Review: Heresy with a Nordost Heimdall 2 ethernet cable

It has bothered me a bit and it took some thinking on what to do with the Nordost Heimdall 2 ethernet cables. The problem? They make a huge difference over my Supra Cat8 cables, let alone the generic Cat6a I was using before the Supra ones. And if someone dares to say “ethernet cables make a difference in sound reproduction” he will immediately get labeled as a fool, a crazy, or even worse, an audiophile.

The story goes back a few months, I was asked if I would be interested in “auditioning” some ethernet cables. The answer was, sure, as long as I don’t have to write a rave if they make zero difference over my Cat8 cables.


No such thing as Cat7 or Cat8 ethernet without the use of specialized connectors such as GG45  or TERA. The so-called “audiophile” ethernet cables who utilize plain RJ45 connectors don’t meet such criteria but they still proclaim themselves as Cat7 or 8 because, and this is my best possible explanation, the bandwidth they allow through meets the required specifications which are 600MHz and 2GHz for the 7 and 8 respectively, far above the audio requirements for even the heaviest DSD or DXD files.

Da system

My digital front end consists of a Synology NAS in raid0 configuration, a TP Link Archer router with linear power supply, an Odroid C2 based Linux streamer with linear power supply running Archphile OS and the Rockna Wavedream DAC (the first generation with the MSB Platinum modules) connected to the streamer with the Das Klang Richter USB cable. I normally run 11m of ethernet cable from the router to the streamer divided into two patches, the long one was a custom-built from Nordost as no 10m Heimdall 2 is readily available, then a Baaske medical grade network isolator placed on the last meter takes care of possible noise before entering the streamer’s gigabit input. Same 10 & 1m lengths were also purchased a few years back from a no name brand, then when Supra announced their Cat8 cables I moved on and got them as a “final” solution. Supra is a serious company, making value-for-the-money cables that would guarantee my peace of mind for years to come.

For test purposes, and since I wanted to eliminate as many bias factors as possible, the isolation filter and last meter of cable were taken out of the chain, so what I used was the vanilla plain Cat6a, the Supra Cat8 and the Nordost Heimdall 2, all running the exact same 10m from the TP Link router straight to the Odroid streamer.

As few of you will be familiar with the Odroid C2, this is, to the best of my knowledge, the most sophisticated single-board computer out there, with specs far above the Raspberry Pi 3, Udoo or Cubox. It packs a 1.5Ghz, 64bit, quad core ARM Cortex CPU, 2Gbyte DDR3 SDRAM, the OS runs on a blazing fast eMMC5.0 HS400 Flash Storage and it also has separate bus controllers for the Gigabit ethernet and the USB ports. There are a few available OS systems but my choice since back the days of Rpi2 remains Archphile, a stripped to the bone Archlinux based operating system. I run Archphile with the largest possible buffer size and always use the latest kernel available. The WiFi option is permanently turned off while the entire system, including the streamer, is powered from a dedicated power line coming straight from one of the three phases, used exclusively for that unique line. Not much left in chance.

Just to make sure it wasn’t something to do with my usual hardware additional tests were performed with the Auralic Aries femto clock & linear PSU streamer along with the latest Mytek Brooklyn DAC+. As a back up USB cable I use the dirty cheap Belkin Gold, which remains an excellent entry level proposition but comes nowhere close to my Das Klang Richter.


And no matter how you put it, no matter what the rest of the hardware was, there is a clearly audible difference between the Supra Cat8 and the Nordost Hemdall 2 Cat7 cables.

Trolls, unleash Hell

But give me a second first. The differences are so big in terms of soundstage depth and width, detail and air among the instruments that far exceeded my expectations. I am not sure where to attribute the differences and honestly I don’t even care that much, they are there. I went on and called some friends for a listen, not just some seasoned audiophiles or what some would call “believers”, I also invited my geekiest, DIYers, “bits are bits” guys I could think. They tried to explain the reason for the obvious decline in soundstage dimensions with the Cat8 cable, they inspected the Metz connectors on the Norodst cable and went on telling me that it probably is a matter of noise, grounding and shielding. Going through the net I stumbled upon a simple but well written page on eBay, where the author explains what are the fundamental differences between the various ethernet cables. The 6a must have grounded connectors, the 7 are fully shielded and each of the shielding layers must be grounded.

The other point that got me thinking is the fact that higher specs guarantee transmission for greater lengths, as ethernet cables are prone in picking interference.

For category 6a cables running at 10 Gbps speeds, 55 meters is the maximum allowable length, and even this length is only allowed in very good alien crosstalk conditions, or areas of low interference, such as when the cable is located far away from other cables that could cause interference

Could that be a logical explanation on why such differences exist? Could it be attributed to the thicker solid copper conductors vs the stranded and thinner ones of the Supra? Again, I don’t give a hoot. Not to mention that on paper the Supra cables rated @Cat8 are superior to the Nordost Cat7. Unfortunately, that is true only on paper, sound-wise there is an abyss between them. The price difference is also significant, the first meter of the Nordost comes at $700 while the Supra sells for approximately $50/m.

A heresy? Probably. A huge step up for my system? Most definitely so.

Now its time to face the music Hell. Feel free to bring it … as long as you have a first-hand experience.

About Panagiotis Karavitis 212 Articles
Doctor and Editor @ Part-Time Audiophile Publisher @


  1. Well – trying to be as civil as possible while expressing my point.

    Any functioning, up to standard ethernet cable is an expression of solid science and engineering principles with the purpose of transmitting error free *data* from A to B, without the nature of the data mattering at all. Whether music or a patient’s vital stats, a functioning up to standard cable will deliver the exact data over to B. Given that A== B, the interpretation at the target (B) will be the same. Whether is audio or anything else. The science behind this is not a belief; we know based on current science what is the accepted true of the behaviour of the system. The nature of science allows for it to be challenged and updated based, of course on solid scientific evidence – which is the opposite of anecdotal. So far that hasn’t happened on this subject. It is fine that you hear a difference. It cannot be accepted as evidence that there is a difference.

    That said I think there is a bigger issue in play that is exposed when expensive “ethernet cables for audio” are discussed. What are we reading? Is it journalism? is it punditry? is it an opinion? Is a journalistic institution? or is it industry advocacy? There are standards that define the behaviour of journalistic institutions. There is a tone of impartiality (whether true of not) that defines them. With that I am willing to go with industry advocacy. The angry and sarcastic tone of replies from “Editorial Staff” to skeptics, to me is evidence of that. Is that wrong? Of course not. That’s what “trade publications” do – industry advocacy is their reason for being. In the days before the internet it was a closed environment – trade publications catered to the involved and the interested, with little or interaction with the rest of the world.

    But the internet changed that – when audiophile trade publications went to the internet they crashed head on with the rest of the world. And yes, that included scientists/engineers that challenged claims on which the industry is based.

    I do commend parttimeaudiophile – all the contrary comments seemed to be here, to me doesn’t look like they are censored – good – compare that to the heavy censorship over at the Stereophile sites/blogs. At least the discussion here is lively.

    That is the great part. The arguing as long as it is civil is a lot of fun.


    • We like free expression as long as it stays within acceptable levels (and this wasn’t the case for all approved comments on this specific article, still 100% of those comments were approved).

      And I knew that such a review would generate a ton of reactions, the only reason I went on and wrote it was that the differences between Ethernet cables were so obvious.

  2. Fraud is a strong accusation. Out of bounds actually. If PTA has one thing it is integrity. Yes $700 for a Ethernet IMHO is excessive but no one is forcing you to buy anything.
    What I want to know is if cable hs a colored sound or it is neutral.
    Give me neutral as I want my components to shine through not the cable.

      • The plural of anecdotal is not data. Now if you are open to bias controlled testing then we are getting somewhere.

        Ethernet is a data cable. As an example current Naim streamers can cache ~2 1/2 minutes of audio where you can pull the plug and music still plays. Tidal will cache the entire track on Windows.

        So the easy test would be to have someone remove the cable while playing back and see if anyone can hear that.

        Basically, I have $5000 to anyones $1000 when their sighted bias is removed from the equation that they are going to fail to hear the difference. I can always provide full details to any interested parties.

      • The only proof I need is to listen to my system which I know very well. Blind test , this test, now wagering, ENOUGH already.
        cables ALL can sound different. Belden can actually measure why what design sounds the way it does.
        If all cabling was $20 this conversation would be missing.

      • I never trust anyone that is scared of the proverbial blind fold. If they don’t trust their ears neither do I.

  3. Is there anyone in this industry that can sit down with a $12 certified cable, and a boutique cable, both with the same type of general assembly, and do any of this blind?

    I’m finding this lack of intellectual honesty disappointing.

    • “Lack of intellectual honesty” is not the same thing as “blind testing”. “Blind testing” is neither necessary nor sufficient to intellectual honesty — or to “an aesthetic judgment” or “rendering of personal experience”.

      You’re projecting.

      You want extra? Feel free to go ahead and do it. Absolutely nothing is standing between you and your quest for truth except, perhaps, your keyboard.

      • Bottom line is that Ethernet is a data standard, not an audio standard and if you haven’t validated your ears, then I have to take it with a grain of salt.

        Bottom line is I simply don’t place any of this on faith. Trust but verify.

        We will have to agree to disagree on what each of us considers intellectually honest.

      • “We will have to agree to disagree on what each of us considers intellectually honest.”

