LAiV Harmony uDDC Re-clocker/DDC

[mikefc]

Quick question, Mike: Do you think a squarewave clock can be amplitude modulated? Sinewave yes, but it produces the same sideband skirts as a squarewave. It means a different kind of modulation is in place, you didn't mention it.

Amplitude modulation produces two sidebands on the FFT plot, but a fundamental difference is that there is a gap between base frequency and both skirts with zero energy. Here is an example from Wikipedia:

Disregard shape, it is just demonstration of distribution of modulation frequency. A gap is characteristic, as there in very little energy close to zero Hz. In nature and technical implementations as well.

Usually amplitude modulation is not covered in publications, focus is on jitter. It is modulation of a clock frequency or a phase or combination of both. It is exactly our concern.

Thank you for letting us know that FFT plot skirts can be manipulated artificially by changing number of sample points, I don't think anyone here has such intention.

Ok. Good point about the square wave clock not being able to be AM modulated. You are saying it isn't AM modulation of the clock, but FM modulation due to jitter. I'm not that familiar with how FM modulation manifests on an FFT. That's why I generalized to *some* type of modulation. You think that it is FM modulation of the clock by jitter (as the modulating source signal) that can produce the type of skirts we are seeing in the AP FFT then?

Thanks, Mike

 

[sajunky]

You think that it is FM modulation of the clock by jitter (as the modulating source signal) that can produce the type of skirts we are seeing in the AP FFT then?

Yes, I said so already. Either a phase or frequency is modulated. In nature there is always a mixture of both. The same result.

 

[mikefc]

Yes, I said so already. Either a phase or frequency is modulated. In nature there is always a mixture of both. The same result.

I heard you say it. I'm not sure. FM modulation doesn't require the characteristic slope that the skirts are showing on the plots. Thats another reason why I said *some* type of modulation.

However the higher audio frequencies shown later in the video do show the same characteristic slope as the skirts. So that says to me that the jitter is actually not low frequency jitter, but higher up in the midband ?

 

[sajunky]

I heard you say it. I'm not sure. FM modulation doesn't require the characteristic slope that the skirts are showing on the plots. Thats another reason why I said *some* type of modulation.

However the higher audio frequencies shown later in the video do show the same characteristic slope as the skirts. So that says to me that the jitter is actually not low frequency jitter, but higher up in the midband ?

Perhaps it demonstration of frequency modulation by a deterministic signal, it is what FFT plot does well. They wouldn't demonstrate what FFT plot doesn't do well (modulation with a random noise), but I don't watch videos.

 

[mikefc]

Perhaps it demonstration of modulation by a deterministic signal, it is what FFT plot does well. They wouldn't demonstrate what FFT plot doesn't do well (with a random noise), but I don't watch videos.

Thinking about it more, I don't think random noise is the cause of the slope in the skirts. If there is a PLL involved or something resembling a low pass filter in the clock path, the slope of the skirt would make sense I think. Higher frequency jitter would get attenuated more relative to the corner frequency of the low pass filter. For FM modulation that would show up as the characteristic slope in the skirts.

 

[sajunky]

Higher frequency jitter would get attenuated more relative to the corner frequency of the low pass filter. For FM modulation that would show up as the characteristic slope in the skirts.

I repeat, please do not speculate. A wider skirt shows how deep is modulation (amount of frequency deviation). It doesn't carry information about frequency of noise which is modulating a clock. It does, but only in the case of modulating with a deterministic noise where FFT is valid.

 

[mikefc]

I repeat, please do not speculate. A wider skirt shows how deep is modulation (amount of frequency deviation). It doesn't carry information about frequency of noise which is modulating a clock. It does, but only in the case of modulating with a deterministic noise where FFT is valid.

I'm talking about slope of the skirts. How far away in amplitude the jitter is from the center frequency is how FM modulation encodes amplitude. If higher frequencies are attenuated due to low pass filtering, there is less deviation from the center frequency. Not speculation. Thats how it works. A low pass filter doesn't care what kind of modulation it is being fed or whether the signal is periodic or not. As long as any frequencies it is being fed are above the corner frequency.

