Effects of DACs defaults on our listening experience.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
...And how to improve them.

Hi folks,
while most of us knows how the analog defaults (distortion, IM, signal/noise ratio etc.), translate on our listening experience, may-be it should be good idea to discuss of the impacts of the various digital defaults on our listening experience.
How jitter problems translate on the reproduction ?
What are the most important parts to improve when we want to mod an existing DAC (Clocks, Power supplies, separate PS for digital/analog etc.)
Do different brands of DAC ICs (AKM VS ESS, by example) have a different "character" ?

The idea being to help unwashed people (like me ;-) in the digital world to know better how to improve or buy their ideal DAC, and the priority of the mods that can be done (from the most obvious improvements to the most subtil ones).

On my side, I have seen this DAC, witch seem to have a good quality price.
TOPPING D50 DAC ES9038Q2M x2 32bit/768kHz DSD512 XMOS U208 Argent - Audiophonics
The measurements are here:
Review and Measurements of Topping D50 DAC | Audio Science Review (ASR) Forum

I have too a DCX2496 (analog modified with a separate PSU and ultralow distortion+noise analog ICs for input/output)

And I don't know what i can do (in an economical way) to improve, if possible, both of them further. And if this could make any major differences
 
Last edited:
The question implies the answer.
That I don't understand ;-)
Let-me be more precise. As an old electronic designer (analog) When there is something I dislike in the reproduction of asystem, I know in witch direction to go to improve this. With digital, I"m blind. As, I believe, lot of people.
Separation, micro-dynamic, fluidity, sharpness, warmness, sound stage etc... Where are the sources in the digital side, what to modify to go one direction or an other ?.
 
Last edited:
I understand the question, but really you can't improve the digital part of a DAC, IMHO. All you can change is the surrounding analog parts: I/V converter, power supply, decoupling capacitors, and I count the master clock and receiver/PLL also here. Not talking about the oversampling filter, that is a different story. I don't believe you can swap it easily in a given DAC.
 
Disabled Account
Joined 2019
Greetings Tournesol,

Your music experience involves... your experience. You may like qualities of analog another people looks as a default. Measurement could be the only objective way to go but halas it doesn't say all about musical qualities nore it takes your preferences into account. And when it comes to measure a dac for tweaking purpose, it's certainly not easy to find always the culprits to improve.

All is important from the layout to the passive parts with all in between you wrote. Moving something may imply to move an other thing for compensation when it comes to check with ears your tweaks and choose the "defaults" you prefer, you also makes compensations in relation to the rest of your system : sort of EQ.
 
I think you are mixing up faults and defaults in the title of this thread.
my poor English, forgive-me.
Measurement could be the only objective way to go but halas it doesn't say all about musical qualities nore it takes your preferences into account.
After 40 years both in audio design and a parallel career as a sound engineer, in the absence of serious studies on our audibility thresholds and their correlation with our auditory experiences, I no longer take the measurement numbers for anything else than a playoff.
With >-115 dB of noise+distortion ratio of the Topping D50, hard to believe that a 10dB improvement can change a lot of what I will hear with it ?
On an other side, My DCX Berheinger was not measuring so bad neither, (0.0035%, ~ 100dB of signal noise ratio) but was far to sound nicely. Hence, my (naive) questions and my quest for totally subjective answers ;-).
 
After 40 years both in audio design and a parallel career as a sound engineer, in the absence of serious studies on our audibility thresholds and their correlation with our auditory experiences, I no longer take the measurement numbers for anything else than a playoff.
With >-115 dB of noise+distortion ratio of the Topping D50, hard to believe that a 10dB improvement can change a lot of what I will hear with it ?
You must be true.
So just keep your DCX.
 
And do you think that the measurements are really good ? :
Not on your opinion ?
Apart this strange IM accident (and I don't understand the vertical scale: Is'it under the signal level or under the 0db level ?).
What could be the reason of this strange behavior ?

You must be true.
So just keep your DCX.
Ironical, or serious and based on personal experience ?

In fact, I need would like good and cheap DACs to put into my (future) active 2ways enclosures+1 in my sub with digital links from the DCX.

For the moment, I am all passive.

