Schiit Eitr Preview Thread

Is it just digital data? I suspect that the noise floor would be much higher if that signal was sent from the laptop through a DDC that did no isolation at all. The Singxer SU-1 and F-1 do offer some isolation but it is done in a different way than the Eitr. I would expect the noise floor to be higher on those devices as well. I wonder if atomicbob has data on those devices from the same test.

 

Issue resolved. Swapped out Eitr. The second Eitr works perfectly. Eitr upstream from Modi Multibit is night and day better than Modi Multibit without Eitr. Thank you Schiit.

 

Color:
1. The Eitr was Coax out (not BNC) vs. the BADA using Coax out through a bnc adapter and the results were still ranked the same. The BADA AES beat all.
2. The Aries USB to BADA in was not as good as the Microrendu to the BADA.
3. The BADA using the BJC AES cable and the Aries using the BJC AES cable were closer. Using the Kimber Orchid cable, they were not.
4. We tried a PC->Bada vs Aries comparison a few months ago with @brencho's Yggdrasil, and didn't really hear much. But the room wasn't really set up at the time, and its since sounded a lot cleaner.

Having had or tested the Regen, Wyrd4Sound Recovery, Mutec 1.2, Mutec 3+USB, Singxer SU-1 and the BADA, the Eitr was the best value by far. We really didnt listen long enough or to enough music to flesh it out completely, but at its price, c'mon.

I was pulling for the Aries, too, so no ownership bias.

Digital is still inferior to Analog, but I had to do a lot to analog to get past this setup. And yes, people will have different results with their setups, I know.

 

Just got Eitr in today primarily for 2nd system. Primary system is much more expensive Rednet D16>Mutec>Yggdrasil. Tried it in primary system. Prefer the Rednet> Mutec combo but it was close and I'm sure a lot will depend on how one responds to the "sound" of the Mutec. The Rednet> Mutec seems a bit more tonally rich while the Eitr seems a tad bit more lively. In either case the value of the Eitr is phenomenal and it's really a no brainer. Since this is a hobby for me I like changing things up and can see me using both pretty regularly. Eagerly waiting on Manhattan project...

 

That is mighty impressive no matter how you slice it.

JJ

 

I wonder if John Atkinson will be covering the Eitr.

 

"Schiit uses an output interface from the 80s, obsolete! USB is newer and betterer. Muh MQA/DSD256." Atkohn Jokingson

 

I remember reading Bob's measurements about the Breeze, Gustard, and Singxer transports quite a few months ago. I'm really not sure how to interpret the graphs, but the Eitr doesn't have the extraneous spikes that were seen in other USB transports. How did they do that while still having a lower noise floor? I'm not really knowledgeable regarding these sort of things.

http://superbestaudiofriends.org/in...time-technical-measurements.4178/#post-130285

 

While you may have quoted Willy, you also came up one cowbell short of three Christopher Walken quotes both from the same movie, if you can name it...

Anyhow, too bad the Aries comes up short pretty much across the board. It is an otherwise useful wireless Roon endpoint if nothing else. I remain happy with Lynx. It doesn't give me the Aries wall of sound effect, and aside from beauty and ultimate utility, it has not compelled me to think about digital gear upgrades in a long time. #WaxedChess

Congrats to @schiit & co for what looks to be a real market disruptor.

 

You would only see that kind of noise if the signal were converted to analog. AB's setup is pure digital, so as long as the bits all arrive, there will be no differences due to Eitr's isolation features. We're looking at the noise floor of an FFT of the actual data being sent, so the only sources of noise would be in the original signal or in the computation of the FFT.

A loopback test would confirm this. A bit-accuracy test would also confirm this.

In that test, AB was measuring the analog output of the Modi MB, which is why the spikes appear: the spikes are an analog phenomenon. The Eitr test is entirely in the digital domain.

 

Refer to the axiom about theory and practice.

Consider the two cases of
1. analyzer loop back from digital output to digital input
2. analyzer sending data to traverse the incredibly tortuous audio Win(x) maze through USB and then to a DDC and finally back to analyzer digital input

Case 1 - simple analyzer loop back:

It may be hard to see the loop back from DO to DI in the diagram. This will be the purest path possible for the analyzer.

Case 1 - results:

Note how it appears there is only one channel showing. Both Left (blue) and Right (red) are depicted but they overlay so closely as to only display one color.

Now consider case 2 - look at the signal path that must be traversed in Win(x) to transport digital audio from application to digital endpoint.

