Watch out :
Read down to the end before jumping to (my) conclusions.If we start looking at the normal audio oscillators, I must now "admit" that I have been experimenting with ultra-low jitter clocks. Think 27fs of RMS jitter (with adjacent superb low phase noise response, so for real).
I couldn't get this working. but let's look at two pictures from a while back :
So what was this about ?
The difference in jitter on the audio clock(s) while being silent (first picture) and while doing some "excersises" on the floor.
All with the notice that the first picture actually shows the scope's limits for time resolution (shows 13ps of peak-peak jitter) and the second picture shows jitter of 36ps which is NOT beyond the scope's limit. IOW, the 13ps will in practice be (way) lower, while the 36ps is for real.
What I want to say with this is that moving a bit up and down on the floor causes (max) 36ps of peak to peak jitter.
Ok, don't move, or bring the lot down to the basement we'd say.
Back to the start of this post - please do recognize that what I had at hand there was around 1000 times less of jitter (not completely comparable because the 27fs is "RMS" which implies more peak to peak than 27fs. But make that 100fs and it is still 250 times less.
Now the crux : this is totally out of control. Or at least that was my conclusion of it, not being able to "create" a decent sound, merely meaning : a consistent sound. One of which I could "see" which direction to go to improve where it wasn't right. And this is not so normal (for me).
What really happens is that while any base higher jitter is just that, but can be influenced to some degree by movement of the floor and such, this is only a relative "somewhat". To understand this : if the first picture already would have shown 36ps, the movement of the floor would not have shown more, but the distribution of the jitter would have changed. Will be audible as well, but not so much (as how I envision it). With the 27fs as the base though, breathing at 1m distance could already be influencing. So whatever happens, the ultra low jitter gets way higher and this undoubtedly in some oscillating fahsion. So think the music (sound pressure) hammering on to the clock, that influencing the clock, that influencing the music, that influencing the clock. Oscillation.
Bring it to the basement and you're good.
Yes ?
Let's think PC now;
Pull out your screwdrivers, hold the pin to the PC chassis and put your ear on the other end (better try not to put the screwdriver the other way around). What do you hear ?
"Sound" you may say. Ya ya, but sound is vibration. Period. So you hear vibration.
Of course, when there's no single fan in there, and only SSDs and no, ehm, transformers, then all should be completely vibration free. But maybe take out the capacitors as well.
Oh.
So you see, if I really can relate this to the audio clock experience, on which I completely gave up (which such crazy super detail in all kind of (inconsistent) aspects), then go figure.
And thus, all we need to do is map this phenomenon on to the MoBo's oscillator(s) - or the one on the USB card for that matter - and ... then what.
First we need to "think" is that the jitter the vibration and everything implies (including that from the listening room, the PC now assumed to be in there which is the case for most of us) is high enough to change the sound for whatever further reason.
Remember, I still assume that mangling that clock (per Nick's ideas) do influence the sound indeed, so from there it is logical to to think that just jitter does - no matter for now how (but see first post about (de-)skewing).
So yes, I have seen pictures of people having their spinning disks in rubber bands outside of the PC's cabinet. That sort of thing would be required. But more generally, and as already said, no moving parts are to be around.
Nice PeterSt, "Eureka". Yeah, well, I don't care, but it would be nice when "we" would be able to find the reason, instead of just following other's noses and be still out of control. So if the sheer reason is the influencing oscillators (crystals) then maybe we can do something about *that*.
How ? yes, nice question.
After the only few experiments we ran through yesterday, it (very preliminary) looks like the PC needs a sort of opposite damping as the DAC. Ok, can't avoid to be general (but like to) - and need to mention something such as the NOS1a. And let's make that "a" specific, because of the way better jitter specs than the normal NOS1. Anyway :
Say, generally, that the NOS1a requires a stiff damping, think hard wood-like. "Requires" ? well, to my liking for the accuracy and e.g. no stumping fuzzy bass. Make that rubber and the latter happens. Now put the same wood-like under the PC and all gets gray and with less bass. Easily audible. Music doesn't work any more either; things get flat.
Now put rubber under there and the very opposite would happen from doing that under the NOS1a : transients improve, bass is right
I know, not exactly the opposite as what happens for the NOS1a, but mere opposite from the wood vs the rubber under the PC and what I would expect from it.
And please watch out, because the situation will be very specific to mine and where the PC recides in the first place (floor, rack, closet, etc. etc.).
And yes, my PC has a running fan for the CPU cooling and contains one (2.5") spinning disk. Not more. Ah, active power supply cooling (fan). Yes.
Especially because these kind of vibrations are high frequency, I can expect that to be more hammering on "jitter" than the floor-excersise experiment. So worse.
I can extend the whole thing to the PC nicely vibrating in the cabinet (say my audio rack) while the NOS1a is in that same cabinet. Now things get more complex because what I did with the rubber under the PC is decoupling it from the DAC.
Ha ! So now we are confused. Bring your PC and DAC down to the basement and you achieved nothing.
Being at this stage of reasoning, I now tend to believe more in the PC influencing the DAC than the PC influencing itself for SQ reasons. So what would be more logical or easy to believe ?
a. The vibrations of the PC influencing the normal audio clock that causing additinal jitter;
b. The vibrations of the PC influencing its own clock that causing jitter (which would be true) to further
reason out that it can influence sound.
Ad b.
I took this for granted because of Nick's experiments. Do notice though that it almost takes "nothing" to change a vibration situation towards the DAC. And might it be difficult to envision how : a USB connecting cable, that being not sufficiently free-floating, will couple the both; things in this realm will change the most easily after turning a screw or whatever change, just because you don't take them into account. So do in the future. Sound waves onto the USB cable, or interlink for that matter, already couple (to say, the DAC). And to understand and believe this, all it really takes is to try such 27fs RMS jitter clock and see that all suddenly matters like day and night. As "we" (NOS1a) have it, it is only less influencing because the jitter base is higher. So still under control, but we must think carefully how.
Now I hope this expands to something valuable some time.
Peter
