Les switchs audiophiles + ponts optiques

[wgb25]

Trouvé et aussi acheté un en occase modèle :

MS220-8P-HW

Réception début de semaine je pense.

[Bachman]

Voici la traduction google en anglais(plus fiable qu'en français de nos amis néerlandais

The test was carried out completely blind for the test panel. The first idea was to do it double blind. But in the end that didn't work, because every switch has a recognizable appearance. Even if there was a cloth over it. In addition, the Ansus switch had to be disconnected from the power, as it can affect the display. That would immediately tell which is going on.

For the test, we pulled a - solid solid, copper - SFP-CAT6a cable of approximately 16 meters from the server room to the Metrum Acoustics Ambre. Each switch to be tested is linked to a Netgear GS724T v3. This is again connected via fiber to our core switch. The ROON core also indirectly depends on the Netgear. The NAS with music is directly connected to the Netgear. Each switch is supplied with the standard adapter. This is because we want to test everything "as-is". Not everyone has an Sbooster, IFI or other luxury food.

The test set consists of:

Pass Labs XP-12
Bryston 4B SST3
Metrum Acoustics Ambre
Metrum Acoustics Pavane (I2S connected)
Focal Sopra No1 speakers (with isoacoustics feet)
Cables from Grimm Audio, Audioquest
Kemp and Isotek filter (titan)
We played three tracks:

Paul Simon - Diamonds on the Soles of Her Shoes
Massive Attack - Angel
Steven Wilson - Pariah (live Royal Albert Hall)
We play every track - except for Pariah - for 30 seconds. And that 2x to get a good idea. After the three tracks, we switch the switch and the list starts again.

Five plus three

We let five switches play in the first round. Then we took the winner from the first round to play with that one more time. Each round starts with a base-line test with a standard switch. So the order is:

standard Netgear (getting used to the sound / baseline)
Fidelizer EtherStream (399 euros)
Ansus switch (1999 euros)
Fiber (250 euros complete)
Netgear standard (149 euros, 24 ports)
Bonn N8 (€ 395)
Standard Netgear (getting used to again after a break)
Bonn N8 (€ 395)
SoTM - basic model (999 euros)
Meraki enterprise 8 port with fiber and PoE (799 euros)
We were able to deploy a group of 5 and then 4 people for this test. The "break group" was able to relax with a drink and food.


After a day of testing and full forms you have a lot of data. We have received a form with + and and -en from one reader / tester. We could not include that data, but we did compare it with the averages. Fortunately it would not affect the outcome. We have also included the comments on the forms in the conclusion of this article. That is perhaps even more interesting.

Difficult

Of course we asked what everyone thinks. It appears that differences are audible, but that they are small. And sometimes very difficult to explain. This is also apparent from the scores. Some models are very close together. However; everyone agreed on the top 3. And that is also apparent from the scores.


We have looked at the data in various ways: average, median and distribution of the scores. The scores for the Top-3 are fairly close to each other. The least appreciated switch in this test are two readers who don't like it. The rest is around average. The Meraki scores high and the most coherent. The SoTM idem and the Silent Angel have a major outlier.

To give you insight into how close it is to each other, we have put all the data in graphs. It is up to you to come to a conclusion. Our conclusion for the time being: it is audible. But make sure the basics are right before you start working on this. It is a dot on the "i". No more.

Averages per switch

Score per Track
Scores average
Total points - final score
What is striking here is how consistently the Cisco Meraki scores. The SoTM also scores fairly equally among all participants. The Bonn N8 slightly less, but that outlier is mainly upwards. The Netgear, Ansus and the Fiber have a big dip in the score. These three are the only ones who have failed, although the fiber is assessed as less "erratic". However, the Ansus again has two failings. Remarkable, since with 2000 euros that is the most expensive switch in this test. We read "sharp", "flatter" and "hard" back there. For the Bonn, Meraki and SoTM especially: "Nice tonal balance", "YES!", "Large" and "Deep low".

The Fidelizer EtherStream is average, but, oddly enough, scores slightly lower across the board than the standard Netgear and the Fiber solution. It is a crazy world ... Anyone who looks critically at the data will see that the Fidelizer scores more consistently. We would therefore set it to 4 and not to 6, given that it did not pass.

Spread of scores

Spread Scores
Spread scores - per switch
Scores all readers
As you can see here, the spread is still fairly limited. Down on a few notable outliers at the Ansus and Netgear. Apparently those switches do something. The Fiber solution is not everyone's taste either. The rest of the test field is fairly stable in the assessments and is fairly close to each other. With a consistently higher score among the winners.

Scores per reader - per model and track

Score reader 8
Score reader 7
Score reader 6
Score reader 5
Score Reader 4
Score reader 3
Score reader 2
Score reader 1
We have indicated the score per reader for your insight. This way you can see how the assessments are individual. You can also read from this that it is extremely difficult for some readers to observe differences. And that is it. Certainly blind!





We streamed live this entire day to let you watch. And grateful use has been made of that. Very nice! This day tastes for more!

Today's conclusion: yes ... a switch makes an audible difference. However, the differences are small. The basis of the set and the acoustics must be correct. Otherwise it is a shame to invest so much money. But the fact is: it is audible. The data in this test also show that.

[Bachman]

Les schémas avec les notes n'ont pas pu être intégrés

[a supprimer merci]

Supprimé

[Vincent.]

(12-04-2019, 09:09 PM)paulw a écrit : ...

On peut se demander si un cable de 16 mètres entre le switch et le lecteur reseau est une bonne idée...

En effet

[bbill]

(12-04-2019, 09:09 PM)paulw a écrit : La conclusion est inchangée : les differences ne sont pas très significatives.

le meilleur à une cote de 8,5
le moins bon a 6,4 (et coûte 1995€)
ce n'est pas assez significatif ?

(12-04-2019, 09:09 PM)paulw a écrit : On peut se demander si un cable de 16 mètres entre le switch et le lecteur reseau est une bonne idée...

ce n'était pas une idée mais une necessité pour le faire "en aveugle"
tu crois que les écarts auraient été encore plus importants avec un petit câble ?

[a supprimer merci]

Supprimé

[bbill]

(12-04-2019, 09:24 PM)paulw a écrit : Non, les notes ne veulent rien dire sans le contexte, et le contexte est bien expliqué: les differences perçues sont faibles.

non, ils parlaient de différences faibles pour les 3 premiers (et la les notes sont proches et surtout pour les 2 premiers ou tous les scores sont cohérents)
quant aux notes, ce n'est pas une explication du "contexte" mais des statistiques sur 168 écoutes, pas moins.. de 6,4 à 8,5

[Mannneu013]

(12-04-2019, 09:09 PM)paulw a écrit : La conclusion est inchangée : les differences ne sont pas très significatives.

On peut se demander si un cable de 16 mètres entre le switch et le lecteur reseau est une bonne idée...

Chez moi il y a 5km de ligne cuivre avant la box. Imagine le bordel qui peut arriver. Ben j'ai enregistré en temps réel ce qui arrivait d'un fichier audio que j'avais mis en ligne : nickel. Bon ça dépasse le cadre des switchs vous allez me dire...

[zaurux]

Et confirmation, il s'agit bien du Meraki MS220-8P (fibre + POE).
Pourquoi ne pas avoir choisi la version sans POE ?? Peut-être qu'il traînait dans l'immeuble ? 
Pour le câble de 16m, effectivement c'est assez surprenant mais en même temps, s'il y a switch, il y a réseau et c'est assez représentatif de ce que l'on peut trouver (Box à 15m du switch ou Nas à l'autre bout..).

[EDIT] Pour le PoE

Citation :It also features 2 PSUs, one outputs 12V for the main board, and another 56V 2.6A unit for PoE. The switch will still boot when the PoE power supply is disconnected from mains voltage.

Donc ça doit se déconnecter pour ne garder que l'alim 12v.