EC Designs

[a supprimer merci]

Le nouveau site Web (même adresse) sera lancé samedi, et proposera donc leur nouvelle gamme.
https://www.ecdesigns.nl

[a supprimer merci]

J'ai passé commande du DAC en version 16 bits, plus l'adaptateur USB-Toslink.

https://www.ecdesigns.nl/en/info/mos16

Livraison prévue aujourd'hui

363 € pour le DAC 
.. .. C'est pas assez cher pour être pris au sérieux  !

[ajls]

super tu nous diras 

[0000]

16 bit. C'est un choix technique ? pour quelle raison ?

[a supprimer merci]

Je viens de louper la livraison en sortant, à 20h30, 5 min pour aller faire une course

Voici les explications pour le choix du 16 bit...

Why offer a 16 bit DAC?
Each bit represents a dynamic range of 6dB. Perfectly healthy auditory system can detect up to 18 bits or 108dB where the lowest bit is just barely audible. With 16 bits we have 96dB dynamic range. Our auditory system has a protection mechanism that changes the gain depending on sound pressure levels. So we don't hear 18 bits all of the time, just for a brief moment before our automatic gain control sets in. There is always some ambient noise in the listening room, typically 30dB that will mask some lower bits. So these bits become inaudible or very difficult to hear. In general it is not comfortable nor safe to listen at high SPL for extended time periods, so we simply have to turn down the volume. This also lowers the practical dynamic range and the related number of bits we can still hear well below 16 bits.

Audio sets are not perfect and often fail to resolve 16 bits at top volume setting. So in practice it makes very little sense to use more than 16 bits for playback. For digital sound editing it makes a lot of sense to have more bits and related higher dynamic range.

Why offer a 24 bit DAC?
When we don't need more than 16 bits for playback, why a 24 bit MOS DAC?

We can play 24 bit digital audio files without the need for converting these to 16 bits.
And in general a 24 bit DAC has to be constructed with more accurate parts. So a 24 bit DAC can offer slightly lower bit errors because of this. Some customers simply want to be on the safe side and therefore choose a 24 bit DAC..

---

Pour l'aspect "source insensitive", j'ai un peu extrapolé par rapport aux explications initiales de ECDesign. Visiblement, la source a toujours une incidence - mais plus faible...  Ils élminent les intérférences liées aux signal électrique avec l'utilisation du Toslink couplée à un circuit "PLL" à très faible jitter (qu'ils ont développé). Les interférences internes au DAC sont limitées par plusieurs facteurs: alimentation batterie, et la segmentation des DAC (mais cela me dépasse complètement). Reste les interférences dans la source même lors du traitement du signal en amont du DAC.

Pour "benchmarker" les sources, ils ont développé un lecteur de fichiers WAV avec clé USB, qui fonctionne sur batterie également - voir ici: https://www.ecdesigns.nl/en/info/upl16

Quelle sera l'incidence "résiduelle" de la source ? Il faudrait faire des comparatifs entre différentes sources (leur UPL, des streamers HDG, et des streamers basiques...). Je les ai questionné sur le sujet mais ils ne m'ont pas (encore) répondu.

En tout cas, je pense que ce sera déjà une nette amélioration par rapport à ce qui est disponible en terme de "conversion" (DAC). On bénéficie notamment de l'utilisation du Toslink sans ses inconvénients. A suivre...

[0000]

Merci. C'est bien argumenté.
Connais-tu des arguments opposés sur le web ?

[a supprimer merci]

Non, je n'en ai pas connaissance.

[a supprimer merci]

Voici quelques explications sur leur DAC (certains points ayant déjà été cités précédemment):

Toslink provides perfect galvanic isolation with zero pF coupling capacitance between source and DAC. This completely eliminates ground loops between source and DAC and related ground loop noise. Ground loop noise usually causes a weak buzzing or humming sound so the background is not ghostly quiet and the ground loop noise will mask low level detail. Ground loop noise can also find its way into the (pre) amps, injecting noise even when the DAC is not playing.

The MOS16 and MOS24 DACs are equipped with a novel type of 4 segment D/A converter with bit power scaling. It basically consists of a combination of two segmented D/A converters. The 8 MSBs are generated by a "high power" D/A converter that offers low output impedance, very fast transient response and very low bit errors. This is a result of the low RDSon resistance in the driver chips and relatively low Ohmic resistors.

