What was the type of the improvement? I remember that my listening impressions many years ago used to be "less fogginess" associated to less jitter, but what happens with the likes of Accusilicon, Crystal, NDK etc?
Jitter can sound foggy, but it can sound like other things too. Depends on the particular jitter type, and its detailed behavior over time. Also it depends on the dac topology, including different variants of Sigma Delta.
Generally though, IME reducing jitter tends to have multiple and complex effects. All for the better, again IME. For one example, it possible to get cymbals to actually sound real. That is a tough one, IME. On most systems every cymbal sounds to some degree like a cymbal type that is described by percussionists as a "trash cymbal."
Generally though, IME reducing jitter tends to have multiple and complex effects. All for the better, again IME. For one example, it possible to get cymbals to actually sound real. That is a tough one, IME. On most systems every cymbal sounds to some degree like a cymbal type that is described by percussionists as a "trash cymbal."
Last edited:
Do you think there is anything about the R2R DAC sound reproduction that is lost with DS DACs, subjectively?
Soekris dam1121 was still best DAC I've heard, that is in terms of realistic sound reproduction, but sounded less clean than DS DACs I used... still unsure if a good enough DS DAC could capture that same realism and fully replace R2R...
Funnily, from what I tried I do not think old commercial R2R/multibit chips can really compete with modern DS chips at all, If I had not tried soekris I would have completely dismissed R2R/multibit as being obselete.
I cannot really say. And I cannot generalise. Maybe there is something correlated with the theoretically unbound settling time in D/S converters, but it is also true that modern D/S converters are actually multi-bit (5 or 6 bits).
The Soekris DAC1541 has a more upfront presentation, the D90 is a bit laid back, and this can influence the perception of many other parameters. Maybe there was a hint of very fine graininess of the highs in the Soekris, giving it a sound that I compare to very moderate sharpening a picture with noise - whereas if anything the D90 sounds almost over-polished, which can also be the result of being too much used to grainy sources in general.
Imaging on the D90 is slightly smaller but way more precise. Better layering. The mids are excellent, not significantly better than the Soekris, and the bass great, even easier to follow than on the Soekris despite the latter being excellent in that area.
You should consider sign-magnitude R2R chips – I think the PCM1704 was one of those – because they have tricks to reduce noise and distortion that a simple R2R ladder cannot. And in fact many swear that the PCM1704 was the best DAC chip ever made.
I would suggest....
SDM DAC chips often do seem to have some limitations, hence my preference for the direct DSD mode of the AKM 4499 chip. These limitations appear to come from the limited capability of the DAC chip's SDM itself. In other words, the modulator incorporated into the DAC chip is not good enough, due to the limited processing power available on chip. When you shift to the approach I currently recommend: using the AKM 4499 in its direct DSD mode, and only sending it DSD 256, with all files converted to DSD 256 using computer based oversampling in HQPlayer, all previously apparent limitations of SDM style conversions disappear. Of course HQPlayer requires a very powerful computer to do these conversions at its highest level of performance (with the ASDMEC7 modulator), basically you need a gaming machine to do it! So clearly Jussi (HQPlayer developer) has coded some very sophisticated modulator algorithms, which could never run internally in a DAC chip.
R2R has some inherent limitations on its performance (resistor precision being one) which are difficult to overcome, although the Holo Audio May DAC appears to have overcome them, at least according to its measurements. Owners of the May, who also have the ability to run HQPlayer, can compare directly DSD conversion via HQPlayer and sophisticated SDM modulation and R2R conversion in the same system. Another DAC which allows for such comparisons is the Bricasti M21 as it has onboard separate R2R converters and a discrete DSD conversion stage. Having listened to R2R vs DSD (both with various HQPlayer oversampling settings for input to the DAC) I tend to prefer the SDM conversion (HQPlayer oversampling to Bricasti M21 or M3 discrete DSD stage) to the R2R, but both are approaching Nirvana! That is, at a high level of conversion tech, and with HQPlayer oversampling in a powerful computer, both approaches are getting very close to ideal, as both tend to sound more and more "natural" while still retaining all the details. Indeed, especially HQPlayer oversampling with the ASDMEC7 modulator into a direct DSD conversion style DAC sounds fantastically natural and real, anyone who has an "aversion" to SDM DAC chips should hear this approach, as it demonstrates that SDM is not the "problem" the problem is the limited processing power in the DAC chip's own SDM.
Perhaps someone even more clever/gifted than Jussi (HQPlayer developer), might be able to produce a really good SDM which could run with less processing power than is required for HQPlayer? If so, such might be made low profile enough to work on a DSP chip inside a DAC, maybe... Until then, my new server build is running an i9-9900K processor in order to take advantage of HQPlayer's fantastic ASDMEC7 modulator in concert with my Bricasti M3 DAC's discrete DSD conversion stage.
SDM DAC chips often do seem to have some limitations, hence my preference for the direct DSD mode of the AKM 4499 chip. These limitations appear to come from the limited capability of the DAC chip's SDM itself. In other words, the modulator incorporated into the DAC chip is not good enough, due to the limited processing power available on chip. When you shift to the approach I currently recommend: using the AKM 4499 in its direct DSD mode, and only sending it DSD 256, with all files converted to DSD 256 using computer based oversampling in HQPlayer, all previously apparent limitations of SDM style conversions disappear. Of course HQPlayer requires a very powerful computer to do these conversions at its highest level of performance (with the ASDMEC7 modulator), basically you need a gaming machine to do it! So clearly Jussi (HQPlayer developer) has coded some very sophisticated modulator algorithms, which could never run internally in a DAC chip.
R2R has some inherent limitations on its performance (resistor precision being one) which are difficult to overcome, although the Holo Audio May DAC appears to have overcome them, at least according to its measurements. Owners of the May, who also have the ability to run HQPlayer, can compare directly DSD conversion via HQPlayer and sophisticated SDM modulation and R2R conversion in the same system. Another DAC which allows for such comparisons is the Bricasti M21 as it has onboard separate R2R converters and a discrete DSD conversion stage. Having listened to R2R vs DSD (both with various HQPlayer oversampling settings for input to the DAC) I tend to prefer the SDM conversion (HQPlayer oversampling to Bricasti M21 or M3 discrete DSD stage) to the R2R, but both are approaching Nirvana! That is, at a high level of conversion tech, and with HQPlayer oversampling in a powerful computer, both approaches are getting very close to ideal, as both tend to sound more and more "natural" while still retaining all the details. Indeed, especially HQPlayer oversampling with the ASDMEC7 modulator into a direct DSD conversion style DAC sounds fantastically natural and real, anyone who has an "aversion" to SDM DAC chips should hear this approach, as it demonstrates that SDM is not the "problem" the problem is the limited processing power in the DAC chip's own SDM.
Perhaps someone even more clever/gifted than Jussi (HQPlayer developer), might be able to produce a really good SDM which could run with less processing power than is required for HQPlayer? If so, such might be made low profile enough to work on a DSP chip inside a DAC, maybe... Until then, my new server build is running an i9-9900K processor in order to take advantage of HQPlayer's fantastic ASDMEC7 modulator in concert with my Bricasti M3 DAC's discrete DSD conversion stage.
That (in bold) is the key point.
Only way to compare the instrument's live sound vs recorded in correct way is to be at the recording session and listen from the spot where the microphone is placed. Also, the comparison has to be done immediately in order to compensate for our short aural memory span. I doubt you've done that.
Actually, there is a full set of drums, two electric guitars with amps, one acoustic guitar, keys, and one bass guitar with amp in the same room as my Sound Lab speakers. I have recorded them, and I do know what they sound like. In fact, I recorded a single cymbal hit for analysis in another thread. Included was highly detailed information about the cymbal, the recording hardware, and the recording setup. Ultrasonics were captured up, I forget exactly, somewhere around 40kHz. Mic was Schoeps CMC-641 which has a frequency response that starts falling off up around there, so the actual harmonics might have gone up higher.
Last edited:
You think you do. You left out "listen from the spot where the microphone is placed. Also, the comparison has to be done immediately in order to compensate for our short aural memory span."
Though if you listened to a good phono system, I have a four box Paradise Phono stage. Against my AK4499 SMSL M400. you might be surprised.
I have vinyl which sounds better than digital, down to the mastering and vice versa.
Though as you say to have such a rig costs £ 10K where the M400 cost far less..
Though if digital is so much better why then do we still chase better than the measures I have with my M400 ?
M400 DAC chip：AK4499, the audio industry's highest performance.
The THD+N is as low as 0.000068% (-123dB)!
When using A-weighted measurement, it is as low as 0.000058% (-124dB)！
Dynamic range up to 131dB!
USB uses the second XMOS XU216, supports native DSD, and 32bit 768kHz
You might then think that I would never play vinyl again.- wrong.
There is no surface noise on a good clean record, no clicks, pops and no inner groove distortion ( 12" carbon fibre unipivot arm).
I keep the nude Shibata stylus bonded to the solid Boron cantilever clean.
And can only conclude that some have never heard a decent vinyl rig, as it sounds on a par with my M400.
The rest of my kit is on par with the above.
M400 DAC chip：AK4499, the audio industry's highest performance.
The THD+N is as low as 0.000068% (-123dB)!
When using A-weighted measurement, it is as low as 0.000058% (-124dB)！
Dynamic range up to 131dB!
USB uses the second XMOS XU216, supports native DSD, and 32bit 768kHz
You might then think that I would never play vinyl again.- wrong.
There is no surface noise on a good clean record, no clicks, pops and no inner groove distortion ( 12" carbon fibre unipivot arm).
I keep the nude Shibata stylus bonded to the solid Boron cantilever clean.
And can only conclude that some have never heard a decent vinyl rig, as it sounds on a par with my M400.
The rest of my kit is on par with the above.
Last edited:
You really should stop digging. Surface noise on the quietest vinyl is about -70dB, THD is as mentioned several percent. Even the cheapest DACs wee all over vinyl for performance. But it's fun to get it to work, as it really shouldn't.
Oh and try to avoid the cheap insult route about us never having heard a decent vinyl rig.
Oh and try to avoid the cheap insult route about us never having heard a decent vinyl rig.
I have a decent vinyl setup, benz lp, kuzma deck and a Paradise phono. I also have a Gustard A18, neither sounds better than the other overall, they each sound better with some recordings, but in truth there's only a hair between them.
Maybe we should stop chasing the last 0.00001% of thd when lowly old vinyl is capable of so much with such poor numbers.
Maybe we should stop chasing the last 0.00001% of thd when lowly old vinyl is capable of so much with such poor numbers.
You may be right. I do feel many people chase the wrong thing and miss the elephant that is the room. But it's a hobby and if someone wants to get obsessional about a parameter or topology that's fine. It's the ramming it down our throats when an inaudible change is suddenly a huge leap of cleaned windows and dropped veils that gets wearing. But with thousands of members there are always a few. And the good stuff on here outweighs that.
With high end audio gear that is highly perceptually rated, but that has less than stellar THD, what really seems to be going on is that the distortions are quite dynamic and nonstationary. We don't see that so well in the stationary HD measurement techniques we usually use. Therefore, a mistaken assumption may be made to the effect that some people really like stationary distortion. Probably not so much on that, unless perhaps a system is really bad and a little edgy distortion helps cut through the mud. My opinions only, of course.
Last edited: