News LAN iSilencer Audiophile Ethernet Dongle Claims to 'Quiet Your Network'

[InvalidError]

When you go to STP the ideal is that the shield acts like an antenna and shunts it to ground. Again because the ground shunts the signal any float up is negated as long as you have a good ground. Any ripple on the ground just becomes common mode noise again which doesn't really do anything.

When you shunt common-mode noise from external cables in your PC's ground planes, it can interact with everything else and get modulated into audible noise. Getting rid of GSM noise was notoriously difficult 15 years ago.

 

[bit_user]

Even if a problem occurs due to some rough solar flair or local EMP blast (that doesn't somehow blow up your computer) the TCP stack takes care of it automatically and any UDP implementations worth their salt will at least run some sort of CRC and dump the packet.

Ethernet has a 32-bit CRC per frame (usually limited to 1500 bytes, but can be up to 9kB).

https://en.wikipedia.org/wiki/Ethernet_frame

Errored frames are simply dropped. This helps protects your link from things like bad cabling.

For errors occuring beyond that link, you're reliant upon the next layer up. In most cases, that's IP. IPv4 has only a 16-bit header checksum (although IPv6 has none).

https://en.wikipedia.org/wiki/Internet_Protocol_version_4

Above that, you usually have TCP, which adds a 16-bit checksum over header + data. Note that IPv4 limits a TCP packet to 64 kB, but TCP over IPv6 has a 32-bit length field and still just a 16-bit checksum.

Transmission Control Protocol - Wikipedia

en.wikipedia.org

Not all internet traffic uses TCP. Realtime communication, like games and video conferencing, tends to use UDP. It has a 16-bit checksum over header + data. UDP packet length are limited by the path-MTU, which is limited by the (usually 1500 byte) Ethernet frame, in practice.

https://en.wikipedia.org/wiki/User_Datagram_Protocol

 

[kjfatl]

I have seen a few situations where this sort of device might actually be useful. When running Ethernet cables between floors in a multi-story office building, or between buildings on different power grids, noise can terrible, particularly if shielded cables are being used. Of-course, the best solution in this environment is an optical link between floors since wired ethernet would hardly be working at all. With an optical link, you not only get better audio, but your data runs 10X faster at the same ethernet speed. (removing 90% of the re-tries)

 

[bit_user]

This is clearly a passive device because Ethernet endpoints (as opposed to network switches) never provide power. The photo shows a 24-pin surface mount package with no other components visible.

I wondered about that. The description mentions something about a "zero-jitter" digital buffer, however. There are also LEDs. Finally, the photo doesn't claim to show the guts of the entire device.

 

[InvalidError]

Errored frames are simply dropped. This helps protects your link from things like bad cabling.

I don't think I would call dropping packets 'protection' against anything besides wasting network bandwidth pushing known-bad data through additional hops only to get dropped elsewhere and possibly forwarding it to the wrong place altogether if it was address bits that got flipped.

 

[lxh89]

Japan’s iFi-Audio has launched a LAN iSilencer dongle to remove noise from your Ethernet music streaming. The firm seems to have missed a trick by charging under $1,000.

LAN iSilencer Audiophile Ethernet Dongle Claims to 'Quiet Your Network' : Read more

For when those 1s and 0s are a little blurry and the DAC gets confused. If this thing sells well, I'll be introducing a better solution using gold contacts which ensures blurry 1s and 0s are impossible. My device will only cost $999.99 so still under the golden $1,000 fee. I'll also include a special buy one get the 2nd one for 50% off.

Regards,
Tony

 

[InvalidError]

I wondered about that. The description mentions something about a "zero-jitter" digital buffer, however. There are also LEDs. Finally, the photo doesn't claim to show the guts of the entire device.

The only thing we see top-side looks like a typical quad transformer package and Pulse is one such device manufacturer too. At least this part of the design looks legit.

I don't think this thing has any active electronics on it. The "LEDs" you are seeing might only be light guides carrying the light coming from jack-integrated lights on either side of the retention tab as is often the case from whatever this gadget is plugged into when present: if you look at the full-sized photo, it looks like it has two round ports next to its plug's tab, which I'm guessing are the light pipes' receivers.

 

[bit_user]

I don't think I would call dropping packets 'protection' against anything besides wasting network bandwidth pushing known-bad data through additional hops only to get dropped elsewhere and possibly forwarding it to the wrong place altogether if it was address bits that got flipped.

You need only compare the relative sizes of Ethernet's CRC and the UDP & TCP checksums to appreciate the role that link-level checksums play in ensuring data integrity. If all they cared about was protecting the network segment from rogue frames getting sent to the wrong MAC, they would've just made it a header checksum.

I've been out of the comms industry for quite a while, but the few link-level protocols I know all have checksums.

 

[kjfatl]

I wondered about that. The description mentions something about a "zero-jitter" digital buffer, however. There are also LEDs. Finally, the photo doesn't claim to show the guts of the entire device.

There is a thing called "Power Over Ethernet" that does provide power over the ethernet connection. Its commonly used for things like office telephones. POE has absolutely nothing to do with this device.

 

[InvalidError]

You need only compare the relative sizes of Ethernet's CRC and the UDP & TCP checksums to appreciate the role that link-level checksums play in ensuring data integrity. If all they cared about was protecting the network segment from rogue frames getting sent to the wrong MAC, they would've just made it a header checksum.

I first mentioned the general case of not pushing known-bad data first though. Seems like you latched on the second part about the special case for wrong address. IPv6 does have header-only checksum, albeit primarily to avoid having to re-compute the entire packet checksum whenever things like TTL are updated in headers or addresses are being rewritten by firewalls for NAT or other reasons.