Have bought a pair of Kef LS50 wireless, very disappointed, but I love the technical way it works.
 
Maybe I can help a little. As I have written before, dacs built to a particular price point and for low measured distortion usually cut corners everywhere possible with the main focus invested in achieving measurement numbers.

Power supplies matter, especially for final stages of regulation which need to be linear for best results. LDO regulators tend to work pretty well for digital and clock circuits. Things like AVCC and output stages need other kinds of voltage regulators, perhaps something like Jung-Didden (there are other good ones, as well).

Commercial ESS Sabre dacs may designed with one or two (sometimes more) 3.3v voltage regulators and some power filtering between different circuits. Audible improvements usually result for using dedicated high quality LDO regulators for each of Clock_module, DVCC, and VCCA (also 1.2v or 1.3v DVDD, for PRO chips). So, 3-4 good quality LDO regulators attached right at the ground plane (no inductive ground wires, please). A linear regulated 5v or 8v supply is helpful as a pre-regulator for the 3.3v LDOs.

For analog audio circuitry power, two good regulators are needed for AVCC (assuming a Sabre dac, or one reference for AKM). For mobile Sabre dacs such as ES9038Q2M, a dual opamp regulator following a low noise 3.3v reference can provide very good sound quality. (Please see ESS website downloads page for more recommendation details, or more updated info available in ES9038Q2M board thread). The output stage (I/V and differential summing/filtering, if Sabre dac) opamps, and AVCC opamp if used, need good quality higher current linear regulators for +-15v power. Jung-Didden would probably be fine, or perhaps Nazar regulators could be built for lower cost.

Most users find that upgrading the clock module also helps sound quality (instructions for removal/replacement available in ES9038Q2M board thread).

Output stage itself tends to sonically benefit from more LP filtering than ESS application note suggests. An output stage schematic known to sound good is available in the ES9038Q2M thread.

Trying to modify the existing output stage on a dac board may or may not be a good investment in time. It seems more likely to work reasonably well with one of the better Chinese ES9028PRO or ES9038PRO dac boards (often available for between $100 - $200). For cheaper boards, may make more sense to start over with a new output stage. (A limitation of the above mentioned Chinese PRO Sabre dac boards is that AVCC_L and AVCC_R are tied together, so one is pretty much stuck with less AVCC stereo separation).

A good galvanically isolated USB to I2S card may be needed for some dac boards. The only low-ish price full-featured one (important for best sound quality, say, if native DSD is desired) is probably this one: I2SoverUSB - I2S over USB Audio

With all the foregoing completed, one might have a pretty good standard asynchronous Sabre dac (with that standard Sabre sound). One could stop there, or one could go for more sound quality improvements, which can very significantly improve sound quality beyond what might be termed Sabre-standard (and having already completed all recommended mods above, the dac would be in a good state to start on such work). Please inquire if more modding information would be of interest.
 
Last edited:
On top of all that, attention to reducing emi/rf from the environment, albeit mostly conducted via ac power lines. Each stage of the operation deserves optimization if you expect to get great results, how that is done, or what your expectations are will vary from person to person. Some people like numbers, usually lots of zeros after the decimal point, while some go for a more realistic, life-like reproduction as a goal.

Brands of dac chips will have different characteristics, but there’s a lot between the chip and the speaker that can change that.

If using usb instead of a dedicated 12s source, then there are things to attend to within the computer that can help as well to reduce noise.
 
You can get the best sound in the world with cheap of the shelves sony cd player / dac, if you have good speakers and amps.

lets make an analogy, the dac is the cartridge, the turntable is the peripheral and phono stage.

each dac has its own character and there is a limit of how you can improve the sound, not much imo. Better dacs will benefit from a superior tailor made buffer/filter stage, but an ok dac will sound not much better.

if you buy a good turntable, a good needle and phono, you will forget about all this dac problems and collect records.

actually, as addendum, the reason dacs sounds bad is that the filters in them use inductors and this smears all the low level sounds and make the complex music sounds like a big mud ball. The Tda1541 has the SAA filter and if you use it correctly with 4x OS it filters most of the problems and leaves almost intact the music. Going from 16 bit 44khz to 32 bit 192khz doesn't solves the filter problem, no more than DSD with the MHZ filters, however we live in a world where the cheapest simplest solution with the best specification gets the deal.
 
