Clock?

[aasbakk]

One thing I don’t quite understand in digital audio is the so called clock. 

 

My digital chain is:

Router -> Ethernet cable -> RPi 4B -> USB -> Bel Canto e.DAC 1.7

Regarding clocks I guess the RPi has one, but the only clock that matters here is the one in my Bel Canto DAC?

 

Please correct me if I’m wrong

 

Additional question:

Is USB the best output on RPi, or is it worth considering a Digi-HAT? 

 

[Snoopy8]

It can open a can of worms and does depend on whether you believe that only the DAC matters or all the components matter in the digital chain.  An example of using a reference clock in the digital chain is in this long review of an audiophile switch.

https://audiophilestyle.com/ca/reviews/uptone-audio-etherregen-review-and-comparison-r887/

 

As for USB versus other outputs for the Pi, this is dependent on the DAC (some favour a particular input) and quality of cable.  Last, but not least, personal preference.

 

I prefer USB for my setup.  If you want a big jump in USB sound quality, consider this.

https://www.stereonet.com/forums/topic/596937-gieseler-audio-new-base-model-usb-isolatorre-clocker/

 

[aasbakk]

14 hours ago, Snoopy8 said:

It can open a can of worms and does depend on whether you believe that only the DAC matters or all the components matter in the digital chain.  An example of using a reference clock in the digital chain is in this long review of an audiophile switch.

https://audiophilestyle.com/ca/reviews/uptone-audio-etherregen-review-and-comparison-r887/

 

As for USB versus other outputs for the Pi, this is dependent on the DAC (some favour a particular input) and quality of cable.  Last, but not least, personal preference.

 

I prefer USB for my setup.  If you want a big jump in USB sound quality, consider this.

https://www.stereonet.com/forums/topic/596937-gieseler-audio-new-base-model-usb-isolatorre-clocker/

 

Thanks for your answer. 
I tried to read about this a couple of times before but think it is complicated. And maybe there isn’t an easy answer to it. 
Link to Gieseler was interesting. Thanks

[Snoopy8]

7 hours ago, aasbakk said:

Thanks for your answer. 
I tried to read about this a couple of times before but think it is complicated. And maybe there isn’t an easy answer to it. 

Complicated sums it up.  And expensive!  

 

There is a simple way to reduce clocking issues in the digital chain.  Use fewer components by getting an all in one streamer/DAC.  But this will take things off topic...

[MTL]

@aasbakk

https://audiobacon.net/2018/09/26/sotm-sclk-ocx10-a-high-purity-10-mhz-master-clock-generator/amp/
 This review may help to explain a little about how a clock fits in

[davewantsmoore]

On 30/12/2023 at 9:10 AM, Snoopy8 said:

As for USB versus other outputs for the Pi

The onboard USB for the Pi is not particularly good, but will be quite dependant on the receiving circuit (as the "not good" doesn't necessarily need to be a problem).

 

The I2S out of the Pi GPIO pins is very bad (and so highly dependant on the circuit in the "HAT" you use, or whatever the I2S receiving circuit does, if you are wiring something directly to the Pi GPIO pins).

 

1 hour ago, MTL said:

 This review may help to explain a little about how a clock fits in

External clocks are typically used to "sync" multiple converters.

 

Using an external clock in an audiophile context (with only one converter) can be tricky and depend a lot on matching the clock to the circuit being clocks (senstive to cables/impedances).

[MTL]

John Swenson suggests that a clock is less sensitive to impedance matching if it outputs a sine wave as compared to a square wave-hope I got this correct. I use a 50 ohm cable to connect to a 10Mhz input 

Edited December 31, 2023 by MTL

[Snoopy8]

2 hours ago, davewantsmoore said:

The onboard USB for the Pi is not particularly good, but will be quite dependant on the receiving circuit (as the "not good" doesn't necessarily need to be a problem).

It was not so good in the Pi's 1 to 3, but I think better in Pi 4 and 5 ?

[Keith_W]

2 hours ago, davewantsmoore said:

External clocks are typically used to "sync" multiple converters.

 

This is a point which needs to be emphasized. External clocks are NOT there to improve the clocking in your DAC. They may be more precise, have less jitter, etc. but ultimately the performance of the clocking signal is degraded by the simple fact that it is external. The best place for a clock is right next your DAC, and by that I mean on the same board, at most a couple of centimeters away. 

