As Marcel explained, it depends on the dac technology. Some very good modern dacs are using DSD or sigma-delta PWM. Examples might be "Topping D90 III Discrete," and the "Mola Mola" dacs.
As the graphic I posted previously shows, DSD dacs tend to be more sensitive to jitter. Question is what is audible about very low looking levels of noise as seen in spectral line skirts? Its not about nonlinear distortion, and its not about plainly audible wow and flutter either. In significant part its about localization in stereo imaging. Time jitter can be significant relative to ITD lateral localization. Amplitude noise which has become part of the audio signal can audibly mask low level distance cues. The true and fully immersive experience of the stereo illusion is for it to sound like the listener is in the same room with the performers, and where the sound of the room itself its reverberations and decays is part of what it makes the experience perceived as more real. Also, the accurate tones and textures of instruments, their subtle dynamics, etc. are an important part of the experience.
Moreover, I know the perceptual experience of "being there" is something manufacturers are now trying to build into upscale automobile sound systems, although I have been told some people find the underlying SQ of those systems to be disappointing.
OTOH, much of common stereo listening sounds much like a flat plane, with three speakers: left, right, and center. There is no fine localization cue detail, tones and textures are approximate, fine details of instrumental dynamics are missing, room echos and decays sound flat and synthetic like a VST reverb plugin. It can be okay, but its not a luxurious perceptual experience.
EDIT: On the question of at what level of jitter performance is there no further audible benefit? Using the best SOA clocks I can find today, I'm not sure I have found a point at which there is no further audible improvement with my DSD dac. What I found previously was an apparent limit which turned out to be an artifact of commonly used squaring circuits for sine wave oscillators.
Last edited:
By frequency offset I meant the frequency difference between the noise and the signal, so when you measure the noise density around 11 MHz of a 10 MHz oscillator, the frequency offset is 1 MHz. As far as I know, it is a normal technical term. At least my colleagues seem to understand it 😉
This might be the explanation why I am so fascinated by the cheap Chinese R2R DAC. It seems to be more immune to jitter. What I hear compared to my other DACs is exactly what you described above. More room information, more space between the instruments, more accurate tones.......
I will try if reclocking has less impact as with my AKM or ESS DACs.
Is there any clock under $200 that you can recommend over the Chrystek 957 and NDK 25?? that I got here?
Iancanada makes some pretty good clocks for $179 each. If you need two clocks and can only afford one at that price, I would suggest to make it the one you would expect to be using most.
Also, you should be aware that Ian's clocks take almost a week of continuous running to settle in to their long term sound. In addition, Ian recommends the clocks always be left powered on for best results.
Also, you should be aware that Ian's clocks take almost a week of continuous running to settle in to their long term sound. In addition, Ian recommends the clocks always be left powered on for best results.
By the why. In the brodcast envirement the golden rule of digital was one clock distributed to all the digital equipment in the chain.
Så actually there I agree. Clocking matters. But then a scope is needed for analysing all the pittfals of HF signaling
Så actually there I agree. Clocking matters. But then a scope is needed for analysing all the pittfals of HF signaling
Just a funny story from when I bought the reference dac with -125db spec.
I wanted to measure it to ensure it was not a counterfeit. And sure enough the Thd was -101dB
Then the seller remembered he had tinkered with the settings to tune the sound to his liking.
He pushed reset and all distortion was gone.
So no counterfeit, just added colouring to please the previous owner
I wanted to measure it to ensure it was not a counterfeit. And sure enough the Thd was -101dB
Then the seller remembered he had tinkered with the settings to tune the sound to his liking.
He pushed reset and all distortion was gone.
So no counterfeit, just added colouring to please the previous owner
That was more in the old days. Word clock isn't used so much anymore. Local crystal clocks and ASRC can sound better than word clock and PLLs.
I've built my own music player with a TI PCM5102 DAC (USD $8) and the sound is beyond anything I've ever heard.
I admit I haven't heard high-end players like the Sony CDP-R1/DAS-R1 (and most likely never will) but I'm pretty sure you don't need ultra-expensive gear for great sound. This whole debate between R2R and Delta-Sigma is abstract and filled with mysticism. Mind you that it's virtually impossible to get 24-bit or 32-bit resolution without Delta-Sigma.
