GPS Disciplined Oscillator

Time and Frequency Standards

With current GPS technology it is possible to have a frequency standard at home in the shack (or portable in the field) that attains 1 part per Billion (1 PPB or 1 in 10e9) (possibly approaching 1 in 10e12) accuracy by tracking the Navstar GPS Satellite Atomic Clocks. This can be used to get your rig on the right frequency, calibrate your frequency counter or keep your computer clocks accurately synchronized. It can be done for less than about $200, maybe quite a bit less if you shop carefully.

The most important part of a Frequency Standard is the oscillator. It needs to be stable yet controlled by a voltage input. Oven Controlled Crystal Oscillators, or OCXOs are very good. The best units contain double-oven crystal oscillators. A GPS receiver with a precise one pulse per second output (or in some cases 10 khz) is used to drive a phase control that generates the correction voltage. Together they can reach about 10e-11 or 10e-12 accuracy, or a little better than a Rubidium Oscillator w/o GPS.

Packaging a GPS Disciplined Oscillator together with some useful tools is a good project. Combine the oscillator with a frequency counter and a Direct Digital Synthesizer (DDS), and you can put a precise signal on a desired frequency. Include a phase comparator and you can set an external signal source to precisely the frequency of the DDS. Or take a modular approach and keep the oscillator and GPS in one unit, and feed 10 mhz to the other boxes. This makes it easier to upgrade part of the system, and complete components independently.

For those who would like a GPS Disciplined Oven Crystal Oscillator there is a new player available at reasonable cost. The Trimble Thunderbolt is about $1600 new, but is available on eBay for about 10% of the original cost. These are pulled out of cellphone sites and are generally in good working order. The Thunderbolt includes the GPS receiver, oven crystal oscillator, and phase lock control system with sophisticated programming. The base unit requires +/-12V and 5V and an amplified GPS antenna. It provides a serial computer interface and both 10mhz and 1pps precision outputs. A packaged model adds a 24V input power conversion supply and a case. Do some searching on eBay for Thunderbolt GPS or Trimble Thunderbolt.

For those who really need even more accuracy, a Rubidium Standard Oscillator can be phase locked to GPS. The Rubidium is good at short-term accuracy (~10e-12), but they drift in the longer term (more than ~10e-10/yr). GPS can be used to correct this. The combination of the two is pretty good, at least. (Rubidium with GPS ~10e-12).

There seem to be three types of GPS receivers:

1. Those with NO accurate time output pulse, such as portable location GPS', which are NOT useful for disciplining an oscillator
2. Those with 1uS accurate time output pulses
3. Those with 15..100nS accurate time output pulses

The GPS' with T in their name are usually the third type above, with high precision outputs. These units are NOT designed for being moved about, they must be stationary. If one is moved a 'survey' must be performed before they will again have the properly synchronized output pulses. The survey takes a few minutes and determines the location of the GPS prior to making accurate time or frequency measurements. The 1uS type GPS receivers do not seem to have this 'survey' requirement.

Some GPS units go a step further and have a 10khz synchronized output in addition to the 1pps. This is much better for feeding a phase lock loop to discipline an oscillator.

Articles

• http://gpsdo.i2phd.com/∞
• http://www.realhamradio.com/GPS_Frequency_Standard.htm∞ HP Z3801A Z3816A
• http://www.rt66.com/~shera/QST_GPS.pdf∞ QST Article
• http://www.rt66.com/~shera/index_fs.htm∞ QST Authors Web
• http://www.jrmiller.demon.co.uk/projects/freqstd/frqstd.htm∞ G3RUH HP Conversion
• http://www.jrmiller.demon.co.uk/projects/ministd/frqstd.htm∞ G3RUH Mini Version
• http://a-aengineering.com/∞ PCB Source

GPS Receivers w 1PPS and 10khz Outputs

• http://dev.navman.com/oem/products/gps_receivers/jupiter_t/∞ NavMan Jupiter T
• http://www.gpskit.nl/documents/rockwell/jupiter-gps-board.pdf∞ Rockwell Jupiter

