Thursday 2 May 2013

MEASUREMENTS: Analogue RCA Interconnects.

Now that I've measured the main digital cables used for audiophile listening these days (USB, coaxial, TosLink), I figure it's time to demonstrate what happens when we bring the testing methodology with analogue cables. [I'm hoping in time HDMI will make some headway into the audiophile DAC world since I look forward to multichannel playback one day through my computer server.]

Remember why we got digital in the first place: robust data storage free from transmission and generational losses - in other words, resistance from corruption.  By transforming data into 1's and 0's, we quantize the data into binary form and complexity is thus encoded in larger quantities and combinations of this quantized binary data which can be saved in a form which makes detection and correction of error possible.

As a result, when digital 'works', it likewise tends to be 'all or nothing'. What we saw with the digital cables is an example of this. Despite some really poor quality cables tested (the Dr. Frankenstein models!), the measurements were essentially identical in each case (they all worked). Parameters like impedance and capacitance of the cable do not affect the transfer of data unless of course they are outside of tolerance for the interface. It's of course possible that the occasional 'bit' of error occurred with poor cables, but obviously it was not large enough to encroach on the measurement results (nor affect the subjective audibility when I was listening). Beyond "bits are bits", in the timing domain, there have been reports of very long cable lengths potentially worsening timing of the digital signal (ie. jitter), what I have seen suggest lengths of 20-30m AES/EBU adding maybe 20ps - totally irrelevant to audio performance and at lengths we generally do not use in the consumer setting.

Let's see now how good ol' analogue interconnects of various lengths fare.

Here are the models being tested:

Cable A:
3' freebie RCA cable that came with an old cheap DVD that has since broken. Connectors not gold plated.

Cable B:
3' Radio Shack shielded RCA cable. Gold plated connectors.



Cable C:
6' Radio Shack shielded RCA cable. Gold plated connectors.



Cable D:
10' poorly shielded composite cable (with stereo audio). Gold plated connectors.



Cable E:
3' pure 4N silver, 4-core braided interconnect. Connectors are gold plated Neutriks. Soldered with Cardas Quad eutectic silver solder. Got this cable about 2 years back and used for my SACD player from here.


Cable F:
With apologies to Keaton & Minoeall-san, I used the 'Super Connectors' from the "Best-Coaxial-Digital" cable along with the two lengths of 10' composite cable and 6' stereo RCA and created an un-audiophile-approved 16 foot RCA "double cable" interconnect.



Setup:
Win8 laptop --> shielded USB --> CM6631A async USB to SPDIF --> 3' coaxial --> AUNE X1 DAC --> *test analogue interconnect* --> E-MU 0404USB --> shielded USB --> Win8 laptop

Summary RightMark 6.2.5 results (24/96):

Frequency Response:

Noise Level:

THD:

Stereo Crosstalk:

Summary:
1. Analogue ain't digital! Although in most ways the measurements are very similar (these are short lengths of interconnects after all), mild differences can be found.

2. Frequency response unchanged among the cables. Interesting. Some people talk about analogue cables as "tone control". I don't see it using these interconnects even with longer length (there is a hint of high frequency roll-off with the 16' cable but really this is trivial) or different conductor material. Using silver interconnects, there are no changes in the frequency response to suggest these cables sound "brighter" as some contend :-).

3. Interesting Stereo Crosstalk performance. Stereo crosstalk looks to be sensitive to cable length. The silver cable had the least crosstalk up to 5kHz and then increased from there - this is possibly a function of the fact that it's constructed as 2 separate cables as pictured above rather than the zip-cord arrangement of the other cables.

4. Measures like THD should not (and in fact does not) show a difference. After all, cables are passive "components" so should not introduce harmonics into the equation. As for noise floor, I suspect if I were to test under conditions with strong RF noise the poorly shielded cables would perform worse (may try this later), but in the home environment where I tested, obviously this was not a problem even in reasonably close proximity to the laptop, DAC, and E-MU ADC.

There you go. Analogue interconnects do make a slight difference and this is quite measurable particularly in terms of stereo crosstalk performance. Remember that these interconnects are of relatively short lengths so minor differences are really not surprising. The obvious question is - would humans be able to differentiate these interconnect cables based on listening tests? I honestly doubt it. Subjective listening using my test setup did not reveal any noticeable change with the long cable vs. the short silver cable. Realize that even with the long 16' cable, stereo crosstalk was still below -75dB which should be inaudible - for comparison, high-end LP cartridges are only capable of 30-40dB crosstalk performance.

