dB SPL white noise.

I measured 3 conditions:
1. Canare 4' cable biwired (speaker jumpers removed, 2 x 14G lengths)
2. Canare 4' cable non-biwired (speaker jumpers in place, effective 11G cable)
3. Frankenstein's "Poor Cable" - 11', 3 segments bare wire twisted together (12G)

Results:

Frequency Response - 1/12 octave smoothing

Time domain impulse response to 10 ms
No difference. The frequency response graph does not suggest any tonal change. Likewise, no timing irregularities were detected on the impulse response measurements.

I listened to music for about 30 minutes with the "Frankenstein's cable" since I didn't want those exposed bare-metal segments out in the open with kids in the home... What I heard sounded fine and I certainly did not perceive any issues like channel imbalance, soundstage anomaly or tonal difference with the other speaker connected using the 4' Canare.

Part II: Canare vs. Kimber 8TC

I have been criticized in the past for measurements of various cables but not using "audiophile approved" wires... In part, I do this because I would prefer that readers just think about whether claims make sense rather than naming names. Note that I have "named names" before, such as looking for differences with the Synergistic Research power products last year.

Well, I was in my local audio store the other day and figured I'll bite the bullet this time! Now in order to keep my conscience clear as I do respect the brick-and-mortar stores, I vowed that I will honestly test this cable I buy and will only return it if I find absolutely no difference with objective testing and spending a decent amount of time listening (I had 1 week to audition for full refund).

I went home then with this box:

As you can see, this is one of the audiophile "classics", a genuine Kimber Kable "TC Series" 8TC (Stereophile recommended, for a thorough review see this Audioholics report). I think variants of this cable have been around for at least 20 years. I hear there are fakes so make sure it's from an authorized dealer. This one is 8' in length, bi-wireable, terminated with banana plugs. The retail price is CAD$599.99 (Amazon.com has it for US$453 + free shipping, the DIY Canares cost significantly <$50 with a little 'elbow grease'). Not cheap but of course not absolute high-end compared to the >$1000 cables but the construction is top notch and it does look pretty - I trust this is "good enough" to look for benefits of expensive cabling. You can see on the web page the various features: "low loss fluorocarbon dielectric" (Teflon), "16-wire counter opposed helix", "Varistrand(TM) copper conductors", "nitrogen assisted hand soldered termination". On the back of the box, we are presented with the following electrical data: 821pF @ 20kHz capacitance, 0.345 uH @ 20kHz inductance, 0.018-ohm resistance, 0.044-ohm total reactance, frequency response DC-300kHz +/-0.5dB. Compared to the Canare or typical zip cord, this cable has higher capacitance (about twice the amount per foot compared to the Canare), but lower inductance. Remember, these are tiny amounts I'm listing here on the order of pico/micro/milli units! The aggregate size of each positive and negative run would be 9G; or 2 x 12G for the bi-wire end.

Unfortunately, there is one snag in using this cable with my Emotiva XPA-1L amplifiers... The Emotiva's binding posts are too far apart for this cable to reach! Yikes, I guess I'm going to have to return these whether I find a difference or not simply because of the incompatibility. But, to make the most of this unfortunate situation, I fashioned a 3" electrical "Bridge" with a new 12G OFC cable to a banana plug to at least run a few tests:

Certainly not ideal, but given the results in Part I, I trust that a short 3" length of wire will not take away substantially from any potential benefit/difference the rest of the last 8-feet worth of cabling before the speakers may provide.

I measured a number of different ways (some not shown here), but I think the following conditions will be adequate:
1. Canare 4' cable bi-wired (as pictured above)
2. Canare 4' cable using only single bi-wire run (essentially 1 x 14G)
3. Kimber 8TC biwire + 3" Bridge (2 x 12G)
4. Kimber 8TC non-biwired (speaker jumpers in place, equivalent to 9G) + 3" Bridge

Although the test condition is similar, these measurements were performed a few days later than Part I so the microphone was not at exactly the same place (still around 65" from the speakers on axis, midrange height). Also, the signal level was a little different from above (signal amplitude set to 65dB SPL with white noise - I didn't realize I dropped the pre-amp volume control until I did a few tests already). Otherwise measurement technique, DAC, ADC, the rest of the cabling the same as Part I.

