Connect with us

10^-9 frequency stability without ovenized crystal

Discussion in 'Electronic Design' started by Andrey, Sep 20, 2004.

Scroll to continue with content
  1. Andrey

    Andrey Guest

    10^-9 frequency stability without ovenized crystal
    How to get it?
  2. John Larkin

    John Larkin Guest


  3. artie

    artie Guest

    A common solution is the HP Z3801A, which is a GPS-disciplined TCXO.
    They still occasionally show up on eBay. There's a Trimble
    more-or-less equivalent which uses a different command set. They were
    developed for the cellllar market, providing time and frequency
    standards at cell sites. Do a google search on Z3801A and you'll get a
    lot of info. Still quite popular in ham radio circles.
  4. Dbowey

    Dbowey Guest

    Andrey posted:

    << 10^-9 frequency stability without ovenized crystal
    How to get it?
    As I recall Stratum 3 is 10E-9.

    Do a google search on "stratum 3 clock." You may gain some insight there.

  5. Mike

    Mike Guest

    Over what time period do you need 10^-9 stability?

    -- Mike --
  6. Mike

    Mike Guest

    You could try this:

    It's a NIST atomic clock in a small package - it's less than 10 mm^3 in
    size, and dissipates less than 75mW. The fractional frequency instability
    is 2.5*10^-10 over a 1s period.

    -- Mike --
  7. Tim Shoppa

    Tim Shoppa Guest

    Actually an OCXO is inside the Z3801A. Typical stability is
    in the range 10^-11 or 10^-12 (Allan variance over 1s to 100s)
    with some doing better.

    TCXO's typically are +/- a few ppm over a wide temperature range.
    Maybe some will be specced to 0.1ppm (100ppb) over a more limited
    temperature range. GPS-locking may indeed give you 10^-9 stability
    over long times but then you're really measuring the Allan variance
    of the GPS, not of the TCXO.

    Over short times and under benign environmental conditions (not
    an oven but ambient temperature not changing much) a really good
    TCXO will show an Allan deviation better than 10^-9. But most TCXO
    modules on the market today are for the cellphone market and have
    enough Vcc and load dependence that you won't see anything this good
    unless you treat them with kid gloves.

  8. Tim Shoppa

    Tim Shoppa Guest

    An example of commodity TCXO's achieving circa this sort of stability
    can be seen at

    He used Connor-Winfield TCXO's treated with kid gloves (very
    controlled loading and Vcc, physical isolation of the TCXO comparable to
    an OCXO but no oven, maybe better than average office temperature

    Some of the units did better than 10^-9, others did not quite that good.

    Note that these were really VCTCXO's, as are most of the commodity
    TCXO's on the market these days (cellphone market). Maybe a non-VC
    TCXO would have less Vcc sensitivity? (Or maybe not!) The nice thing
    about these commodity VCTCXO's is that they are made by multiple
    manufacturers, are readily available through the catalog houses,
    and are cheap ($7-$20).

  9. Ian Buckner

    Ian Buckner Guest

    Why the exclusion of ovenized oscillators?
    As Don says, there are Stratum 3 oscillators out there, for example
    the Piezo Technology Inc XO5363-002 which says 5x10^-10 over
    0.1-30 secs. That is an OCXO, although it is in a package
    (I've not used them, just had the datasheet handy.)


  10. The required heating of in the order of 1 Watt may
    not be doable ? Especially when the device is to be portable.

  11. Ian Buckner

    Ian Buckner Guest

    A bit less than that - 0.75W for 10 sec, dropping to about 1/3W
    or so for reasonable ambient temps. Did the OP state portable?
    I missed that, but even so a couple of NiMH AA's
    would provide enough power for 20 hours. Putting thermal
    insulation round it could improve on that.

  12. I recall one boss of mine who wanted plenty of airflow around an OCXO
    because it got hot. :)

  13. I read in that Allan Herriman <allan.herriman.hat
    > wrote (in <quvvk0lnn8cgak3r1g9rfjqfl0ub3f03
    >) about '10^-9 frequency stability without ovenized crystal',
    I'm surprised that he didn't want to put it in hot water to make a
    nourishing drink.
  14. Andrey

    Andrey Guest

    Thank you very much tothose who replied, it was very helpful.
    Though it looks like to get that 10E-9 stability I have no choice but to use
    ovenized crystal.


  15. John Larkin

    John Larkin Guest

    Expect a good TCXO to have a drift of many parts in 1e9 per degree C,
    and a drift with time on the order of 1e-6 to maybe 1e-7 per year if
    you're lucky. A *very* good SC-cut ovenized oscillator will typically
    drift 1e-9 or so in a week.

    What do you need such accuracy for?

  16. Andrey

    Andrey Guest

    I was thinking to remove ovenized crystal from SARSAT emergency beacon. I
    guess it will not work. You prognosis in your last EMail matches very
    accurately to my findings. Thank you for help.
  17. dd

    dd Guest

    TCXOs have to sweep a wide temperature range and all crystals have small
    purturbations over temperature due to unwanted responses. An OCXO
    crystal sits between the temperatures that give purturbations.
    There was an emergency beacon that employed a crystal under the armpit
    to maintain close temperature control.
    I could envisage other constant temperature locations.
    An SC cut matched to the individual could meet your requirements.
  18. Well, there are 2 reasons trying to avoid an OCXO,
    power consumption and cost. Perhaps size at one point,
    but the current OCXOs are rather tiny compared to what
    size they were 2 decades ago.
    Yes, thermal insulation may help a fair bit on power.

Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day