Best audiophile switch

[audiobomber]

What is the benefit of reclocking with the Mutec REF10 SE120?

Great site, thank you! Mostly over my head though.
If I understand correctly, the external clock ignores any phase noise from the data stream, such that the only phase noise presented to the streamer/DAC is generated from the external clock and clock cable?

 

[audiobomber]

how does it apparently do the same over an optical connection? If it is supposed to somehow be embedded in the data itself, by what mechanism?

From a John Swenson post:
"All the optical does is block leakage, it doesn't get rid of clocking issues at all (it can actually make them worse). The fact that it is optical does not automatically apply some universal quantum time scheme that mystically aligns edges perfectly, If you send in a pulse, then another that is 50ns apart, then another at 51ns, then another at 49, that difference gets preserved at the receiver, the optical does not magically force all of them to be exactly 50ns.

The raw data coming out of the optical receiver goes into a chip that rebuilds the Ethernet signal using its own local clock, that is done with flip flops inside the chip, these flop flops behave just like any other flip flops, again no magic here. I was trying to avoid re-iterating what I have said before on this, but it looks like I'm going to have to do it anyway.

So how come this reclocking with a new clock is not perfect? As edges from the input stream go into a circuit each and every one of those edges creates a current pulse on the power and ground network inside the chip and on the board. The timing of that pulse is exactly related to the timing of the input data. The timing of the input data is directly related to the jitter on the clock producing the stream. This noise on the PG network changes the threshold voltage of anything receiving data inside the chip, especially the local clock going into the chip. This means the phase noise spectrum of the data coming in gets overlayed on top of the phase noise spectrum of the local clock. It's attenuated from what it is in the source box, but it is definitely still there.
THAT is how phase noise gets from one device to the next, EVEN over optical connections."

These are rhetorical questions perhaps (I'm certainly not losing sleep over them myself) and even if answered with crystal clarity and rock solid certainty, I'm not sure it really helps anyone identify their best audiophile switch as per thread title so is verging on "off topic"...

I believe that theories on the best switch design are definitely on topic.

 

[StreamFidelity]

How does phase noise manifest itself?

I understand your skepticism. In fact, John Swenson has not yet been able to show any measurements for the effect of Wander.

My point is, just give it a try. You don't have to invest in it yourself. Maybe there's someone near you who does reclocking and you can listen to it.

I invite anyone in Germany to visit me.

 

[StreamFidelity]

If I understand correctly, the external clock ignores any phase noise from the data stream, such that the only phase noise presented to the streamer/DAC is from the external clock and clock cable?

I think only a chip designer like John Swenson can answer this question. That's way above my pay grade.

My understanding is that the phase noise in the low frequency range up to 10Hz (and only this range) is not stopped by buffers and fiber optics, but is always present. Even when no data is flowing.

In some experiments I have to listen carefully to notice a difference. Not so with reclocking. My first listening impression was that the localization accuracy has increased significantly. The bass is even more full-bodied and to the point. With a well-known track, I suddenly heard a detail that I had never heard before. The music flows around me, there's no other way to put it.

 

[NigelB]

From a John Swenson post:
"All the optical does is block leakage, it doesn't get rid of clocking issues at all (it can actually make them worse). The fact that it is optical does not automatically apply some universal quantum time scheme that mystically aligns edges perfectly, If you send in a pulse, then another that is 50ns apart, then another at 51ns, then another at 49, that difference gets preserved at the receiver, the optical does not magically force all of them to be exactly 50ns.

The raw data coming out of the optical receiver goes into a chip that rebuilds the Ethernet signal using its own local clock, that is done with flip flops inside the chip, these flop flops behave just like any other flip flops, again no magic here. I was trying to avoid re-iterating what I have said before on this, but it looks like I'm going to have to do it anyway.

So how come this reclocking with a new clock is not perfect? As edges from the input stream go into a circuit each and every one of those edges creates a current pulse on the power and ground network inside the chip and on the board. The timing of that pulse is exactly related to the timing of the input data. The timing of the input data is directly related to the jitter on the clock producing the stream. This noise on the PG network changes the threshold voltage of anything receiving data inside the chip, especially the local clock going into the chip. This means the phase noise spectrum of the data coming in gets overlayed on top of the phase noise spectrum of the local clock. It's attenuated from what it is in the source box, but it is definitely still there.
THAT is how phase noise gets from one device to the next, EVEN over optical connections."

I believe that theories on the best switch design are definitely on topic.

This risks becoming a maker vs maker thing so I'll bow out gracefully soon as that would not be helpful to anyone... but before I go:

I'm very very familiar with this piece as so many people refer to it, but to me it is simply a reassertion rather than an explanation.

How can an optical connection make clocking issue worse? If it can, can it also make them better? How? Where is the timing information encoded in the OSI 7-layer model? If the timing is not embedded in the data but, like RFI, is transmitted as an electrical signal alongside it, how does this occur?

What are these edges? How do they mainfest themselves? As modifications to the bits, the 1's and 0's, the digital data? How does the 7-layer OSI model represent and communicate these?

How does the spectrum get transmitted from device to device? In the bits? Or in an electrical signal like RFI alongside the bits? If so, how can this possibly go over an optical connection?

These are all rhetorical questions. For clarity, I'm hugely grateful to Swenson and co for even raising the possibility (against a cynicism at the time that I can only imagine) that a switch, suitably designed and suitably deployed, has any contribution at all to make to sound quality. I've heard an EtherRegen and know it makes a difference. It's the connection between theory and practice which I (and others) have yet to make.

 

[NigelB]

I understand your skepticism. In fact, John Swenson has not yet been able to show any measurements for the effect of Wander.

My point is, just give it a try. You don't have to invest in it yourself. Maybe there's someone near you who does reclocking and you can listen to it.

I invite anyone in Germany to visit me.

Thank you. I hugely appreciate you response.

I can obtain a reclocker and try it on my switch, and will do so. My point though is not about the effect; it's about the mechanism by which that effect takes place. It may for example be that when I introduce an external clock I provide some sort of galvanic isolation which was not previously there, or an improved path-to-ground vs what was already there. Or other stuff I can't think of right now.

All I seek to do is to associate results with causality, as this allows inquisitive audio enthusiasts to make an informed decision about which switches to shortlist and helps switch manufacturers to focus on the design factors which make a difference rather than chasing other factors which don't.

 

[Di-fi]

I can obtain a reclocker and try it on my switch, and will do so. My point though is not about the effect

Respectfully, that’s exactly what I’ve been considering throughout this discussion on switches with and without clocks. Since you don’t believe in their impact and seem quite convinced of your approach, it surprises me that you never tested it. I do understand when you say, "All I seek to do is to associate results with causality." I can only encourage manufacturers to remain open-minded—the proof lies in the fact that we have top-performing switches both with and without clocks.

 

[NigelB]

Respectfully, that’s exactly what I’ve been considering throughout this discussion on switches with and without clocks. Since you don’t believe in their impact and seem quite convinced of your approach, it surprises me that you never tested it. I do understand when you say, "All I seek to do is to associate results with causality." I can only encourage manufacturers to remain open-minded—the proof lies in the fact that we have top-performing switches both with and without clocks.

I need a thread of a theory onto which to hang an experiment, I’ve yet to be persuaded by any theory which relates… oh, and I remembered last night something fairly important: that my switch doesn’t have an external clock IN!

Given reports from others, I wouldn’t be at all surprised if I heard an external clock making a positive difference. So it’s not that I don’t believe in their potential impact, it’s that I don’t believe any impact which does occur can arise from the accuracy/stability of that clock.

 

[treitz3]

Believe what you want, but that belief will coinside with restricting your system's performance.

Have you ever thought about the possibility that the clock could be equal too or more beneficial/substantial than the switch itself?

Tom

 

[MarkusBarkus]

...perhaps it's difficult to isolate the extent to which a better OXCO improves an audiophile switch because products seem to hit the market with multiple enhancements.

I recognize an engineer may be able to measure various attributes to determine the impact of that variable, but as a consumer faced with the prospect of evaluating the sound, we're evaluating the product with all it's characteristics.

I have mentioned my ancient daisy-chained Cisco switch fun, dedicated power, filters, etc. elsewhere, but when I pulled all that stuff out and replaced it with a modified Buffalo (Melco parent) switch, it was a very nice improvement to the network.

I kept that thing in place for years, just recently removing it when streamlining my Taiko-based network to the Olympus.

That switch had isolation feet added to the chassis, the premium PinkFaun OXCO added, some damping material inside, and a conversion to DC: 12v feed for the main board, 5v feed for the clock. Also, half of the ports were turned-off in the software.

In aggregate, it was a notable improvement to the device's performance in my network, but what (if any??) amount could be ascribed to the quality of the OXCO itself, vs. good power, etc.? Maybe none? I don't know...but the switch was very nice.

 

[audiobomber]

How can an optical connection make clocking issue worse? If it can, can it also make them better? How? Where is the timing information encoded in the OSI 7-layer model? If the timing is not embedded in the data but, like RFI, is transmitted as an electrical signal alongside it, how does this occur?

I know for certain that changes upstream from the optical conversion are readily audible. Something capable of changing the sound quality is traveling with the light, and obviously not contained in the data, otherwise one would hear gross distortion or dropouts.

Here is the explanation from John Swenson, while discussing Sonore products (opticalModule and opticalRendu):
"The important thing to understand is that ALL digital signals carry the "fingerprint" of the clock used to produce them. When a signal coming from a box with cheap clocks comes into a box (via Ethernet or USB etc) with a much better clock, the higher level of phase noise carried on the data signal can contaminate the phase noise of the "good" clock in the second box. Exactly how this happens is complicated, I've written about this in detail if you want to look it up and see what is going on.

The contamination is not complete, every time the signal gets "reclocked" by a much better clock the resulting signal carries an attenuated version of the first clock layered on top of the fingerprint of the second clock. The word "reclocked" just means the signal is regenerated by a circuit fed a different clock. It may be a better or a worse clock, reclocking doesn't always make things better!

As an example if you start with an Ethernet signal coming out of a cheap switch, the clock fingerprint is going to be pretty bad. If this goes into a circuit with a VERY good clock, the signal coming out contains a reduced fingerprint from the first clock layered on top of the good clock. If you feed THIS signal into another circuit with a very good clock, the fingerprint from the original clock gets reduced even further. But if you feed this signal into a box with a bad clock, you are back to a signal with a bad fingerprint.

The summary is that stringing together devices with GOOD clocking can dramatically attenuate the results of an upstream bad clock."

The explanation above handily explains why stacking switches can be an improvement, and why devices should optimally be deployed from worst clock to best clock. I follow Sonore and UpTone forums on Audiophile Style, because John Swenson and Alex Crespi are forerunners in the network audio field, and very free with information, from whom I and many others have learned a great deal.

This risks becoming a maker vs maker thing so I'll bow out gracefully soon as that would not be helpful to anyone...

I believe you are the only manufacturer still posting to this discussion?

 

[orange55]

Here is my experience which on clocks in switches.

I run two SOtM switches, the set-up is as follows:
Router > Chord Sarum Super Array Ethernet > Melco N1A (Which also has built in filtering with all traffic going through it) > Chord Sarum Super Array Ethernet > SOtM Switch > Amphenol dac cable SPF+ > SOtM Switch > Chord Music Ethernet to Steamer / Chord Sarum Super Array Ethernet to Apple TV

My SOtM switches are fitted with internal clocks and Master clock BNC inputs, this allows me to do a direct comparison to the internal clock and adding an external clock.

I have the Mutec REF10 which I use to clock both of my switches, I do not clock anything else as no other kit has an external clock option. Adding the Mutec clock makes a dramatic difference for the better, night and day, very easy for anyone to hear whether they are into hifi or not. It also makes a massive difference to the quality of the picture and sound on the Apple TV, so all video on demand services benefit of the upgrade in clock.

 

[audiobomber]

I have the Mutec REF10 which I use to clock both of my switches

I expect that being able to synchronize devices to the same clock as you have done is advantageous, especially with the REF10.

 

[Superdad]

I expect that being able to synchronize devices to the same clock as you have done is advantageous, especially with the REF10.

No “synchronization” is taking place by utilizing the same external reference clock for both of his switches. That’s not a thing.

The clock board in the SOtM switches (same board as offered for their streamer and USB boxes) is based on an early SiliconLabs clock synthesizer. That synth chip has two inputs and 4 outputs (though in their switch they use only one output). When an external clock is not connected, the synth uses as reference a custom crystal that SOtM had made for them long ago. When an external clock is used as reference the crystal is not used, the synth gets reprogrammed to run from 10MHz, and then again outputs 25MHz for the switch chip.
[We do something similar in the EtherREGEN, but with a newer synth and with all differential clock lines; and we use all 4 outputs of the synth because we need 2 each of 25MHz and 2 each of 250MHz—to feed both Ethernet chips and our special ultra-low-jitter reclocking flip-flops on both sides of our active-differential isolation “moat.”]

 

[audiobomber]

No “synchronization” is taking place by utilizing the same external reference clock for both of his switches. That’s not a thing.

I thought the original point of an external clock was to synch the timing among various devices in recording studios.

 

[Di-fi]

I thought the original point of an external clock was to synch the timing among various devices in recording studios.

You might think that connecting a high-quality external clock to both switches would synchronize them, making them work in perfect time together. But that’s not how it works—Ethernet doesn’t rely on synchronization like audio signals do in SPDIF, I2S, or AES.

What actually happens is that each switch stops using its internal clock and instead follows the external clock as its reference. This doesn’t mean the two switches are now "in sync" with each other—it just means they’re both running from the same, more precise clock source. They still operate independently because Ethernet doesn’t require or use direct clock synchronization like a DAC and transport would.

 

[Di-fi]

the synth gets reprogrammed to run from 10MHz, and then again outputs 25MHz for the switch chip.

Since 10MHz needs to be converted to 25MHz—and that conversion isn’t always perfect—does this have a noticeable impact on audio? If not handled properly, can it introduce additional noise or jitter? I assume the eR 1 (and 2) does the same, making this unavoidable. But is it audible in the context of audiophile best switches?

 

[Di-fi]

I had hoped given the thread title people would be relaying their experiences (and comparisons) of the Ansuz switches, Nordost QNET, NA Tempus e.t.c

Pura Power Supplies have a new switch called the Ammonite Elite powered by discrete super regulators of their own design. Melco have a new ultra expensive high end switch.

Ansuz, NA Tempus, and Melco have been brought up in this thread, though maybe not with the level of detail you were looking for. Below is part of a comparison between Ansuz and NA from HiFiAdvice. Ultimately, the two brands have different presentations, appealing to different audiences—just like all the switches discussed here.

Between Ansuz and Network Acoustics, we have two rather unique presentations. In my Ansuz review, I mentioned finding the switches surprisingly neutral, given earlier experience with very rich or warm and slow-sounding switches. But the tempus offsets the balance by providing an even more neutral and transparent perspective. In terms of frequency spectrum, I feel the Ansuz switches (especially the X-TC3) are more voluptuous in the bass, tonally more saturated, and darker in the treble, whereas the tempus is the opposite: light-footed, tonally slightly lean and exceedingly open and airy.

Besides the tonal differences, another distinction is in soundstaging. While both brands enhance the soundstage, they do it differently. With the Ansuz switches, the sonic aura warps all around you, along with everything in, which is hugely entertaining, but it may also be perceived as bordering on supernatural. With the tempus, the stage becomes much deeper and airier, but not so much wider, providing a performance that is less sensational but arguably more “normal” and may be perceived as more natural. The tempus very carefully threads the fine line between adding fluidity, refinement, and air, and not editorializing too much.

I hasten to add that the tempus is very far from a sober-sounding product. More like the opposite! It is more ethereal than grounded or earthy. But ultimately, the two brands have different presentations, which will cater to different audiences.

 

[Di-fi]

From the same site, HiFiAdvice, comes a review of the Silent Angel Bonn NX switch, which is already equipped with two internal TCXO clocks. Interestingly, in the context of our discussion, the author enhances its performance by combining it with a Genesis GX external 25MHz clock. Today, the NX+GX combo is still the author’s favorite switch setup.

As soon as the Genesis GX clock was connected and the music played, I smashed my mental “Instant Like” button. The clock takes immediate control and pulls everything into focus. The presentation becomes tighter, more upbeat, cleaner, more precise, and more transparent. There’s more control but not so much that the music is suffocated. Instead, the inverse happens; the music actually pulls me in that much more.

the Genesis GX clock has pulled the sound significantly more toward neutral

just as with the NX by itself, the GX avoids instilling a sense of listening to a tweaked-out audiophile system. Instead, even more so than with the NX alone, the combo’s presentation is punchy, planted, earthy, direct-coupled, and dynamically expressive, and as a result, actually more akin to being at a live music event.

When the Genesis GX Clock is added, the full potential is unleashed, and the presentation becomes tighter, more upbeat, cleaner, more precise, and more transparent while retaining the tonal saturation and the lyrical and emotional aspects of the music.

 

[Superdad]

Since 10MHz needs to be converted to 25MHz—and that conversion isn’t always perfect—does this have a noticeable impact on audio? If not handled properly, can it introduce additional noise or jitter? I assume the eR 1 (and 2) does the same, making this unavoidable. But is it audible in the context of audiophile best switches?

The many methods and performance issues of clock synthesis is a VERY complex subject!
And indeed some of the variations we see used in some popular products are doing more harm than good. But clock synthesis can also be done exceedingly well. This is why one can not begin to generalize about benefits or not of external clocking. It depends entirely on implementation in the device!
(Though often a device that does poor synthesis is still using the same circuit when running from its internal clock; I won't name names as doing so gets me reported to the management. But obviously if you see a box that has a 10MHz internal clock, it absolutely has to convert that rate even when an external clock is fed.)

To be sure, the PLL or DDS of even the best (expensive) ultra-low-jitter clock synthesizers (from Skyworks, T.I., Analog Devices) will have some additive jitter/phase-noise. But we have studied and measured these and find that--at least with the good ones--the added phase-noise below offsets of about 30KHz (yes, KHz not MHz as I am referring to PN at low offsets from carrier) is only single digits of dBc/Hz greater than whatever the input reference clock measures.

Plus in the case of the EtherREGEN, all the clock lines we run from the synth are differential, and that is very helpful.

@audiobomber asked me in a PM this morning about the 25.0MHz Crystek CCHD-575 we use as internal reference in the EtherREGEN. Part of my reply was:

The CCHD-575 is the lowest-phase noise production XO (not OCXO) available and costs us $10 each in 500 piece quantity.
Ignore the too conservative phase-noise graphs that Crystek publishes. We regularly measure (on our $30K Jackson Labs PhaseStation) these 25.0MHz XOs at -108dBc/Hz to -112dBc/Hz at 10Hz offset.
That performance equates (with 6dBc/Hz per octave compensation) to being equal in performance to 10MHz clock of -119dBc/Hz @10Hz.
I personally find that it takes an external clock at least 12dBc/Hz better than that to begin to be worthwhile--so that's about -130dBc/Hz (@10Hz) or better as a place to start.

--Alex C.