Last edited:
Maybe I can help a little. .
Thank-you very much for all this. And specially the various tracks you indicate about the best kind of regulators etc.
Tried to read your lips, however, i cannot figure out how much those mods can make real difference and on what aspect of the sound character.
I ask this, because I was surprised by the relative quality of my little Chromecast audio, witch is not a good performer, Jitter side.
 
Thank-you very much for all this. And specially the various tracks you indicate about the best kind of regulators etc.
Tried to read your lips, however, i cannot figure out how much those mods can make real difference and on what aspect of the sound character.
I ask this, because I was surprised by the relative quality of my little Chromecast audio, witch is not a good performer, Jitter side.

There is zero consensus on these, despite what others may try to tell you.

Jitter is a fashionable scapegoat for anything that cannot be readily explained. Most of them not understanding that the intrinsic jitter of the clock inside the converter will dominate at some point. These chips don't even have differential clock inputs.

My observation is that the majority of mods done by members here actually make their gear objectively worse.

I'd just buy the Topping and leave it alone.
 
...i cannot figure out how much those mods can make real difference and on what aspect of the sound character...

Fully moddiing a Sabre dac as far as you can take it will sound much closer to the original master tapes than you probably ever thought possible. It depends how good the A/D was that was used for mastering. The best mastering houses had the best money could buy, and they are still surprisingly good even by today's standards.

For one example, I sometimes use the Steely Dan remastered Aja CD for comparing dacs and evaluating changes. There are many gradations of how that can sound. When the male vocalist sings along with multiple accompanying female backup singers, there are many different things that can be heard. With most systems the singer's voices sort of blend together to make a combined chordal sound, something like playing multiple strings on an electric guitar with a little distortion. It becomes a new sound, sort of a texture rather than individual notes.

In Black Cow, for example, when everyone is singing at once it is actually possible to hear every separate voice and come very close to identifying which vocal part belongs to which singer, who it is makes a little error for a small moment, etc. Small inflections in the voice of each singer become more clear and distinct, and yet sound beautifully recorded. The recording was not made with a vocal texture sound there. Also, the recording is not harsh or blurred at all (although maybe a bit of analog 'glue' processing effect).

Somehow you may think you are hearing it all now, but I assure you you are not. You will only believe how much difference is possible when you hear it for yourself.

I strongly recommend to go the whole way if you are up to doing the work and if you want to get started rather than wait for a few months (more below). It doesn't have to be done all at once, and you can quit with as much as you have completed at any time. Either you will be satisfied with the sound you have at that point or tired of modding dacs. Also, you should be aware that going any farther than I described in the earlier post requires changing the programming of the dac registers. One can learn how to do that with an Arduino (I can provide some advice and limited example code), say, or there is Dimdim's ES9038 controller project that can do most things people normally want to do (including synchronous mode, harmonic distortion compensation, DPLL bandwidth adjustment, etc.) Reducing DPLL bandwidth from the default values helps sound quality a lot, which is why there is an adjustment for it (the setting is limited by incoming jitter). HD compensation helps too. Don't know for sure if the HD comes from the dac, they kind of hedge around that by saying it can be used to help compensate for analog output stage HD imperfections. Personally, I doubt the compensation capability would be included if it wasn't needed to correct for dac chip imperfections.

While we have not gone into it in the details of how to do it in the ES9038Q2M thread, making a synchronous USB dac is probably the easiest path to excellent sound. It assumes that once you try upsampling to DSD512 on your PC that you will never look back, and never want to have only some lesser dac again. Sure, a cell phone or car radio can be okay now and then, but when you want to really enjoy the music to the utmost, your modded dac will be essential.

It could also be that the landscape will change over the next weeks and months. AK4499 is coming out and some are saying it was designed to be an ESS killer :) We will see how that turns out.

Also, I know of more than one entity working on good quality USB powered dacs that they hope will set new standards for USB powered dac sound quality.

I am also trying to tweak an Iancanada Dual ES9038Q2M dac more to my liking. If that works as I hope it will, it could take a lot of the work out of modding a dac at low cost into one that I would like (and I am very picky).
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.