FORGETTING the Digital to Analog conversion part, what is BEST Digital source?

[sandyk]

17 minutes ago, beerandmusic said:

Does sonore use rcore?

If not, maybe they should try it?

 

I don't know. You would need to ask Barrows that.

John Swenson likes R-Cores though!

[barrows]

8 hours ago, beerandmusic said:

Does sonore use rcore?

If not, maybe they should try it?

 

Here is the deal on transformers:

 

Toroids are often used because of low cost and a smaller radiated field of magnetic interference.  The problem with toroids is that they allow a lot more AC line noise through to the rest of the circuitry because of the capacitive coupling via the primary and secondary windings being right next to each other.  So the advantage of toroids is the smaller size of the radiated field, this means the chassis can be smaller and the toroid can be placed closer to critical circuits (but still not too close) without it causing problems.  In other words, toroids are cheaper and allow for a cheaper chassis.

 

Other transformers, like R-core, EI core, and flat packs with separate winding bobbins are much better at isolating AC line noise from their outputs because the windings are separated by some space.

 

What Sonore does on the Signature Rendu SE:

 

We use a custom made EI style transformer, which was designed specifically for this product.  This transformer is custom made for our circuit, by Mercury Magnetic in California.  This transformer is superb at rejecting AC line noise, and the cost to use this part is about 9x that of just using a standard toroid.  Because it is an EI style this transformer does have a larger radiated magnetic field than a toroid.  So that the magnetic field does not interfere with any of the circuitry we do two things: we use a large enough chassis that the transformer can be spaced well away from the circuitry, and we add a chassis dividing shield of a carefully chosen metal which shields the circuitry from both the magnetic field interference, and any RF interference from the AC input wiring.

[marce]

11 hours ago, beerandmusic said:

less power supplies and less clocks can also mean less noise and less manufacturing costs...They do advertise torroidal transformers to reduce noise....the biggest question really is, do they get the bits to the dac buffer with less noise, right?

Local POL regulation with LDO's and separate local clocks is actually better for a low noise design... 

[beerandmusic]

46 minutes ago, marce said:

Local POL regulation with LDO's and separate local clocks is actually better for a low noise design... 

is there a way to measure the noise at the point that causes the differences in SQ between the sonore and sotm? since the only purpose of those devices is to transfer the bits to the dac via usb, and the bits themselves are processed perfectly, I would think this would be easy to do?  And since they both sound different and each has it's own benefits, this would imply that both could be better on correcting the noise?  I still have a hard time understanding why the noise couldn't be eliminated from the output of these devices, by isolating the ground from the dac.  Would also love to hear from Barrows on if ethernet over fiber could be better to isolate the noise...not that it is feasible because of market acceptance, but just what could be done to reduce the noise more since clearly the "noise" situation is still not under control?

[barrows]

@beerandmusic, Yes you can measure noise at any point in the circuit, if you have the appropriate gear.

The noise can never be "eliminated" it can only be reduced.  The noise on the output of the Signature Rendu SE is as low as it is possible to be.  USB ground has to connect to the DAC for the USB interface to work, that being said, a properly implemented USB receiver will isolate (galvanic isolation) the USB ground from the rest of the DAC circuitry, all my (DIY) DACs do this.  Remember regarding noise, we are not just concerned about noise coming from the computer, or router, etc.  Noise is present and created in all active electronic circuits.  One has to pay attention to reducing it in every circuit, the DAC included, and the amp, etc, etc.

 

Using fiber Ethernet connection can stop noise from the Ethernet from entering the Renderer, but remember that ethernet interfaces are already transformer isolated and this stops noise as well.  Some are using fiber though... remember that electrical to optical conversion is also lossy, so it is likely that the Ethernet interface will have more errors when using fiber (these errors will be corrected by re-sends of course), there is some speculation that the more re-sends/activity there is the more noise is created in the Ethernet receiver.  Additionally, you still have the problem of the optical to electrical converter at the Renderer side, this will still have its own noise added to the system at the receive end.  To me the added complexity of additional components required to use optical Ethernet makes no sense: it is fixing a "problem" (which does not exist) by adding more components (each component adds more noise).