        What you’re really arguing for is your own logic. And for that, I can safely offer this: NOPE. Big fat truckload of NOPE.

        We all get to play with the same puzzle pieces. That’s the game! But if you insist on relativism, it’s worth pointing out that this is a non-starter. No-play for you. Especially when you enter the room mouthing off about the “proper use and application of epistemic tools”. Relativists don’t get to have opinions about success criteria (or about anything else), much less get to pass value judgments about “who’s doing it right or wrong”.

        And FFS, since when did ANYONE suggest you take ANY OF THIS on faith? The billboard outside, the table with fliers by the door, the dude standing there attempting to tell you something — every one of them are emblazoned with the same damn thing: “CAVEAT EMPTOR”. All we have ever done is relate experiences. Not liner notes. Not spec sheets. Experiences. Will yours differ? Probably. Which is why we aren’t expecting you to blindly trust us (unlike what the haters on this thread are asking us to do — which is disregard our personal experience in favor of blindly trusting their faith in their scientism over our empiricism).


        I am truly baffled why this very simple prescription seems so terrifically difficult to grasp.

        If you don’t like the result, DON’T BUY IT. If you don’t find that you get a result worthy of the price tag, DON’T BUY IT. If you don’t find that it produces a result AT ALL … Congratulations! You just saved yourself some money! Hooray!

      • What I’m debating is that I understand how data works. If I create a 1/2 GB RAM disk and copy an audio track with a $700 cable and a $12 cable I’m guaranteed a few things:

        1. I’ll get the same MD5 hashed file
        2. I’ll get the same SQ/imaging/layering/depth of field

        Mathematics don’t care about your or my relativism.

        I’m willing to put my $$ where my mouth is. This is the way and conditions that I’m stating that I’m willing to be proven wrong about my position.

        If the differences are that huge for you then my offer makes logical sense.

  4. We sell DAC’s and transports. Do not sell cables. When we audition our wares we want such to sound its best.
    Why ? To make the sale.
    If cables made no difference in sound for an audition it would be easy.
    However they do and our customers here such.
    Measure away but remember it is the end result of sonic achievement of the cable.
    Stop the name calling and childish remarks. Thank you.


  5. I empathize for the poor souls that are duped by this farce. It’s made to prey on people who assume Ethernet devices and cables as a system, have the same RF noise susceptibility as the analog audio hardware and cables systems of yesteryear…

    Gigabit Ethernet, as defined by IEEE 802.3 specification, is made to work error-free, with Category 5 or better cable, such as Cat 5e, Cat 6, and Cat 6a, at distances up to 100m. Ethernet equipment manufacturers routinely test their hardware to verify it works error-free in the worst case scenarios, such as with 100m of cheap cable (at least the one I work for does).

    If you have the money, and want to buy cool looking, or super durable cables, more power to you. Spec-wise, I could see getting up to Cat 6a cables, for future-proofing compatibility with 10Gbps Ethernet – but don’t think that’s going to get you better sound when using 1Gb Ethernet hardware.

    When purchasing Ethernet cable beyond Cat 6a, you’re just buying marketing fluff that no current or past Ethernet equipment is designed to work with.

    -Advice from an sound geek, and Electrical Engineer, who has been testing Ethernet equipment for the last 2 decades.

    • Ha ha ha ha ha!!!! Sound geek and EE? Ha ha ha ha ha!!! Testing Ethernet equipment for 2 decades? Ha ha ha ha ha!!!

  6. I’ve had the AQ Vodka 1.5 meter, the WW Starlight 3 meter, the Nordost Heimdall II 1.5 meter all compared to UTP CAT5e at 98 meters.

    One system a Cary Audio DMS-500 and with a Intel 3450 based system acting as a server, and the other client another Intel 3450 based system (with Intel based HP dual RJE server adapter that was $18) feeding an Emotiva DC-1

    The switch is a CISCO SG 200-8 with Ports 7 and 8 in a LAG using LACP.

    Using Tidal or JRiver you can swap between cables without interrupting the playback (this is with the WW):

    Here’s the offer: $5000 to your $1000. In your setup from the USB cable on back. I provide the client, the server, the switch, J River. You provide the entire stack USB cable on back, your music, your room, your system.

    It will be setup so that I can switch cables out of the Cisco SG-200 w/o you being sighted to it. We do two rounds of 10 each and you have to hit 9 out of 10. You can have up to two hours of you listening sighted, swapping out cables etc. Then we will move onto ears only evaluation.

    Loser pays travel costs.

    I’ll make a CAT5E UTP on the spot.

    • Nine out of ten times? Hell, most people couldn’t differentiate Coke and Pepsi nine out of ten. Does that mean there is no difference between the two?

      • The issue is that we are testing claims. The claims in this review represent a large delta of change. Small, granular, changes aren’t being bandied about:

        “there is a clearly audible difference between the Supra Cat8 and the Nordost Hemdall 2 Cat7 cables”

        This should survive bias controlled evaluation. It there is a ‘clearly audible difference’ then I would encourage someone to pony up their $1K to my $5K.

  7. Any actual diff is going to be RF noise (or lack of) entering via the cable. Try going to fiber… and have a very short run of copper as you don’t have a fiber input, though that final dual fiber/copper switch itself might be the source of RF as Ethernet switches are not designed for RF isolation.

  8. Wow you are a tough crowd. I use an Ethernet Rj45 to connect the La Scala transport and Formula DAC from Aqua Acoustic.
    We have noticed a difference in tonality and harmonics using different brand cables. Period. Everything can make a difference good or bad.
    Now I will agree $700 is steep. Actually unfair pricing IMHO as a dealer/distributor. It appears it is the construction method used not the conductor. Grounding of each wire is good engineering practice.
    find me a cable constructed the same way for less and we will try it.
    Thanks for an honest review. NO BS except pricing.


  9. I’ll bite – although I don’t comprehend the stated magnitude of the results. I have never gotten any major gains when I have tried cables, although I have never tried a cable over $200 – for what that is worth.

    1) If you have a modem that feeds Cat 5 into the wall that traverses the house and exits near the audio stack into an Accesspoint and/or Switch, would this eliminate any improvements?

    2) What would be the impact of only looking at the connections from the switch into the various source and end-point devices?

    3) Wouldn’t using Wi-Fi or a Network isolator completely bypass any network cable-related issues?

    • Two items to keep in mind are external (induced noise) and internal coupled noise.

      For the latter T.I’s paper “Reducing Radiated Emissions of 10/100 LAN applications” talks about the real Achilles heal of noise that can disrupt data cabling and that is the single ended nature of power supplies. For that you can simply run WiFi as you have pointed out and be done with it.

      For the former Siemons “The Antenna Myth” correctly points out that un-shielded CAT6 is noise immune (yes immune) to 30 MEGA hertz. That using CMNR, 50 ohm terminations at the jack, that it’s going to effectively deal with any noise we could possibly hear.

      I’ve seen two AudioQuest Vodka’s now unable to cleanly pass the 6A spec. This is a cable that AQ says is a CAT 7 cable, but used Telegartner 6A industrial terminations with hard to obtain NeXT that is clearly in the 6A spec let alone the 7 spec.

      While you can get cabling that costs 2000% less that easily passes.

      I’ve benched the WW Starlight 8, AQ Vodka, Nordost Heimdall II against generic CAT5e. Both listening and instrumented measurement show no difference.

      The fact that reviewers don’t understand the significance of the fact that well designed systems have a large reservoir and can play entire tracks without the cable even being plugged in should be a huge clue. With GBe an entire 16/44.1 track can be cached in 1/2 second.

      If the music still plays with out the cable plugged in then all these reviews are basically on shaky ground at best.

  10. Spending $700 on a meter of Ethernet? That is beyond crazy. You are completely out of your mind. It is technically, scientific and by any definition of common sense absolutely impossible that an Ethernet cable has any influence on the way streamed audio sounds. Unless you have a massive amount of packet loss, which will be audible as pauses, loud hisses and pops, in other words something everyone, not just self-declared audiophiles can hear, the digital stream will be 100% identical.

  11. Any Ethernet cable that produces error free transmission will stream the same data, bit for bit, and therefore, also stream digitized sound, bit for bit. Doesn’t matter if you have cat5e, 6a, or better on your gigabit Ethernet LAN. If the data is the same, so is the sound.

    Don’t believe me? Run iperf overnight and check your interface statistics for any kind of errors, with numerous cables. Even [most] cheap cables run error free for several days at minimum.

  12. I hope you are making money on the ads. Everyone needs a hobby. When you find someone smart enough to appreciate this article, send me an email. I have some prime real estate in Florida that I would be willing to sacrifice at a very attractive price. It has outstanding audio characteristics. No traffic noise, sound deadening provided by live vegetation, with gentle water sounds and insect buzzing noises.

    • We are all billionaires because of the huge piles sweet dough rolling in from our participation in the ongoing campaign to suck your collective brains through a stirrer-straw.

  13. Hahaha! I have just bought various lengths of Supra Cat8… Thank you so much 🤪

  14. What a joke! This is why the hobby is dead and laughable. $700 for an ethernet cable that isn’t even as good as the $50/m…yet this clown swears it’s better.

    First of all, did he buy it? What price did he pay? Is Nordost as advertiser on this site? Was the test done blind?

    Simple questions, but I bet all the answers expose this fraud.

Comments are closed.