USB Nervosa Thread Decrapifiers, pro interfaces, and bears oh my

I think there are lots of network audio endpoint style devices or oven DAC inputs. I'm not convinced that will be the magical goody that will fix digital inputs, especially if we are still needing to use AES or SPDIF. My bet is the praise is coming because it allows to bypass poor USB implementations and suboptimal computer sources. Maybe if they can couple it with lvds i2s it could be convenient for huge librarys.

The other solution I am interested in is SD card minimalist transport, again with some sort of I2S connection.

I think with all the USB bashing some might throw the baby out with the bath water. Good USB implementation has very low jitter and noise, especially when paired with dedicated hifi computer source is not bad at all...

 

I myself am now using a dbl Wyrd setup and the improvement (from just a single Wyrd) was substantial.
And I should be getting an opportunity here shortly to hear the difference between these 2 digital pathways on a well tweaked system in the next few weeks.

I look forward to hearing what the ethernet approach can offer.

JJ

 

I think I see (the light) what you are talking about with AES67 and other audiop over IP solutions. This could be very good in that I thin USB 2 audio is nearing the end of it's life as it is simply running our of bandwidth and usefulness.

For me to jump onboard though, I would want to see as I noted above AES67 I2S cards, but in the meantime it could be a good alternative to USB/SPDIF/AES D/D converters in getting around USB 2 limitations.

I think we may be waiting at least a couple of years to see built in solutions become mainstream though (unfortunately). Even longer to turn up in affordable gear I'm betting.

I only dread needing drivers to use, which would mean a bit of messing around setting up vs what I am now used to with USB and Linux. But I'm not averse to a bit of time in setup if the end result is easy to use.

Luckily Wireworld are also coming out with an ethernet cable, which I already had my eye on.

HOWEVER

From my experience RJ45 network adapters are not automatically benign and noiseless. We still have a packetised transfer, and still need noise shaping, and in my experience regular NIC running regular firmware for normal LAN such as that on computer motherboard produce quite a bit of noise, and are better left off if they are not needed.

I expect AES67 will still be quite implementation dependent, much like, erm USB

 

I agree that ethernet isn't 'ideal', but it is what we've got to work with and it does have several advantages over any of the other options.
I was using ethernet to feed my PWD but I ran into player issues, such that when the Wyrd came out and I could compensate (due to DSP becoming available) for them, and experiment even further with additional tweaks, USB became 'Better' and the digital stream of choice.

If these new ethernet to AES/SPDIF converters work as well as I suspect they will and when coupled with ASIO drivers (which in my case allows Media Center to use it's full featured player and not the limited ethernet player) then we get the best of both options.

I still find it a bit of a mind blower that we still need to spend 1K$+ to run a wire from the computer to the dac in order to deliver the 'best' SQ.

JJ

 

Lucky for you guys that I tried one of the Rednet devices out. The D16 AES to be exact. If anyone cares read on...

If Ethernet represents complete isolation from the PC then I think you can get 99% of the way there with a usb setup provided you have a low power, low TDP processor based pc with a solid linear PS and maybe one stage of further isolation from something like a Regen.

The Rednet was not leaps and bounds better than the sub $200 Xmos F-1. In fact I have spent the last 3 days going back and forth between the two trying to hear substantial differences. The Rednet is a bit better in the midrange - when I think I noticed differences it was there - the mids seemed to have slightly more density and tonality and space but I would not bet money I identify the Rednet over the F-1 in a blind test.

The Rednet uses ASIO and J River could use it without issue. The Rednet does not do on the fly sample rate switching though so you have to set it to 44.1 or 96 etc. Not a great thing for those of us with lots of files with different sample rates.

BUT, the Rednet does isolate your dac from your pc completely. So if unlike me your main pc doubles as a media server the Rednet 3 (I wouldnt buy the D16 if I had to do it all over) would likely save you money and sanity in the long run. In my case the D16 is going back to sweetwater.com.

I have been following the Xmos/Rednet Head-Fi thread and I won't be posting any review of the Rednet there becasue its Head-Fi and flame wars would probably ensue.

 

That's interesting.
There are 3 guys in our local area who have been exploring the use of the F-1 and just recently the RN3 and MUCH prefer the RN3 and one even uses a Mutec 3+ in line as well.
And since we will be having a meet soon we'll be able to compare them (F1 and RN3), and a few others (my dbl modded Wyrd and cooked cables etc.) on tweako systems.

JJ

 

Dude as long as the files are coming from the PC or a NAS there is no complete isolation. One could argue that throwing galvanic isolation in the USB path, or after the receiver and before FPGA reclocking also completely isolates the DAC from the computer... With RJ45 network networks there is also galvanic isolation and greater bandwidth which will prevent ground loops and offer some noise isolation, in a similar way to galvanic isolation in erm USB. 1G network also has higher bandwidth and faster signalling frequency (1GHZ vs 480 MHz of High Speed USB).

