Thanks Mark , so far I do prefer it bypassing the last section so will listen for longer and try reverting back just to be sure . I'd like a little more depth to the sound so look forward to trying any other tweaks . I'm reading through the thread trying to find any ideas posted to try out .
I might try a couple more shunts and run the output stage board +/-15v separate
I might try a couple more shunts and run the output stage board +/-15v separate
It is good practice to make some measurements as baseline before embarking on modifications. Then after each modification new measurements can be made and compared against the baseline. This just to ensure that the modifications have not resulted in performance degradation. Ears can easily be fooled.
Personally, I would not overly trust standard measurements when it comes to dacs. Some years ago ESS gave a seminar in which they briefly explained some things about how audiophiles hear, and that standard PSS measurements do not capture everything people hear. I can post a link to the slide deck and or post quotes from the relevant part of the presentation if you would like.
That said, I would mention that @bohrok2610 has posted some good measurements of PSS spectral line noise skirts.
That said, I would mention that @bohrok2610 has posted some good measurements of PSS spectral line noise skirts.
Link to Yumpu Presentation:
https://www.yumpu.com/en/document/read/23182504/noise-shaping-sigma-delta-dacs-ess-technology-inc
"WHY AUDIOPHILES KNOW BY LISTENING IF ITS A SIGMA-DELTA DAC" starts on page 28 of the presentation. From there I would suggest to read through the next several pages to the end. If there are any questions and or any need for clarification, and there have been some before, please do let us know.
https://www.yumpu.com/en/document/read/23182504/noise-shaping-sigma-delta-dacs-ess-technology-inc
"WHY AUDIOPHILES KNOW BY LISTENING IF ITS A SIGMA-DELTA DAC" starts on page 28 of the presentation. From there I would suggest to read through the next several pages to the end. If there are any questions and or any need for clarification, and there have been some before, please do let us know.
It's a pity they don't use the normal terminology, such as non-stationary noise instead of none-PSS noise. They are also abusing the term dither at one point, in a manner that unfortunately seems to be getting in fashion in the field of ADPLL design.
I wonder what they mean by unconditional stability. I know two definitions of that, stable for any combination of passive source and load impedances (does not seem applicable) and remaining stable when the loop gain is reduced while the open-loop poles and zeros stay where they are. The latter would very much restrict the feasible noise shaping, to second order when you don't use techniques such as MASH converters.
I wonder what they mean by unconditional stability. I know two definitions of that, stable for any combination of passive source and load impedances (does not seem applicable) and remaining stable when the loop gain is reduced while the open-loop poles and zeros stay where they are. The latter would very much restrict the feasible noise shaping, to second order when you don't use techniques such as MASH converters.
I took unconditional stability to mean exactly as you suggested.
Regarding non-stationary versus non-PSS, non-stationarity seems mostly defined in terms of statistics, whereas PSS is terminology used by RF engineers. IMHO, the concept of PSS is more directly relatable to what is easily seen on an FFT, as opposed to what is not.
Also, dither is a term that has been applied to any signal, including steady state signals, which has the effect of obscuring/decorrelating truncation distortion. UV-22 dither was such a steady state signal. https://audio.rightmark.org/lukin/dither/dither.htm
Regarding non-stationary versus non-PSS, non-stationarity seems mostly defined in terms of statistics, whereas PSS is terminology used by RF engineers. IMHO, the concept of PSS is more directly relatable to what is easily seen on an FFT, as opposed to what is not.
Also, dither is a term that has been applied to any signal, including steady state signals, which has the effect of obscuring/decorrelating truncation distortion. UV-22 dither was such a steady state signal. https://audio.rightmark.org/lukin/dither/dither.htm
Last edited:
That's typical of what happens when you start approaching very high performance levels of devices. There are lots of problems to solve and each one only gives a small improvement. Only by fixing all or most of them does the sound slowly start getting to the point that people can't believe you are playing something from a CD because it sounds so good. Also, it takes fixing things in the whole system, not just the dac. Otherwise the rest of the system will limit being able to hear improvements in the dac. That's one reason why really good dacs and really good systems tend to be expensive (which is not to suggest that expensive implies good, it doesn't). Just means fixing a lot of stuff can be complex and costly.
Another thing you may find is that solving some problems may sometimes require taking a step backwards before taking two steps forward in another direction. Using SE outputs may to be one of those things. Something is gained but something else is lost. The problem is then how to get back what is lost while keeping what is gained from SE. Turns out its doable, but figuring out a solution from scratch can take a lot of time and effort. That said, you can buy a dac board and support circuitry that solves most of the known problems. It just costs more than $1k to assemble them into a diy dac. That's what I am doing at the moment; the dac system includes boards from Andrea Mori, clocks from Acko Labs, and other devices which for now remain under NDA.
EDIT: One of things I did with Marcel's dac that gave one of the biggest improvements was to replace the Amanaero with the latest version of I2SoverUSB. The latter takes two isolated 5v power supplies for best performance. One thing it has that Amanero doesn't have is galvanic isolation. It helps with reducing common mode noise coming in through the USB connection.
Last edited:
I keep looking at the I2SoverUSB because I'd like to try the simultaneous mode with my AYA2 dac too as well as using it with the Marcel FIRDAC . Hopefully it's not too fiddly to use with the simple pcm2DSD , the Amanero plugs straight in to it . I mainly use a MacBook Pro
I'll have a look at the options for Andrea's dac , maybe something to try after Marcels
It possible to make short adapter cable (or adapter PCB), so I2SoverUSB can plug into an Amanero format pin header. Also I made a little board that can hold an I2SoverUSB and output Amanero format which I used when I had one of Marcel's dacs here on loan. The board wasn't really designed for that, but it works. Maybe I could send you an unpopulated one if you don't have any other solution.