Would it be an advantage to have SFP ports for all Antipodes servers?
1 Like
While SFP ports on servers are not very common, can they offer significant benefits in certain scenarios?
@matthias maybe you should clarify what specific use cases or environments you are referring to? This will help in understanding how many users might benefit from SFP ports and what the specific advantages could be.
I think it might be beneficial to use Direct Attach Cables:
I understand your perspective and share your interest in that discussion. It seems that a conversion to analog (and back to digital) with a 1 or 2 meter DAC cable in the last leg of a fiber connection to a server can significantly enhance sound quality, but this improvement is only achievable with specific devices designed around an SFP cage.
This suggests an improvement relative to an existing fiber connection. As far as I understand, no comparisons have been made with ābetterā Ethernet connections.
Could it be itās too early to draw conclusions or to generalise? Antipodes developed with an Ethernet connection to a server for a reason.
But I do believe we should continue to monitor this development.
At least for another great server manufacturer (Taiko) the SQ ranking is:
DAC > Ethernet RJ45 > Fiber.
I admit there might be grounding issues as there is no galvanic isolation between devices with DAC cables, but DAC cables are not very expensive so worth a try.
And those who want to spend more money can go the audiophile route:
Again, as with the Signature Rendu the Taiko example vividly demonstrates the distinctive nature of this particular connection. Only when using a DAC cable between a high-end Taiko-designed switch and their specific network card for the Extreme server Emile ranks it as the top choice. However, upstream from this setup, the importance of the above cable hierarchy diminishes significantly.
I hoped that an inexpensive $50 DAC cable could work wonders in any last leg with SFP cages, but unfortunately, it seems to be more complicated than that 
I canāt speak for Antipodes of course, but since Taiko started using a standard and affordable DAC cable for high-end audio two years ago, Iāve been imagining how this is possible and how well this cable would perform in my setup.
@PaulW is correct to point out that context matters here. There is no benefit from using a DAC cable to connect a streamer to a noisy switch. But, there could be benefit in using a DAC cable over other options when connecting to low noise switch. The key consideration is the amount of noise that fiber SFPs generate. With the Taiko switch, these are far noisier so it would be a mistake to use them. And when connecting the B side of the ER to a streamer that already has an SFP input, a DAC cable can be a better choice as well.
Getting the best from a system can be better achieved through considering things end to end than by just chasing the latest buzzword.
Yes, my point is that DAC cables might be superior in some configurations with Antipodes servers as well BUT you can try this only when SFP ports are available. That means Antipodes could offer RJ45 and SFP for comparison.
Any brand could pursue this, but I believe stronger arguments are needed beyond the few specific examples mentioned before they invest significantly in R&D and redesigning the boards.
Although the DA conversion must occur within the server, in the last part of your fibre chain you could try connecting a DAC cable to something like a Sonore optical module, linked with a very short Ethernet cable to your Antipodes server. It might reveal some benefits. Why not give it a try and see what you discover (and share please)?
The topic of this thread is SFP ports not fiber and like @NickBacon I am not very fond of fiber at all.
Now more and more routers, switches and servers offer SFP ports.
So my question is directed to both Antipodes users and Antipodes as manufacturer.
According to the feedback from others it might be an uplift in SQ to connect the server via a DAC cable, for example:
https://the-ear.net/review-hardware/melco-c1-d20-sfp-a-direct-connection/
Superdad recommendation to try DAC was in relation to EtherRegen because it has isolated circuits that donāt pass on noise from Side B to A.
It seems the Pros and Cons are:
-
The potential benefit is DAC is less circuitry (no PHY) imprinting noise/jitter.
-
a DAC cable will probably pass on noise from upstream unless the downstream isolates it like in EtherRegen ā¦ or somehow filters it.
-
a DAC cable will probably collect noise (EMI etc) enroute unless it has something noise repelling and/or the downstream isolates it like in EtherRegen ā¦ or somehow filters it.
-
fibre doesnāt pass on or collect noise, but the SFP module circuits generate a bit (perhaps less with SFP+/10G and upwards
-
āwiredā ethernet blocks some noise due to isolation transformers, but passes on some and also collects some enroute and passes that on too unless it has something noise repelling/absorbing and/or the downstream isolates it like in EtherRegen ā¦ or somehow filters it.
Iād guess its hard to predict which has least noise and jitter consequences because it depends on the devices being connected, the DAC cable and what SFP modules are being compared with.
If Antipodes was able to include an ethernet isolator/reclocker like it does for USB in some models that would probably be very good if its circuits generate less noise than incoming ethernet. Instead of this the next best thing is a switch with lowest possible jitter and noise output and the shortest available cable with highest noise rejection/absorption possible.
Alternatively a well implemented and isolated wifi that is lower noise and jitter than wired or fibre.
1 Like
I like this alternative, do you have any recommendations?
In short, no. And thereās probably not a one size fits all solution but moreso a more cost effective solution. Some explanation ā¦
We are talking here about an alternative to connect ethernet network to a server, and with wifi that means no physical connection - I dont have experience with this application although have some experience improving the connection to my server (and I am not an engineer in this stuff,)
My CX Oladra is connected to my network via ā¦ fibre > EtherRegen Side A > wired cable > CX. That is quite isolated, but could be improved if say I was able to get the server connected via a EtherRegen Side B. (which I will try someday when time permits, I have a 2nd ER waiting). I have also observed improvements when eliminating noise from the server (both the current CX and previously EX). I use ānoise stopperā caps in all unused ports, a Telos MacroQ USB and grounding. This suggests that disconnecting a potential noise source (eg. wired ethernet) from the server would be beneficial.
With wifi ā¦ I have experienced products were the wifi is done really well. Firstly Auralic Aries G1 and now I use Devialet Pro, but both of these are players rather than servers.
The Side B of the ER (mentioned above) goes to a WAP for dedicated connection to Devialet. Wifi beat $1000s worth of wired connection cables and switches hands down last time I tried, and now the WAP is powered by battery and exceeds the performance of a good LPS powered via excellent power cord and conditioner (the AC power source and LPS could have been improved, for more $$).
So this is not to say wired canāt outperform wifi, but Iād guess the wired option would involve $1000s more to get close.
But it follows that wifi could be a very good solution if someone could engineer a very low noise wifi module (ideally with a decent phase noise clock), and route its ground directly to a ground point (not via the ground plane) and isolate its data signal from the server motherboard like done with the USB reclocker. Probably not terribly difficult.
In the absense of that, it might be beneficial to use this ā¦
network > WAP > āwifiā > WAP > cable > server
I use a Mikrotik mAP lite because its low power and low output, and power it with a USB power bank (with just 2 ports) which has an Akiko USB stick in the extra port. The power cable from the battery to the mAP is wrapped around an Akiko Universal tuning stick right up to the mAP and there is a Shakti on-line adjacent to both the power and RJ45 inputs. All this sits on a Synergistic Research Tranquility POD. Incidentally that whole wifi set up is connected to Side B of a EtherRegen via a Shunyata Sigma which does a very good job of reducing noise it collects/transmits - so the whole wifi set up is isolated like āfloatingā!
Basically I am accepting a probably diminished clock performance of the mAP to eliminate wired connection the the noise it might (um, probably does) carry into the Devialet.
In the case of a server and doing network > WAP > āwifiā > WAP > cable > server, the degradation from 2 x mAPs might have worse effect than the noise of a wired connection depending on how good that wired connection is. Further the mAPs might struggle with the data throughput either causing more heat (and noise) and/or drop-outs.
But that doesnāt address the direct connection inside K30, K50 and Oladra between the server and player boards which might benefit from isolation also. That connection seems to be using a pair of something very similar to this ā¦
Iād guess the connection to the player would yield greater bang for buck in efforts to isolate/improve it.
1 Like
Ah, but thatās the challenge! Personally I find a good switch and a copper Ethernet cable connected to my Antipodes provides divine sound! Recently a friend and I spent an afternoon comparing six ethernet cables from the cheap generic Cat6 up to audiophile ones costing thousands. As it happens the most expensive one sounded worst to our ears. What does worst mean? To me and my friend it meant that the music with the worst cable acquired noise artefacts that coloured the music and seemed to over emphasise percussion or plucked notes to the point of making it very difficult to hear some of the delicate notes that we knew were there from listening with some of the other cables. As an aside, that audiophile cable (yes I took it apart to have a look) ignored all the accepted methods of noise mitigation in network cabling and looked more like an RF aerial to me than a noise reducing ethernet cable.
This thread is an interesting discussion but I am always interested in what exactly is being heard that enables one to say that one network connection or collection of devices is better than an alternative. Can I ask what you were hearing that you liked with the wifi you mention above and also with the other options you discuss in your post? I ask because elsewhere there have been discussions about different switches and fundamentally I disagreed that what they were hearing was actually better because what they were describing seemed to me to be more like noise artefact effects on the music.
Indeed, but one that it appears few have taken on.
We have crossed paths on WBF and it seems my listening experience is similar to yours, although I donāt have the calibre of gear you do. Since I chose to prefer wifi as I outlined above years have passed and I have learned a lot about optimising ethernet and how the impacts of ānoiseā sound in music, but had a quantum of experience up to that point. It is too long ago to remember what I was hearing then but it was so obvious it was a revelation and contrary to the $1000s of network gear Iād accumulated to that point.
I have been ātrickedā by things sounding higher res and more detailed due to noise and have had many noise reduction experiences both in the ethernet and elsewhere in the system to understand it better. Aside from the few technical tell-tales that can be focussed on, I have tried to translate how noise reduction influences how the musical performance is conveyed as well as how much more real the instruments (and venue) sound. Here is something I wrote previously that explains further.
What I mean by improvement
There are various tell tales I hear that are not technical or analytical listening. These help steer me. My main objective is it means for the music which the following sets out roughly from least to highest importance:
- The soundfield (venue or virtual space) is more defined
- The placement of instruments in the soundfield is more defined
- The sound of instruments is more real and natural
- The distinction between instruments is improved
- The performance of individual musicians is more appreciable
- The interaction between music parts and instruments is more appreciable
- The cohesion of ensembles is improved
- The expression of the entire composition is more appreciable.
Based in these, there is now a significant improvement from the situation before the beginning of this post.
I have found definition and resolution of low frequencies improves when my network is improved. Typically, overall the sound is less āloudā or could be described as leaner, but the notes sound stronger, more weight and impact. Another way of putting it is the individual notes and instruments become more distinct and defined, and in between these notes and sounds there is more āspaceā (or quiet) and less sound. The individual notes of electric bass or synth are more separated, kick drums more impact with more distinct overtones and decay of drum heads. The articulation and timing of bass music notes (and time/space between the notes) improves which leads to more rhythmical expression.
From that foundation other instruments are portrayed with more contrast and distinction in terms of sound character and placement in the soundfield. As an analogy, it is like bit by bit revealing and lifting a discovery like treasure from the ground, cleaning off the dirt and illuminating it. There is far more purity and ārealā texture, for instance in vocals which I find to have natural texture that can easily be blurred by distortion in reproduction to sound like distortion. Other examples are wind instruments (eg. saxophone), string instruments (eg. violin) and particularly cymbals and chimes and their resonances, harmonics and overtones.
And then fine details in musical performance are also revealed. For example, greater contrast of dynamics and amplitude (Dynamics (music) - Simple English Wikipedia), subtle variations are more apparent and the difference between very loud and very soft is greater and faster transients (Transient (acoustics)). In vocals this enhances emotional expression. It is also apparent in very fast playing where what previously seemed to be one notes is now actually several successive notes ā on bass guitar this is not only the sound of fingers playing the strings but sometimes (depending on the notes and presentation of the instrument) also the accompanying low frequency sound which is more felt or perceived than heard. It is like the musicians now include new techniques ā as if theyāve been away for some time to learn them and then come back.
To come back to SFP ports:
Do exist WAPs with SFP ports?
If so maybe worth a try to connect a router via DAC cable to the WAP and then the downstream WAP via DAC to server/streamer.
Iāve seen a couple of WAPs with SFP. or SFP+. Probably not common place because a lot of installations rely on PoE rather than separately powering each WAP and SFP canāt pass on PoE.
But Iād say a big consideration is how ānoisyā the WAP is (aside from its power source and what is connected to it). If a WAP has SFP it might also include other features and therefore be doing other functions that increase its workload and generate noise. The total sum of noise using a WAP is the noise of the WAP, plus noise from the power supply and noise coming via a cable (both from the device on the upstream end and picked up from the environment by the cable).
A DAC cable will pass on noise from the device at the upstream end and the environment (unless it has some special noise repelling/absorbing features). A fibre cable wont, but its SFP modules generate a bit of noise. which has impact at least at the downstream end. The total sum of noise from a fibre cable may be a lot less than a DAC cable depending on what it connected at the upstream end. The total sum of noise using a wired ethernet (RJ45) might be (Iād guess usually is) less than a DAC cable thanks to isolation transformers.
Your suggestion is ā¦ router > DAC > WAP āwifiā WAP > server.
My suggestions for wifi was to put it in a server like putting SFP in a server. Your suggestion is what could be done if wifi was not include din a server. But the DACs will probably pollute the WAPs with noise from the router and server and might be higher noise than RJ45 WAPs.
My guess is the ranking of lowest noise per dollar:
- fibre plus SFP modules
- very good - high end wired ethernet cables plus WAPs
- really high end wired ethernet cables
- DACs plus WAPs
Somewhere in this ranking is really high end wired ethernet cables plus some other type of isolator or filter (etherregen, Waversa, Muon, etc) because it depends how much they cost and the impact they have (assuming āusuallyā higher cost = less noise).
DAC cables seem to have different impacts depending on what they are connected to.
Getting back to your original idea, in the absence of an Antipodes with SFP, the solution could be network > FMC > fibre > FMC > wired cable> server. Cost will depend on what FMC, SFP modules and wired cable are chosen.
SFP into my MSB Ref DAC sounds sensational.
Iād love SFP straight out of the Oladra.
Of course this is all down to personal opinion and evaluation of the sound quality but I will just reiterate that in my own personal judgement the worst sound I have experienced with my Oladra was using FMC - fiber - FMC. Worse indeed than just using a copper Ethernet cable between the Oladra and a cheap Cisco 2960 from eBay.
And yes I know I have repeated this several times but equally I have repeated the experiment more times than I care to remember and also with many different SFP and FMC (including some exotic audiophile ones), also with a variety of power supplies.
On the other hand I do want to try the DAC connection route but the thought of using fiber into an Antipodes streamer / server with SFP ports does not fill me with glee unless that is Antipodes find a better way of handling the FMC process than any of the FMC I have tried.
I assume you use MSB ProISL Interface which is not typical.SFPs. does MSB supply a soecial FMC?
So your not usung Oladra player and reckocker?