As is the case with USB galvanic isolation, this is not a magic fix-all. With USB it was probably better to put the galvanic isolation AFTER the interface receiver and follow this with clean reclocking. Some noise will still make it through the galvanic isolation. Galvanic isolatoin should offer quite effective suppression of high frequency noise, however it will also inheretly add some jitter. It's not a free lunch. Packetised digital receivers produce quite a bit of electrical noise of their own, and RJ45 is no exception here.

Lastly in my mind external D/D converters, especially using intermediate format such as AES or SPDIF/TOSLINK are not an ideal solution. SPDIF has quite high inherent jitter (I usually see around 50 ps) while the best USB receivers are around 10 ps (Amanero). SPDIF is also limited in the bandwidth and signal frequency so cannot transfer the new super high res formats, or for example DSD upsampled material from HQPlayer. Built in USB or IP receiver would be much better candidate for an optimal solution. As such I would personally not want to invest too heavily into any more external D/D type devices.

 

That's interesting…

Have you been able to measure the noise & jitter on AES?
It strikes me that since it uses a differential pair, it might be 'Better' than SPDIF or USB in this regard.

JJ

 

Just going by the specs from the receiver chips - I think that number (50ps) is down to the design of the PLL so could be meaningless number. IIRC AES is basically balanced version of SPDIF designed for use over long distances. Before when I was using a D/D converter I was using AES on recommendation of Wireworld. I think the balanced connection might have some better common mode noise rejection. SPDIF and AES generally are limited in that the clock must be on the same signal as the data, but are designed to work over distance unlike I2S. Basically though I don't think jitter or noise are going to get any lower by converting the signal from I2S to SPDIF back to I2S. I think the SPDIF is a necessary evil to get around the poor design of a lot of USB inputs, or the lack of better input standard like AES63 - and even that I suspect although step in right direction will not be automatic solution to all problems of computer audio.

I don't think any magic box by itself will fix the problems of dealing with a source (the computer) which is not really designed from the start with hifi in mind. It's a bit like expecting a DAC with reclocking to be indifferent to the CD player. After that, expecting asynchronous USB or galvanic isolation to cure evils of synchronous USB. These are all good things and step in right direction but I am skeptical in the degree of faith being invested that this new technology will solve all our problems.

 

I understand.
But I don't see anyone thinking that these AOIP solutions will "solve all our problems".

However the reports, which are increasing, are saying that the resultant step up in SQ is not subtle and is big time compelling.
Enough so that the SuperGlue syndrome is in full effect which I take as not just a significant improvement but is, for those who seek to be sucked into their music, becoming a necessity, right along with other recent major steps up (Bifrost Multibit, Gungnir Multibit, Yggdrasil etc.).

I for one am looking forward to hearing for myself and to compare these different digital signal paths.

JJ

 

I've kept following the relevant threads on the other site. Most seem to be reporting massive gains with the Rednets (whether 3 or D16). Finally one guy (atomicbob) set up an A/B with the Rednet vs. Mutec. Here were his chains:

His conclusion? No discernible difference between the two. This seems to jive with what @Aklegal says above.

Other users are putting together gear to compare different chains so apparently there will be more subjective data soon. I'm interested in hearing johnjen's impressions after his meet. The jury still seems way out. The Rednet relies AFAIK on proprietary drivers and some sort of virtualized sound card - who knows how long these are going to be supported - so I'm not really inclined to jump in at this point.

On a side note, does anyone know how much latency all these various USB devices add?

 

Nice to know that I wasn't going crazy lol.

 

Note that he's using a Gustard U12 (cheap < $200 Chinese converter) with the Mutec reclocking the spdif output so it's not a direct comparison between Rednet and Mutec. The really over the top guys are getting Mutecs to reclock the Rednet (so well over $2000 before they even get to the dac).

 

I do see that a lot. So much misinformation from all these posts keep the money churning. And the resales going, too. Mutecs to reclock a Rednet may be audible, but maybe to the top 1% of systems. In a hobby of diminishing returns, and if the whole thing front ends a system over $ 50K, or $100K who's to say?
Why they just don't buy a good transport is beyond me. Or just buy a Berkeley USB and be done with it, stop bitching about how much it costs. Cause when you factor in Intonas, Regens, linear power supplies, Wyrd, Reclockers, Mutecs, Curious cable, Rednet, et al. you still have a turd. And the specs alone don't translate to how it will sound, you are screwed anyway.

 

Just curious, if all the packets are right, the bits are right and the timing is right, why would latency matter in a playback system?

 

Synching with video. Especially if you have some intensive DSP processing (like OOYH) already slowingthings down.

 

Using f1 in my chain and it has zero latency while gaming or watching movies in 4k.

 

ah, I get it. I did not know folks went in for all this decrapifying usb audio to watch videos. I am in the stone age when it comes to video I am sure.

 

Well I'm not going to switch up all my gear just to watch a movie.

I'm just contemplating various downsides to these intermediaries (besides the price, of course).

 

I had to ship back my Mutec for an exchange due to a loose USB jack--no big deal. But what is more revealing of its reclocking prowess, and of my USB source's impurity, is that I can't listen to my system for very long now without cringing from its sharp, unmitigated digital edge.