[time-nuts] FEI-5660 Rubidium Oscillator

Tom Van Baak tvb at LeapSecond.com
Thu Mar 27 19:10:23 EDT 2014


> Recently I happened across an eBay listing for an Antelope Audio Isochrome, 
> a device that apparently packages an SRI-PRS10 rubidium oscillator and  
> distribution amplifier in a box and sells to audiophiles for a price in the 

Bruce,

There have been threads about this on time-nuts every few years. The consensus is that audio companies that use atomic clocks are naive. It makes good marketing, though.

Then again, speaking from experience, many of us make the same mistake: first thinking that precise time is the goal, then thinking that precise frequency is what counts, and later thinking that stability is what really matters, and only eventually realizing that all of these metrics are functions of tau, and that tau ranges from MHz/microseconds to years. Phase noise plots along with log-log ADEV plots start to tell the whole story.

In the case of digital music, as far as I know, L(f) phase noise in the audio band and ADEV(tau) frequency stability from microseconds to seconds is far more important to the fidelity of digital recording and playback than absolute SI-accurate frequency or long-term timekeeping. Consequently, most atomic frequency standards are actually a poor choice as a sampling reference clock -- because their jitter (short-term noise) is no where near as good as a free-running, undisciplined, high-end OCXO.

True, the PRS10 is a better choice than other cheap telecom rubidium's but none of these comes close to the performance of a premium OCXO. For the ultimate audio reference clock you want to avoid Rb, or GPSDO, or Cs for that matter. Instead pick a 1e-12 or 1e-13 stable OCXO, strap it to a 100 pound block of granite, and leave it alone.

/tvb




More information about the time-nuts mailing list