Devialet USB input

[Antoine]

Intona was the first to succesfully implement real USB galvanic isolation for high speed USB (480Mbit/s). The technology didn't exist yet when the Regen was designed ....

It did at the point of the microRendu design though.
That said having now looked inside the rendu and the Intona, it's obvious who are leagues ahead in circuit design so maybe it's far from trivial for anyone outside that specific field - the Intona board looks pretty complex.

I'm too much of a layman but I can't quite see why the same technology used on an Ethernet connector can't be used on USB? The Intona seems so complex by comparison.

True, only Sonore can tell us why they've chosen to not implement it inside the microRendu. "Even" the Singxer DDC's have it.


The layer 1 (electrical) specs of ethernet and USB differ greatly.

On 10BaseT standard each bit that the computer wants to transmit is physically coded into a single transmitting bit, i.e., for a group of eight bits being transmitted, eight signals will be generated on the wire. Its 10 Mbps transfer speed means that its clock is of 10 MHz, but just because each clock cycle a single bit is transmitted. On other standards this is different.
100BaseT uses a coding scheme called 8B/10B, where each group of eight bits is coded into a 10-bit signal. So, differently from 10BaseT, each bit does not directly represents a signal on the wire. If you make the proper math, with a 100 Mbps data transfer rate, the clock rate of 100BaseT is of 125 MHz (10/8 x 100).
So, Cat 5 cables are certified to have a transmission speed of up to 125 MHz.
What Gigabit Ethernet does is to change the coding. Instead of making each bit to be coded into a single signal like 10BaseT or to code each 8-bit group into a 10-bit signal, it codes two bits per signal. So, a signal over a Gigabit Ethernet cable represents two bits, instead of a single bit. In order words, instead of just using two voltages on a signal representing merely “0” or “1”, it uses four different voltages, representing “00”, “01”, “10” and “11”.
Also, instead of using just four wires of the cable, Gigabit Ethernet uses all wires.
On top of this, all pairs are used in a bi-directional fashion. As we’ve seen above, both 10BaseT and 100BaseT uses different pairs for transmission and reception; on 1000BaseT, as Gigabit Ethernet cabling is also called, the same pairs are used for both data transmission and reception.
The beauty of Gigabit Ethernet is that it still uses the 100BaseT/Cat 5 clock rate of 125 MHz rate, but since more data is transmitted per time, the transfer rate is higher. The math is quite simple: 125 MHz x 2 bits per signal (i.e., per wire pair) x 4 signals per time = 1.000 Mbps.
This modulation technique is called 4D-PAM5 and it actually uses five voltages (the fifth voltage is used for its error-correction mechanism).
So it is a mistake to say that Gigabit Ethernet runs at 1.000 MHz. It doesn’t. It runs at 125 MHz just like Fast Ethernet (100BaseT), but it achieves a 1.000 Mbps because it transmits two bits per time and uses the four pairs of the cable."

(from: http://www.hardwaresecrets.com/how-gigab...net-works/)


USB 2 high speed is 480Mbit/s over a single wire pair (four for gigabit ethernet) and bi-level only so here the signalling rate is the same as the data rate thus 480MHz. That's why USB is short length as well. There's no transformers for isolation "fast" enough to work at those speeds. The Intona uses FPGA's to "parallellize" the data transfer through isolators which are specced at a max of 150Mbps. That's why the design is so complex compared to the "simple" transformer based isolation found in ethernet.

Edit: I just realized this isn't the whole story as there's 10gbit ethernet for example that's still galvanically isolated while the signaling rate is 833MHz. These probably simply are not compatible with the USB electrical requirements. Perhaps the reason is simply that no semi conductor company like (Pulse Electronics) has seen the need to develop an integrated solution for USB. The standard doesn't dictate it so the target market is small and limited to niche markets only.