 

The reason pros use external clocks is if they have multiple ADC's, DAC's, mic preamps, mixers, etc. and need to sync all of them up. Even then, some protocols like Dante and Ravenna have embedded clocks, or are able to sync all devices to one clock. We might like to think of pros as being more rational than average audiophiles, but in reality they are as prone to magical thinking as some of us. 

[davewantsmoore]

On 31/12/2023 at 12:16 PM, Snoopy8 said:

It was not so good in the Pi's 1 to 3, but I think better in Pi 4 and 5 ?

AFAIUI it still has a pretty bad 8khz glitch... but to be fair it seems, almost everything does.

 

As always it's' really about what happens to the digital audio later.  Anything and everything can be overcome if there's a good clock re-applied once you get out of USB land.

[davewantsmoore]

On 31/12/2023 at 11:01 AM, MTL said:

John Swenson suggests that a clock is less sensitive to impedance matching if it outputs a sine wave as compared to a square wave-hope I got this correct. 

Yes.

But you need a square wave clock for audio devices.

 

Super high enders sometimes use a sine to square converter, so they can use sine wave clocks.   For the gear I use (iancanada) it is said to be a tangible benefit... I haven't tried it though.  

[Keith_W]

10 minutes ago, davewantsmoore said:

Yes.

But you need a square wave clock for audio devices.

 

No you don't. You only need the amplitude to cross the threshold for it to be interpreted as a 1 or 0. Whether the top of the wave is square or round is irrelevant. Square wave is only for nice illustrations in textbooks. 

Edited January 1 by Keith_W

[MTL]

I’m using an Sotm which appears to output a sine wave they say using analogue circuitry-not sure how they do it ,but still using a Shunyata 50 ohm clock cable to avoid impedance issues

I used an esoteric g01 prior to this and it as well as all the other esoteric clocks output sine waves

Edited January 1 by MTL
Added info

[rand129678]

On 31/12/2023 at 10:09 AM, davewantsmoore said:

The onboard USB for the Pi is not particularly good

 

It is perfectly fine. The only jitter that matters is at the point of D to A (and A to D).

 

Hopefully the DAC handles this - easy to measure with a Cosmos ADC.

 

USB Ground noise can show up in DAC analogue output. Use balanced output.

 

If DAC doesn't have USB galvanic isolation, simply add an Intona or such device.

 

 

[Addicted to music]

2 hours ago, Keith_W said:

 

No you don't. You only need the amplitude to cross the threshold for it to be interpreted as a 1 or 0. Whether the top of the wave is square or round is irrelevant. Square wave is only for nice illustrations in textbooks. 

 

Not correct. 

All digital circuits are designed to respond to a negative or positive edge that occurs within a specific time frame, a sine wave was never in the equation.  Therefore the digital input is designed to accept only to a almost perfect low to high or high to low ( only a square wave will do this and satisfy the condition) within a specific time frame,  anything outside that time frame its regarded as inactive.   Anyone claiming to the use of a sine wave as a clock signal are leading you down the wrong path.  An example: There is a known university lecture who makes his mark on these pages claiming replacing a clock circuit on a OPPO device with a clock that generated a saw tooth improves SQ of the OPPO device,  stay clear of any "clock modification" that can deviate and worsened the performance. 

 

To the OP dont play with things you read on the WWW if they dont give clear measurements of before and after to indicate improvements.   

[Keith_W]

3 hours ago, Addicted to music said:

 

Not correct. 

All digital circuits are designed to respond to a negative or positive edge that occurs within a specific time frame, a sine wave was never in the equation.  Therefore the digital input is designed to accept only to a almost perfect low to high or high to low ( only a square wave will do this and satisfy the condition) within a specific time frame,  anything outside that time frame its regarded as inactive.   Anyone claiming to the use of a sine wave as a clock signal are leading you down the wrong path.  An example: There is a known university lecture who makes his mark on these pages claiming replacing a clock circuit on a OPPO device with a clock that generated a saw tooth improves SQ of the OPPO device,  stay clear of any "clock modification" that can deviate and worsened the performance. 

 

To the OP dont play with things you read on the WWW if they dont give clear measurements of before and after to indicate improvements.   