 

With the Signature Rendu SE I am confident that I have the best USB source possible (lowest noise) for my DAC, without resorting to what I consider extreme measures (optical fiber Ethernet, etc).  There is no problem here, it is solved and sounds amazing.

[beerandmusic]

15 minutes ago, barrows said:

@beerandmusic, Yes you can measure noise at any point in the circuit, if you have the appropriate gear.

The noise can never be "eliminated" it can only be reduced.  The noise on the output of the Signature Rendu SE is as low as it is possible to be.  USB ground has to connect to the DAC for the USB interface to work, that being said, a properly implemented USB receiver will isolate (galvanic isolation) the USB ground from the rest of the DAC circuitry, all my (DIY) DACs do this.  Remember regarding noise, we are not just concerned about noise coming from the computer, or router, etc.  Noise is present and created in all active electronic circuits.  One has to pay attention to reducing it in every circuit, the DAC included, and the amp, etc, etc.

 

Using fiber Ethernet connection can stop noise from the Ethernet from entering the Renderer, but remember that ethernet interfaces are already transformer isolated and this stops noise as well.  Some are using fiber though... remember that electrical to optical conversion is also lossy, so it is likely that the Ethernet interface will have more errors when using fiber (these errors will be corrected by re-sends of course), there is some speculation that the more re-sends/activity there is the more noise is created in the Ethernet receiver.  Additionally, you still have the problem of the optical to electrical converter at the Renderer side, this will still have its own noise added to the system at the receive end.  To me the added complexity of additional components required to use optical Ethernet makes no sense: it is fixing a "problem" (which does not exist) by adding more components (each component adds more noise).

 

With the Signature Rendu SE I am confident that I have the best USB source possible (lowest noise) for my DAC, without resorting to what I consider extreme measures (optical fiber Ethernet, etc).  There is no problem here, it is solved and sounds amazing.

 

Ok, thanks for info on why enet over fiber is not a good idea (e.g. the resends necessary would add additional noise).  And thanks for your opinion that you have best usb source possible (although i don't buy that it can't be further improved due to the fact that sotm has different sonics (not going to speculate which is better)), just that if one can sound better in some aspects and the other can sound better in different aspects, then the noise cold be further improved in both.'

 

In another thread I started, there is talk about i2s if you want to comment on your opinion on the feasibility of that?

 

 

[barrows]

Current implementations of I2S are mostly flawed.  It does not have to be this way, I2S could be "perfect", but the way most use it is not.

 

The PS Audio standard for I2S places the masterclock in the source component, then creates a balanced signal for output.  the devices which create the balanced signal add jitter (and again at the receiving end converted it back to unbalanced).

 

One could make the I2S system "perfect" by placing the masterclock in the DAC (where it should be), and sending it back to the source from there.  There would still be extra jitter created by the conversion to balanced and back, but it could be eliminated by a simple re-clocking stage in the DAC (POTATO flip flops), as the re-clocking could be done done directly from the (now local) masterclock.  I believe that MSB is the one company who implements their I2S interface this way (but I am not absolutely sure), of course this makes their products only work together, not with other I2S devices, which generally follow the PS Audio standard.

 

USB already does this the right way, with the masterclock in the DAC.  It appears that you keep thinking the USB interface is flawed: IT IS NOT.  Asynchronous USB is a great interface design, and is capable of producing the absolute lowest jitter possible when implemented correctly.   The only problem is that some do not implement it as well as it should be (and the same can be said for any other interface as well, they all need proper execution to work well).

[beerandmusic]

24 minutes ago, barrows said:

Current implementations of I2S are mostly flawed.  It does not have to be this way, I2S could be "perfect", but the way most use it is not.

 

The PS Audio standard for I2S places the masterclock in the source component, then creates a balanced signal for output.  the devices which create the balanced signal add jitter (and again at the receiving end converted it back to unbalanced).

 

One could make the I2S system "perfect" by placing the masterclock in the DAC (where it should be), and sending it back to the source from there.  There would still be extra jitter created by the conversion to balanced and back, but it could be eliminated by a simple re-clocking stage in the DAC (POTATO flip flops), as the re-clocking could be done done directly from the (now local) masterclock.  I believe that MSB is the one company who implements their I2S interface this way (but I am not absolutely sure), of course this makes their products only work together, not with other I2S devices, which generally follow the PS Audio standard.

