It is not about finishing second. Please remember that what Markw4 claims is "the truth" is not the truth for everyone. It is of course to be expected that he prefers the dac he already has since he has invested in it. It is very difficult to remain neutral.
Listening session tentatively scheduled for tomorrow. Hard to get people organized on short notice is all. We will see what the consensus is with Marcel's dac setup as it is now.
Later when the slightly long u.fl cables arrive, I will reconfigure the combination of Andrea FIFO and Acko clocks to drive Marcel's dac in Native DSD mode. Depending on how that goes I may take a look at driving the dac directly in differential mode as Marcel has specified. Seems likely the dacs will start to sound closer to alike the more their surrounding circuitry becomes alike.
EDIT: By the way, over time I have invested in a number of dacs including a Benchmark DAC-3 and a Topping D90. IMHO at this point Marcel's dac as presently configured sounds better than any of those. IOW, IMHO and IME if Andrea's dac is an "A+" then none of those other dacs is better than a "B" or maybe a "B+" at best.
That said, I built some custom dacs that sounded better than any of the commercial dacs I owned (mainly because of better and more costly surrounding circuitry). I think Andrea's dac sounds better overall than the dacs I built, but maybe not in every specific way. IME dacs with DEM/DES can have smoother sounding HF. Could be one or both of the AK4499 dacs would be at a similar overall level to Marcel's in terms of SQ (although they would surely sound different), would have to compare them directly to say more. Guessing Marcel's has more potential for improvement though.
Later when the slightly long u.fl cables arrive, I will reconfigure the combination of Andrea FIFO and Acko clocks to drive Marcel's dac in Native DSD mode. Depending on how that goes I may take a look at driving the dac directly in differential mode as Marcel has specified. Seems likely the dacs will start to sound closer to alike the more their surrounding circuitry becomes alike.
EDIT: By the way, over time I have invested in a number of dacs including a Benchmark DAC-3 and a Topping D90. IMHO at this point Marcel's dac as presently configured sounds better than any of those. IOW, IMHO and IME if Andrea's dac is an "A+" then none of those other dacs is better than a "B" or maybe a "B+" at best.
That said, I built some custom dacs that sounded better than any of the commercial dacs I owned (mainly because of better and more costly surrounding circuitry). I think Andrea's dac sounds better overall than the dacs I built, but maybe not in every specific way. IME dacs with DEM/DES can have smoother sounding HF. Could be one or both of the AK4499 dacs would be at a similar overall level to Marcel's in terms of SQ (although they would surely sound different), would have to compare them directly to say more. Guessing Marcel's has more potential for improvement though.
Last edited:
Since you have many people coming why not arrange the session so that listeners do not have visual cues to what device is playing? No need to "use eyes" if you trust your ears.
We can achieve the same type of result using a different perceptual testing methodology. A good overview of perceptual methods is given in a paper that happens to be on non-audio perceptual testing, although the same principles apply.
That said, sometimes we do double blind too. This time will probably focus on descriptive analysis by independent expert/trained observers.
That said, sometimes we do double blind too. This time will probably focus on descriptive analysis by independent expert/trained observers.
Last edited:
That’s funny.
George also noticed my comment and asked if I was willing to start a new thread on this specific item.
Hans
Rough comparison with Bohrok`s NDK:
https://www.ndk.com/en/products/upload/lineup/pdf/NDKT02-00008_en.pdf
LF noise is a bit different...
https://www.ndk.com/en/products/upload/lineup/pdf/NDKT02-00008_en.pdf
LF noise is a bit different...
Attachments
Nobody has claimed that NDK2520SDA (2.5 EUR) has lowest close-in phase noise. Why not compare it with the "audiophile" CCHD-957 that is 25 EUR. Also as has been discussed many times measuring clock phase noise separately says nothing about the phase noise at dac output which is what you hear.
Sorry, but I haven't expressed any desire to compare clocks or their phase noises. But as you can see NZ2520SDA has much lower phase noise than CCHD-957 which is highly touted by audiophiles
I used to work in the same building as a company that did taste tests. When, for example, testing Brussel sprouts, the sprouts were first put in glass jars that looked exactly the same, except for a label with a number that helped to identify what was what afterwards. That is, they were tested blind.
Some here seem to confuse blind testing with blindfolding. Just removing visual cues is sufficient.
Perceptual test methods depend on the research question being posed. If the question that needs to be answered is, can anyone at all reliably hear any difference between two things without knowing what they are, then ABX or A/B DBT can be a good choice.
In this case we already know ABX will be trivially simple for trained/expert listeners. After a few minutes of practice, if they can't tell which dac is which blind after hearing one or two 5 to 10 second passages of very familiar recordings then they don't belong at this listening session. Seriously. We have already tested a couple of people. Differences are obvious to them. So we don't think that's the question we need to ask here. We need to know things that will help get Marcel's dac more precisely rated in its present state, and we want to know what about the dac we think could most benefit from some work and or what we think is important to get the most attention. We also want a perceptual snapshot of where it is now for comparison with what future experiments may produce.
