Megatest speaker cables – real measurements, samples and blind test!


We can say without a doubt that this test of speaker cables is the most intensive test Alpha Audio has done to date. We have invested a tremendous amount of time and energy in this comparative test. We have also acquired two new, beautiful measuring devices to carry out this test. And all this to determine once and for all if there is a relationship to be made between measuring and listening. Whether we succeeded you can read in this mass test of no less than twelve speaker cables.

We’re starting to build up a serious arsenal of measurement equipment at Alpha Audio: a Prism dScope III, a versatile Rigol digital scope and 60 MHz function generator, and a Sourcetronic LCR meter. Then there are many special adapter plugs, measuring cables, dummy loads, variable resistors and other accessories.

The reason we invest in this is twofold: we want to know if a manufacturer communicates honest specifications and it is very interesting and educational to (try to) make a link between objective measurements and subjective listening tests. Because often we hear differences, but are they measurable? And explainable? And the same applies the other way around: for example, we measure interface jitter, but is it really audible? Establishing those relationships has become almost your author’s mission.

Speaker cables… the test procedure

There is much debate about cables. Do they actually do something positive to your sound? Is it all nonsense? Should you invest in a decent (speaker) cable? Or can you just use coat hangers.

Without a doubt, we hear a difference. But can we prove it in a blind test? To find out, we’re going in completely blind. At least: fellow authors Martijn ten Napel and Yung Lie are going in blind. Your author has measured and sampled all cables and is therefore no longer objective. His only task on the listening day is to change the cables and drink coffee…. ah, well: someone has to do that.

Yung and Martijn both receive a piece of paper on which they write down their findings. We use defined characteristics (detail, timbre, imaging and overall feel of the cable). Just like the switch test. Your author has created a top three cables from the measurements. Yung and Martijn are going to do the same, so that afterwards we can also see if the best measuring cable is also the best sounding cable.

Measurements were taken on a Sourcetronic ST2829A LCR meter (20 Hz – 300 kHz, medium speed sweep at 2.8 Volts). A Rigol digital scope and function generator was also used to send signals through the cables (20 kHz square wave at 2.8 Volts rms). And the Prism dScope III was used as a spectral analyzer and response measurements of the speaker cables.

The Prism dScope has again proven its flexibility. What a fine device that is. With the dScope spectral analyzer you can easily map the effectiveness of the shielding. And the sweep functionality of the Prism flawlessly shows what a cable can do in the chain.

You can find most of the set-ups in the photos. But to explain one thing or another…. The Sourcetronic is largely automated. For all tests except the capacitance measurement, the cable is shorted to our ‘cable panel’. Between each change of one we calibrate the meter again to make sure all measurements are correct. This setup is pretty bulletproof we have noticed. The only important thing is that we place each cable the same way and calibrate in between. Know that the graphs/measurements are logarithmic. That is why there are more bumps an dips to be seen at the beginning of the measurements; the resolution is much greater there.

For the noise test we took a 19V / 3A switch mode adapter and connected it to the 4 Ohm dummy load. This way a constant current flows. The speaker cables are shorted at one end (source) and connected to a 4 Ohm dummy at the other end (speaker). From there, measurement cables also go into the Prism.

Then we placed the adapter close to the cable (equal distances). This way we can test shielding. After all, if the spectral analysis – set to 256K FFT for optimal accuracy – does not change, no noise is picked up. Without fail, this test gave surprising results…. It is important to realize that cables also pick up noise without the adapter next to it. Also, they are not all measured on the same day. So look especially at the difference in noise.

The cable response measurement was done using a Bryston ProPower power amplifier (very stable) and a Teufel Definion 3S monitor speaker (average difficult speaker). We also used a dummy load to see what the differences were in response. The Prism feeds a sweep to the Bryston which sends the signal through the speaker cables. So the cable goes into either the Teufel Definion or the 4 Ohm dummy load. At the same time – either from the speaker or the dummy load – a measurement cable goes into the Prism. That then produces a nice little graph. We have put the data in one overview at the end of the article. Perhaps unnecessarily: measurement cables are kept as short as possible in all cases.

Type test
Tested price class
Price cheapest product: €
Price most expensive product: €
  • Brand and model: Audioquest Type 9 DBS
  • Build quality: Solid. Bit stiff
  • Overall impression: Rich sound. Tight, round. Brings peace
  • Price: €999
  • Brand and model: Audioquest Type 9
  • Build quality: Solid. Bit stiff
  • Overall impression: Rhythmic, a bit restless. Focus on the high
  • Price: €700
  • Brand and model: Chord Company Epic XL
  • Build quality: Smooth. Beautifully finished.
  • Overall impression: Very energetic. Focus on detail and rhythm.
  • Price: €1500
  • Brand and model: Driade Flow
  • Build quality: Smooth. Fine to use. Slightly fragile plugs.
  • Overall impression: Layed back. Fascinating display. A little rounded high.
  • Price: €800
  • Brand and model: Supra Sword 3
  • Build quality: Stiff. Not very special. Typical blue colour.
  • Overall impression: Round, warm sound. Sufficient detail. Pleasant sound.
  • Price: €975
  • Brand and model: John Van Gent M.L Reference
  • Build quality: A bit stiff. Beautifully made. Beautiful materials
  • Overall impression: Rhythmic, rich in detail. Fine balance in reproduction.
  • Price: €940
  • Brand and model: Ricable Dedalus Elite
  • Build quality: Remarkably supple. Very nicely finished
  • Overall impression: Lightweight. Large stereo image.
  • Price: €840
  • Brand and model: Van den Hul Inspiration Hybrid
  • Build quality: Supple. Typical VdH colour. Beautiful plugs.
  • Overall impression: Round and complete sound. Large stereo image.
  • Price: €1250
  • Brand and model: Transparent Plus
  • Build quality: Smooth. Box in the cable.
  • Overall impression: Neutral. But a little polished and less involved.
  • Price: €1239
  • Brand and model: MIT EVO Three
  • Build quality: Smooth. Two boxes in the cable.
  • Overall impression: Large stereo image. Analytical sound. Slightly mid-prominent.
  • Price: €1295
  • Brand and model: Shunyata Venom X
  • Build quality: Supple. Beautiful materials.
  • Overall impression: Very balanced. Everything 'fits' in this cable.
  • Price: €1800


  1. Some very nice work here, thanks so much for sharing. Especially the listening comments. Very nicely done!

    This is meant constructively. I’d like to point out the square wave testing here is misleading at best. The wave being used appears to have a very fast rise time (Mhz range edge speed). Thus this is high speed digital (at this cable length) and puts everything into the high speed digital cabling realm. Ie this is now a transmission line system operating well above the audio band. It’s no longer a lumped system where all we need is Ohm’s law.

    T-line is a whole ‘nother advanced electrical ball game that not everyone is aware of. The scary looking overshoot and ringing are due to transmission line effects with mismatched source, load and cable impedance (at the edge speed), not the cable per se as in ~ audio band Ohms law. At this edge speed it’s all relative to the length, the cable impedance (at the high freq), the source and the load impedance.

    In short if the cable, load and source impedances do not match properly there will be overshoot and ringing in any transmission line..

    And load side, terminating the line with 4R does not match most transmission line cable impedance properly at that speed anyway. For example regular old coax, generally around 60-70R would match it properly. The boutique cables in this case will be all over the map as to what their MHz range impedance is.

    At high speed definitely even the resistor used matters. A regular large load test R for example (or a cheap sand cast xover R) will have a MHz range impedance that is complex and not straightforward to match properly for example (not just 4R). Terminate instead for example with a small SMD R with short leads to the cable and the picture will likely change dramatically.

    This also means your source termination has to match as well. Most high speed generators have a a 50 Ohm output impedance setting. This will also likely not match well with most of your setup.

    The easy solution here for your testing would be- don’t be square wave testing this setup in the high speed digital realm. These interconnect are not for a high speed digital system. Instead, launch a square wave into the test setup that has a more relevant audio bandwidth edge speed (rise time setting). Your test will then also not show scary looking overshoot and ringing result, but it will at least show the true picture for audio.

    1. Hi!

      Thanks for your extensive comment. I have to look into the SQ-wave tests I used a 20 kHz square wave, generated by the Rigol function generator. I’ll look up the specs.
      I have a question though… why do the Ricable and Shunyata transmit it perfectly? Also the transparent seems to be doing well. All cables are 3 meters… Input was terminated at 100 kOhm.

      Best regards,


  2. Again thanks for sharing all that information.
    Yes, please check the rise time of the waveform out. From the scope shot it looks like it could be quite fast. A cable system will begin to exhibit transmission line effects when its length is about 1/6 the “length” of the rising edge Thus my commentary. If I’m wrong on that edge speed, please forgive me.
    Coax cable for example has a propagation velocity of about 66% the speed of light and a corresponding propagation delay of about 1.5 ns seconds per foot. So for example if you had a 10 ns rise time the electrical length of that in a coax cable would be 6.7′. 1/6 of that is 1 1′.
    If your signal generator is capable of say a 50 MHz square wave, that signal rise time is also going to have to be much much smaller than the fundamental frequency of 50 MHz (Tp=20ns). Ie, it would be much much faster than 20ns..
    And using 50 ohms output impedance from the generator would be a better match for most sources that are driving the cables. Eg, a preamp out. And also wondering why the load side is terminated in 4 ohms rather than say, 10K (these are not speaker cables).
    From here I’m not sure about the shunyata waveform for example, but it could be related to the impedance it’s presenting in the MHz range. We would have to know some more about these things.
    I do believe that you’ll find that, once you slow down the edge rate (if you can) to a more appropriate audio bandwidth rate you’ll probably see about perfect square waves on all of them.
    You can check the link below for some more basic info. Take a look at the lumped vs. Distributed section for example:

  3. I also think the use of psuedo-square waves is problematic! Fourier transform theory says that a square wave is made up of all the sinusoidal odd harmonics of the base frequency, up to infinite frequencies! But there is an odd mathematical effect – the combination of these sine waves produces an overshoot of about 9%. If the rise time of the wave is not instantaneous (and in practice it never can be) high frequency ringing artifacts are produced MATHEMATICALLY. A good music source for near square waves is a CD, but the rise time would be 1/44,000 th of a second. SuperAudio CDs or DSD files, which can only nudge the sound up one notch a few million times a second, cannot produce anything like square. Similarly, a vinyl record with a groove with a perpendicular right angle is un-trackable.

    The speaker used is described as an average difficult speaker. This means that its impedance varies substantially with frequency – probably dropping in the cross-over regions which requires corresponding more current from the amplifier. In the good old days, it was standard practice for speaker tests to include graphs showing impedance again frequency. Add a purely resistive load (or cable) and the overall impedance curve flattens a bit, changing the relative loudness especially in those troublesome cross-over frequencies where more current is needed. This means that any cable changes the sound for any practical loudspeaker, and different cables (or cable lengths!) will change the sound differently.

    The most thorough blind audio test I have read about pitted three types of amplifier (valve and solid state) made by Quad against each other. Each test played the same four studio-quality music tracks, and six experienced reviewers were asked whether A sounded better than B, or worse, or the same. Sometimes A and B were the same amplifier! Statistically, nobody could reliably tell the difference …

    1. Can you make explicit what your message is? For me, there is an implicit message you are trying to make, but I fail to pick it out of your text.

      The word ‘statistically’ is (ab)used incorrectly in a lot of cases. Just for illustration purposes: With a variable x=3 and the number of subjects n=6, you cannot speak of any form of statistical inference. The number of uncontrolled variables at play is even greater, so any form of conclusion is anecdotal.

      As is our test, by the way. But we do not pretend or claim to have setup test conditions that would yield any statistical validated conclusion.

      The feedback we got though is that no one has ever tried to do a test like this, so at least there is more information than there has ever been. And you can listen back.

      But we encourage you to draw your own conclusion from this test, because we know there isn’t a definitive conclusion and much of it is down to personal preference in the end. We will never convince the skeptics, but that isn’t our goal. We just hope this will help to guide people in the purchasing decisions they want to make and to give some guidance in a world where hyperbole is the norm.

      1. Thanks Martijn

        I was trying to make three separate points.

        1 – True square waves do not exist in nature or audio recording systems. IN MATHEMATICAL THEORY, when transformed into all their sinusoidal frequencies and summed to recreate the original square waveform, an overshoot of about 9% appears. Inconvenient but also present with real-life near-square waves together with ringing artefacts.

        2 – Real-world speakers do not have the same impedance at all audio frequencies. Speakers may have 20 times as much impedance at some frequencies as at others.
        Therefore adding any impedance in series, even a simple resistance, will change the current flow in different ways at different frequencies, which we hear as a tonal change. So any speaker cable will change the sonic character of any speaker. The change will depend on the impedance curve of the specific speaker and the preference of the listener. But a cable that may ‘improve’ one speaker is just as likely to ‘impair’ another.

        3 – I gave an example of just how hard it is to reliably report on differences in listening tests. In this context, statistics is an exact science and answers the question “what is the probability that the observed results could have arisen by chance?”. The mathematics is based on the result being perturbed by a large number of uncontrolled variables! The perturbations are ‘averaged out’ by repeating the test – the more repetitions, the better.

        The six listeners were presented with 96 A-B tests where A could have been any of three amplifiers, and B could have been any of the amplifiers including the one played in A! This is 9 possible combinations so each test was replicated about 10 times. Each listener only had to report better, worse or no preference.

        I am sure you would agree that if any listener reliably got the same result for the same 10 tests, it would be highly significant. But in reality, no individual’s results differed significantly from pure chance. As a group, 68% of results showed no preference when the amplifiers were the same, but remarkably 64% also had no preference when the amplifiers were different. The scientific conclusion: the expert panel could not hear a difference between the valve and solid-state amplifiers under test.

  4. Hi Martijn

    I don’t have the original AES paper, but here is a handy link to a Wireless World article:

    There have been several other articles about this test. You can discover the names of the ‘golden ears’ who spent a full day listening.

    Quad also used to carry out listening demonstrations of live accoustic music and recordings of the same. A curtain hid the source.

  5. A good old tweak is to mount a resistor equal to the characteristic impedance of the speaker cable across the speaker terminals to eliminate reflections if the speaker is not impedance linear.
    Be aware that the sound may take on the flavor of the sound signature from the resistor used.

Leave a Reply