SPDIF vs USB

A lot of this is over my head.

But my Proacs are like a window. I've been listening really acutely to Eric Bibb and Habib Koite - Blowin' In the Wind.

Zenith > Supra USB cable > Hugo

Zenith > Supra USB cable > Gustard U12 > Optical > Hugo

You could be tricked into thinking direct USB is better as it sounds slightly louder, alluding to more detail. But it adds a false accentuation or push to the sound of both singers voices and instruments. Jabbing at your ear 'oles.

Compared with the Gustard U12 SPDIF in the chain which is more organic, no forced edges or notes. It glides. Softly softly like a whisper.

Now, I have a Melco coming this week for a listen hopefully. I'll tell you of I hear the same. But I won't be getting a Rendu. Too much hassle and cost. Especially now the Uniti Core is announced which will be easier to demo. If the Uniti Core turns out to be weaker then I'll be damned and then try a Rendu.

Oh and the Regen in the loop might also taken it to the same level as the optical? We'll see this week (If the postie gets a move on).

I'm sure USB can work but it seems like a lot of effort.

Loudness and forwardness is deceiving at first.

fatcat posted:

Simon-in-Suffolk posted:

fatcat posted:

Simon-in-Suffolk posted:

However, as I believe I might have said a few times, cross system coupling means transport clock jitter will manifest itself as noise in connected circuitry. This noise can modulate digital clocks or analogue ground planes etc.. it will be incredibly low level, but typically the perturbations caused I understand our brains are most sensitive to.

Simon.

It’s the other way around, noise produced in the transport manifests itself as jitter. Reclocking doesn’t eliminate the noise, it eliminates the effect caused by the noise.

 I think I’ve mentioned this a few times before.

Fatcat, not sure I follow what you are trying to say..

Noise in the output signal of the transport, and in the digital processing circuitry appears as jitter when viewed in the digital domain (but remains as noise when viewed in the analogue domain) and this acts in addition to the jitter that's already there due to other causes.  Reclocking substantially reduces the jitter (in the digital domain).  However, the noise in the signal and the digital processing circuitry, and the sidebands of the remaining clock jitter will still be coupled to other parts of the circuitry (to some extent).

Hi Fatcat,

In my understanding the "digital domain" is another definition of information theory and the way it can be transferred and stored as information (0 & 1's). So I don't think it is possible to explain jitter in the digital domain; it has no meaning.

Jitter is a deviation of the "ideal" electromagnetic wave (analogue of course because digital signals don't exist) that can be defined as noise. The challenge of the reclocking proces is to recreate this "ideal" electromagnetic wave without introducing other noise. I fully agree this is a very tough challenge because one way or the other, all circuitry is (more or less) coupled electromagnetically. 

Peter

fatcat posted:

 ...Reclocking substantially reduces the jitter (in the digital domain).  However, the noise in the signal and the digital processing circuitry, and the sidebands of the remaining clock jitter will still be coupled to other parts of the circuitry (to some extent).

well exactly - you have to reclock SPDIF to create a sample stream  - or at least in the olden days put the transport clocks through a clock frequency modifier - because SPDIF is not a serial bit stream - its a protocol of framed data with headers and so has to be extracted and rebuilt into a bit stream. Because clocks are relatively cheap and accurate these days its easier to reclock rather than adapt from the source clock - and it is also a better solution... so system noise coupling aside the new clock theoretically should be decoupled from the transport clock - but of course in closed systems and through system coupling this is not the case in the real world.

Yes any noise in a clock frequency is  jitter. However clock oscillator accuracy noise will typically I believe have a non linear distribution (possibly Gaussian i can't remember without looking it up) around the fundamental - where as white  noise modulating the clock will have a linear distribution around the fundamental. I think in audio circles most tend to regard jitter as the former.. and yes most of these perturbations we hear I believe are caused by closed system coupling - i.e. its a case of applying good old fashioned closed system theory. 

Peter I think fatcat when he was referring to 'Digital Domain' he was meaning  a discrete data series in the Time Domain.

Simon

The term "digital domain" can have different meanings dependant on context.  In the context of electronics, (specifically digital signals in mixed mode electronics as distinguished later from "viewed in the analogue domain"), it refers to the electrical signal carrying the information (and this is the context in which I used the term, and later reused by FatCat).