I admit I haven't heard high-end players like the Sony CDP-R1/DAS-R1 (and most likely never will) but I'm pretty sure you don't need ultra-expensive gear for great sound. This whole debate between R2R and Delta-Sigma is abstract and filled with mysticism. Mind you that it's virtually impossible to get 24-bit or 32-bit resolution without Delta-Sigma.
I don't think R2R or delta sigma matter. What matters is OS vs NOS. Any idea if your DAC may have interpolation filters bypassed or what the filter settings are? Any link to your specific DAC?
One can only imagine the aural ecstasy to be achieved when you add a Markw4 approved clock.
What does that mean: OS vs. NOS?
I was gonna post a link to my music player on the Teensy forum but it seems the site is down at the moment.
Oversampling vs Nonoversampling. The PCM5102 has a 8x oversampling digital filter, but it can be bypassed. I thought the PCM1794 could be used as a NOS DAC in mono mode? The DDDAC1794 uses the PCM1794 DAC chip. The "1794" in the DDDAC1794 name suggests that it is based on the PCM1794 DAC chip, and it is indeed a non-oversampling DAC design that uses the PCM1794 as its core conversion component.
Last edited:
Maybe an AI hallucination? Question: Can the PCM5102 be used in a NOS DAC?
"Yes, the PCM5102 can be used in a Non-Oversampling (NOS) DAC. The PCM5102 is a 32-bit, 384kHz, Burr-Brown (now part of Texas Instruments) DAC chip that can be operated in a NOS mode.
In NOS mode, the PCM5102 can be configured to bypass its internal digital filtering and oversampling, allowing it to convert the digital audio signal directly to an analog signal without any intermediate processing. This approach can provide a more direct and unprocessed sound, which is often preferred by audiophiles.
The PCM5102's NOS capability makes it a popular choice for DIY DAC projects and high-end audio applications where sound quality is paramount. By using the PCM5102 in a NOS configuration, designers can create a DAC that provides a unique sound signature and a more direct connection to the music."
"Yes, the PCM5102 can be used in a Non-Oversampling (NOS) DAC. The PCM5102 is a 32-bit, 384kHz, Burr-Brown (now part of Texas Instruments) DAC chip that can be operated in a NOS mode.
In NOS mode, the PCM5102 can be configured to bypass its internal digital filtering and oversampling, allowing it to convert the digital audio signal directly to an analog signal without any intermediate processing. This approach can provide a more direct and unprocessed sound, which is often preferred by audiophiles.
The PCM5102's NOS capability makes it a popular choice for DIY DAC projects and high-end audio applications where sound quality is paramount. By using the PCM5102 in a NOS configuration, designers can create a DAC that provides a unique sound signature and a more direct connection to the music."
@hifiamps - I'd be convinced if the AI came up with a reference to the PCM5102 datasheet and mentioned precisely how to configure that chip into NOS mode. I'm aware that PCM179X has a 'NOS mode' which DIYers have used (DDDAC I believe is one) but that only bypasses the first FIR filter upsampling stage, not all of the oversampling in the chip. The bulk of the oversampling is handled by simply repeating the previous sample rather than with a filter.
Thank you abraxalito for the clarification. This is the DDDAC1794 link. No reference provided for the NOS PCM5102.
AI produced the following. No references:
"The Pavouk is an example of an NOS DAC that uses the PCM5102, but it achieves NOS operation through the use of an external digital filter and a unique clocking arrangement, rather than the PCM5102 itself operating in a true NOS mode. The Pavouk's design allows it to bypass the internal digital filtering and oversampling of the PCM5102, and instead uses an external filter to remove high-frequency noise and images, resulting in a NOS-like operation."
Searching for Pavouk PCM5102, it is an OS DAC. Not NOS. Pavouk did make a TDA1543 based NOS DAC. AI hallucination.
"The Pavouk is an example of an NOS DAC that uses the PCM5102, but it achieves NOS operation through the use of an external digital filter and a unique clocking arrangement, rather than the PCM5102 itself operating in a true NOS mode. The Pavouk's design allows it to bypass the internal digital filtering and oversampling of the PCM5102, and instead uses an external filter to remove high-frequency noise and images, resulting in a NOS-like operation."
Searching for Pavouk PCM5102, it is an OS DAC. Not NOS. Pavouk did make a TDA1543 based NOS DAC. AI hallucination.