 

USB already does this the right way, with the masterclock in the DAC.  It appears that you keep thinking the USB interface is flawed: IT IS NOT.  Asynchronous USB is a great interface design, and is capable of producing the absolute lowest jitter possible when implemented correctly.   The only problem is that some do not implement it as well as it should be (and the same can be said for any other interface as well, they all need proper execution to work well).

 

I don't think the usb interface is flawed so much as how it is implemented in audio. I even believe in the concept of your product....I have always said, back when dsd was first introduced, that enet sounded better and I understand now that was because of the noise coming from the source.  Either way, i have moved onto enet.  I "STILL" believe that the dac "should" be able to minimize the noise, whether the input is enet or usb without having to use a "bandaid".  Again, I believe in your concept, and believe it works as advertised.

 

The fact that neither enet->usb option is "clearly" a victor over the other,  and that they both have noise affecting sq, I am not certain, that the noise the teac (or other streamer/dac) would have, would be any better or worse than the differences between the sonore & sotm...

 

I am also still confident that "someday" we will be able to get the bits to the d->a buffer where there is no sonic difference by swapping the digital input....it may just be my digital background of 1's and 0's that prevents me from thinking in an analog world.  I am willing to accept there will be differences, with the analog out, just not the digital in, that i believe the digital input can be "perfected" as much as possible, such that no sonic difference on the input....just like no differences in the digital file itself.

 

But i will keep researching....i wake up at oddest hours trying to figure it out in my head anyway...it has to be possible.  I have found interesting stuff as it relates to digital video transmission, and it seems like they have their issues too, but that they are further along.

 

Oh, and regards to I2s, kind of disappointing news....i was hoping something more since there is talk about something new in months ahead, but if its related to i2s, it sounds like it will just be another failure, but thanks for sharing your insight.

 

[barrows]

My main point is that there is not any single "magical" interface which solves all interfacing "problems".  All interfaces are quite capable of excellent performance, WHEN THEY ARE WELL IMPLEMENTED.  The differences are in the quality/engineering of the implementation.  Point being, we have already solved the "problem"-for example, in my system, I do not fret about this stuff at all, I listen to music and it is glorious sounding.  My current digital interface provides way better performance than I ever got from a spinning disc player, and way more convenience in use as well.

In fact, i am much more interested in perfecting other aspects of my system as there is much more in terms of performance to be gained elsewhere.  For example, I am currently engaged in working on different vibration isolation/control devices and the results of this are making for much bigger improvements in SQ than anything I could do with my digital interface (which is so close to "perfection" as to be a non-issue).  Next step, trying speaker de-coupling vs. coupling.

[sandyk]

5 hours ago, barrows said:

EI core, and flat packs with separate winding bobbins are much better at isolating AC line noise from their outputs because the windings are separated by some space.

 

Agreed.  They have much less capacitive coupling between the Primary and Secondary windings, but a small drawback is  their lower efficiency.

[Speedskater]

I want to apologize to Ayre, They are not one of the poorly designed products that I was referring to. (had to dig out my old magazines)

But for an example of a poor design, see the Feb. 2018 Stereophile magazine. (and two good designs)

[Superdad]

4 hours ago, barrows said:

Current implementations of I2S are mostly flawed.  It does not have to be this way, I2S could be "perfect", but the way most use it is not.

 

The PS Audio standard for I2S places the masterclock in the source component, then creates a balanced signal for output.  the devices which create the balanced signal add jitter (and again at the receiving end converted it back to unbalanced).

 

One could make the I2S system "perfect" by placing the masterclock in the DAC (where it should be), and sending it back to the source from there.  There would still be extra jitter created by the conversion to balanced and back, but it could be eliminated by a simple re-clocking stage in the DAC (POTATO flip flops), as the re-clocking could be done done directly from the (now local) masterclock.  I believe that MSB is the one company who implements their I2S interface this way (but I am not absolutely sure), of course this makes their products only work together, not with other I2S devices, which generally follow the PS Audio standard.