GPS Receivers w 1PPS Outputs

• http://www.garmin.com/products/gps18oem/∞ Garmin GPS 18
• http://www.trimble.com/tmg_resolutiont.shtml∞ New Trimble type T GPS receiver
  • Trimble Manual∞
  • Trimble Brochure∞

GPS Antennas

• on Ebay∞

Oven Controlled Crystal Oscillators

• http://www.isotemp.com/ocxo134.htm∞ IsoTemp Oven Controlled Crystal Oscillator
• http://www.bliley.com/index_003.htm∞ Bliley
• Search EBay for "OCXO"∞

Rubidium Standard Oscillators

• http://www.thinksrs.com/products/PRS10.htm∞

Synthesizers and Counters

• http://www.amqrp.org/kits/dds60/index.html∞ 60mhz Synthesizer
  • http://www.analog.com/en/prod/0,,770_843_AD9851,00.html∞ 9851 DDS Data
• http://kd1jv.qrpradio.com/msdds60/msdds60.HTM∞ Display for above (note, for some reason Steve deleted this from his web when he ran out of kits)
• http://www.norcalqrp.org/fcc1.htm∞ Norcal 50mhz LCD Counter
• http://www.norcalqrp.org/fcc2.htm∞ Norcal Synthesizer 20mhz for above Counter
• http://kd1jv.qrpradio.com/ddial/Ddial.HTM∞ 40mhz LED Counter
• http://home.att.net/~jacksonharbor/prescale.htm∞ 500 mhz Prescaler

PLL Systems and Parts

• http://gref.cfn.ist.utl.pt/cupido/reflock.html∞ Reflock I, II
• TI CDC7005 phase lock IC

Desired Features in a Disciplined Oscillator for the Shack (AlanB, WB6ZQZ)

• GPS receiver, external pre-amplified antenna
• Oven Controlled Voltage trimmed Crystal Oscillator
• 10 mhz disciplined output (buffered)
• other fixed-frequency disciplined outputs?
• 10 khz GPS output (buffered)
• 1 pps GPS output (buffered)
• RS232 I/O to GPS (db9, buffered, w/ 1 pps to xx serial line, for cpu time)
• Dual Time of Day display (clock) (both local and GMT displayed)(also date, day of week)
• 12 volt input for power (less than approx 2A)
• Internal rechargeable battery to keep GPS memory up
• 1 part in 10e9 stability and accuracy (or better)

Separate Boxes w/ 10 mhz inputs

• Disciplined synthesizer, 0.1-30+ mhz by 1hz steps, FSK input
• Disciplined frequency counter, 0-30+ mhz, prescaler, dual input differencing?
• Phase Comparator (DDS to counter input)/Frequency Difference readout

Frequency Schema

How Long to measure Phase Error with 1uS accurate GPS time ticks

• 10e-6 in 1 second
• 10e-7 in 10 seconds
• 10e-8 in 100 seconds (1.7 min)
• 10e-9 in 1k seconds (17 min)
• 10e-10 in 10k seconds (2.8 hr)
• 10e-11 in 100k seconds (28 hr)
• 10e-12 in 1m seconds (11.8 days)

Integrating Correction Algorithm

1. In a counter count the 10 mhz OCXO (most convenient if counter wraps at 10e7)
2. Configure the 1hz GPS pulses so they capture the 10 mhz count (need xx low bits only)
3. After x min warmup and GPS lock delay capture the count, this becomes offset
4. At each subsequent GPS pulse (interrupt on these):
   1. compare the just-captured value with the original offset
   2. Report the magnitude of the error as the apparent accuracy
   3. When the average error exceeds 2x the uncertainty of the GPS signal perform a correction
      1. start a new cycle. when the error is observed smaller than the previous reported accuracy, increase the accuracy display.
   4. Anytime GPS lock is lost return to step 3

Other Time/Frequency News

• New Mercury Clock∞