There is an astounding amount of misinformation and misunderstanding in some of the threads here about clocks.
Having followed a lot of discussion on this topic for years and years over on diyaudio I would like to clarify some things and hopefully inform people better before spending thousands on some of this gear.
This is purely technical information but I'm neither an 'objectivist' or 'subjectivist' (hate these terms...).
As we all know sound quality can have no little or relation to technical performance, some people could well like the sound of jitter... lots of people like the sound THD, that is a fact.
There are also so many variables in audio equipment that measurements may not be broad or accurate enough to really guarantee or predict how something will actually sound as a whole.
A common topic is 'Internal' and 'external' clocks. There is no actual difference between them, aside from their location. Connect your 'external' clock to your DAC, put all of into a cardboard box, I suppose you can now call that external clock an internal clock in your new 'cardboard box edition' DAC.
The most common thing you see now is the use of a single external 10MHz clock (external simply meaning located in a seperate chassis).
In order to generate the correct clock frequencies for digital to analogue conversion at standard audio sample rates (multiples of 44.1kHz and 48kHz) this 10MHz clock signal must be processed because it cant be evenly divided to these sample rates.
This process, which happens inside the DAC with an external 10MHz input, is what determines the jitter/phase noise of the clocks signals used in the actual DA conversion. The jitter/phase noise of 10MHz clock alone means nothing, in fact this conversion process is typically the dominant source of jitter.
With Internal clocks DAC clocks you typically have 2 seperate clocks dedicated to each audio sample rate multiple, most common clock frequencies being 45.1584MHz (44.1kHz x 1024) and 49.152MHz (48kHz x 1024). They can be divided evenly and therefore do not to be processed in the same way as the 10MHz external clock.
In 99.9% of DACs these are low cost, off-the-shelf clocks, although there are some lower jitter options available.
Accusilicon, NDK and Crystek are some examples of those which you find in some higher end DACs.
A Very low jitter/phase noise 10MHz external clock + very well designed circuit to generate the audio rate clocks is definitely possible to outperform these low cost, off the shelf audio rate internal clocks.
Now some clarification needs to be made about 'phase noise' and 'jitter', words you see thrown around a lot in relation to clocks.
Jitter and phase noise are simply different way of looking at the same thing.
A very good analogy is that phase noise is like looking at FFT spectrum of THD while jitter is like a simple THD percentage.
A phase noise graphs tells you far more about the performance of clock than the jitter figuree.
Low 'close-in' phase noise (around 10Hz and below on phase noise graph) is considered by many to have the biggest influence on sound quality of clocks.
Designing a clock with very low close-in phase noise is very difficult and expensive.
Gustard's published phase noise plot for their 10MHz clocks show they perform very well in that regard, they have definitely designed a high performance clock. Definitely much better than the off-the-shelf clocks mentioned above, you can find published phase noise measurements for some of them if you look around if you want to compare.
Gustard's clock processing in their DACs for the 10MHz is presumably also very good, but I have not seen published data on it.
However the most optimal to way to design a clocking system for DA conversion is to build 2 high performance, low phase noise clocks, on the same level as Gustard's 10MHz clocks, but at a multiple of the native audio sample rates (44.1kHz and 48kHz) like typical low cost internal clocks, which allows the clocks to be used directly in DA conversion without bottlenecking the high performance of the clocks with that processing stage needed for the 10MHz clock.
Currently, the only options available for this that I'm aware of are the products of Well Audio.
They have designed a range of ultra low phase noise clocks at audio rate frequencies with published phase noise plots.
They have also designed a range DACs and a FIFO digital interface whose purpose is to effectively isolate the audio system from all source interference and jitter (e.g a PC) and allow their high performance clocks to be interfaced with DACs correctly without degrading their excellent performance.
Although these products are very reasonably priced, they unfortunately are not readily available to buy due to small scale and are heavily DIY oriented so will not be accessible to most people... at least that is the current situation.
And again, while Well Audio are the only ones I'm aware of handling clocks in this technically 'optimal' way, there is no guarantee that everyone will automatically think this way sounds best.
Human hearing is too subjective and inconsistent, and there are far too many other variables in an audio system to ever make any definite conclusions about sound quality.
I have personally not heard any Well Audio products either, I have just been following them on diyaudio for years and was very impressed by their technical achievements and innovation in the DIY scene.
Hopefully other big manufacturers, such as Gustard, follow suite and offer designs more like this
Hopefully this post helps inform people on the topic of clocks
Having followed a lot of discussion on this topic for years and years over on diyaudio I would like to clarify some things and hopefully inform people better before spending thousands on some of this gear.
This is purely technical information but I'm neither an 'objectivist' or 'subjectivist' (hate these terms...).
As we all know sound quality can have no little or relation to technical performance, some people could well like the sound of jitter... lots of people like the sound THD, that is a fact.
There are also so many variables in audio equipment that measurements may not be broad or accurate enough to really guarantee or predict how something will actually sound as a whole.
A common topic is 'Internal' and 'external' clocks. There is no actual difference between them, aside from their location. Connect your 'external' clock to your DAC, put all of into a cardboard box, I suppose you can now call that external clock an internal clock in your new 'cardboard box edition' DAC.
The most common thing you see now is the use of a single external 10MHz clock (external simply meaning located in a seperate chassis).
In order to generate the correct clock frequencies for digital to analogue conversion at standard audio sample rates (multiples of 44.1kHz and 48kHz) this 10MHz clock signal must be processed because it cant be evenly divided to these sample rates.
This process, which happens inside the DAC with an external 10MHz input, is what determines the jitter/phase noise of the clocks signals used in the actual DA conversion. The jitter/phase noise of 10MHz clock alone means nothing, in fact this conversion process is typically the dominant source of jitter.
With Internal clocks DAC clocks you typically have 2 seperate clocks dedicated to each audio sample rate multiple, most common clock frequencies being 45.1584MHz (44.1kHz x 1024) and 49.152MHz (48kHz x 1024). They can be divided evenly and therefore do not to be processed in the same way as the 10MHz external clock.
In 99.9% of DACs these are low cost, off-the-shelf clocks, although there are some lower jitter options available.
Accusilicon, NDK and Crystek are some examples of those which you find in some higher end DACs.
A Very low jitter/phase noise 10MHz external clock + very well designed circuit to generate the audio rate clocks is definitely possible to outperform these low cost, off the shelf audio rate internal clocks.
Now some clarification needs to be made about 'phase noise' and 'jitter', words you see thrown around a lot in relation to clocks.
Jitter and phase noise are simply different way of looking at the same thing.
A very good analogy is that phase noise is like looking at FFT spectrum of THD while jitter is like a simple THD percentage.
A phase noise graphs tells you far more about the performance of clock than the jitter figuree.
Low 'close-in' phase noise (around 10Hz and below on phase noise graph) is considered by many to have the biggest influence on sound quality of clocks.
Designing a clock with very low close-in phase noise is very difficult and expensive.
Gustard's published phase noise plot for their 10MHz clocks show they perform very well in that regard, they have definitely designed a high performance clock. Definitely much better than the off-the-shelf clocks mentioned above, you can find published phase noise measurements for some of them if you look around if you want to compare.
Gustard's clock processing in their DACs for the 10MHz is presumably also very good, but I have not seen published data on it.
However the most optimal to way to design a clocking system for DA conversion is to build 2 high performance, low phase noise clocks, on the same level as Gustard's 10MHz clocks, but at a multiple of the native audio sample rates (44.1kHz and 48kHz) like typical low cost internal clocks, which allows the clocks to be used directly in DA conversion without bottlenecking the high performance of the clocks with that processing stage needed for the 10MHz clock.
Currently, the only options available for this that I'm aware of are the products of Well Audio.
They have designed a range of ultra low phase noise clocks at audio rate frequencies with published phase noise plots.
They have also designed a range DACs and a FIFO digital interface whose purpose is to effectively isolate the audio system from all source interference and jitter (e.g a PC) and allow their high performance clocks to be interfaced with DACs correctly without degrading their excellent performance.
Although these products are very reasonably priced, they unfortunately are not readily available to buy due to small scale and are heavily DIY oriented so will not be accessible to most people... at least that is the current situation.
And again, while Well Audio are the only ones I'm aware of handling clocks in this technically 'optimal' way, there is no guarantee that everyone will automatically think this way sounds best.
Human hearing is too subjective and inconsistent, and there are far too many other variables in an audio system to ever make any definite conclusions about sound quality.
I have personally not heard any Well Audio products either, I have just been following them on diyaudio for years and was very impressed by their technical achievements and innovation in the DIY scene.
Hopefully other big manufacturers, such as Gustard, follow suite and offer designs more like this
Hopefully this post helps inform people on the topic of clocks
Last edited:
