I can't answer this because I don't understand the question. Mark has answered the question, but you will never know his answer, as you have blocked him.
Even it would be only about filter effects there would be things that are easier to measure (e.g. frequency response, impulse response) and things much harder to measure (e.g. numerical precision artifacts in multi stage filters, clipping effects with various kinds of source material).
But I would think everything matters if you want to create a top notch DAC... Technology, chips used, IV and buffer stages, components, phase noise, power supplies and its filtering, vref, digital and or analog filtering, muting, layout, PCB tech, decoupling, sample rates, bit depths, numerical precision, modulators, dithering, power supply, temperature consistency, lack of vibration etc.
(In no particular order)
Last edited:
By the way, measurements could mean various things. It could be restricted to output of DAC only or could also be internal (supplies, vref, phase noise etc.). 2nd would already be more insightful...
Last edited:
You are on slippery ground here. If the output of a dac measures as essentially perfect, who cares what goes on internally? Internal measurements wouldn't seem to matter, at least superficially ("superficially" being the key word).
The real issue is whether or not Amir's SINAD measurements are sufficient to be sure two dacs are audibly transparent. IMHO that's highly unlikely. And Amir has still not provided any "proof" that his theory is true in practice. All it takes is the existence of two dacs that measure well according to Amir's measurements, yet sound even slightly different to falsify his hypothesis.
Last edited:
Just be glad that someone does all that work and publishes it for others to read in an overseeable format. Pleasant to read and for readers good to compare devices. Can be handy to check what Amir has to say. The real bad ones can then be avoided. That is more to many than fragmented information with strong personal preferences.
It is about disposible entertainment/consumer stuff not lab equipment at a university.
It is about disposible entertainment/consumer stuff not lab equipment at a university.
Last edited:
The good news here is that ESS ADCs (ES9822pro/ES9842pro) offer very effective true brickwall filters, at least up to 384kHz. Not perfect ones but still way better than the available filters of any other ADCs I'm aware of.
Well, university labs need excellent audio equipment, too ;-)
But in general I do agree, ASR is fine as a database for expert users even with its severe restrictions and omissions. However, it is not a suitable platform for lay people, and that is definitely a problem.
Also ES9039Q2M has a linear phase apodizing fast roll-off filter that has full suppression at 0.5*fs. Filters in AKM dacs are mostly half-band filters. Still many claim that AKM dacs sound superior to ESS dacs so filter cut-off does not seem to be the culprit.
Even in consumer circles the terms "AKM DACs", "ESS DACs" and "mostly" are too broad. Is the ES9039Q2M the first or maybe the only DAC chip by ESS that has a linear phase apodizing fast roll-off filter with full suppression at 0.5*fs?
Which of the AKM DAC chips can then be compared to that ES9039Q2M? It would seem logical to only compare to the specific AKM chips that also have a linear phase apodizing fast roll-off filter with full suppression at 0.5*fs. Or I am I mistaking?
Which of the AKM DAC chips can then be compared to that ES9039Q2M? It would seem logical to only compare to the specific AKM chips that also have a linear phase apodizing fast roll-off filter with full suppression at 0.5*fs. Or I am I mistaking?
Last edited:
Filters in e.g. ES9038Q2M are slightly different but the brickwall filter is probably close to linear phase apodizing fast roll-off in ES9039Q2M. Strangely the filters in ES9039PRO have different profiles compared to ES9039Q2M.
I'm waiting for someone to check if the two DACs sound the same, of course both with good measurements. 🙄
The first step would be to verify that different filters in either dac can be reliably identified in AB test. This controversial video claims audible differences in dacs that use different filters:
Good measurements ≠ good sound. Look at this for example, do you like those graphs ? But this is just measurements and this is not a representive of how this sounds. And I know and I hear how the same DAC chip can sounds much better than in that realisation.
https://www.audiosciencereview.com/forum/index.php?threads/smsl-su-1-stereo-dac-review.44029/
Nixie62, we will be very old men when this is confirmed. Maybe something I can take to my grave.
I can imagine what might be a convincing test of whether there is an audible difference due to on-chip versus PC s/w, brick-wall FIR interpolation filter implementations, which, I suspect there is. It would, however, require a purpose designed DAC, and two sets of the same music test tracks, one CD standard, the other set pre-interpolated via PC s/w. Certainly, there are other details which I haven’t thought through here.
The DAC would feature a ubiquitous on-chip, half-band, equiripple digital interpolation filter which can be bypassed via switch selection. In one setting it is internally OS, and in the other it is NOS (internal interpolation filter bypassed). No, this won’t be a comparison of OS with NOS, so bear with me.
One set of test tracks would be normal CD rate of 44.1kHz for playing via the DAC’s internal interpolation filter. The other set would be interpolated via a sophisticated PC s/w interpolation filter, such as Remastero’s, ‘PGGB’, but there are others, and played through the DAC via its NOS mode. This would enable the subjective comparison of the two significantly different interpolation filter implementations, while keeping the other DAC circuits elements identical. Both interpolators accurately correct for SINC aperture roll-off, so there shouldn’t any concern over audio-band frequency response difference between the interpolators, although will be ultrasonic differences.
Any difference in subjective character should primarily be due to the difference in how each interpolator (one chip h/w, and the other PC s/w) is designed and implemented. Yes, any subjective objective difference between the interpolators would result from some objective difference, and be measurable under the right conditions (no audio magic), but only after it is known which parameter(s) to measure. Which are important, that hadn’t seemed important before. Of course, this is a test of the suspicion that a specific DAC functional block which is thought to be in-audible, a brick-wall FIR interpolator, is actually audible in most consumer DAC implementations.
As a rough example of the sort of test DAC I’m imagining, long ago, I created a simple experimental DAC featuring the AD1896 ASRC chip, the DAC had no other interpolation function. Essentially, just the AD1896 (acting as the digital interpolation filter), and an AD1865 DAC IC, plus analog circuits. It was, admittedly, quick and dirty. The AD1896 enabled on-the-fly switching between OS, and NOS (bypassing the interpolator) while listening to music, which is why it was chosen. The purpose of that DAC was to subjectively evaluate OS versus NOS, however, not compare different OS interpolators.
Last edited:
- Status
- Not open for further replies.