Best audiophile switch

[Di-fi]

Categorizing the best switches and understanding expected improvements

We've seen many impressions in this thread, and it's clear that audiophile switches can significantly impact sound quality. Users report improvements in clarity, soundstage, timing, and smoothness—all enhancing the listening experience.

Based on the thread, can we identify the best switches that consistently lead to notable SQ improvements? Some seem to perform less well, (sometimes due to a lack of isolation, the wrong ethernet cable or power supply quality) or simply because they were replaced by a better-performing switch.

When categorizing the best switches, it's important to consider audible improvements you’re aiming for, as technical features (like noise reduction, jitter control, and optimized streamer processing) are often known but not directly linked to the SQ to expect. The improvements are often subjective and vary depending on the system synergy and implementation.

While there’s no universal technical explanation for why switches improve SQ, some common traits emerge in the best switches. The way switches are implemented—power supplies, isolation methods, and clocking precision—can often make or break their impact.

Maybe we don’t need a new thread. As we continue discussing switches, I think it’s important to ask: What are you specifically hoping to improve in your system? Since technical features are mostly well known, but audible improvements are harder to predict, understanding which traits users value most will help keep this discussion moving forward. Just my 2c.

 

[NigelB]

Categorizing the best switches and understanding expected improvements

We've seen many impressions in this thread, and it's clear that audiophile switches can significantly impact sound quality. Users report improvements in clarity, soundstage, timing, and smoothness—all enhancing the listening experience.

Based on the thread, can we identify the best switches that consistently lead to notable SQ improvements? Some seem to perform less well, (sometimes due to a lack of isolation, the wrong ethernet cable or power supply quality) or simply because they were replaced by a better-performing switch.

When categorizing the best switches, it's important to consider audible improvements you’re aiming for, as technical features (like noise reduction, jitter control, and optimized streamer processing) are often known but not directly linked to the SQ to expect. The improvements are often subjective and vary depending on the system synergy and implementation.

While there’s no universal technical explanation for why switches improve SQ, some common traits emerge in the best switches. The way switches are implemented—power supplies, isolation methods, and clocking precision—can often make or break their impact.

Maybe we don’t need a new thread. As we continue discussing switches, I think it’s important to ask: What are you specifically hoping to improve in your system? Since technical features are mostly well known, but audible improvements are harder to predict, understanding which traits users value most will help keep this discussion moving forward. Just my 2c.

A good way of seeking to take stock of the last 36 pages.

However, if you're seeking to attribute (sonic) effect to a cause, jitter and clocking precision should not be on this list. Ethernet is asynchronous (data packets, error correction, retransmit if necessary etc); timing information in the ethernet domain doesn't effect sound quality because these packets are completely rebuilt by the streamer into a continuous bitstream the DAC can recognise. The streamer is the first point at which clock precision/jitter can be controlled with any effect on sound quality. Clock quality perhaps (eg the amount of noise it produces) but not clock precision.

If "best" relates purely to sound quality, the focus will be on noise, primarily RFI noise but also noise from other sources: power supplies and vibration as you suggest.
In RFI terms, I'd suggest the best switches would do three things:
- pass on via their output connection as little as possible of the conducted RFI noise they receive on their input connection (multiple design factors),
- resist absorbing as much radiated RFI as possible from the local environment (case design), and
- minimise the amount of RFI noise the switch itself produces in doing its job (internal components/circuitry).

So much for theory! Looking at such criteria is one way of narrowing down to a shortlist but I suspect what most of us here value above all is hearing from fellow enthusiasts who've compared at least 2 switches and landed on one or more they prefer, and your post is a great reminder of that.

 

[Di-fi]

However, if you're seeking to attribute (sonic) effect to a cause, jitter and clocking precision should not be on this list. Ethernet is asynchronous (data packets, error correction, retransmit if necessary etc); timing information in the ethernet domain doesn't effect sound quality because these packets are completely rebuilt by the streamer into a continuous bitstream the DAC can recognise. The streamer is the first point at which clock precision/jitter can be controlled with any effect on sound quality. Clock quality perhaps (eg the amount of noise it produces) but not clock precision.

I believe it’s important to stay open to the idea that different designs can lead to similar or complementary results. Your approach, focusing on noise and interference without clocking, works well, alongside improvements like better Ethernet cables and power supply. However, based on what users are sharing here, external clocks, optical isolation, and other methods also seem to contribute to perceived improvements. Whether it’s due to clock quality or another factor, the effect is clearly audible for many.

 

[agisthos]

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.

 

[NigelB]

I believe it’s important to stay open to the idea that different designs can lead to similar or complementary results. Your approach, focusing on noise and interference without clocking, works well, alongside improvements like better Ethernet cables and power supply. However, based on what users are sharing here, external clocks, optical isolation, and other methods also seem to contribute to perceived improvements. Whether it’s due to clock quality or another factor, the effect is clearly audible for many.

I am always very clear NOT to suggest that external clocks, clock quality etc can't make a difference; if people hear differences in switches, external clocks, better quality clocks or whatever then who are we to deny that?

But I am happy to stand my my assertion that clock accuracy (which is what you said in your earlier post) can NOT make a difference as there is simply no mechanism by which it can do so.

My point is only that it would be incorrect to attribute any sound quality improvement one hears to clock accuracy, whether the clock in question is internal or external. That's all.

I say this in part to stop people wasting their money on external clocks which promise better sound due to higher accuracy and to focus their search on better clock quality; if the clock designer understands what can and can't make a difference, they get my vote.

I also say it in part to challenge the the unfortunate extrapolation some folk (including some manufacturers) make from the post-streamer digital domain to the pre-streamer digital domain. There are 3 domains in most streamed audio setups: analogue (post-DAC), digital (from streamer to DAC) and network (pre-streamer); they all operate under different constraints.

 

[NigelB]

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.

Indeed. If I recall correctly, there are some posts which do this (though I may be mixing this up with another thread nominally about affordable switches), but the off-topic posts make it a hard slog to dig these out. @Di-fi 's suggestion that we take stock is a welcome one.

 

[MarkusBarkus]

...a very reasonable reply, @NigelB. Cheers, fella.

 

[orange55]

@NigelB Before I post my experience's with clocks and switches, how do you define clock accuracy? That way I can check the stats of my clocks. Thanks

 

[NigelB]

@NigelB Before I post my experience's with clocks and switches, how do you define clock accuracy? That way I can check the stats of my clocks. Thanks

I'm not sure I'm the best-placed guy to answer this, actually, as my own focus is primarily in the pre-streamer domain where this is not important (vs in a streamer or a DAC where it absolutely is)... but generally speaking people tend to rate OCXO over TCXO and they do so based on accuracies typically measured in ppm and expressed as stability over time and/or over temperature range. So you'll see figures like ±1 x 10-7 to ±1 x 10-9 or ±0.20 ppm to ±2.0 ppm.

There may well be a correlation between higher accuracy/stability and lower noise, and if there is then this would at least partly explain why some people implementing higher accuracy clocks report sonic improvements.
Perhaps at least as common as discussions of accuracy are discussions of non-accuracy usually called jitter; so sometimes you'll read about a low jitter clock rather than a high accuracy/stability clock. After the streamer, in the bitstream world, higher clock accuracy does indeed mean lower jitter and this is universally acknowledged as A Good Thing.

But ethernet jitter means something else entirely: it's a measure of the stability in a corporate network based on variability in the timing of the arrival of data packets, and higher jitter typically suggests a heavily loaded or overloaded network. See this https://www.ir.com/guides/what-is-network-jitter for example. Jitter is a classic example of where terminology from one domain has been misunderstood and misapplied in another. This kind of jitter, network/ethernet jitter, is simply not a concern in our audio-dedicated networks.

Hope this helps.

 

[agisthos]

I spent a lot of time doing clock mods back in the Oppo days and it seemed not accuracy but phase noise was the important factor. The power supplying the clock itself also a huge factor.

But here was a surprising result, it wasn't OCXO that was the best clock but SAW clocks (surface acoustic wave), which are barely ever used or mentioned in the audiophile world. But it was hard to get the values you wanted it was only Epson that seemed to manufacture them for industrial use.

 

[PYP]

Maybe we don’t need a new thread. As we continue discussing switches, I think it’s important to ask: What are you specifically hoping to improve in your system?

There are 3 domains in most streamed audio setups: analogue (post-DAC), digital (from streamer to DAC) and network (pre-streamer); they all operate under different constraints.

Since my setup is now static from streamer to speakers (2 or the 3 domains mentioned above), I focused on the pre-streamer. When the setup was a bit different, I had accumulated quite a complex combination of boxes (switches, clock, LPSs) and lots of little wires. With the Grimm streamer in place, I could eliminate all of that stuff because I didn't hear a difference.

I subsequently discovered that a passive filter helped refine the HF somewhat and thought it would be interesting to experiment with a switch again for my current setup. It was just curiosity rather than a goal for better sound and I was skeptical. The criteria were more physical and pragmatic: small form factor, no or little heat generated, all-in-one box, extended in-home demo. Just as that thought was buzzing around my brain, I received and email about the Synergistic Research UEF switch.

With this active switch (proceeded by the passive filter) there are improvements in depth, density, tone and timbre. It isn't subtle but it is natural sounding to my ears.

 

[NigelB]

I spent a lot of time doing clock mods back in the Oppo days and it seemed not accuracy but phase noise was the important factor. The power supplying the clock itself also a huge factor.

But here was a surprising result, it wasn't OCXO that was the best clock but SAW clocks (surface acoustic wave), which are barely ever used or mentioned in the audiophile world. But it was hard to get the values you wanted it was only Epson that seemed to manufacture them for industrial use.

"In the Oppo days": was this in the pre-streamer network domain or in the post-streamer digital domain? Oppo suggests post-streamer (non-ethernet) but I may be missing something as I only really know their verging-on-legendary players.

 

[PYP]

In RFI terms, I'd suggest the best switches would do three things:

- pass on via their output connection as little as possible of the conducted RFI noise they receive on their input connection (multiple design factors),
- resist absorbing as much radiated RFI as possible from the local environment (case design), and
- minimise the amount of RFI noise the switch itself produces in doing its job (internal components/circuitry).

So much for theory!

Are these three types of RFI measurable or is the theory confirmed by listening? Bolding added...

 

[Di-fi]

I am always very clear NOT to suggest that external clocks, clock quality etc can't make a difference; if people hear differences in switches, external clocks, better quality clocks or whatever then who are we to deny that?

Sure, so the impact of clocks isn’t about correcting the Ethernet signal itself but about how the network environment affects the streamer’s processing. Well-implemented external and internal clocks seem to reduce unwanted interference, lowering noise that interacts with power supplies and downstream circuitry (and possibly more). Users consistently report audible improvements, suggesting that clock quality plays a role beyond just shielding and noise reduction.

But I am happy to stand my my assertion that clock accuracy (which is what you said in your earlier post) can NOT make a difference as there is simply no mechanism by which it can do so.

My basic clock characteristics understanding:

Accuracy refers to how close the clock's time or frequency is to the true or desired value. An accurate clock’s timing is close to the intended or ideal value.

Precision refers to the consistency of the clock's timing (or jitter). A precise clock will consistently produce the same result, even if that result is slightly off from the desired value. For example, if a clock's frequency is consistently 1 MHz but it’s actually 1.0001 MHz, it is precise but not necessarily accurate.

Clock quality includes both accuracy and precision and other characteristics.

Audiophile switch quality goes beyond basic networking functionality, incorporating factors like clock quality, a clean and stable power supply (including AC or DC cables), and effective noise control to minimize EMI/RFI interference. Isolation and shielding play a crucial role in preventing unwanted interactions, while high-grade internal components contribute to overall signal integrity. Additionally, at least one Ethernet cable to the streamer can influence the overall result, either enhancing or limiting performance.

 

[agisthos]

"In the Oppo days": was this in the pre-streamer network domain or in the post-streamer digital domain? Oppo suggests post-streamer (non-ethernet) but I may be missing something as I only really know their verging-on-legendary players.

This was upgrading the clock on the mainboard of the Oppo blu-ray players. It would improve both video and digital audio output quality for anything streamed through its ethernet input. I focused on this aspect and would use these players as a pure digital streamer, I did not bother to use the Oppo audio DAC boards (which I used to remove).

 

[orange55]

This was upgrading the clock on the mainboard of the Oppo blu-ray players. It would improve both video and digital audio output quality for anything streamed through its ethernet input. I focused on this aspect and would use these players as a pure digital streamer, I did not bother to use the Oppo audio DAC boards (which I used to remove).

Agree, I have an Audiocom upgraded player with all their mods and compared to the standard model it’s on another levels for picture and sound. I had a 205 standard to compare side by side and it was very obvious. I went for 203 upgrade as never used the analog outputs of 205 so online paying the extra.

Long live my Oppo.

 

[StreamFidelity]

Jitter is a classic example of where terminology from one domain has been misunderstood and misapplied in another. This kind of jitter, network/ethernet jitter, is simply not a concern in our audio-dedicated networks.

I agree. If jitter comes into play, the packets are written to the memory too quickly or too slowly. This is because every clock has Synchronization fluctuations, which of course depends on the quality of the clocks. But this doesn't matter because the data packets are retrieved at the right time at the receiver based on its own clock.

There may well be a correlation between higher accuracy/stability and lower noise, and if there is then this would at least partly explain why some people implementing higher accuracy clocks report sonic improvements.

Each of my visitors heard the effect when I switched the reclocker Mutec REF10 SE120 on or off. One explanation could be Wander. Wander is low frequency jitter down to 10 Hz! To clarify, most jitter numbers only take into account the phase noise in the range of 12 kHz to 10 MHz, which is well outside the range that seems to affect digital audio. Wander can demodulate the audio signals and cause a "broadening" of frequencies. For example, a 1 kHz signal would be "fanned out" by 10 Hz jitter to a frequency range of 990 Hz to 1010 Hz.

John Swenson's theory is that the phase noise spectrum of the incoming data overlays the phase noise spectrum of the local clock (receiver clock). This is how phase noise travels from one device to the next, including via optical connections. This phase noise propagates - both with interfaces such as Ethernet, USB, S/PDIF, I2S, and in the chips on the boards. Even a DAC with a perfect clock sitting right next to the DAC chip is affected by all the upstream jitter/phase modulation that took place in the chips and on the circuit board just before.

If you are interested: I have published a review on my homepage: What is the benefit of reclocking with the Mutec REF10 SE120?

 

[NigelB]

Are these three types of RFI measurable or is the theory confirmed by listening? Bolding added...

I'm not sure anyone could measure them separately. I've had my devices measured for both conducted and radiated emissions for CE compliance purposes.

Measurements won't tell you what you need to know; listening will. People interested in a thread about the best audiophile switch already know this of course!

 

[NigelB]

This was upgrading the clock on the mainboard of the Oppo blu-ray players. It would improve both video and digital audio output quality for anything streamed through its ethernet input. I focused on this aspect and would use these players as a pure digital streamer, I did not bother to use the Oppo audio DAC boards (which I used to remove).

Yes, that makes perfect sense as it is in the core digital (bitstream) space not in the ethernet space.

 

[NigelB]

I agree. If jitter comes into play, the packets are written to the memory too quickly or too slowly. This is because every clock has Synchronization fluctuations, which of course depends on the quality of the clocks. But this doesn't matter because the data packets are retrieved at the right time at the receiver based on its own clock.


Each of my visitors heard the effect when I switched the reclocker Mutec REF10 SE120 on or off. One explanation could be Wander. Wander is low frequency jitter down to 10 Hz! To clarify, most jitter numbers only take into account the phase noise in the range of 12 kHz to 10 MHz, which is well outside the range that seems to affect digital audio. Wander can demodulate the audio signals and cause a "broadening" of frequencies. For example, a 1 kHz signal would be "fanned out" by 10 Hz jitter to a frequency range of 990 Hz to 1010 Hz.

John Swenson's theory is that the phase noise spectrum of the incoming data overlays the phase noise spectrum of the local clock (receiver clock). This is how phase noise travels from one device to the next, including via optical connections. This phase noise propagates - both with interfaces such as Ethernet, USB, S/PDIF, I2S, and in the chips on the boards. Even a DAC with a perfect clock sitting right next to the DAC chip is affected by all the upstream jitter/phase modulation that took place in the chips and on the circuit board just before.

If you are interested: I have published a review on my homepage: What is the benefit of reclocking with the Mutec REF10 SE120?

Thanks for the link re Mutec, I checked it out. I used to use an MC-3+ USB on my Bluesound Node2 to positive effect. I'm afraid I'm still here (from your concluding remarks): "In the basics, we discussed that jitter is usually not a problem with asynchronous data transmission. Unless the jitter is so high that crackling or dropouts occur due to data loss. In this case, you can try to get the problem under control with buffer settings."

I am familiar with John Swenson's theory as many people refer to his paper on the subject; however I'm simply not persuaded. I've yet to understand how phase noise manifests itself in a signal made of of data packets, let alone how that same phase noise makes it all the way through the streamer to the DAC after the packets have been ripped open and rebuilt as a continuous digital signal.

RFI I understand. Radio Frequency Interference is a subset of the broader spectrum of EMI ElectroMagnetic Interference; I understand what it is, how it is conducted from one device to another along conductors and/or shields, how it is picked up when radiated, and how it interferes/interacts with the analogue circuitry of the DAC through intermodulation to create distortion.

One could stick a switch or other device in a chamber and get it measured for radiated and conducted RFI; my point is not about the measurements themselves (as I'm not convinced these are sufficient to assess the overall performance of one switch vs another), it is that RFI clearly exists and its mechanisms, impacts and controls are generally understood (oh, and subject to legislation).

How does phase noise manifest itself? I presume it is conducted not radiated. Is it a wave in a different part of the EMI spectrum? Does it travel along the conductors and shields of a copper cable? If so, 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?

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"...