This is a review and detailed measurements of the Silent Angel (Thunder Data) Bonn N8 8-port gigabite audiophile Ethernet switch. It is on kind loan from a member. It seems to cost US $399. As a way of comparison, a brand name switch without the "audio grade" designation costs US $35 so about ten times cheaper.
From outside, the Bonn N8 looks like the most generic switches:
Look at the cheap labeling underneath:
The external power supply from respected Mean Well company is a bit step above what you get with these cheap switches.
Company advertises better clock, better power supply filtering and better PCB layout as the improvements. Some generic eye pattern and noise improvement graphs are shown by the company but of course, no graphs showing the improvement out of any audio streamer. Unless you listen to digital bits from Ethernet cable, I am not sure what is behind this common and critical failing.
So let's do what they should have done, i.e. measure the real output of a streamer and see if the switch makes a difference.
Audiophile Ethernet Switch Audio Measurements
Let's agree that you are not going to buy a $400 switch and try to improve the performance of a $40 Raspberry Pi streamer. And that any test with a low-end DAC and streamer will be dismissed out of hand for not having "enough resolution" to show the difference. As with my prior tests of such devices (see EtherRegen review) I will be tested these network devices with my favorite streamer, the Matrix Element i. As a baseline reference, let's use the Roon to push a 1 kHz tone to it through the backbone of my home network which is a 48 port Netgear smart Gigabit switch:
Performance is identical to using USB input on Element i which is reassuring. It means that it has total isolation of the networking subsystem which is usually a very busy subsystem with a processor, operating system and boatload of software to handle streaming.
Note that my switch is in heavy use while the above test is going on so if it were to transmit "noise" that would be impacting the audio, it would be there. Yet we see no evidence of that. The only thing in the output is the harmonic distortions of the 1 kHz tone.
Now let's route the same network link through Bonn N8 and see what we get:
Nothing. Not a thing is out of place other than ordinary tiny run to run variations. Our clock frequency is dead on either way at 1 kHz to five decimal places. Output voltage has not changed. Distortion, noise and harmonics as translated into SINAD remain the same.
An implication is made that if the Ethernet switch generates Ethernet signals with better click (read as lower jitter), that must translate into lower jitter in the DAC. So let's test that theory:
No difference whatsoever. As noted in the slide, due to advanced signal processing, our measurement noise floor is at -155 dB which goes beyond the dynamic range of even 24 bit audio. We see a tiny jitter source at +- 8 kHz relative to our 12 kHz tone which remains there with or without Bonn B8 switch in the loop. There is also a power supply spike to the left and it also remains there independent of the switch in use.
Let's see if we see something in the noise spectrum. I opened the gates by increasing the bandwidth to 96 kHz and measured the residual noise out of Element i with and without Bonn N8 switch in the loop:
Not a thing is different.
In order to give the switch a better chance to make a difference, my son suggested that we create large amount of traffic on the link and see what happens (audio is very low rate as network rates go). Using a traffic generator program, we pounded on UDP port 7 (ping protocol) pushing the network rate as measured by perfom on Windows to above 700 mbit/sec. I then ran the above test again (not shown) and it made absolutely no difference either way.
As another way to test the noise theory, I captured the noise level out of the Element i continuously once a second. First, with no Ethernet cable into it, then my Netgear switch and finally, routed through Bonn N8. Can you tell when I made these changes?
You cannot of course because nothing changes in the output of the DAC inside the Element i streamer.
Listening Tests
I queued up my reference playlist in Roon and streamed them to Ethernet i. Baseline was with the Bonn N8 in the loop. I switched to just the Netgear and it sounded better! More detail. More air. Just higher fidelity. I then reversed the AB sequence, starting with the Netgear and going to Bonn N8. Same difference! More air, more analog like, etc.
What is going on? Simple: you frame of mind makes a huge difference in what you perceive. When performing comparisons, your brain gets into instrumentation mode and analyzes music differently than when you lean back and just enjoy music. In doing so, it hears more detail, air, etc. It does so even when nothing is changed in the waveform. In other words, the Ethernet switch is not changing sound. It is you that is changing.
This is why we test people blind. When your brain doesn't know what is being tested, then its variations factor out (statistically speaking).
Discussion and Conclusions
I honestly go into every one of these reviews with a clear slate, hoping to find a difference. Any difference. Even though I am usually successful in such things as USB cables and such, I have yet to find any measurable difference whatsoever with these "audiophile switches." As it should be. Ethernet is designed for long distance communication (100 meters or 330 feet) so it already has provisions to deal with heavily distorted signals, noise, ground loops, etc. If it did not, there would be hell to pay as our foundation of networking would fall apart.
