Jitter clarifications


#1

Hello,

Being a long time employee of a major player in the so-called “Clocks & Timing” semiconductor industry, (i.e. a manufacturer of electronic chips which create low-jitter clocks), I am interested to know more about the jitter values we see on Auralic products’ pages, such as:

  • Altair, 120 fs
  • Vega G2, 72 fs
  • LEO, as a specialized clock generator, should be even better.

How are these numbers measured? Which definition of jitter do they refer to? (period jitter, cycle-to-cycle, RMS phase jitter etc?)

For all of you interested to know more about jitter, I find this application note by SiLabs very informative and well written:

Please note that this question comes purely out of personal curiosity, no business purpose or connection with my employer (or any other manufacturer in the industry) is meant or intended by any means here.

Thank you,
Etienne


#2

" The LEO GX clock is so precise that existing benchmarks aren’t detailed enough to accurately represent what it can do. Instead, we use Allan deviation to describe the resolution of the LEO GX, which is like looking at phase noise closely enough to detect shifts of +/-1Hz or even +/-0.1Hz. The Allan deviation of the LEO GX Reference Master Clock comes in at 2E-12 (at 1 second), which is equal to a 10MHz rubidium atomic clock with phase noise of +/-1Hz at -110dBc/Hz, or an amazing 500 times less jitter than an 82fs Femto clock oscillator."

From auralic website:

The jitter measuring is much defined by the bandwidth. The advantage of LEO GX is the extremely low amount of close in jitter (below 10Hz) which is usually not been included in jitter measurement but contribute a lot to the final listening result.