However, in the context of information theory (the information as carried by the digital electronic signal), yes, jitter is meaningless (although too much jitter can corrupt the data, but in the information context this is seen as corruption not jitter).

The other interaction between noise and jitter worth mentioning is the effect on the timing of the Lo/Hi and Hi/Lo transitions, due to the finite nature rise and fall times.  If noise adds to the signal voltage during a Lo/Hi transition, then the signal will cross the upper threshold earlier than would be the case without the noise.  Similarly, if noise reduces the signal voltage during a Lo/Hi transition, then the signal will cross the upper threshold later than would be the case without the noise.  The reverse happens at the lower threshold for Hi/Lo transitions.  This is a mechanism by which noise on a digital line can cause jitter in the digital signal, even when Schmidt triggers are used.

Hi, I am doing an experiment on sound quality.  1) connecting my PC USB + Audiphilleo 2 with pure power -> DAC SPDIF; and, 2) streaming from the PC (Server)  to a Raspberry Pi 3 (Client) as end point for Roon via Ethernet and connecting the Audiophilleo 2 to the USB port in the Pi 3 and to DAC via SPDIF.   I am using a linear power supply for the Pi.

I feel that the sound quality via the combination Pi + Audiophilleo is cleaner, less noisy. Though the PC+Audiphilleo -> SPDIF is fantastic too.  So difficult to decide.  I may be just trying to split hairs here.   I don't know the technical reasons, but if may have to do with the low voltage (5v) of the Pi thus generating less electrical noise.  Maybe this is the same principle behind of the MicroRendu.

Experimenting...

Hi BPou, yes as you're finding, trying to explain the more subtle of interactions between components is, well 'complex'!

Huge posted:

The term "digital domain" can have different meanings dependant on context.  In the context of electronics, (specifically digital signals in mixed mode electronics as distinguished later from "viewed in the analogue domain"), it refers to the electrical signal carrying the information (and this is the context in which I used the term, and later reused by FatCat).

The other interaction between noise and jitter worth mentioning is the effect on the timing of the Lo/Hi and Hi/Lo transitions, due to the finite nature rise and fall times.  If noise adds to the signal voltage during a Lo/Hi transition, then the signal will cross the upper threshold earlier than would be the case without the noise.  Similarly, if noise reduces the signal voltage during a Lo/Hi transition, then the signal will cross the upper threshold later than would be the case without the noise.  The reverse happens at the lower threshold for Hi/Lo transitions.  This is a mechanism by which noise on a digital line can cause jitter in the digital signal, even when Schmidt triggers are used.

Hi Huge,

IMO we are talking about streaming media so the D/A output should be the analogue stereo signal as originally recorded with no added noise or other artefacts.

For understanding how the streaming media is processed it is convenient to use "digital" and "analogue". But there are no digital signals. F.E. the "jitter" means different timing of transitions; This means different timing of energy-supply to the trigger; This means a different timing in the demand for energy-supply to the trigger and thus in supply voltages and groundplane. This all will have it's effect on the "cleanliness" of the information (the electromagnetic waves) that are processed and the electrical signal output of the streamingproces.

I think this is the same as Fatcat stated with "be coupled to other parts of the circuitry (to some extent)"

Bowers posted:

Huge posted:

The term "digital domain" can have different meanings dependant on context.  In the context of electronics, (specifically digital signals in mixed mode electronics as distinguished later from "viewed in the analogue domain"), it refers to the electrical signal carrying the information (and this is the context in which I used the term, and later reused by FatCat).

The other interaction between noise and jitter worth mentioning is the effect on the timing of the Lo/Hi and Hi/Lo transitions, due to the finite nature rise and fall times.  If noise adds to the signal voltage during a Lo/Hi transition, then the signal will cross the upper threshold earlier than would be the case without the noise.  Similarly, if noise reduces the signal voltage during a Lo/Hi transition, then the signal will cross the upper threshold later than would be the case without the noise.  The reverse happens at the lower threshold for Hi/Lo transitions.  This is a mechanism by which noise on a digital line can cause jitter in the digital signal, even when Schmidt triggers are used.

Hi Huge,

IMO we are talking about streaming media so the D/A output should be the analogue stereo signal as originally recorded with no added noise or other artefacts.

For understanding how the streaming media is processed it is convenient to use "digital" and "analogue". But there are no digital signals. F.E. the "jitter" means different timing of transitions; This means different timing of energy-supply to the trigger; This means a different timing in the demand for energy-supply to the trigger and thus in supply voltages and groundplane. This all will have it's effect on the "cleanliness" of the information (the electromagnetic waves) that are processed and the electrical signal output of the streamingproces.

I think this is the same as Fatcat stated with "be coupled to other parts of the circuitry (to some extent)"

 

Hi Bowers,

It's impossible to have the original analogue signal without "added noise or other artefacts"; electrical noise at the output is inevitable and the jitter at the D/A transfer clock input (which is also inevitable) also causes unwanted modulation of the output.

There is a "digital signal", it's the serial input to the DAC chip, extracted from the buffered version of the input data stream.  The jitter is variation in the clock timing of the D/A converters's transfer clock (i.e. transfer from the input shift register to the D/A processing stages - for instance a Sigma/Delta converter).  On the other hand, the modulation of the supply lines and ground planes by the digital switching in the digital circuitry is an energy based effect, but this is quite different from the effects of jitter at the clock input of the D/A.

"...be coupled to other parts of the circuitry (to some extent)" is actually from Fatcat quoting me, so I have to agree with that don't I!  
Yes, this is how the modulation of the supply lines and ground planes by the digital switching in the digital circuitry influences the analogue circuits.

Huge posted:

"...be coupled to other parts of the circuitry (to some extent)" is actually from Fatcat quoting me, so I have to agree with that don't I!  

Which was an expansion of my statement

It’s the other way around, noise produced in the transport manifests itself as jitter. Reclocking doesn’t eliminate the noise, it eliminates the effect caused by the noise.

 So, I'll also have to agree with it.

Huge posted

 

It's impossible to have the original analogue signal without "added noise or other artefacts"; electrical noise at the output is inevitable and the jitter at the D/A transfer clock input (which is also inevitable) also causes unwanted modulation of the output.

Is it actually jitter that causes the modulation at the output, would the pulses cause modulation even if the jitter was infinitely small or zero.

Frank F posted:

The problem with Hugo is that it uses micro usb input and the supplied usb cable is crap as are most usb micro to usb cables.  I found that SQ dramatically improved using a proper, high quality usb A to usb B cable with an adaptor.

Case solved?

FF

I think usually of mre significance than the connector is that the galvanic isolation was not implemented due to the original mobile purpose requiring it to work with devices such as phones, associated with power capability of their usb outputs, IIRC.

Hugo's designer, says that the best sound quality is via optical, while some people report that electrical SPDIF sounds better than optical -  which may depend on the source as optical of course ensures complete isolation.

I've tried my Hugo on pretty much every input, have to say that despite repeated statements on here I prefer it with USB and a regen/jitterbug with Audirvana.  My USB cable cost all of £80 and was the least most obvious improvement compared to the others.  I think there is far too much naval gazing with some of this stuff, just find something that works for you and then forget about it and listen to some music 

fatcat posted:

Huge posted:

"...be coupled to other parts of the circuitry (to some extent)" is actually from Fatcat quoting me, so I have to agree with that don't I!  

Which was an expansion of my statement

It’s the other way around, noise produced in the transport manifests itself as jitter. Reclocking doesn’t eliminate the noise, it eliminates the effect caused by the noise.

 So, I'll also have to agree with it.

It's not the jitter that's coupled to other parts of the circuitry, it's the noise itself (and other electronic artefacts, most specifically the noise cause by switching in the digital circuitry) that's coupled to other parts of the circuitry.  Often (but not always or exclusively) this is coupled via modulation of the supply lines and ground planes.  This coupling (the effect) isn't eliminated by reclocking.

fatcat posted:

Huge posted

 

It's impossible to have the original analogue signal without "added noise or other artefacts"; electrical noise at the output is inevitable and the jitter at the D/A transfer clock input (which is also inevitable) also causes unwanted modulation of the output.

Is it actually jitter that causes the modulation at the output, would the pulses cause modulation even if the jitter was infinitely small or zero.

 

To what pulses are you referring?

Huge posted:

Hi Bowers,

It's impossible to have the original analogue signal without "added noise or other artefacts"; electrical noise at the output is inevitable and the jitter at the D/A transfer clock input (which is also inevitable) also causes unwanted modulation of the output.

There is a "digital signal", it's the serial input to the DAC chip, extracted from the buffered version of the input data stream.  The jitter is variation in the clock timing of the D/A converters's transfer clock (i.e. transfer from the input shift register to the D/A processing stages - for instance a Sigma/Delta converter).  On the other hand, the modulation of the supply lines and ground planes by the digital switching in the digital circuitry is an energy based effect, but this is quite different from the effects of jitter at the clock input of the D/A.

"...be coupled to other parts of the circuitry (to some extent)" is actually from Fatcat quoting me, so I have to agree with that don't I!  
Yes, this is how the modulation of the supply lines and ground planes by the digital switching in the digital circuitry influences the analogue circuits.

Hi Huge,

So with a "digital signal" you mean electrical charges that are transferred to another location in a chip (or elsewhere) and the timing of the transfer is controlled by a (oscillating) clock. However the charges are not digital, the voltages are not digital, the (transition) currents are not digital and time is not digital. This proces is based on electromagnetic theories as modellized by Maxwell.

I fully agree that it is impossible to have the orginal analogue stereo signals without added noise or other artefacts. I personally believe that when the outputsignal contains no signal related noise (f.e. only pink or white noise)  our brain cannot distinguish the result from the original.  

Bowers posted:

Hi Huge,

So with a "digital signal" you mean electrical charges that are transferred to another location in a chip (or elsewhere) and the timing of the transfer is controlled by a (oscillating) clock. However the charges are not digital, the voltages are not digital, the (transition) currents are not digital and time is not digital. This proces is based on electromagnetic theories as modellized by Maxwell.

...

Hi Bowers,

Yes, purely from the perspective of physics there's no such thing as a digital signal.  From the perspective of electronic design, there is such a thing - it'd be near impossible to design digital electronic equipment without using the concept of a digital signal.

Actually I was referring to the digital signal at the 'input pin' of the D/A device (be that a physical pin or a connecting track in an ASIC or FPGA etc.) that's used to load the 'bits' to be decoded into the input sift register, rather than in the operation of the converter itself, and considering this from the perspective of digital electronic engineering.

As a complete aside, with technical creativity, it's possible to make the concept even more remote: from the perspective of quantum mechanics then looking at the transfer of charge, one cannot know both the speed of the electrons and their location at the same time, so the process is merely statistical anyway (c.f. shot noise vs Johnson noise).  As everybody knows quantum mechanics is the only thing that really exists, reality is just a side effect of quantum mechanics.  

Bowers posted:

...

I fully agree that it is impossible to have the orginal analogue stereo signals without added noise or other artefacts. I personally believe that when the outputsignal contains no signal related noise (f.e. only pink or white noise)  our brain cannot distinguish the result from the original.  

I also believe that (with current technology at least!), it's impossible to have analogue stereo signals that are reproduced free of artefacts, even if you exclude the effects of uncorrelated (i.e. statistical) noise.

Each charge is of course quantized (the charge on the electron). But the digital signal in the real world is analogue - namely a step change in voltage between 1s and 0s. The change in voltage has to have a rise time, and the point where it starts or stops can be delayed by varying amounts depending on the circuit it is connected to.

Huge posted:

 

Hi Bowers,

 

As a complete aside, with technical creativity, it's possible to make the concept even more remote: from the perspective of quantum mechanics then looking at the transfer of charge, one cannot know both the speed of the electrons and their location at the same time, so the process is merely statistical anyway (c.f. shot noise vs Johnson noise).  As everybody knows quantum mechanics is the only thing that really exists, reality is just a side effect of quantum mechanics.  

 

I also believe that (with current technology at least!), it's impossible to have analogue stereo signals that are reproduced free of artefacts, even if you exclude the effects of uncorrelated (i.e. statistical) noise.

Hi Huge,

Maybe even more aside, I think our reality/universe is caused by quantum mechanics; You might call this a "side effect" indeed. But in addition; We know that we think to understand no more then 20% of the universe so what else is there ? 