You can try to clean up Ethernet signals if you wanted and get longer distance propagation from it. But better fidelity in a streamer? The only way that would happen is if the DAC in a streamer was sensitive to such things. And if this were to be the case, then the mere inclusion of streaming functionality in a DAC would destroy its performance. The noise from the CPU and operating system activity is so much more significant than any switch upstream.
Measurements conclusively prove what I explain above. That the Ethernet switch makes zero difference in the performance of a decent streamer. So let's not talk about jitter and noise changing. They absolutely, positively are not. If they were, the manufacturer would be showing that and shouting from the top of the mountain.
So there is really one argument left in favor of such devices: user testimony. Many people buy these switches, immediately notice improvement to the sound and call it good. When pushed on why, the talk about jitter and noise but per above, that is a made up fantasy. And justification after the fact by a person who doesn't understand the system design.
The only thing we need to deal with then is that "it sounds good to my ears." Well, it did that for me too expect that I could make either my standard Netgear switch or Bonn N8 sound better! Clearly that can't be the case. Logic would say that the testing is flawed. That using one's ear in a sighted situation is at fault. Lest you want to agree that both switches can be better than each other at the same time!
This is an uncomfortable situation for audiophiles. Their perception says one thing. Our data and science says the opposite. They chose to believe their perception. If you are hear and think that way, I implore you to have a loved one test you blind. Have then do the test as I did with the switch being in the loop or not (randomly) once a day. Do this for 10 days and you see if you can guess correctly which switch is in play 8 times or more.
My strong conviction is that you will not be able to get anywhere near 8 right.
Conversely, I can take two identical switches, change the label on one, and have people listen. I guarantee you that I can get testimonials in favor of whichever switch I want even though they are identical! Please don't make me do that. It will be embarrassing for many people involved. Do the test I suggest above in your own home with no one looking. Once you see the truth, you can sell the switch to someone else and have money for something worthwhile in audio or otherwise.
Needless to say, I cannot recommend the Silent Angel Bonn N8 audio grade switch (or any other switch like it).
-----------
As always, questions, comments, recommendations, etc. are welcome.
My wife made a bunch of masks for the pink panthers. Alas, they are complaining that the colors don't match their complexion! So need to buy more fabric to do that and would appreciate some donations for it using : https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/
From outside, the Bonn N8 looks like the most generic switches:
Look at the cheap labeling underneath:
Company advertises better clock, better power supply filtering and better PCB layout as the improvements. Some generic eye pattern and noise improvement graphs are shown by the company but of course, no graphs showing the improvement out of any audio streamer. Unless you listen to digital bits from Ethernet cable, I am not sure what is behind this common and critical failing.
So let's do what they should have done, i.e. measure the real output of a streamer and see if the switch makes a difference.
Audiophile Ethernet Switch Audio Measurements
Let's agree that you are not going to buy a $400 switch and try to improve the performance of a $40 Raspberry Pi streamer. And that any test with a low-end DAC and streamer will be dismissed out of hand for not having "enough resolution" to show the difference. As with my prior tests of such devices (see EtherRegen review) I will be tested these network devices with my favorite streamer, the Matrix Element i. As a baseline reference, let's use the Roon to push a 1 kHz tone to it through the backbone of my home network which is a 48 port Netgear smart Gigabit switch:
Performance is identical to using USB input on Element i which is reassuring. It means that it has total isolation of the networking subsystem which is usually a very busy subsystem with a processor, operating system and boatload of software to handle streaming.
Note that my switch is in heavy use while the above test is going on so if it were to transmit "noise" that would be impacting the audio, it would be there. Yet we see no evidence of that. The only thing in the output is the harmonic distortions of the 1 kHz tone.
Now let's route the same network link through Bonn N8 and see what we get:
Nothing. Not a thing is out of place other than ordinary tiny run to run variations. Our clock frequency is dead on either way at 1 kHz to five decimal places. Output voltage has not changed. Distortion, noise and harmonics as translated into SINAD remain the same.
An implication is made that if the Ethernet switch generates Ethernet signals with better click (read as lower jitter), that must translate into lower jitter in the DAC. So let's test that theory:
No difference whatsoever. As noted in the slide, due to advanced signal processing, our measurement noise floor is at -155 dB which goes beyond the dynamic range of even 24 bit audio. We see a tiny jitter source at +- 8 kHz relative to our 12 kHz tone which remains there with or without Bonn B8 switch in the loop. There is also a power supply spike to the left and it also remains there independent of the switch in use.