This path is both complex and susceptible to all manner of data corruption. Most obvious would be data under or over flows. Less obvious is a potential trip through any of Win(x) sample rate conversions, mixers, enhancers, etc. Not easily understood through the various places one may interact with audio device and driver properties.

Now consider a DDC that attempts valiantly to traverse the gauntlet only to result in the following:

Not bad but not nearly as close to the analyzer loop back path. Also note how the Right channel (red) is now appearing as there are differences between left and right channels.

Here again, for convenience is the Schiit Eitr measurement:

Again note how there are differences between left and right channels showing as both blue and red are now displayed. These are quite minor but also demonstrate the difference between theory and practice.

 

At the risk of asking a redundant / stupid question (and infuriating @Torq further): Are there *any* sonic differences between the Eitr (BNC input on DAC) and the USB Gen 5 on the Gungnir Multibit / Yggdrasil? Or is the difference only a matter of flexibilty (Eitr) vs neatness (USB upgrade)?

I did read near the beginning of the thread that the upgrade was recommended for Gungnir Multibit / Yggdrasil but don't remember if my specific question was previously answered. Hopefully I'm not a doofus

I'm finding the digital output of M-Audio Audiophile 192 to be too sharp and artificial (and unpleasant to listen to for long periods). The standard USB input of Gungnir Multibit smears and blurs these attributes but they are still there, it doesn't fix them. I was wondering if this is an attribute of digital sound? Hopefully Gen 5 USB can fix this for me.

 

That's my understanding of it. To quote @Torq:

It seems as if there is no audible benefit to use e.g. AES from Eitr (via converter) as opposed to default RCA Coax. And there doesn't seem to be a difference between using external Eitr vs USB Gen 5.

Which brings us to @Torq's TLDR Executive Summary:

 

As I have mentioned elsewhere, I plan on opting for the 'self install' option on the Gen-5 board for my Yggdrasil.
That way I'll be able to easily flip back and forth (assuming I can set up a new zone from JRiver to feed both outputs simultaneously.
Synching the 2 feeds is the tricky part.

It should be a rather interesting series of tests, methinks.

JJ

 

@atomicbob Not sure what to make of that last graph; almost looks like least significant bit dithering.

 

I lamented elsehwere the Eitr not having optical as I'd not be able to use it with the Behringer DEQ2496 digital parametric equalizer. I just remembered the thing has AES and that @baldr suggested the use of a Sennheiser transformer for coax to AES.

I cannot find any mention of a Sennheiser unit on google. Can anyone point me to it, or recommend a well regarded alternative?

With my setup I'll be better off with an Eitr over Gen 5 so I can use the DEQ2496 at all times.

Cheers!

 

Thank you to @atomicbob for expanding on the difference. Now I can truly says "Wow." It doesn't get better than that (at least, not in a reasonably priced consumer box)!

I created a comparison to show the difference between loopback and Eitr. Very small indeed; just a few dB. Just mouse over and off of the image to flip between them.
http://screenshotcomparison.com/comparison/216969

EDIT: Bob actually said "Noise floor with -40 dBFS 1 KHz sine is below -175 dBFS !!!!:" So when I say "just a few dB" it is likely 1-2dB of difference as the graph bottoms out at -180dbFS. So just 29.1666 bits instead of 30
Bits aside, the complete lack of harmonics or distinct peaks is excellent. I really think we are looking at the bare minimum amount of noise that is introduced by going outside the device (see Bob's great chart on signal flow above).

 

Neutrik and Sescom both make SPDIF to AES/EBU transformers. You will need a RCA to BNC converter as well, but that can be even simply a RCA -> BNC cable.

 

He meant Sescom. That should help you find the adapter / transformer.
Sescom SES-AES-EBU-2 BNC Female to XLR-M AES/EBU

 

I wish to avoid turning this thread into a tutorial. There are some excellent publications which provide useful reference:
Fundamentals of Modern Audio Measurements, Richard C. Cabot, AES Fellow, Audio Precision
Audio Measurement Handbook, Bob Metzler, Audio Precision

It is not dithering but the averaging from complex mathematical operations involving FFT and window functions. Further, the signal generator is asynchronous to the analyzer thanks to the trip through Win(x) device driver and USB. Lastly, the 1000 KHz sine is not nudged to a slightly higher or lower frequency to line up with the FFT bins. Note the following settings on the dScope setup: