How Hi-Fi Magazines Write about Cables, Part 24: Ethernet

So here we are again, with yet another magical hardware device to make music sound better. This one is an Ethernet switch; it’s what you use to, for example, connect to a router then connect a bunch of Ethernet cables, with then connect to other devices. They could be computers, wi-fi access points, other switches, or even audio equipment. Some receivers and amplifiers have Ethernet jacks to received digital audio.

First, a brief primer about Ethernet. It’s a technology used to send data over cables, and most networking uses TCP/IP, very robust protocol that has been around for decades, and has been standardized. It is what runs the internet and most networking. A Wikipedia article explains the main element of Ethernet that makes it so robust:

Systems communicating over Ethernet divide a stream of data into shorter pieces called frames. Each frame contains source and destination addresses, and error-checking data so that damaged frames can be detected and discarded; most often, higher-layer protocols trigger retransmission of lost frames.

The bit about error-checking data and retransmission of frames is what’s important. You see, unlike analog data – think of an old TV where you get your signal over the air, and there can be static – each frame, or unit of data, which can be from 64 bytes to 1518 bytes, is checksummed before it is sent, then after it is received. If the checksums don’t match, the frames are resent.

This is very important. Let’s say you’re reading this article on a computer connected via Ethernet to your router. Upstream from there, the router gets data from a series of network devices, from your local exchange, from other switches, finally from the original server sending you data. If data correction is not used, it’s possible that some data gets corrupted during transmission, and you might lose some of it, or some might be garbled. In which case, you would be reading some random sequence of characters instead of what I’ve written; or some of the words and sentences would be missing.

So think about how that works with music. When you send music from a server – be it Apple Music, Spotify, or your own computer – it is sent the same way. Your hardware devices don’t know that it’s music; they only know that it’s frames of data that has to perfectly match what was sent. If not, it is resent. This happens very quickly, and data is buffered to allow for resent frames to catch up and be correctly reassembled. I’m simplifying a bit, but you can be certain that, with working hardware, the data sent is exactly the same as the data received.

Anyway, back to our hardware hawker. According to the venerable What-HiFi:

The Chord Company has relaunched a Great British brand to front its audio electronics business, English Electric, and has demoed its first product – a hi-fi grade network switch.

Beyond the question of whether this product serves any purpose, this company is capitalizing on a recent trend in the UK, among a certain demographic, of wanting “great British brands,” and “blue passports.” (Search the term “gammon” on Google.)

So, what is this device?

English Electric announced the device at the Bristol Hi-Fi Show with claims that the English Electric 8switch can act as a filter for streamed audio to remove unwanted noise which regular computer industry-made switches and routers never consider.

Let’s highlight the magical thinking above:

unwanted noise which regular computer industry-made switches and routers never consider

They are saying that there is “unwanted noise” on network hardware, and that “regular computer industry-made switches and routers never consider.” Let’s think about this. When Netflix streams 4K video, do you think they “never consider” “unwanted noise?” Or that banks, for example, don’t care about noise that could cause data loss or corruption?

Actually, they don’t, because it doesn’t exist. Ethernet, and the protocols it uses, don’t have “noise.” They have 1s and 0s, and error checking, and retransmission, to ensure that the data sent matches the data received. It’s true that you can have cables that malfunction; when that happens, they simply don’t work, because there are too many errors to correct. (You can see this if you ever have a bad HDMI cable; it’s a mess, with lots of artifacts and pixelation.)

But, says the magazine:

The 8switch will come with its own unique power supply and feature better separation with the idea of “making nastier sounding tracks better”.

Now this is something different. Not the bit about “making nastier sounding tracks better,” which is just marketing copy written by someone who didn’t attend much school, but having it’s own “unique power supply” could make a difference to noise; at least the noise that you hear if you put your ear really close to the device. Because while a bad power supply could create a hum, or ground loop, in an amplifier, it will not do anything to an Ethernet switch. If the power works, the data is sent and received. It doesn’t matter if there’s a hum, or if the power is “dirty” as people in the audio fantasyland like to say. If the cables aren’t broken, the data is sent correctly. If the cables are broken, then there’s too many frames that aren’t received correctly, and data won’t flow. It’s not like data comes through looking like this:

Gur 8fjvgpu jvyy pbzr jvgu vgf bja havdhr cbjre fhccyl naq srngher orggre frcnengvba jvgu gur vqrn bs “znxvat anfgvre fbhaqvat genpxf orggre”.

Oh, and it costs £400. You can get an 8-port Ethernet switch for about £21 on Amazon.

But back to the brand. This is, apparently, a truly great British brand:

English Electric was originally founded in 1918 in the armistice of the First World War by amalgamating five companies which had been used in the war effort to manufacture munitions, armaments and aeroplanes. It became defunct in 1968 after some notable successes including the English Electric trains and the supersonic English Electric Lightning jet.

And as the article ends:

Expect to hear from English Electric later in the year.

I shall hold my breath.

13 thoughts on “How Hi-Fi Magazines Write about Cables, Part 24: Ethernet

  1. Being one of the gammon, nothing would please me more than to see consumer electronics manufacturing repatriated to the UK. But I’m also a realist. Even if the English Electric 8switch is assembled in the UK, as the price tag suggests, all of the components, save possibly the casework, will be manufactured in the Far East.

    I agree that a “hi-fi” network switch makes absolutely no sense. Any problems in transmission can be easily corrected by buffering the data on the receiving end. And since the received data must exactly match the sent data, no intermediate transport device such as a switch or router should attempt to change the data. A network device could receive the data, process it, and send out a modified copy, but then it wouldn’t be a switch.

    Incidentally, my network streamer is fed via an 8-port Netgear switch (made in China) which I picked up for less than £30.

    It’s also odd that Chord has decided to resurrect the English Electric brand. My dad worked for English Electric building heavy industrial switchgear after the war. The company was never known for consumer electronics, never mind hi-fi equipment. These days anything overtly English is (wrongly) treated with a certain degree of disdain, so I can’t see the brand gaining much traction, even among the gammon. But then having a network switch nobody needs as your first product does not bode well.

    • There are some British brands that seem popular and respected. Naim, Cambridge Audio (though I don’t know how much they are respected outside the UK; and they get a big boost in the UK because of being owned by Richer Sounds), KEF, and others.

      But, as you say, having a first product being an Ethernet device seems very odd.

  2. Hi, slightly off topic but within the hi-fi ‘improvement’ area. A friend of mine with many years of listening experience has a very good hi-fi set up playing mostly CD’s (he has probably spent around 4-5k, £1500 on speakers alone) told me the other day that he had heard a system demonstrated before; plugged in to a standard 4 gang 13 amp power strip and then after attaching the same system into a high-end Naim power supply. He described the difference in sound like lifting a heavy curtain from in front of the speakers.

    The Naim was an unbelievable price so he looked for a cheaper alternative for himself and settled on a ‘Russ Andrews YelloPower’ at around the £200 mark. In his words ‘the improvement in the overall sound is amazing. The sound is cleaner, with more clarity and the bass stronger without being over powering. Separation of individual instruments is just incredible. A major improvement.’

    This article is sceptical

    https://www.theguardian.com/science/2006/jan/07/badscience.uknews

    however, I certainly know that good quality inter-connects can make a difference and that I have read before how a good power supply can as well but what do you think of that Kirk… or anyone else?

    • It’s not an uncommon belief. But here’s the thing: AC power is converted to DC inside audio equipment. With that in mind, I can’t see how anything in the AC power could have an effect on the sound. Sure, if there’s a ground loop, which can create a hum, that’s one thing, but that’s an all or nothing problem.

      As the article points out, that power has come from miles away, through a large number of cables of different types.

  3. The mumbo-jumbo in that device description is amazing. I am astounded at the lack of understanding of how fundamental technologies work. I remember briefing a major university technical staff, back in the day when the Internet was just getting underway, on how (and why) TCP/IP works. I described TCP/IP as a protocol for inherently unreliable physical networks. My point was that the noise and errors in transmission were compensated for by the protocol. The alarm from the term “inherently unreliable” was visible in the group countenance. Maybe I shouldn’t have used that term? They wanted a highly reliable network, not something “inherently unreliable.” Anyway, the frames are also numbered sequentially because they may or may not be received in the proper order. Each frame travels through the connections independently and they may not all take the same path from source to destination, leaving them out of order. The numbering allows the receiving site to reassemble them in order. The numbering also allows for transmissions to continue even as a given frame is being repeated due to errors. The sender just keeps sending, resending when asked, but not stopping the overall transmission. As a result, the overall time spent sending the data is reduced.

    One quick story on Ethernet: We installed Ethernet in the dorms and classroom buildings, with a vertical backbone connecting branches for each floor. In one classroom building we had a strange problem where the network worked fine for a while, then had a temporary spate of huge error rates and repeated packets, slowing the entire building down. It was relatively predictable in that we would have about 50 minutes of stability, then ten minutes of chaos, then 50 minutes of stability, etc. It turned out that the cable installers decided that instead of drilling holes in the concrete floors for the vertical backbone, as designed, that they would just run the ethernet in the elevator shaft and avoid the work to do it as planned. The electrical noise of the elevator running at breaks between classes was filling the ethernet with errors for that 10 minutes between classes, but when the elevator was idle, the network was fine. We had them come back and do the work over, correctly this time. TCP/IP can only handle so much noise before it gets really slow.

    • It can get slow, but this doesn’t affect the integrity of the data. If you have an elevator in your home, your music won’t sound any different. (Plus, lossless audio is only about 100 KB per second anyway, so it’s not much data.)

      • True, the error rate is not critical, as long as the number of repeated packets isn’t excessive. What the elevator caused was a repeat rate of over 90% of the frames, which bogged down the network to where it was impossible to use in practical terms. The data received was correct, but it took thousands, possibly millions, of retransmissions to get it through in that 10 minutes of chaos.

  4. Thanks for another fun look into audio equipment marketing. I really enjoy these articles, and the ensuing discussions.

  5. I really don’t enjoy these articles although I am a follower of everything else Kirk writes. This one is just another variant of the ‘bits are bits’ argument and a basic lecture on TCP/IP protocol. For me, it is akin to shooting fish in a barrel everytime Kirk picks on audio prose. Yes, it can be awful prose. So what, let’s move on.
    Unlike everything else Kirk writes, he has not actually listened to the device – or any of the audio devices being reviewed in articles he has criticised in the past. Where is the value in pontificating from an armchair? Go out and listen to this stuff, then come back and tell us what you think.

    • Tastes vary, of course. I enjoy the articles, and I always learn something new about audio and/or computer technology. These articles may have a higher entertainment to technical information ratio than many others that Kirk writes, but I like to read them a couple of times a year.

  6. Amazed at how many audiophiles simply relinquish their reason, abandoning all possibility of persuasion or meaningful discussion. Reminds me of bulimia or some such.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.