 

USB already does this the right way, with the masterclock in the DAC.  It appears that you keep thinking the USB interface is flawed: IT IS NOT.  Asynchronous USB is a great interface design, and is capable of producing the absolute lowest jitter possible when implemented correctly.   The only problem is that some do not implement it as well as it should be (and the same can be said for any other interface as well, they all need proper execution to work well).

 

+1000.  Barrows reiteration of the facts should be kept in mind by everyone!

 

Remember:

For these packet data interfaces--internal to a DAC where they belong for proper master/slave clocking--it will be either USB>I2S/DSD or Ethernet>I2S/DSD.  There is a PHY with both, and blocking external leakage matters with both.

 

Ethernet has a little electrical advantage in that there is always a tiny transformer behind each RJ45 jack, so most (but not all) leakage is blocked.  Beyond that, and the fact that you can run a very long cable to it, there is not a huge advantage.

And as with USB, it has to be done right!  Power supplies, isolation, clocking and reclocking, etc.

 

USB presently has a convenience advantage in that all the major computer operating systems directly support USB ports for audio output.  Yes, Ethernet has a range of possible "endpoint" modes--varying with the hardware (DLNA, RAAT, NAA, AES67, AVB, SqueezeLite, etc.)--but truly mainstream adoption (i.e. all new DACs having an Ethernet instead of USB) won't happen until some "virtual sound card" (VSC) software is either freely available on all platforms or better yet, built into popular computer OS.

 

I've been saying the above for at least 5 years, and we are still waiting.  Hopefully not too much longer... 

 

[Ralf11]

barrows, I'll forward to hearing about your  different vibration isolation/control devices ...

 

 

There is another aspect to interfaces besides SQ (or noise reduction) at any price, and that is to bring prices down to a point where high SQ can be implemented easily in most products.

 

Do you have any thoughts on USB-C as providing any easier path for high SQ?

 

 

PS: DeFlatriots SUCK

[mansr]

1 minute ago, Ralf11 said:

Do you have any thoughts on USB-C as providing any easier path for high SQ?

USB-C is just a different connector. Apart from permitting higher power transfer, it is electrically identical to other connectors.

[fas42]

4 hours ago, barrows said:

In fact, i am much more interested in perfecting other aspects of my system as there is much more in terms of performance to be gained elsewhere.  For example, I am currently engaged in working on different vibration isolation/control devices and the results of this are making for much bigger improvements in SQ than anything I could do with my digital interface (which is so close to "perfection" as to be a non-issue).  Next step, trying speaker de-coupling vs. coupling.

 

Yes, the way to approach system optimisation. Hint: make the cabinet of the speakers be the equivalent of a bank safe, in terms of their stability in the listening environment - whatever you can do to increase their effective mass will be a plus, IME.

[beerandmusic]

6 hours ago, barrows said:

My main point is that there is not any single "magical" interface which solves all interfacing "problems".

 

this also sounds logical to me, and the reason that i think that the less interfaces, the better.....ultimately, don't create an interface to interface.....we may not be there yet, but i believe some day they will figure it out.

[beerandmusic]

7 hours ago, barrows said:

 my digital interface....(which is so close to "perfection" as to be a non-issue). 

 

Ok, well that is good to hear.  by reviews comparing the two, it sounded as though they both had their shortcomings of the other.

 

If you believe that both are so close to perfect, and even if there was zero noise introduced, that the sonic differences would be very subtle as well (e.g. your inferred "close to perfection"), then I can rest that enet interface in other equipment is likely very close to perfect as well, and there is no need to wait for any improvements in the digital domain.

 

I know the other areas are far more important than the circuitry prior to the conversion.  It was just very difficult for me to accept that the digital interface cannot come as "close to perfection", that there would not be ANY sonic difference...but if yours is close to perfect, and many believe the sotm is better, than i can assume we have reached the plateau in this regard.

[beerandmusic]

7 hours ago, barrows said:

  All interfaces are quite capable of excellent performance, WHEN THEY ARE WELL IMPLEMENTED. 

 

I know i have seen you state many times that USB is fine also, as long as it is well implemented.

But you also state the enet is better because it allows you to keep the noisy pc interface away from the dac.

 

So the two statements kind of contradict one another?