 

[sajunky]

I'm talking about slope of the skirts. How far away in amplitude the jitter is from the center frequency is how FM modulation encodes amplitude. If higher frequencies are attenuated due to low pass filtering, there is less deviation from the center frequency. Not speculation. Thats how it works. A low pass filter doesn't care what kind of modulation it is being fed or whether the signal is periodic or not. As long as any frequencies it is being fed are above the corner frequency.

You are speculating that clock had been already conditioned by PLL. It is not seen of the FFT, no indication whathever.

Shape of the skirt, so a slope (as you say) is determined by the statistical distribution density of modulating noise. It can be Gaussian function (typical for a thermal noise), can be any other. By example, if a very small deviation of a clock frequency happens very frequent, the beginning of a skirt will be flat. Once a bigger deviation happens less frequent, it has a lower value of energy, it builds a falling slope. If a maximum deviation is sharply limited and dominant, skirt will be very flat at the top and ending sharp dropping to zero instantly (a square shape).

Did I mention above anything about frequency of noise? Not a single word. It is all about statistical energy distribution.

 

[mikefc]

You are speculating that clock had been already conditioned by PLL. It is not seen of the FFT, no indication whathever.

Shape of the skirt, so a slope (as you say) is determined by the statistical distribution density of modulating noise. It can be Gaussian function (typical for a thermal noise), can be any other. By example, if a very small deviation of a clock frequency happens very frequent, the beginning of a skirt will be flat. Once a bigger deviation happens less frequent, it has a lower value of energy, it builds a falling slope. If a maximum deviation is sharply limited and dominant, skirt will be very flat at the top and ending sharp dropping to zero instantly (a square shape).

Did I mention above anything about frequency of noise? Not a single word. It is all about statistical energy distribution.

Yeah, don't know where you get this. Reference?

PLLs are modeled with their output going thru a low pass filter.
Look at the May plots. Very low corner frequency (.05Hz). Shows in AP FFTs.
https://www.magnahifi.com/shop/holo-audio-may-l2-holoaudio-may-dac-level-2-r2r-dsd1024-24
Very low jitter and slope. All attenuated by a low pass filter. Did I mention anything about noise or its statistical distribution. A low pass filter does not care.

 

[Fluty]

I just received my Laiv DDC and it's paired with the Harmony dac. I'm frankly shocked about the difference in sound quality across the board. I need a bit more time to process but whatever is in that little box is doing something very impressive. Key things right off the bat are greatly increased clarity, detail and soundstage. Bass also seems to be tighter overall with more impact. It sounds like I just bought a much more high end dac! My front end is a wryed4 sound music server/streamer. (their highest tier version so no slouch)

At last! Someone else commenting on the wonderful improvement in the sound wrought by this uDDC!

 

[gtman55]

At last! Someone else commenting on the wonderful improvement in the sound wrought by this uDDC!

I just got mine also ..listening atm. Was running a gustard u18 prior and can say it's better than my gustard. I'm not going to use the typical hifi cliche, but this crap takes my music next level. I'm sorry Goldy, I trust my ears over your numbers...can't wait for full burn in

 

[Sound Dragon]

I noticed it on day one and said as much. But after a while, I can honestly say that I prefer my set up with the uDDC than without it. It is not a fluke at all. Whatever this combination is doing, it is doing it really well. So I have to say that I disagree with the review as well. It is not matching my experience listening to the uDDC connected to the Harmony DAC at all.

 

[gtman55]

Another observation is that I can hear so much more detail, air and instrument separation at lower level listening. This wasn't the case with the gustard u18.

 

[theveterans]

Another observation is that I can hear so much more detail, air and instrument separation at lower level listening. This wasn't the case with the gustard u18.

At the same time, you crank it up and not feel any hardness in the presentation at very loud volumes as well

 

[Jake2]

That all sounds most excellent, thanks guys for sharing your impressions. It also sounds to me very much like the characteristics of greater temporal precision, aka less audible jitter in places where it really matters to our hearing perception, perhaps even lower levels of that tricky-to-measure unicorn, close-in phase noise…