Of course around the same time Intona brought their isolator to market another company named Silanna Semiconductors was close to releasing its solution: http://www.silanna.com/usb. It's being used in cheaper than Intona isolators now like the Coolgear released in April this year. http://www.coolgear.com/news/usb-isolato...debut.html

[Sharone]

Hopefully I am not intercepting abruptly in this thread.

I have the Mutec 3+ USB in my Devialet setup.
Personally i find the AES/EBU interface superiority as compare to the USB interface in Devialet helps to reduce the digital sonics instead of the USB galvanic isolation. The reclock capability helps in this as well, especially getting better bass definition.

Nevertheless, i tend to evaluate various configurations in my setup.
I do have 2 set of balanced isolation transformers in my setup, one for the digital source (Aurender, Regen) and one for Devialet.
As i am aware that Mutec has galvanic isolation for their USB input interface, my Mutec was supplied from the transformer tied to the Devialet. Somehow I gotten a better defined sound from this setup, rather than supplying it to my digital source during my experiment.

Nevertheless both supplies are tied to a common AC point in order to avoid loops from the electrical supply.

[GuillaumeB]

Some fascinating discussions going on here! 

When I home demo'd the N1Z I found it to be absolute reference level. The best I had ever heard, and I pay a lot of attention to bass. And this was naturally using the USB input of the Devialet.

The N1Z would have connected to a dedicated 10 or 15A circuit, I can't remember now.  

My guess (and it is just that!) is that we are possibly dealing with a multi factorial situation and possibly everything, including quality of cables and power supply need to be taken into consideration. At the time my Devialet units would have been connected to an Entreq box so we possibly need to take this into consideration too.

Guillaume

[ogs]

Some fascinating discussions going on here! 

When I home demo'd the N1Z I found it to be absolute reference level. The best I had ever heard, and I pay a lot of attention to bass. And this was naturally using the USB input of the Devialet.

The N1Z would have connected to a dedicated 10 or 15A circuit, I can't remember now.  

My guess (and it is just that!) is that we are possibly dealing with a multi factorial situation and possibly everything, including quality of cables and power supply need to be taken into consideration. At the time my Devialet units would have been connected to an Entreq box so we possibly need to take this into consideration too.

Guillaume

Absolutely fascinating! An asynchronous protocol like USB should easily be on par with AES/EBU and SPDIF. but lack of isolation on USB may well be one of the  reasons (maybe THE reason) why results differ so much between setups. Your experience with the the N1Z proves that a USB input is capable of stellar performance.

[Hifi_swlon]

And more importantly, sounds like the USB input on the Devialet is capable too. (Oh, that might be what you meant )

[Hifi_swlon]

Thanks for the ref on the transformer stuff Antoine. I have to say from my physics days I thought transformers were to all intents and purposes instantaneous, so I'm even more stumped now. Am now reading up on transformers!

Isn't the AES input using transformer isolation?

[octaviars]

Thanks for the ref on the transformer stuff Antoine. I have to say from my physics days I thought transformers were to all intents and purposes instantaneous, so I'm even more stumped now. Am now reading up on transformers!

Isn't the AES input using transformer isolation?

 
I think in the discusion about the RCA-XLR cable Devialet told me that the AES input on the Devialet is transformer coupled so that gives isolation

[ogs]

Thanks for the ref on the transformer stuff Antoine. I have to say from my physics days I thought transformers were to all intents and purposes instantaneous, so I'm even more stumped now. Am now reading up on transformers!

Isn't the AES input using transformer isolation?

 
I think in the discusion about the RCA-XLR cable Devialet told me that the AES input on the Devialet is transformer coupled so that gives isolation

Correct. I got the same confirmation from Devialet a long time ago.

"...there is a transformer on the AES/EBU input - this is a fully differential input.."

[ogs]

One thought.. Rufus is still using grounding boxes (says so in his sig...) and the N1A. If the leakage loop is a problem, temporarily removing the grounding should effect bass quality in his system. Depends of course if the USB output of his N1A is similar to the N1Z.

[Rufus McDufus]

True. I'll give this a go...