Musical selection tonight was Rachel Podger & Brecon Baroque's renditions of Bach's Violin Concertos on Channel Classics (SACD converted to 24/88). Sounds great with the 16' Frankenstein cable with my Sennheiser HD800 headphones off the E-MU 0404USB.

7 comments:

  1. For those not familiar with interpreting RMAA plots.. the Frequency range of cables is not limited to 30kHz (-0.5dB) nor rolls off at the bottom end of the spectrum, but the used soundcard is the limiting factor here.
    Bandwidths from DC to 1MHz are easy to reach for all cables. In the 10MHz range differences may start to show.
    When you use audio interlinks as video cables you can literally 'see' where they start to differ.
    Basically the measurements (except crosstalk, which shows the resistance of the screen differs) all basically show the limits of the used soundcard and not of the cable.


    ReplyDelete
  2. True. I'm of course looking just at the transmission within the audio spectrum here.

    ReplyDelete
  3. Glad to see your blogs. You do a good job of showing where things differ and where they don't.

    A note about noise pickup, which is treated like voodoo magic by many audiophiles. I have done a bit of testing and expected to find something in unshielded interconnects. But there is pretty much nothing there. I had a setup with test signals run through DA to AD, and the results stored. The noise floor of this was around -135 db. Finally I wrapped analog cable 3 times around a computer switching power supply. Literally wrapped right around the PSU itself which I had moved outside the PC box. I finally got a few noise spikes with a couple reaching -85 db. Unwrap it and move it 6 inches away and one spike was just barely above the -135 db noise floor though I might not have noticed had I not located it earlier. Move it any further and there was nothing.

    Another interesting result was even wrapping balanced cable around the switching PSU 5 times the balanced cable never picked up anything. I ran this test with the attached PC running normally, and while having it run video test software and CPU stressing software to put the biggest load on the PSU possible. Still nothing.

    Lots of people who are using computers as music sources these days seem to worry about noise in analog and digital cables. I have used shortwave, and AM radio to listen to the radiated noise around PC's. Different operating systems have clearly different radiated signatures as do different hardware configurations of PC's. But doing testing, measuring, null testing etc. I can find no impact of that on the resulting reproduced analog sound. All this worry about PC noise seems a waste. And all the extra cost, PS upgrades to some of this equipment hardly seems to matter if the resulting analog sound is already uncorrupted with the stock "dirty" power supplies.

    Oh well, in any case, good work, keep it up.

    ReplyDelete
  4. Well - you pretty much used a bunch of low end - bottom of the line cable - so your results don't surprise me. Would have like to have seen your test results with some high end cable. You know - compare Monster's best with AR vs Kimber etc etc. Also - besides digital measurements - how about a real world listening test ? What Hi-Fi Magazine seems to indicate that there are indeed differences in the way these cables sound. THX

    ReplyDelete
    Replies
    1. Low end or high end cable as long as there is nothing added like little boxes of electronics for example will all sound the same. Electrons do not change how they behave just because the cable costs more money. Its all snake oil.

      Delete
    2. I've made all my own cables so far from Maplin here in the UK. All cost less than £20, £1/m shielded cable, quality jacks and I cannot notice any difference between anything I have created and any cheap RCA cables I've used. The biggest differences have been the amplifiers and preamps I've been trailing. They make a difference. Valves and solid state make a difference. I have a Leak Stereo 20 and that was such a level up in stereo imaging and frequency response over my Quad 303. A passive and active preamp also makes a notable difference, but a cable, over a short half metre run, reviewed in non test conditions, using emotionally responsive subjects.... I'm sorry but ridiculous euphemism driven conclusions get drawn into everything.... I don't think so.

      Even though I am relatively knew to all this I have extensive musical creation understanding and audiophiles pay such little regard to mixing / microphones / placement / mastering / format - I am firmly on the side of the sceptics and your tests show this.

      Most audiophiles don't even know what 10k sounds like or where it is in the sonic frequency range to make accurate assumptions - most have never even used a signal generator to test their own hearing let alone understand room acoustics.

      Delete