The "low power" D/A converter section is used for the remaining 8 or 16 bits. These are based on driver chips with relatively high RDSon and higher Ohmic resistors for high precision and very low bit switching noise. The segmented configuration of both, high and low bit power sections ensures that only those bits are activated that are actually needed for a specific signal level. This greatly reduces jitter sensitivity of the D/A converter and offers superb low level performance without (re)trimming. The noisy bit flip at the zero crossing that occurs with many multi-bit D/A converters is also eliminated by the segmented configuration.


These measures are necessary for highest performance but are not a cure for everything.

There still is the required bandwidth of roughly 12 MHz for audio data to enter from the source. These interfaces cannot differentiate between data and interference that falls within this bandwidth. In short, we cannot prevent some source noise from entering the DAC. It enters together with the S/PDIF signal in the form of ripple voltage and phase modulation.

So some source dependency will always remain. So when we use a cleaner digital audio source, less interference will enter the DAC and sound quality improves. That's why we developed the UPL, it is a very clean digital audio source because it uses low clock frequency (42 MHz typical @ 44.1/16) and low power (approx. 220mW @ 44.1/16) compared to RPI (5 ... 10W, 1GHz), dedicated streamer or PC (50 ... 100W @ 2 ... 4 GHz).

It reads WAV files from the USB memory stick and simply routes the data to the S/PDIF transmitter chip using dedicated hardware circuits on the single, low power CPU. Since no conversion from FLAC or Apple Lossless is needed, noise levels are even much lower as there is no conversion related interference.

The UPL also guarantees bit-perfect playback, always, without any configuration. Are you 100% sure that streaming providers, a RPI or a PC always offer bit-perfect playback?

Playback is not bit-perfect when using software volume control, software mixing, software equalising or software upsampling. 44.1/16 albums are end products that include extra information (more than 16 bit) in the form of a carefully calculated dither pattern. This dither pattern and the extra resolution it provides is instantly destroyed when we perform further digital manipulations on this end product. That's why it is so important to maintain bit-perfect playback.

Similar, there is low and high quality software for manipulations in the digital domain. On a consumer product with cheap or free digital manipulation software the quality will not be that great.

[0000]

Merci  !

(12-06-2018, 08:24 AM)paulw a écrit : The UPL also guarantees bit-perfect playback, always, without any configuration. Are you 100% sure that streaming providers, a RPI or a PC always offer bit-perfect playback?

De la même manière, "Are you 100% sure that streaming providers, a RPI or a PC UPL always offer bit-perfect playback?"

(12-06-2018, 08:24 AM)paulw a écrit : Playback is not bit-perfect when using software volume control, software mixing, software equalising or software upsampling.

Ce n'est pas nouveau. La communauté le savent depuis longtemps.

(12-06-2018, 08:24 AM)paulw a écrit : 44.1/16 albums are end products that include extra information (more than 16 bit) in the form of a carefully calculated dither pattern. This dither pattern and the extra resolution it provides is instantly destroyed when we perform further digital manipulations on this end product. That's why it is so important to maintain bit-perfect playback.

Pas complètement juste même si je le préfère pour mon usage.
C'est une approche locale.
Il faudrait place la réflexion globalement sur l'ensemble de la chaîne HiFi. A un moment des corrections peuvent intervenir (volontaire ou involontaire). Si on pousse un peu plus loin les deux extrêmes, l'ingénieur de son ne fait pas de bit-perfect avec les musiciens, et la pièce d'écoute a une plus grande influence. 

(12-06-2018, 08:24 AM)paulw a écrit : Similar, there is low and high quality software for manipulations in the digital domain. On a consumer product with cheap or free digital manipulation software the quality will not be that great.

C'est les hommes qui créent le software, il n'y a pas de lien avec "cheap et free".

Ça sent la pub sans arguments techniques (comparaison de codes source) vers la fin même si leur produit pourrait être excellent.

[a supprimer merci]

Oui leur présentation des "soft" est un peu schématique mais ils sont sérieux et ils maîtrisent leur sujet.

Pour le "bit perfect" ils ont des outils pour le vérifier--leur DAC Mosaic UV contient d'ailleurs un test, que l'on peut réaliser soit même pour vérifier si l'application de lecture est bit perfect. Donc pour l'UPL je leur fait confiance sur le fait que ça l'est.