I agree that there are no analogue stereo signals that are reproduced free of artefacts. And our brain can hear this artefacts whatever minor they seem to be.  It makes this "audio" hobby to a never ending project but I think we all enjoy the "ride".

Peter

ThatsNotMyNaim posted:

Brilliant posted:

I am sure you've heard or read (even on this forum) about a dedicated computer in a little box by Sonore  - the microRendu which has good USB. For sure when set up right, it improves over my HP 8440 laptop direct USB to DAC-V1. I am powering it with a TP 7 too, iFi 9V power on the TP-Link switch and using a mix of CAT 5e (router-switch)/CAT 6 (everywhere else) and digging it (CAT 6 alone resulted in too 'dry' a sound for some reason or another)!

I did not have the mR when I ran a Hugo - but its USB never really impressed as it's strong point, though it did get better with the Regen. It was  better on coax SPDIF.

Where did you go after Hugo?

Cheers

Adam

I have not replaced it yet. Hoping for a version 2 perhaps.

Huge posted:

fatcat posted:

Huge posted:

"...be coupled to other parts of the circuitry (to some extent)" is actually from Fatcat quoting me, so I have to agree with that don't I!  

Which was an expansion of my statement

It’s the other way around, noise produced in the transport manifests itself as jitter. Reclocking doesn’t eliminate the noise, it eliminates the effect caused by the noise.

 So, I'll also have to agree with it.

It's not the jitter that's coupled to other parts of the circuitry, it's the noise itself (and other electronic artefacts, most specifically the noise cause by switching in the digital circuitry) that's coupled to other parts of the circuitry.  Often (but not always or exclusively) this is coupled via modulation of the supply lines and ground planes.  This coupling (the effect) isn't eliminated by reclocking.

If the jitter is removed or reduced, it doesn’t remove the noise that produced it, that is an effect of the noise, noise is still present.

Huge posted:

fatcat posted:

Is it actually jitter that causes the modulation at the output, would the pulses cause modulation even if the jitter was infinitely small or zero.

 

To what pulses are you referring?

The ones in the analogue wave transferring the digital information.

fatcat posted:

Huge posted:

fatcat posted:

Huge posted:

"...be coupled to other parts of the circuitry (to some extent)" is actually from Fatcat quoting me, so I have to agree with that don't I!  

Which was an expansion of my statement

It’s the other way around, noise produced in the transport manifests itself as jitter. Reclocking doesn’t eliminate the noise, it eliminates the effect caused by the noise.

 So, I'll also have to agree with it.

It's not the jitter that's coupled to other parts of the circuitry, it's the noise itself (and other electronic artefacts, most specifically the noise cause by switching in the digital circuitry) that's coupled to other parts of the circuitry.  Often (but not always or exclusively) this is coupled via modulation of the supply lines and ground planes.  This coupling (the effect) isn't eliminated by reclocking.

If the jitter is removed or reduced, it doesn’t remove the noise that produced it, that is an effect of the noise, noise is still present.

Jitter is most often a caused by temporal variation within the clock circuit rather than being caused by external noise (and noise that's only internal to the clock circuit doesn't get coupled to anywhere - by definition!).  Jitter can be caused by external noise affecting the digital signal, but very often that isn't the primary cause.

fatcat posted:

Huge posted:

fatcat posted:

Is it actually jitter that causes the modulation at the output, would the pulses cause modulation even if the jitter was infinitely small or zero.

 

To what pulses are you referring?

The ones in the analogue wave transferring the digital information.

Transferring data from where to where (the source and destination have a strong influence on the effect of jitter)?

Are you considering the frame clock and/or master clock in this, or just the pulses of the data stream?

Huge posted:

Jitter is most often a caused by temporal variation

But jitter is a temporal variation.

Huge posted:

fatcat posted:

Huge posted:

fatcat posted:

Is it actually jitter that causes the modulation at the output, would the pulses cause modulation even if the jitter was infinitely small or zero.

 

To what pulses are you referring?

The ones in the analogue wave transferring the digital information.

Transferring data from where to where (the source and destination have a strong influence on the effect of jitter)?

Are you considering the frame clock and/or master clock in this, or just the pulses of the data stream?

I'm open to consider all pulses.