Results:
Frequency Response - 1/12 octave smoothing

Time domain impulse response to 10 ms
Again, no difference here other than very minimal inter-test variation.

Some people claim that speaker cables can be "burned-in" just like any other cable/interconnect. Some even pay for this. Well, since I realized I needed to return the cable due to the binding post spread length compatibility issue, I didn't really want to run them for too long nor did I want to waste electricity if nobody is around to listen.

Furthermore, just in case the 3" 12G 'bridge' wire was an issue for some readers, I made a measurement running both speaker cables in parallel with bi-wire connection to the speakers (so one whole 8TC cable for just the + and the other for -). Of course this leaves one of the channels disconnected in each cable. It looked like this:
(Note the black cable at the bottom is my DIY 4' Canare.)

So the next set of measurements are as follows - including post-6 hours "burn-in" playing a random selection of rock and pop and with parallel 8TC runs:

1. Canare 4' cable bi-wired
2. Kimber 8TC bi-wired + 3" Bridge (new out of box)
3. Kimber 8TC bi-wired + 3" Bridge (after 6 hours playing time)
4. Kimber 8TC parallel bi-wired (no 3" Bridge)
5. Kimber 8TC biwire + 3" Bridge - empty box on top

I did measurement 5 as a kind of "sensitivity test"! For fun I wondered if I can show the effect of putting the empty Kimber box on top of my speaker as shown below with the edge of the box sticking out about 1" over the tweeter.

Empty plastic box on top... Sensitivity test!

Results:

Frequency Response 1/12 octave smoothed
Impulse Response to 10ms (note the difference with the "box" tracing starting at 4ms).

As you can see, one should not be putting an empty box on top of speakers :-); it changes the sound slightly. Listening for a few minutes, I can't hear a difference until I push the volume high enough that the box rattled, but the measurement system is sensitive enough to show the change in treble frequency response with that box hanging over the tweeter (black tracing). Clearly the only tracing that looked different out of the 5. The rest of the tracings just show slight inter-test variability. Impulse response suggests a change with the amplitude of the 4ms reflection with the box; appropriate given that the direct sound would not be affected by the box, rather it changes the quality of reflections and probably provided some absorption.

No evidence that 6 hours of cable "burn-in" made a difference to these measurements, nor there being a difference between the Canare vs. Kimber cables (either as 2 parallel runs or with that 3" wire "bridge" which should deteriorate performance). Sure... Some folks claim you need 100-300 hours of burn-in, but I'm just not going to keep these cables that long nor waste electricity. Realize that my Canares are well burned-in with many many hours since I made them last year and there's no difference with these new Kimbers.

I spent about 3+ hours over a couple nights listening to the Kimber Kable in my system (with two 3" bridges constructed for each channel) to evaluate the sound subjectively. Among the new & old albums, selections from: George Ezra's Wanted on Voyage (2014), Taylor Swift's 1989 (2014), Björk's Vulnicura (2015), Miles Davis Kind Of Blue (1959, 2001 SACD rip), Muddy Waters Folk Singer (1964, Classic Records 1999 24/192 HDAD release), Jimi Hendrix The Ultimate Experience (1993, dBPowerAmp HDCD decoding makes a big difference due to Peak Extension on this disk) and Lilo & Stitch soundrack (my kids visited me in the sound room and wanted to hear this, some classic Elvis on here with Hawaiian-ish pop). Basically, I do not believe I heard any significant change. For about an hour, I even ran the Canare to the left channel and Kimber to the right... No difference in channel balance, tonal change, nor thought I heard any issues with the sound stage. The kids didn't think they heard any difference either (and they're very familiar with that soundtrack!).

Conclusions:

1. I trust that no self-respecting music lover would be satisfied with that 11' zip cord's construction in Part I. Despite the bare-wire twist, basic banana plug connection, zip cord wire, copper of different ages, and a slight measurable increase in overall resistance, this made no appreciable difference to the measured output of my speakers in the frequency or time domain compared to a relatively new 4' speaker cable that was bi-wireable, of slightly greater gauge, locking banana plugs and wires configured with Star Quad geometry.

2. With a highly regarded speaker cable like the 8' Kimber Kable 8TC in Part II, I was again unable to show a difference compared to my 4' Canare. Admittedly, many of these tests were suboptimal in that my amplifier binding posts were too far apart for the cables to reach so I had to extend it with that 3" length of 12G cable. I would still expect that if there were a difference, something would show. Nothing of significance showed up within the limits of the measurement resolution (the measurement system was sensitive enough to demonstrate the situation with the box on top of the tweeter!). Obvious differences between the Kimber and Canare include geometry (multistrand braiding vs. Star Quad), locking "closed-screw" banana plugs vs. welded banana plugs, and different aggregate wire gauges (11G, 14G, 9G depending on test condition). No evidence that "burn-in" makes any difference (either the 6 hours with the new Kimber or there being a difference with the Canare which has hundreds of hours playtime over months.)

3. There was no difference for either the 4' Canare or 8' Kimber cables whether they were bi-wired or not (speaker jumpers open or closed or even if just single run of bi-wire ends connected which brings the aggregate gauge down to 14G for the Canare). I do not believe I have ever seen an explanation for bi-wiring that would suggest any appreciable electrical difference; hence, I would question why anyone would expect a difference in sound. (Remember, I'm talking about bi-wiring here... Bi-amping clearly is a different matter!)

I suppose I could throw some more resistance into the cables to demonstrate signal loss; maybe add some capacitance in parallel to create a filter network and show frequency/phase roll-off... But I think these results already speak for themselves.

Of course, there could be other factors I'm missing. For example, is the braiding with the Kimber 8TC or Canare Star Quad geometry more resistant to noise compared to the zip cord? Highly unlikely as speaker cables - speakers are low impedance devices and it takes a lot of radiated noise to induce enough current to create audible distortion in speakers compared to low-level interconnects which need shielding. Could there be difference in the sound with different volume levels due to power handling? Maybe. But if you look at typical forum "reviews" like this one here or this one comparing all kinds of cables (not to pick on these, they just came up on Google), you see strong suggestions that the reviewers think cables somehow change the tonal balance ("warmer", "more neutral", "good deep extension"); an idea that a cable can somehow act as a "tone control". Also commonly, there's talk of time-domain effects like "tightness" of the sound and related phasic effects like soundstage differences. I see neither frequency effects nor time-domain differences just by changing wires. These reviews are interesting and even entertaining for both the reader and writer. But let's keep in mind that we as humans are also very prone to cognitive biases, and are creative, imaginative creatures who enjoy sharing with our fellow humans the things which give us pleasure whether objectively true or not.

Compared to the speaker grill results from a couple of weeks back, I think we can say with quite good confidence that speaker cables have less of an effect (if any) than whether you decide to listen to your speakers with the grill on or off. Or at the very least, whatever differences there may be, the effect is below the threshold of my speakers, amplifiers, DAC, and ability to measure. I trust I have done my due diligence in experimenting with and evaluating whether speaker cables make a difference. I therefore returned the Kimber 8TC to the dealer due to the incompatibility with my amp and the lack of any appreciable sonic difference objectively or subjectively. As usual, I am open to evidence showing otherwise beyond just purely uncontrolled subjective testimony. I'd love to know if anyone has seen measurements like this elsewhere looking at actual speaker output using different cables to show if any difference is demonstrable, with what test methodology, and with what gear (?tube amps)... Links to tests would be great!

