The point in doing measurements on these devices is to accumulate enough datapoints whereby one can discount certain areas from further investigation &/or focus on certain areas which hint at possible correlations to SQ differences. There is no more expectation than a hint can be ascertained at this stage.
Yes indeed - I'm going to start by looking at USB Packet jitter (and Eye pattern) + RF levels, these should be easily measurable - but its not easy to equate these result with sound quality.
One problem that I face is that lowest jitter might not be the best sound quality - as its the spectrum of the jitter that's important.
I believe it was Michael (Fusion) who pointed me to a CA thread where the designer of the Regen inserted resistors in the Ground connection of the USB lead and this resulted in "much improved Bass".
My first thought was that once the impedance of the Earth connection was increased Earth leakage currents found the next easiest route which was through the USB Data Lines themselves - thus superimposing higher Mains related harmonics on the Jitter spectrum - "Mains Hum" if added in small doses adds "bass weight". I've seen this effect before - it even sounded better, but it’s not technically correct.
The problem with looking at an interface in isolation is that striving for the best "individual" result might not be the best option when the whole system is considered.
I've found that in many instances when the system is non linear (when we cannot have Zero interface jitter) then its better to add "noise" to the Clock to "Bury" any discrete spurie components. Our hearing is sensitive to "discrete" components but not so much to random noise.
For example – a gentle air-conditioning blowing in the background does not disturb me as it produces purely ‘White noise” but if the bearings in the fan are warn then the fan can rattle & rub which produces discrete tones / noise, this noise is no longer clean “White Noise” and becomes “noticed” by our brain and rather irritating.
The same happens with the Phase noise spectrum, we know the USB devices are not designed for low noise and produce complex noise spectrum (even as a second order effect in the system) – I find its better to bury these discrete components under noise….
While the “total” integrated jitter level result is worst – atleast the phase noise plots no longer have a complex spectrum of spurie components, sure everythings “buried” by the higher noise floor
This is the reason I’m adding the Spread spectrum mode – I was even toying with the idea of using a tube based clock oscillator, to use the inherent noise of the Tube to produce a “nice” noise spectrum – its “Clean” noise…
The problem is that I need to balance my credibility with what I believe people are prepared to accept. I believe a Tube base oscillator would be better because its noise would bury (Mask) a whole spectrum of discrete noise components unavoidable in the USB interface, but it would be a leap too far for many….
I do respect Adamdea’s point of view on such matters, but I believe that there is a good reason why an idea like this is not so crazy… in a perfect world Zero jitter would be best, but with the USB interface so much is beyond my control – it’s better to bury the unavoidable discrete components with noise.
Maybe the Detox should be renamed “White Washer”
I might be sitting up there with the “crazies” with a tube based Detox
– but hey sometimes I have to put the measurements aside and just listen to the Music