Let's see if we see something in the noise spectrum. I opened the gates by increasing the bandwidth to 96 kHz and measured the residual noise out of Element i with and without Bonn N8 switch in the loop:
Not a thing is different.
In order to give the switch a better chance to make a difference, my son suggested that we create large amount of traffic on the link and see what happens (audio is very low rate as network rates go). Using a traffic generator program, we pounded on UDP port 7 (ping protocol) pushing the network rate as measured by perfom on Windows to above 700 mbit/sec. I then ran the above test again (not shown) and it made absolutely no difference either way.
As another way to test the noise theory, I captured the noise level out of the Element i continuously once a second. First, with no Ethernet cable into it, then my Netgear switch and finally, routed through Bonn N8. Can you tell when I made these changes?
You cannot of course because nothing changes in the output of the DAC inside the Element i streamer.
Listening Tests
I queued up my reference playlist in Roon and streamed them to Ethernet i. Baseline was with the Bonn N8 in the loop. I switched to just the Netgear and it sounded better! More detail. More air. Just higher fidelity. I then reversed the AB sequence, starting with the Netgear and going to Bonn N8. Same difference! More air, more analog like, etc.
What is going on? Simple: you frame of mind makes a huge difference in what you perceive. When performing comparisons, your brain gets into instrumentation mode and analyzes music differently than when you lean back and just enjoy music. In doing so, it hears more detail, air, etc. It does so even when nothing is changed in the waveform. In other words, the Ethernet switch is not changing sound. It is you that is changing.
This is why we test people blind. When your brain doesn't know what is being tested, then its variations factor out (statistically speaking).
Discussion and Conclusions
I honestly go into every one of these reviews with a clear slate, hoping to find a difference. Any difference. Even though I am usually successful in such things as USB cables and such, I have yet to find any measurable difference whatsoever with these "audiophile switches." As it should be. Ethernet is designed for long distance communication (100 meters or 330 feet) so it already has provisions to deal with heavily distorted signals, noise, ground loops, etc. If it did not, there would be hell to pay as our foundation of networking would fall apart.
You can try to clean up Ethernet signals if you wanted and get longer distance propagation from it. But better fidelity in a streamer? The only way that would happen is if the DAC in a streamer was sensitive to such things. And if this were to be the case, then the mere inclusion of streaming functionality in a DAC would destroy its performance. The noise from the CPU and operating system activity is so much more significant than any switch upstream.
Measurements conclusively prove what I explain above. That the Ethernet switch makes zero difference in the performance of a decent streamer. So let's not talk about jitter and noise changing. They absolutely, positively are not. If they were, the manufacturer would be showing that and shouting from the top of the mountain.
So there is really one argument left in favor of such devices: user testimony. Many people buy these switches, immediately notice improvement to the sound and call it good. When pushed on why, the talk about jitter and noise but per above, that is a made up fantasy. And justification after the fact by a person who doesn't understand the system design.
The only thing we need to deal with then is that "it sounds good to my ears." Well, it did that for me too expect that I could make either my standard Netgear switch or Bonn N8 sound better! Clearly that can't be the case. Logic would say that the testing is flawed. That using one's ear in a sighted situation is at fault. Lest you want to agree that both switches can be better than each other at the same time!
This is an uncomfortable situation for audiophiles. Their perception says one thing. Our data and science says the opposite. They chose to believe their perception. If you are hear and think that way, I implore you to have a loved one test you blind. Have then do the test as I did with the switch being in the loop or not (randomly) once a day. Do this for 10 days and you see if you can guess correctly which switch is in play 8 times or more.
My strong conviction is that you will not be able to get anywhere near 8 right.
Conversely, I can take two identical switches, change the label on one, and have people listen. I guarantee you that I can get testimonials in favor of whichever switch I want even though they are identical! Please don't make me do that. It will be embarrassing for many people involved. Do the test I suggest above in your own home with no one looking. Once you see the truth, you can sell the switch to someone else and have money for something worthwhile in audio or otherwise.
Needless to say, I cannot recommend the Silent Angel Bonn N8 audio grade switch (or any other switch like it).
-----------
As always, questions, comments, recommendations, etc. are welcome.
My wife made a bunch of masks for the pink panthers. Alas, they are complaining that the colors don't match their complexion! So need to buy more fabric to do that and would appreciate some donations for it using : https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/