 

Or do you believe that a USB only dac can be implemented to isolate noise where it would be "equally capable" as an enet interface? 

 

Understanding of course that it may cost a lot of money and well design....do you feel the usb->usb interface can equal or better an enet-usb design?

[barrows]

48 minutes ago, beerandmusic said:

then I can rest that enet interface in other equipment is likely very close to perfect as well

Then you would be making a big mistake, that is a huge assumption.  Remember all the details I talked about in the Sonore Signature Rendu SE, I guarantee that, for example, the Teac device you constantly refer to does not have those details (dedicated transformer for the Ethernet interface, ultra low phase noise clock, multiple ultra low noise linear regulators).  When I said close to perfect, I was referring specifically to the Signature Rendu SE, not any other devices.

[barrows]

33 minutes ago, beerandmusic said:

Or do you believe that a USB only dac can be implemented to isolate noise where it would be "equally capable" as an enet interface?

Yes.  But you will still need a low noise high quality source component to provide the best possible USB signal, not a consumer grade computer.  Remember, we want to get the consumer grade computer gear away from the audio system.

It appears that you keep wanting to find a simple and cheap solution to achieving the best possible sound quality, and I keep pointing out that this is not going to happen, all the details are going to matter no matter what you say.

For example, I am working on a DIY DAC right now.  I am trying to make everything as good as it can possibly be, while still keeping it (somewhat) simple.  today I built a dual output power supply for it, this is just two of the 7 supplies I will be building.  There are about $230.00 in parts just for this supply alone, plus another $100 for its dedicated transformer.  I am building a really good supply, but this is just 2 of 7 such necessary really good supplies and this cost does not include my labor.  all these details matter, there is no simple, cheap solution if you want to achieve really good results.

[Ralf11]

3 hours ago, fas42 said:

 

...make the cabinet of the speakers ...

 

 

What kind of pervert would build a cabinet around his speakers???

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Did you ruin Quads or Maggies??

[beerandmusic]

10 minutes ago, barrows said:

Yes.  But you will still need a low noise high quality source component to provide the best possible USB signal, not a consumer grade computer.  Remember, we want to get the consumer grade computer gear away from the audio system.

 

Ok, thanks for this information.  The only thing i have tried as a usb source is a regular pc, which i already speculated was the problem back when i found that enet via dlna using a $50 blueray sounded better.

 

....i also found that toslink out of same computer sounded much better (but then couldn't get dsd).

 

I may try more usb options in future.

 

[fas42]

3 hours ago, Ralf11 said:

 

 

What kind of pervert would build a cabinet around his speakers???

 

Did you ruin Quads or Maggies??

 

Around? Think in these terms - that the structure that supports the driving mechanism of the speakers, however they are constructed, remains as motionless as possible.

 

A big part of my first good sound was that the chipboard boxes used for the B&W bookshelf speakers were very strongly coupled to the concrete sub-floor - gave me very tight, subjectively intense bass quality - never have the slightest desire for a subwoofer ...

[Ryan Berry]

10 hours ago, Speedskater said:

I want to apologize to Ayre, They are not one of the poorly designed products that I was referring to. (had to dig out my old magazines)

But for an example of a poor design, see the Feb. 2018 Stereophile magazine. (and two good designs)

Thank you for the follow-up, I like to think we work hard to do things right. We often wish right didn't mean expensive too!

[beerandmusic]

4 hours ago, barrows said:

 I guarantee that, for example, the Teac device you constantly refer to does not have ...dedicated transformer for the Ethernet interface, ultra low phase noise clock, multiple ultra low noise linear regulators).  When I said close to perfect, I was referring specifically to the Signature Rendu SE, not any other devices.

 

My guess is that most streamer/dacs (even high-end) don't have a separate transformer for the enet intfc alone, that the engineers don't feel it is necessary and probably would just add more noise to have a 3rd one...You have a dedicated one because it wouldn't work without one.  As far as "ultra low phase noise" clocks, i don't know, they do market saying "high precision low noise clocks", and option for external clock for even better clocking..In one file they even state "noiseless reference internal clocks".  As you and others say its impossible to be noiseless, but the fact that they would use the term, i would tend to believe they have "really low noise", and debatable as to who has lower noise?