All properly designed DACs must sound the same (SNR and such).
In theory the only difference (audible) can be:
type of filter (or NOS)
chip (ESS, AKM,…)
To answer this basic question, two questions must be answered.
Filter type is clearly measurable but is it really audible? Do you hear any difference by filter used? If so could you share track, source and specify what and when is audible using which filter?
Is there any audible (or even measurable) difference between chips? DA conversion is just mathematical formula (which is different for each filter) so there should be no difference.
Do you hear difference between filters?
YES, I hear
NO, I don’t hear
0voters
Do you think there is audible difference between DAC chips?
The first vote was relatively easy, although I prefer linear phase fast roll-off filters, since they are the closest to the ideal/theoretical reconstruction filter defined by the sampling theorem.
The second vote was not as decisive. I voted ‘no’ based on the assumption that the measurable differences between modern DAC chips are below audibility. If you include boutique DACs, that assumption may not stand. Also, if I were to build my own DAC, it would sound bad for sure.
I’m not voting on these, but like to offer my thoughts.
1
If you look at the time domain behavior (impulse response) of typical multi driver loudspeakers with crossovers and compare that mess - worse even when measured in a typical room - with the subtle change that different type DAC filters effect, it becomes rather obvious they will be extremely difficult to reliably audibly distinguish.
In the frequency domain, appreciable differences are mainly in the ultrasonic range - pretty inaudible as well, usually.
2
DAC chips can only be auditioned in conjunction with their analog output stages, so that seems like a rhetorical question without practical relevance to me.
But if DACs measure transparent to the source, i.e. all distortions somewhere below -120dB or so while driving a real amplifier of equally high resolution, there won’t be much left to audibly differentiate.
I don’t get it. Let’s say there is ideal absolutely transparent board. And on this board I’ll only swap latest ESS for latest AKM. Is there going to be any measurable difference or not? If yes then why? Both chips should convert to analog exactly same, shouldn’t they?
Well, I think it can be conceded that there is a discernible difference between sigma delta DACs and R2R DACs. So, if that’s true then the way any two DAC chips are designed must also imply a different sound. As long as by sound, one means soundstage, air, presence, etc. and not just tone.
Chips are tuned according to the designer’s ears, not some hypothetical listener. That’s why the voice chips, as an example, coming out of the Texas Instruments plant had a definite Texas twang.
Come on Jan, nothing’s perfect or ideal, we’re in the real world after all.
So yes there will be measurable differences, but you asked about audibility, right?
Were you there, or where did you get that story from?
I worked at a TI fab for 10 years in equipment engineering - been in that business for some 20 years - and can assure you, their designers are working with measurement equipment and not their ears.
Time is money, so numbers are what everyone’s looking at there - no audiophile hype time-consuming listening tests while rolling some silicon design.
Why would that be? A DAC can be called “good” if the measurable differences between its output and the ideal output are not audible. The way the actual D/A conversion is done internally is certainly interesting, but not a differentiating factor.
Over the years I’ve owned more than a dozen DACs from various manufacturers with many chips (Burr-Brown, ESS etc) I think the question is over-simplified and a somewhat reductive. I have never heard a ‘raw’ DAC chip. All have been part of a larger components from NAD, Schiit, Gustard, Anthem and others. Each one sounded unique and distinct from the DACs in CD/DVD players, AV receivers or tube amps they were connected to.
How these differences are achieved is for an engineer to explain.
Generally speaking, R2Rs measure worse than delta-sigmas due to the extremely narrow tolerances required by the resistors, so that may account for the differences people hear. Getting it right amounts to a waste of resources and money. If they do get it right and measurements are comparable, I’ll lay it down to confirmation bias. After all, people claim they can hear power cables. And what does “more analog” mean anyway?
If they do, I avoid them.
Different DACs in different rooms:
Topping D70
SMSL M8A
SMSL SU-8
I do have an R2R DAC BTW, but that’s because I’m a geek who likes to collect misguided tech. That’s why I have a Pono.
You have different DACs in different rooms because you listen in several places, not because a given DAC will sound different depending on the room, right?
If all DACs sound the same, why don’t you just have 3 DACs, all the same model?
Correct. The rooms and speakers are different anyway, so there’s no way to compare directly. (To be exact, one of the “rooms” is actually a pair of headphones).
I guess I just like variation. We use four cars in our family of five and they’re all different makes, for example. It’s more fun.