Cheapest True Atomic Clock

From Net127 notes

Jump to: navigation, search

"I set my watch to the atomic clock"
--Bright Eyes - Easy/Lucky/Free


A clock is a device for keeping time, . An atomic clock "is a type of clock that uses an atomic resonance frequency standard to feed its counter."

Page 4 of this document has an excellent block diagram of a Rubidium Frequency Standard.

(I've just moved the bits and pieces of my clock to a new office. Hopefully, they will be hooked up and online soon. -glenn 2006 Dec 13)


Contents
[edit]

Here is my design for the Cheapest True Atomic Clock:

  • First, you'll need an oscillator.
    • Try searching ebay for an oscillator.
    • I got an EFRATOM LPRO-101
      • I paid $250 for it, Symmetricon quoted me $2,250 for a new one.
      • Manual from Symmetricom
      • It has a Rubidium standard.
      • Symmetricon says is should last a very long time. (Link to PDF Here)
    • SRS offers the PRS10 for $1,495
      • "The PRS10 can time-tag an external 1 pps input with 1 ns resolution. These values may be reported back via RS-232, or used to phase-lock the unit to an external reference (such as GPS) with time constants of several hours. This feature can provide Stratum 1 performance at a very low cost."
      • Qty 1 pricing from SRS for the PRS10 is $1,495.


  • Second, you'll need a way to count the oscillations:
    • First, you'll need a way to get 1 PPS if your oscillator doesn't have it already: TVB Divider from Brooke Clarke
      • The TVB uses the 74AC04 for the output counter
    • You need a way to get the 1 PPS into the computer serial port. Try a Maxim CP233.
    • You may also want to use a 1-Wire Dual Counter for backup.


  • Third, you'll need to be able to display (or use) the data:
    • The Precision Clock from Brooke Clarke will be great, when they are available.
    • To run an NTP server, you'll need:
      • An old computer with serial ports
      • FreeBSD is recommended for running NTP servers.
      • ntpd (included in FreeBSD)
    • Software:
    • You may also want a mechanical display. I got a six digit, 24VDC mechanical counter out of an old HPLC. I'm working on getting it hooked up now. I think I'm going to connect it to the 100s output of the TVB divier, which will give me an 11+ year epoch. Since the design life of the LPRO is 10 years, an 11 year epoch seems resonable.


  • Fourth, you will need Time Transfer
    • GPS
    • UTC
      • Coordinated Universal Time (UTC), maintained by the BIPM, is the time scale that forms the basis for the coordinated dissemination of standard frequencies and time signals. The UTC scale is adjusted by the insertion of leap seconds to ensure approximate agreement with the time derived from the rotation of the Earth. These leap seconds are inserted on the advice of the International Earth Rotation and Reference Systems Service (IERS).
      • Physical realizations of UTC – named UTC(k) – are maintained in national metrology institutes or observatories contributing with their clock data to the BIPM.
      • Differences between these physical realizations of UTC for these institutes or observatories are published monthly in Circular T.


  • Fifth, you may want to calibrate things
    • HP 5334A/B's are recommended by time-nuts
      • Manual from HP
      • You can get 0.1 femto second resolution on them
    • You'll need a GBIP<>USB converter, like from Prologix
[edit]

Photos

  • 01010019.jpgLPRO in the Box


[edit]

Design Considerations

  • Power Sources Requirements
    • The LPRO requires 13W @ 24v, dropping to 9W @ 19V at 65C. So, I'm going to try and keep that thing warm. Being able to connect it to different power sources is also a good thing. The idea is to keep the clock running as long as possible. A battery in the Osc. box should help with transistion between sources.


  • Power Sources
    • Solar
      • Here, you could opt for a 24V system (panel, charge/load controller, batteries), or a 12V or a 48V system with the appropriate converters. Try AstroDyne for AC/DC and DC/DC converters. Remember, converters are only ~75% efficient.
    • Wind
      • Pretty much the same deal as solar, works better in some areas.
    • Grid Electric
      • Great when you've got it. You'll still want to go for the AC->24VDC->Charge Controller->battery, so you'll always have a backup. Please don't put an AC/24V converter on a UPS and think that's done. It's one of the most energy wasting solutions available.
    • PoE
      • Yes, I think you can run an LPRO from Power Over Ethernet. It might be pushing it a little, but if the Osc is kept warm, it's within spec. You'll need a 12V to 24V converter.


  • Counting
    • Lately, I've been leaning toward the 1wire dual counter. It's certainly not the cheapest, but it may be the easiest. You could also add temperature sensors to the 1wire bus. BTW, counter one counts the number of power cycles while counter two counts the 1 PPS. It's very important to count both of these things. You need the power cycle number to know if an offset is still valid.


  • Installation: Here's the setup I'm working toward
    • Motorola Mag Mount GPS Antennta attached to our DISHtv dish
      • The dish makes a decent mount and the ant. stays stuck, even in storms!
    • 6m ant. cable down to a "nice box" in the bedroom w/LPRO, UT+, TVB-DIV and interface board
    • 3x ~30' coax down to the 5334A in the server closet
      • 1: 10 Mhz Ref. to 5334A ext. osc.
      • 2: 1 PPS from GPS to INPUT A
        • Since I'm testing the TVB-DIV and the UT+ doesn't shut off the 1 PPS, this is the more reliable of the 1 PPS's.
      • 3: 1 PPS from the TVB-DIV
  • 1 CAT5 cable, also down to the server closet, to the NTP server
    • GPS TX/RX
    • TVB-DIV STOP signal / 1 PPS (RS232 level)
    • +12V/GND
    • 1-wire interface (for temp. monitoring and a separate counter)
  • GP-IB/USB adaptor from the 5334A to the NTP server.

This should allow me to measure time and freq. offset and drift.

[edit]

References

[edit]

Current Research

Due to a new development, it is now possible to put a complete Rubidium atomic clock together with a pulse per second generation fully synchronized to reference signal such as GPS /PPS time signal or any other signal extracted from telecom ISDN frame synchronization pulse inside a volume of 90 cm3

[..] At present, the room temperature atomic hydrogen maser is the most stable time and frequency source available for intervals from seconds to days. [..] We are developing the double-bulb rubidium maser (DBRM) as a new high-stability active oscillator. The DBRM will operate at 6.835 GHz on the 87Rb hyperfine transition. The DBRM has the potential to provide frequency stability better than that of a room temperature hydrogen maser, but in a unit that is smaller, lighter, more robust, and less expensive to manufacture and maintain.

We have built and begun characterization of an engineering prototype rubidium atomic fountain at the USNO for use in its clock ensemble. We achieve short-term stability of 1.4×10-13/τ1/2, which meets the desired performance for characterizing maser drift. We will discuss the design, operation, and preliminary characterization of this device.


[edit]

Other Links

[edit]

Possible Controller Hardware

[edit]

Measurements

  • 2006-07-07: Connected Rb Osc. to TVB Divider to 5334A; Counted Seconds
  • 2006-07-08 Measurements


[edit]

Things to do with an Atomic Clock

  • Impress Your Friends
  • Possibly, know the correct time.
  • Start a radio navigation network. (Some assembly required.)
Retrieved from "https://net127.com/notes/index.php?title=Cheapest_True_Atomic_Clock"
Views
Personal tools
  • Log in / create account
Toolbox
  • What links here
  • Related changes
  • Special pages
  • Printable version