Finally, manufacturers have suggested that well formed crystal structure, special dielectrics, and complex wire geometries make the system sound "better", yet I do not believe there has been any objective evidence (either measured or from controlled blind testing) to prove these factors matter in a home high-fidelity audio system. As consumers then we should expect manufacturers to produce evidence of claims.

-------------

Hey everyone, last couple weeks of the "Digital Filters Test"! Anyone with high-resolution audio playback capability and good hearing - I WANT YOU TO TELL ME WHAT YOU HEAR! :-)

Remember, survey / study closes on June 25th!

15 comments:

  1. Kimber Kable 8TC Speaker Cable Measurements and Analysis

    -> http://www.audioholics.com/gadget-reviews/kimber-kable-8tc-speaker-cable/measurements

    ReplyDelete
    Replies
    1. Those articles are complete nonsense. Did you even look at the graphs? The frequency bands affected by the cables is in the 100,000hz range. My dog couldn't even hear that.

      Delete
  2. Can Good Amplifier Design Mitigate Sonic Differences Between Cables?

    -> http://www.audioholics.com/audio-amplifier/amplifier-and-cable

    ReplyDelete
  3. This comment has been removed by the author.

    ReplyDelete
  4. Nice work, Archimago! Goes to show the the power of marketing behind these "audiophile" cables.

    In other news, Philae woke up. Now we may find out how comets can affect the sound quality of one's audiophile system.

    ReplyDelete
    Replies
    1. Hi Vijay,
      No joke! I wonder if we were to add together all the advertising budgets for companies selling boutique cables, just how much money has been expended on promoting the merits of these "components" to the audiophile over the decades.

      Putting aside all the subjective testimony, I wonder in all these years, just what objective data do we have at all when companies like Kimber, Synergistic, Nordost, Monster, etc... claim to be doing "research"!? As far as I can tell, if any of this is science, where do we read about the advances and where are the papers to show validation of claims or replication of experiments?

      Some really serious credibility issues for this "industry".

      Yeah. Philae's awake! Hope it gets enough sunlight to fully recharge. Maybe it'll discover some new metal for speaker cables that'll add another few hundreds to the price tag per metre...

      Delete
  5. I've thought all along it should be easy to demonstrate the audible properties of these cables via actual sound measurement in the room, but this is the first time I've sen it done.

    All the inductance / capacitance / resistance, whatever, measurements in the world mean squat if the sound that comes out of the speakers is the same ;)

    ReplyDelete
  6. Cool article! No magic! :)

    May be if change path placing of cable, we get different results under 2 kHz for the cable.

    ReplyDelete
  7. This comment has been removed by the author.

    ReplyDelete
  8. And by the force of this test, one more magic unicorn of audio was taken down! Good post!

    I'm really happy with my sub US$ 50 speaker cable!

    BTW, you wrote about Synergistic Research and tested some of their products, and their products really are a bunch of garbage! Some dull products and technologies hidden behind a well done marketing. I'm really tired of manufacturers promising "darker black levels, better color saturation and a more 3 dimensional picture" and "more focused sound and a better 3D soundstage, as if a veil was removed from the speakers". Oh, God, how more darker the dark can be, and how much veils have my Hi-Fi system?

    I have the impression that if i buy all of these "amazing" cables and tweaks, the sound and image will be more real that the real thing! Gimme a break!!

    As always, thanks for the measurements!

    Best regards!!

    ReplyDelete
  9. Hi,
    Thanks for the great details about wires & stainless steel, i have bookmarked your blog for the future references.

    We are leading manufacturers of PTFE Wire, Teflon Wire, Stainless Steel Wires, MIG Wire Welding, TIG Wire.

    ReplyDelete
  10. On the subject of Bi-wiring, I heard a massive difference with my Missions which was entirely due to poor quality link plates. I don't think the cables (Audioquest SlipFlex) made much difference but removing the metal links, transformed the sound. Obviously replace them with short wires at least.

    ReplyDelete
  11. good blog.it is very useful information.We are mi cable manufacturers

    ReplyDelete
  12. This comment has been removed by the author.

    ReplyDelete