In this case we already know ABX will be trivially simple for trained/expert listeners. After a few minutes of practice, if they can't tell which dac is which blind after hearing one or two 5 to 10 second passages of very familiar recordings then they don't belong at this listening session. Seriously. We have already tested a couple of people. Differences are obvious to them. So we don't think that's the question we need to ask here. We need to know things that will help get Marcel's dac more precisely rated in its present state, and we want to know what about the dac we think could most benefit from some work and or what we think is important to get the most attention. We also want a perceptual snapshot of where it is now for comparison with what future experiments may produce.
Last edited:
No, but by popular demand and by chronologic chance I thought a reference was welcome.
In fact, I thought the NDK looked rather good;-)
In fact, I thought the NDK looked rather good;-)
When I tried NDK SDA, and then Crystek with film bypass caps and clock buffers with film bypass caps, and with well optimized voltage regulation, I decided to go with Crystek. At that point I wasn't able to get NDK to sound as good in my application. If X7R bypass caps then I always preferred NDK SDA.
That said, if I went back and did it all again then maybe I would come to a different conclusion. I don't know because I didn't find a reason to go back and do it all again. BTW, I was using some hand selected NDK SDA I got from Jocko. Never got them sounding as good as Crytek done right.
That said, if I went back and did it all again then maybe I would come to a different conclusion. I don't know because I didn't find a reason to go back and do it all again. BTW, I was using some hand selected NDK SDA I got from Jocko. Never got them sounding as good as Crytek done right.
Last edited:
I don't believe the company I shared a building with tested whether people could taste a difference between several types of Brussel sprouts, but rather what the differences in taste were and which tasted best. That didn't keep them from testing them blind.
@bohrok2610 Regarding the ES9038Q2M: if I understand the datasheet correctly (which is a big if), the ES9038Q2M has an asynchronous sample rate converter that is on by default. When the input signal is DSD, it can only be switched off by putting the ES9038Q2M in master mode (register 10 bit 7). You wrote you used a DSD input signal, so assuming your ES9038Q2M worked in slave mode like the RTZ shift register DAC, the ASRC must have been on. Far-off phase noise on the bit clock is then almost entirely eliminated, far-off phase noise on the XI clock is not.
Datasheet I looked at: https://eu.mouser.com/datasheet/2/1082/ES9038Q2M_Datasheet_v1_4-3074379.pdf
Does anyone happen to know the patent number of the ESS HyperStream II DAC?
Datasheet I looked at: https://eu.mouser.com/datasheet/2/1082/ES9038Q2M_Datasheet_v1_4-3074379.pdf
Does anyone happen to know the patent number of the ESS HyperStream II DAC?
Last edited:
There is preference testing and discrimination testing. ABX is only for discrimination. Sometimes blinding is unnecessary and more trouble than its worth. It really depends on how obvious something is.
For example, there are two sound files at: https://purifi-audio.com/blog/tech-notes-1/doppler-distortion-vs-imd-7 Without caring about particular focus of the article, can you hear any difference at all between the two files if listening with headphones? Yes or no?
For example, there are two sound files at: https://purifi-audio.com/blog/tech-notes-1/doppler-distortion-vs-imd-7 Without caring about particular focus of the article, can you hear any difference at all between the two files if listening with headphones? Yes or no?
You seemed to be mixing up ABX and double-blind in posts #1064 and #1073. All the methods I saw in the document you referred to in post #1064 can be executed double-blind or not. Hence the Brussel sprouts example.
Regarding the sounds, I think I hear a difference, and that I prefer the phase-modulated version over the amplitude-modulated version. That qualitatively matches the graph I showed here: https://www.diyaudio.com/community/threads/phase-noise-in-ds-dacs.387862/post-7066124 , which shows that amplitude modulation sidebands are more audible than phase modulation sidebands at frequency offsets below about 90 Hz. How is it relevant?
Regarding the sounds, I think I hear a difference, and that I prefer the phase-modulated version over the amplitude-modulated version. That qualitatively matches the graph I showed here: https://www.diyaudio.com/community/threads/phase-noise-in-ds-dacs.387862/post-7066124 , which shows that amplitude modulation sidebands are more audible than phase modulation sidebands at frequency offsets below about 90 Hz. How is it relevant?
Last edited:
Its relevant because spectral analysis shows the files to be identical. Therefore, I am skeptical. I would like to see DBT test results with sufficient trials to show you didn't just get lucky or imagine something. Would you be willing to submit blind test results for review by the group?
EDIT: Okay, I think you get the point. You know blind testing isn't needed don't you? The files obviously sound different. Why should you go to the trouble?
Also, I already told you differences between the dacs are obvious to trained/expert listeners. If you are here its obvious.
Last edited: