I am considering an Arylic Up2Stream mini board, that has I2S output besides analog and SPDIF. Would it be possible to build this board in a CD player and connect it directly to the I2S bus? What can be used to switch over between internal I2S source and this board automatically?
I don't know your specific components, but in general you can easily switch I2S sources using digital multiplexer ICs. As for the autodetection, I'd recommend to have a look at the behaviour of your I2S sources using an oscilloscope. Knowing this, an autodetection shouldn't be too much of a challenge.
I am aware that digital multiplexer can increase jitter 🤔 ... maybe a relay based switch can works better 🤔
Especially when you use 1 multiplexer IC for both the clock lines and the data line, you are bound to get some data to clock crosstalk. Whether that matters is hard to say.
It is possible to reclock the I2S signals (only the LRCK is needed to reclock IMHO) after the multiplexer.
Thanks for the idea of the MUX anyway.
Thanks for the idea of the MUX anyway.
I would try it with the multiplexer only (as many people do). Then, you can check with an oscilloscope if there is the need to improve the signal qualities.
Jitter, in particular close-in phase noise, can't be measured with a typical scope. Some rather expensive scopes offer an optional jitter analysis package, but that is more for far-out phase noise statistical characterization (say, maybe >12kHz offset from the carrier).
For some systems and or some listeners, a modest amount of jitter may not make much if any audible difference. No worries in that case. For a very high performance system, then smaller aberrations may become more audible, at least for some people. Threshold of hearing for whatever parameter (including for jitter audibility in a given system) is a statistical thing, it varies from person to person.
For some systems and or some listeners, a modest amount of jitter may not make much if any audible difference. No worries in that case. For a very high performance system, then smaller aberrations may become more audible, at least for some people. Threshold of hearing for whatever parameter (including for jitter audibility in a given system) is a statistical thing, it varies from person to person.
Last edited:
It was mentioned above that the jitter would arise from bad signal shapes. These should be obvious in a scope. At least, this was my thought... 😉
Most scopes are not well suited for accurate visualization of clock signals. It takes a very wide bandwidth scope, a special low-capacitance probe, and special probe grounding technique. A standard 2 or 4 channel, 100Mhz scope is not up to the task. Just the probe capacitance will distort the waveform.
Even with a proper scope, they are not the best instrument for measuring dac clock phase noise (phase noise is another way of characterizing jitter). What clock jitter is amounts to slight variations in the timing of the clock edges. Their timing can be slightly random, or there can be some slight pattern to the timing variation. Sometimes there is a combination of both. Since the variation is small and to some extent random, it may take statistical analysis to get a good idea of its effective magnitude. One instrument that can be used is called a Time Pod. They aren't cheap and they are only good for one thing, so most people don't have them.
As a practical matter, most of us are left to use very careful design techniques and then experimental methods for whatever level of validation is possible (except for whatever can be deduced from using something called J-Test). Maybe one thing for sure, its hard to get phase noise very low, and its very, very easy to make it worse.
Even with a proper scope, they are not the best instrument for measuring dac clock phase noise (phase noise is another way of characterizing jitter). What clock jitter is amounts to slight variations in the timing of the clock edges. Their timing can be slightly random, or there can be some slight pattern to the timing variation. Sometimes there is a combination of both. Since the variation is small and to some extent random, it may take statistical analysis to get a good idea of its effective magnitude. One instrument that can be used is called a Time Pod. They aren't cheap and they are only good for one thing, so most people don't have them.
As a practical matter, most of us are left to use very careful design techniques and then experimental methods for whatever level of validation is possible (except for whatever can be deduced from using something called J-Test). Maybe one thing for sure, its hard to get phase noise very low, and its very, very easy to make it worse.
That's an extraordinary claim. It's only fair to accompany it with extraordinary evidence.