I've been out of the comms industry for quite a while, but the few link-level protocols I know all have checksums.

RS232 doesn't have checksum... at least not a mandatory one if we count parity as such.

In the telecomm space, I don't remember there being checksums or even parity bits in ISDN through Tx - the networks were designed to carry primarily voice and worst case on virtual channels so there is no routing data to propagate once a connection is set up to mess things up, flipped bits would only cause a slight pop when they land on MSBs. Dropping entire frames worth of voice connections due to one bad bit somewhere isn't exactly an option

 

[bill001g]

Most ethernet ports are already optically isolated from the equipment they are in. They then use small magnetic transformers to send the data over the wire. So there are already 2 levels of isolation built into the ethernet.

Again products for audio guys who think analog rather than digital. You have multiple software levels that can detect and correct or re transmit data before it is even presented to the end user.

 

[samopa]

I reckon building your own hi-fi amp got a lot easier when integrated OpAmps became a thing.

Integrated OpAmps already a thing. 😊
Since high school (40+ years ago), I already used 741 in many applications. Heck, even integrated PowerAmps is already a thing with STK series IC. 😉

 

[digitalgriffin]

I reckon building your own hi-fi amp got a lot easier when integrated OpAmps became a thing.

Lol. I'm not that old!

Regular opamps are noisy as the devil and can lead to easy runaway / oscillation with the negative feedback loop.. I built class a (using nos Russian triodes) and ab amplifiers. Those days are gone though. Finding those old tubes is getting hard.

 

[Murissokah]

People claiming this doesn't work clearly lack the gold plated optical cables to go along with it.

 

[goriot99]

A dictionary definition of ”trashing one’s reputation”. Well done, Tom‘s Hardware.

Signed, every other hardware site on the Internet.

 

[bill001g]

A dictionary definition of ”trashing one’s reputation”. Well done, Tom‘s Hardware.

Signed, every other hardware site on the Internet.

They have posted a number of these type articles lately I suspect it is more of a joke article to allow us forum users to talk about the crazy stuff some people buy.

 

[eriksquires]

LAN Isolators pre-date audio and in very rare cases are required in patient care settings, there is actually a standard: IEC 60601. They provide (I think) around 4kV of additional galvanic isolation than what the Ethernet devices themselves have. The one other use for these devices is, according to an online EE magazine I read, better surge protection inside a building than shunt-type devices. I would definitely recommend one between say your outside fiber adapter and your inside router or if your Ethernet run is 30' or longer in a lightning prone area.

Per the article, shunt-type Ethernet surge protectors create a ground path at 100V or less and that can cause a fire on the connected Ethernet cable. Less equipment gets damaged by isolation, and forcing the lightning surge to find a path upstream than encouraging a path downstream. Overall the cases of lightning entering via Ethernet remains rare, so the evidence is still coming in, but the idea that enabling the surge current is at the root cause of downstream failures makes sense to me.

Also, after I installed mine Michelle Yeoh won an Oscar, and I don't think that's coincidence.

Tripp Lite and Everstar make units they claim meets spec. I have no idea if the iFi unit has been certified or not.

 

[JTWrenn]

Cheaper bs is still bs.

 

[JTWrenn]

Oh and ps, headlines that don't completely trash this...sell this. Stop giving these things airtime unless you are going to come down really hard on them being full of crap.

 

[dave.rara66]

Silly, cash-grab product. Reminds me of heavy gauge, "one way" RCA Monster Audio cables from the 80's. Total BS.

 

[Deleted member 431422]

I hate snake oil like this. It be one thing if these products actually worked... heck I'd get one even. But testing again and again on similar products show they didn't help things and were more likely to degrade your audio than clean it up.

I don't know how it can help if audio is sent as digital signal. Perhaps with signal integrity, but definately not audio quality.

 

[wr3zzz]

Is it just me thought iFi is pronounced "iffy"? I almost checked the calendar to see if it is 4/1 and the name is the clue.

 

[InvalidError]

I don't know how it can help if audio is sent as digital signal. Perhaps with signal integrity, but definately not audio quality.

A standard Ethernet port will shunt cable ground and the line-facing transformer center tap to ground, which will dump all the common-noise that may be on the cable into your PC's ground. If this dongle breaks the ground connection, then it would block that common-mode noise from getting in your PC and reduce the overall amount of external electric noise getting in your PC by whatever amount that is.

As I wrote earlier, it has at least a tiny bit of probable merit.

 

[bill001g]

A standard Ethernet port will shunt cable ground and the line-facing transformer center tap to ground, which will dump all the common-noise that may be on the cable into your PC's ground. If this dongle breaks the ground connection, then it would block that common-mode noise from getting in your PC and reduce the overall amount of external electric noise getting in your PC by whatever amount that is.

As I wrote earlier, it has at least a tiny bit of probable merit.

There is no connection to ground on ether end of a ethernet cable, unless you are running shielded cable or if the port supports POE. This is done intentionally to prevent any mismatch of grounds on ether end of the ethernet which could cause unwanted current to flow between the 2 locations. Each pair of wires in a ethernet port is on a small transformer output.

 

[InvalidError]

There is no connection to ground on ether end of a ethernet cable, unless you are running shielded cable or if the port supports POE. This is done intentionally to prevent any mismatch of grounds on ether end of the ethernet which could cause unwanted current to flow between the 2 locations. Each pair of wires in a ethernet port is on a small transformer output.

Cable runs need to be referenced to ground in some way, be it through a high-value resistor, TVS diode, MOV or other at at least one end to prevent charge buildup from eventually exceeding the isolation voltage.