A dive into digital interlinks – coaxial cables

The jitter measurements

A quick look at the measurements. We measured ten different cables, counting the Audioquest Diamond as two cables for a moment, since we tested it with and without DBS. The results can be found in the table and graph below. First we will go through the salient points per cable.

Baseline measurement

We took our Hirschmann BNC of 50cm. This is an official, coaxial measurement cable with BNC connectors. With this we also calibrate the Wavecrest if necessary.

You can see that the internal 10 MHz calibration-clock of the Wavecrest gives nice values. Note that this clock is not used for measuring jitter. This machine does that differently; for example, with asynchronous random sampling. However, this is not the place to go into that.

We see around 2.5 ps jitter, -69 dBc/Hz phase noise at 10dB offset and 2.5 ps (1-sigma) LFM (Low Frequency Modulation) jitter. We’ll take this as a baseline for a moment.

(At the bottom you can see Freq (The input frequency) – Rise (rise time of the square wave) – Fall (the fall time/fall time of the square wave) – Width (width of the square wave in time)).

When we put the square wave on our scope, we don’t see anything crazy compared to the rest of the test field. The differences tend to be very subtle. With two – actually three – exceptions. More on that later.

The rest of the test field!

So we have nine more cables where we measure the Audioquest Diamond twice: with DBS connected and without voltage on the DBS system. This is to chart any influence.

You will see that with decent, coaxial cables, the differences are extremely small. With coaxial cables with RCA connection – which is NOT perfectly 75 Ohm, which is measurable and also influenced by the adapter – it is between 3 and 5 ps. We estimate that the differences come from shielding, connector material and perhaps a piece of geometry. But the fact is: there are sometimes external influences.

Note: the length of the cable also has an influence. Know that the Audioquests are 1.5 meters each. The competitors are 1 meter. That also has some influence on jitter. Keep that in mind.

Audiomaat BNC – 1 meter

We had some cables from Audiomaat left over from a reader of ours. These had been sent and did not need to be returned. Now it is always handy to have cables lying around. Especially for research.

Now most BNCs measure fine, as you can see. This Audiomate, however, shoots through the roof. And not in a positive sense. The jitter goes up by a factor of five to even seven! Strange.

Since we have two equal cables lying around, we decided to sacrifice one for science. Turns out: it’s not a coaxial cable at all. It’s just two normal stranded cables soldered to a BNC. It looks like lamp wire….

You will understand that these cables are totally unsuitable for transporting a 10 MHz clock signal. And also totally unsuitable for processing other digital signals (again, just think about the characteristic impedance!). And this test also shows that. Very bizarre that this is sold as a BNC interlink.

This isn’t going well at all! Look at the noise on the square wave! This is the result of an unshielded cable…. Something unthinkable with an official coaxial cable. Next, looking at the rise and fall times, we see something crazy there too: the fall time is significantly longer than the rise time. That’s not supposed to be the case.

Audioquest Forest – 1.5 meters

The entry-level Audioquest performs really nicely! In fact, it is at the top of the field. And that with 39 Euro for 1.5 meters. That is a very nice performance Audioquest shows here.

We see an average jitter of about 3.6 ps. Considering the fact that they are RCA connectors and we have to use adapters, that’s neat. You can see the rest of the measurements in the table below.

All in all, a stellar performance from this entry-level Audioquest!

We didn’t expect the Forest to score so neatly. For a coaxial cable with rca plugs, we think this is pretty impressive! It is comparable to the BNCs if we look closely. Ringing is even less than with the Sommer BNC. This could be due to a little more attenuation, or the fact that the BNCs have 50 Ohm characteristic impedance and we terminate with 75 Ohm. Still. good score.

Audioquest Cinnamon – 1.5 meters

The Cinnamon is next in line from Audioquest. Now we see a bit of a wacky trend emerging. The higher up the line we go, the higher the jitter gets. Even though it’s tenths of picoseconds we add. You may not hear that; we can measure it. And we honestly can’t explain it very well.

The Cinnamon, however, still does nicely when we look at overall performance.

The Cinnamon and Forest – and Carbon – perform almost identically here. We see no crazy overshoot or ringing. Rise and Fall times are well proportioned. Excellent score.

Audioquest Carbon – 1.5 meters

As with the Cinnamon, we see a slight increase in jitter values. We are guessing that this is also the plugs, since they are also different. Who knows: we can’t cut all the cables of course!

However, again the values are just neat, considering that we are dealing here with RCA coaxial interlinks with an adapter that we are forced to use. The price difference with the Cinnamon is significant. Also with the listening test in mind.

We see a neat square wave, free of noise or distortion. This is a very similar picture to those in the Forest and Cinnamon; slightly more attenuation, but almost negligible (we’re guessing this has to do with the slightly higher capacitance; see the LCR measurements below).

No crazy overshoots or ringing. It’s even a bit less than with the Sommer and Ixos. All in all, a good score.

Audioquest Diamond – DBS – 1.5 meters

The Diamond fits the bill for more expensive but slightly higher jitter. It shows that one number does not paint the total picture. Also, these measurements show that we need to dive deeper. Take a closer look at the graphical representation of Low Frequency Modulation. In the version where we pull the plug, it’s actually higher, but apparently the Wavecrest counts it as lower on average….

However: all in all, the Diamond scores better without the system active. It’s subtle, but reproducible. And it’s also audible. What you like better we’ll leave to you.

The square wave is very clean – as with the rest of the Audioquest cables. However, we see the influence of the DBS system. First of all, the square wave is smaller; as soon as we plug in the Diamond, the amplitude drops. (Just look at the other cables). That may have to do with the slightly higher capacitance in this cable.

Now we must say that Audioquest has completely solved the “DBS problem” we had discovered earlier. Indeed, the block waves are identical when the DBS-system is energized and not energized. However, in both cases the fall time is longer than the rise time. This can also be seen in the “skirt” at the base. And that’s kind of strange…

Audioquest Diamond – No DBS – 1.5 meters

Albedo silver BNC – 1 meter

Wow… this is going badly wrong too! A jitter value that doubles on average: that’s just not possible with a decent BNC cable. Unless… it’s not a coaxial cable. Our suspicions are confirmed as soon as we strip the cable. Shame on you, Albedo…

Again, we see noise on the square wave. And an extremely poor ratio of rise to fall time. This is disproportionately bad. What else should we say about this: avoid these cables.

Sommer BNC – 1 meter

Not entirely surprisingly, we see good scores equivalent to the Hirschmann measurement cables. These are very tight scores that are probably at the limit of the calibration clock. Maybe with a direct sma-connection we can get sub 2ps scores…maybe.

No crazy things here either, although there does seem to be a bit more ‘overshoot’ on the square wave when we compare it to the Hirschmann and Audioquests. Whether that is ‘audible’ remains to be seen. Other than that, the square wave is free of noise or other distortion and the rise and fall time is nicely balanced.

Ixos 105 coax – 1 meter

This cable is kind of surprising since it measures incredibly well. But does not sound very good. How that is possible, is the big question. We don’t see anything crazy in the measurements. Everything seems fine when we look at the scores. In other words: perfect measurement cable…. but not as perfect as digital interlink.

Again, neat scores. Everything is in order: no noise, no excessive overshoots and rise and fall time is perfectly balanced.

Wavecrest jitter measurements overview

Overview in graph form

You can see that the bulk of Audioquest cables lie together in terms of jitter and phase noise. That means the cables have little to no impact on jitter. So technically, it’s all right. Although we do find it odd that the more expensive the cable, the higher the jitter. Even though it is only 2ps difference in total; it is reproducible and therefore constant.

The Albedo and Audiomat Mway stand out negatively. These are not coaxial cables. The Albedo is a twisted pair cable and the Audiomaat is just a kind of lamp wire: the wires are parallel to each other. That does not work at all… as you can see. The jitter increases enormously.

Cable properties Audioquest coax interlinks

For your – and our – imaging, we have linked all Audioquest cables to the LCR. The reason we grabbed only AQ is that the lengths are the same. Length affects measurements.

You can see that differences can be measured and sometimes they are reflected in the other measurements. Consider difference is capacitance.

It is noticeable that the Forest has a higher conductance than the Cinnamon. And the Carbon again above the Diamond. The Cinnamon responds a bit odd in the impedance/phase test. Now this is not directly linked to the “weird focus,” since this is not analog signal.

In terms of impedance, we see that Diamond and Carbon are close to each other. Forest is a bit higher and Cinnamon relatively very low again. Know that we are talking about mOhms; practically nothing. Although it does increase at 10 MHz of course; impedance is frequency-dependent.

More than jitter?

Our listening tests show that there is more to it than pure jitter. Although the Ixos 105 scores well, it sounds less pleasant than the Sommer BNC and Audioquest Forest. And where the Forest plays very well, the Audioquest Diamond again brings a little more calmness and feels more complete….

However, the listening test also reveals that the Audiomaat just overshoots in jitter. The cable is also unshielded and does not adhere to standards…. so there are limits. And the Wavecrest shows them….