 

You might like to watch this. 

 

 

[Addicted to music]

5 hours ago, Keith_W said:

 

You might like to watch this. 

 

 

 

Not required, I’ve already seen it.   Amir just touches on the subject on digital fundamentals and it’s basic,  note what he saids about signal transition regardless of level is correct,  my post covers what Amir leaves out.  This is why digital is so good,  what’s been discussed at lengths of dirty noise that can effect transmission Amir covers that in great lengths,  regardless of what you read here or on other sites or what’s been marketed,  digital is literally immune to any noises that audiophiles dream up!  And can never prove with live measurements,  If the noise was that great to disrupt the transmission you are going to know about it;  because your little experiment on another thread confirms “buffering “ shows you exactly what happens.   And if someone tells you you don’t need a clock producing a square wave I suggest you move on to stop wasting your time.

Edited. For a digital signal to work where a transition from low to high;  a 90%  level or higher of that signal is required,  and when a negative edge is required where it goes from high to low,  less than 10% of that level is required.   This is also applicable to ethernet  and other digital transmission, it’s generic.   Unlike analog transmission the quality of the media and noise can severely degrade signal quality.

Edited January 1 by Addicted to music

[davewantsmoore]

11 hours ago, Keith_W said:

No you don't. You only need the amplitude to cross the threshold for it to be interpreted as a 1 or 0. Whether the top of the wave is square or round is irrelevant. Square wave is only for nice illustrations in textbooks. 

 

Will you post an example of a sine shaped clock waveform for audio (that doesn't use a square converter) ?

 

[davewantsmoore]

5 hours ago, Keith_W said:

You might like to watch this. 

That is all correct... but it doesn't make your comment about the clock correct.

[davewantsmoore]

11 hours ago, Addicted to music said:

Anyone claiming to the use of a sine wave as a clock signal are leading you down the wrong path.  An example: There is a known university lecture who makes his mark on these pages claiming replacing a clock circuit on a OPPO device with a clock that generated a saw tooth improves SQ of the OPPO device,  stay clear of any "clock modification" that can deviate and worsened the performance. 

You both have this somewhat wrong.

Sine and SAW clock inputs are then converted to a square wave.... you do not just "input a sine", as the "detection" of the threshold is not precise enough (it is a galaxy far far away from that).

 

You can see this in the design of the interface between the input clock (sine, saw, whatever) and the audio clock, which are effectively high gain RF amplifiers.

 

It is (can be) better to do it this way, because the source clock can be higher performance (jitter, especially low frequency), and the impedance match between the source clock, and where the audio clock goes can be much much better designed...... so it is very misguided to say "stay away from any clock modifications".    If really what you mean is that you "need to be very good at what you're doing to even measure the performance let alone harm it" ... then sure, that is true.

 

Doing these sort of things need very serious measurement, and RF isolation, pretty crazy power supplies, ovens, silly power supplies, and even designing out the piezoelectric effect.

 

 

[Keith_W]

4 hours ago, davewantsmoore said:

 

Will you post an example of a sine shaped clock waveform for audio (that doesn't use a square converter) ?

 

 

All of them I am aware of use a square wave. I was pointing out that the square wave does not need to be perfect. 

[stereo coffee]

9 minutes ago, Keith_W said:

 

All of them I am aware of use a square wave. I was pointing out that the square wave does not need to be perfect. 

Well it does, otherwise jitter is the byproduct. more so you need to allow the clock to interface with flip flops other slower lines, and  fully design to be rid of SPDIF as well. .   https://www.enjoythemusic.com/magazine/viewpoint/0401/deficienciesofspdif.htm

[davewantsmoore]

12 minutes ago, Keith_W said:

All of them I am aware of use a square wave. I was pointing out that the square wave does not need to be perfect. 

You need a square wave clock for audio purposes

[MTL]

17 minutes ago, davewantsmoore said:

You need a square wave clock for audio purposes

So youre saying cybershaft , esoteric and sotm etc are all incorrect?

[Keith_W]

1 hour ago, davewantsmoore said:

You need a square wave clock for audio purposes

 

You don't need a perfect square wave clock for audio purposes. 

 

This can go back and forth forever arguing whether you need a perfect square wave or not like a school playground saying "yes it is!" ... "no it isn't!!!".