[audioengr]

I decided to try some different cables to see if their jitter can be seen in the plots.  Driven by the Synchro-Mesh and using a 44.1 track. Cables range from 1-1.5m in length.

First, here is how I connected the Synchro-Mesh directly to the scope:



Synchro-Mesh output jitter, no cable:



The two humps indicates that the driver pulling high has a slightly different output impedance than the drive pulling low.

Now, Jitter plot with Empirical Audio Standard 1.25m BNC-BNC cable, which is silver with expanded Teflon dielectric.  Aerospace cable with true 75 ohm BNC connectors.  Retails for $275 plus shipping and PayPal fee.  Dynamo LPS supply powering it.



The two plots look very close.  The two humps are more similar in magnitude. 

Now, the spectrum of the Synchro-Mesh reclocker alone:



Very clean, low noise floor, no low-frequency jitter at all, and one small spike at around 1MHz and another smaller one near the big one.

Now, the Empirical Audio Standard BNC-BNC cable spectrum:



Very close to the spectrum with no cable at all, so the Standard BNC-BNC cable adds nothing significant.  Both of the smaller spikes have decreased.  I think we are seeing here the benefit of a minimum cable length.  I always recommend 1.25-1.5m.

Now, jitter plot from cheap BNC cable with 50 ohm connectors on it (like $20):



The two peaks are fatter and much shorter (see difference in scale) and the space between them is less than the silver cable, but not bad.

Now Jitter plot of Belden 1694A cable with RCA connectors on it.  Homemade, soldered.



The Belden 1694A is not so good.  The two peaks are not even visible and shorter than the silver cable, but taller than the cheap BNC cable (see vertical scales).  Does not sound good either.  See listening correlation tests lower in the thread.

Here is the jitter plot of my old manually-built all silver twisted-pair RCA cable, the Bitmeister:



Pretty bad jitter compared to the silver cable I sell now.  At the time, it sounded decent though.

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

This shows without a doubt that one can make a qualitative measurement as well as a quantitative measurement on digital cables that will correlate to sound quality.

Listening tests below have the following conclusions:

Our ratings for overall musicality:

1) Silver reference BNC-BNC cable
2) Bitmeister silver cable
3) Cheap BNC cable
4) 1694A RCA cable.

Steve N.

[Speedskater]

Do you have any higher resolution screen captures?

[audioengr]

Do you have any higher resolution screen captures?

No.  What is not legible?  I can read all of the measurement data.

Notice that the vertical measurement scale is not the same in each plot, so the results are even more different than I thought. The peaks on the silver cable are actually much higher than the other two cables.   I cannot make all vertical scales identical.  They are the numbers of samples at each specific time, so it scales automatically.

Steve N.

[audioengr]

Updated the plots to add one more cable, and show the difference with the Dynamo LPS.

Steve N.

[audioengr]

Reference silver cable, using the upgrade Dynamo power supply to power the Synchro-Mesh.




Cheap BNC cable with 50 ohm connectors on it (like $20):





The two peaks are fatter and much shorter (see difference in scale) and the space between them is less than the silver cable, but not bad.

After non-blind listening tests, here are the conclusions of my wife and myself.

The Cheap BNC cable has the following audible attributes compared to the silver cable:

1) Less bass focus
2) some loss of harmonic richness and texture
3) crescendos are less impactful
4) vocalist image is wider
5) decent depth

Belden 1694A cable with RCA connectors on it.  Homemade, soldered.





The Belden 1964A is not so good.  The two peaks are not even visible and shorter than the silver cable, but taller than the cheap BNC cable (see vertical scales).

The Belden 1694A cable has the following audible attributes compared to the silver cable:

1) Much more forward, almost no depth
2) Instrument images are very wide
3) Positions of some instruments are actually out of place
4) vocalist sonics are okay, but forward
5) bass image okay
6) my wife says "musical soup"

Here is my old manually-built twisted-pair RCA cable, the Bitmeister:



At the time, it sounded decent.

The Empirical Audio Bitmeister cable has the following audible attributes compared to the silver cable:

1) Decent depth of image, but not as deep as the silver cable. Better than the 1694A cable, more musical than the Cheap BNC cable
2) Instruments and vocalist are live sounding, but a bit forward.  Pretty sounding.
3) Crescendos are decent, better than the Cheap BNC cable
4) High-frequencies are problematic.  Sound-booth used by vocalist is not obvious, just blends with instruments.  Echoes from the venue are absent
5) Bass not imaging well
6) Harmonics and richness is there, but the texture a bit confused

I believe Bitmeister was a popular cable because of the richness and lush vocalist sonics.  I unfortunately don't make them anymore.

Conclusions:

The graphs are helpful to determine what will deliver detail of the venue and instruments, but it's not that obvious which will deliver good dynamics or smooth natural vocals, both critical for good SQ.

Even though the jitter profile for the 1694A cable looks better than the Bitmeister, the Bitmeister sounds a lot better.

Our ratings for overall musicality:

1) Silver reference BNC-BNC cable
2) Bitmeister silver cable
3) Cheap BNC cable
4) 1694A RCA cable.

It is obvious to me that the silver in the cable is just as important as the jitter caused by the cable.  The reason I think the Bitmeister sounds so good with relatively more jitter is that it is 100% silver.  The reference cable has a silver center conductor and silver-plated copper shielding with aluminum foil.

Steve N.

[artur9]

Thanks for those listening impressions. 

One question.  If you were to draw the ideal measurement, what would it look like?  2 really thin peaks with no connection between them?

[audioengr]

Thanks for those listening impressions. 

One question.  If you were to draw the ideal measurement, what would it look like?  2 really thin peaks with no connection between them?

It would be one thin peak.  I think the two peaks are not an issue because they are so close together.  It probably causes a tiny shift in audio frequency that is not audible because the shift is so small or because it happens so quickly.

In the other thread on Renderer jitter, the plot with 4 peaks spaced wider apart still sounds very good, but causes a slight "hardness" to the sound that is not evident with one peak or two closely spaced peaks.

Steve N.

[audioengr]

I was asked to post the TIE or Time Interval Error measurements as well, so here they are:

Reference silver cable:



Reference silver cable time trend:



Cheap BNC cable:



Cheap BNC cable time trend:



Belden 1694A RCA cable:



Belden 1694A RCA cable time trend:



Bitmeister cable:



Bitmeister cable time trend:



These histograms look a bit different, even better.  I don't think they are that helpful for correlating SQ though, although it seems that flat-spots in the time trend may correlate to poor imaging.

Steve N.

[audioengr]

BTW, because of the results above, I made some modifications to my reference silver cable and now it delivers the liquid vocals of the Bitmeister, but with superb pin-point imaging.  Best S/PDIF cable I have heard yet.

I think I will still sell the basic $275 version of the BNC-BNC cable, but I will be adding the reference version at maybe $499.  For those that have a BNC-BNC cable in good condition, I can accept a trade-in and give a $200 credit towards the reference cable.

Steve N.

[paul79]

Do you have plots for this new SPDIF Cable? Maybe a comparison with your regular SPDIF Cable?

[audioengr]

Do you have plots for this new SPDIF Cable? Maybe a comparison with your regular SPDIF Cable?

I have a special plug adapter on order, so I will post no cable, the standard BNC cable and the reference BNC cable when I get it.

Steve N.

[paul79]

This might help correlate the visuals with the sound as well, so I look forward to this one.

[audioengr]

This might help correlate the visuals with the sound as well, so I look forward to this one.

Measurements completed and the first post has been updated to include no cable at all, only Synchro-Mesh. I figured out how to get accurate spectrum plots too, so these are included.

Steve N.

[brj]

Interesting results, Steve!

Two questions, if I may:

1) Can you clarify which of your Reference Cable plots, if any, used the updated version of the Reference Cable vs. the original?

2) How do things change at higher bit-depths/sampling rates, such as 24-bit/192 kHz?

Thanks!

[audioengr]

Interesting results, Steve!

Two questions, if I may:

1) Can you clarify which of your Reference Cable plots, if any, used the updated version of the Reference Cable vs. the original?

2) How do things change at higher bit-depths/sampling rates, such as 24-bit/192 kHz?

Thanks!

Standard BNC cable jitter spectrum



Reference BNC cable jitter spectrum



Notice the difference in vertical scale, so its not as bad as it looks.

Higher incoming sample-rates don't affect the Synchro-Mesh output, but I can use other sources.

I'll post the higher sample rates comparison when I get a chance.

Steve N.

[audioengr]

Some more cables to compare, both measurements and listening tests:

Synchro-Mesh was driving all cables using BNC connectors on both ends for measurements.

Interchange was driving all cables to Overdrive SX DAC using BNC connectors on both ends for listening tests.

First, my Standard BNC jitter distribution:




Canare LV-77S distribution:



Very close, but the "tails" are a bit elevated.

Belden 8281F distribution:



Almost identical to the Canare cable.

Now for the Spectral plots:

Standard Empirical BNC spectrum:



Canare LV-77S spectrum:



About 50% more jitter across the spectrum and wider main pulse.

Belden 8281F spectrum:



About double the jitter across the spectrum.

Listening tests:

Belden 8281F - bass very wide unfocused.  Some percussion undeciferable, like chimes, bells and tambourine etc.., dry and "pasty" sounding compared to the Empirical cable.

Canare LV-77S - good imaging and positioning, but dry and pasty sounding compared to the Empirical cable.  My wife would say it's a "wet cardboard" sound.

The Empirical Audio Reference BNC cable delivers a presentation that is much more "3-D" and all of the instruments sound correct and live.

Steve N.