Return-to-zero shift register FIRDAC

[acko]

Both Amanero and I2SoverUSB can work with external master clocks, both for PCM and for DSD. Difference is Amanero requires 22/24MHz whereas I2SoverUSB uses 45/49MHz.

Used a clock multiplier chip to externally clock I2SoverUSB from my 22/24MHz oscillators, then reclocked the I2S output to de-jitter the clock multiplier phase noise.

EDIT: Reason for using a multiplier chip was so that the ratio could be set with jumpers. That way I could use 11/12Mhz clocks or 5/6MHz clocks without a redesign of the board. Just move a jumper.

Thank you, Mark
Another great setup, like SOTA to me. Again, interestingly no FIFO needed. So why were you planning to use Andrea’s FIFO to drive Marcel’s DAC?

 

[Markw4]

...why were you planning to use Andrea’s FIFO to drive Marcel’s DAC?

Because right now it has the Acko clocks hooked up and the lowest phase noise. Its also the most convenient to try at the moment, with Simple DSD converter and I2SoverUSB already connected.

 

[Hans Polak]

In hindsight, using the Amanero for testing was the easiest to connect but looking at what Bohrok achieved, not the best choice at least not without galvanic isolation and external MCK.
The I2SoverUSB board with galvanic Isolation and several optional add-ons would most likely have produced much better results.
A bit of sad feelings for all the time invested in testing.
You could call this a learning curve

Hans

 

[bohrok2610]

Another great setup, like SOTA to me. Again, interestingly no FIFO needed.

In practice asynchronous UAC requires an internal FIFO within the USB-I2S interface. That FIFO is kept in balance (i.e. without over/underruns) with UAC asynchronous feedback. External FIFOs are used to provide clean clock (i.e. separate from source clock) but most external FIFOs don't have such feedback mechanisms so they will eventually have over/underruns since source clock and FIFO clock are bound to have a small difference in frequency. Of course with large FIFO size and tiny frequency difference it will take ages for over/underrun to happen but theoretically external FIFO without feedback is flawed.

 

[acko]

@Mark, I am very pleased with all the efforts you put in for this setup for the listening tests. I think it has reached a stage where further work will require additional contributions and resources if further scope can be identified. Sometimes I wished if we could somehow teleport Hans & Co to your place for all the measurement tests but I reckon we can also accept what has already been presented so far. Many thanks also to all the others who contributed to this discussion🙏

 

[Markw4]

Regarding Andrea's FIFO, it resets to half full every time a pause in the music is detected. It can also be configured for a depth of up to 8Mb. Therefore underruns/overruns are not a practical problem for normal music playback use.

 

[Markw4]

...if further scope can be identified.

Haven't we already identified further scope? What about the effects of lower phase noise clocking on this particular dac? How might that affect sound stage? Do anyone know how to measure sound stage with electronic instruments, or are biological sensors the only practical way we have to assess performance of that type today?

 

[bohrok2610]

How do you assess the performance regarding sound stage? Do you know what was the sound stage used in the recording?

 

[TNT]

As it was probably assessed using studio recordings it will fall in the "nice" and not in the "accuracy" bucket. "Nice" is subjective and can never me measured. My guess is that it was created by a less accurate unit actually.

//

 

[bohrok2610]

That was my point. If this is about subjective preference then "performance" is inappropriate terminology.

 

[Markw4]

We can tell if better clocking tightens up left/right localization.

"Localization accuracy is 1 degree for sources in front of the listener and 15 degrees for sources to the sides. Humans can discern interaural time differences of 10 microseconds or less.[9][10]"

https://en.wikipedia.org/wiki/Sound_localization

 

[bohrok2610]

So this is about subjective perception, not performance.

 

[Hans Polak]

Isn’t listening to music not all about subjective perception

Hans

 

[Markw4]

Its about biological sensors in practical applications.

 

[bohrok2610]

Do you have layouts for the room, speaker placement and especially the jig that holds listener's head tightly to prevent movement?

 

[bohrok2610]

Isn’t listening to music not all about subjective perception

Absolutely! It has nothing to do with performance.

 

[Markw4]

Your position is noted. Thank you.

 

[gpapag]

Haven't we already identified further scope? What about the effects of lower phase noise clocking on this particular dac? How might that affect sound stage? Do anyone know how to measure sound stage with electronic instruments, or are biological sensors the only practical way we have to assess performance of that type today?

Recording space spacial information can be evaluated when playing back through loudspeakers.
Different loudspeakers/room interaction give different effect.
When playing back through headphones most of this information seems to vanish (brain unable to recreate the recording space?), despite the same biological sensor(s).
George

 

[MarcelvdG]

It can hamper the enjoyment of music when the musicians perform poorly.

 

[Hans Polak]

George,
The same is of course true for the unique acoustics in the space where the life music was produced and not to forget the very important position you occupied in that space.
There is simply no such thing as absolute correct sound.
Too many variables.

Hans