The Global Positioning System (GPS) maintains extremely accurate system clocks and communicates this to all receivers.  GPS receivers internally keep time better than nanoseconds and the time output at the 1 PPS dedicated time port is typically better than 1 microsecond.  (NOTE, however: the LCD display on  consumer handheld receivers  generally does not show the precise time  - see below).  Excellent background information on the GPS system can be found on Samuel Wormley's site  GPS Resources  and a more technical web site by  Sam Storm van Leeuwen.

The GPS system as a source of time has the following advantages:
 -  Extremely good accuracy and traceability,
 -  Available reliably anywhere in the world, 24 hours per day,
 -  Provision of positional information as well,
 -  Expected to be stable for many decades to come.

For the new user, a real problem is that  at present, the availability of GPS timing equipment with output directly suitable for astronomical timing is extremely limited  -  and often quite expensive.   However, some groups are now working to remedy this situation - see the discussion below of the KIWI  design from New Zealand.


As many scientific observations are made using video recordings there is a growing interest in "Video Time Inserters" which add time information to a video signal stream.  A number of active individuals have developed low cost time inserters and much technical information is available from the respective web sites and on-going discussion can be found in the  IOTA Occultations email group.  Please follow this link   to the video time inserters paragraph   lower on this page.

A special word of thanks to Geoff Hitchcox in Christchurch, New Zealand for his technical advice to the RASNZ Occultation Section on GPS Timing applications.    However, this web page was put together by Alfred Kruijshoop, who is fully responsible for any errors and omissions.

A lot of discussion on GPS timing techniques is accessible through the IOTA Occultations email group:


It has been found by many users in the USA, New Zealand and Australia that the LCD time display on navigation type GPS receivers (Garmin, Magellan, etc.) is often "pretty good", but  can be early or late by as much as 1 or 2 seconds.  This can be readily shown by comparing the LCD display on a handheld GPS receiver with another source of known accurate time.  This discrepancy  is probably caused by the relative priorities of the computing tasks performed by the microprocessor in the unit, which gives highest priority to computing position and velocity.  The time discrepancy for any one receiver is not necessarily constant but may vary with the internal computing workload of the device, e.g. the number of satellites tracked at the time.  In some models it has been shown to also depend on the power mode ('Battery Save', 'WAAS', or 'Normal'). 

It is important to only consider using the display on these receivers once they have a full 'lock' and advise that they are receiving signals from enough satellites for navigation.  Note that before the GPS receiver achieves lock, the 'display' simply reads an internal crystal clock, using a time setting left over from the last time it was turned on.  Only upon achieving lock is the internal clock synchronised with the GPS data, and even then there is a possibility of the above mentioned 1 or 2 second discrepancy.

Therefore, the LCD display on a standard handheld GPS receiver should not automatically be accepted as being 'accurate'.  You have to get to know the characteristics of your particular receiver, and do so in a range of different circumstances.  On the other hand its long term stability over extended periods is excellent, because even after many years it will still be within that same +/- approx. 2 sec range.

The Japanese 'MICON-GHS'

One of the first published (2001) successful GPS based timing devices for remote field use is the  Japanese 'MICON-GHS Clock'  developed by Geshiro Hiroyuki (Oishikohgen Observatory), Hayamizu Tsutomu (Sendai Uchukan), and Soma Mitsuru (National Astronomical Observatory) in Japan (initials G, H, S).  This is a well engineered GPS time receiver with a microprocessor, visual display, electrical and audio outputs, very suitable for occultation applications, and a number have been produced in Japan.  Refer to this   message by Mitsuru Soma   in the IOTA occultations email group.

More recently, Mitsuru Soma advised in 2004 that Tsutomu Hayamizu made a versatile GPS clock using a more readily available GPS receiver (Garmin 15H-W).   The web site:   gives circuit diagrams and explains how to synchronize the internal clocks of your PCs and how to get the geodetic coordinates
using Takashi Setoguchi's free software "Satk".   Related information can be found at   A number of these units have been built and used extensively by Japanese observers in the field.

Full credit is given to these Japanese groups for demonstrating at an early stage what 'can be achieved with this technology'. 

For timing purposes, a GPS receiver (board) should have:

(a)  A separate dedicated 1 PPS (pulse per second) time output on the receiver  AND  the receiver specifications needs to state explicitly that this is synchronised to UTC and is continually updated as the receiver is running.  The leading edge of the pulse on this 1 PPS output is typically aligned with the beginning of the UTC second to better than 1 microsecond.  Such a 1 PPS output is present on many GPS circuit boards (including some that are inside consumer handheld units); HOWEVER, the handheld units do NOT normally give us access to the 1 PPS port - although we have been advised that they do exist.  Therefore (regrettably) the most commonly available handheld units can NOT be used for timing.

If any readers know of affordable handheld GPS receivers with a 1 PPS output port (in addition to the usual serial port) please advise us through the email address on the  NZ&A Time Resources home page.
(b)  A serial port on the GPS receiver to output the identifier data string.  This is a common feature of all receivers and commercial (consumer) units.  Note, however, that in most GPS units this 'serial' port is not identical to the common PC serial port, and will require one chip conversion from CMOS/TTL before connecting to a PC.  But the serial port by itself is not enough to acquire accurate time - the 1 PPS port is also required.  The receiver can be instructed to transmit each second over this serial port a data string that contains selected information, including position, number of satellites tracked, etc. and the full date and time identifier of the preceding second pulse that came from the 1 PPS port.  However, the start of this data string itself is not accurately aligned with UTC, but often half a second late, depending on the processor workload.

Sources of OEM GPS Receiver Boards

GPS OEM (original equipment manufacturer) receiver modules with 1 PPS output and an active antenna can be purchased new from a range of suppliers.  Do check that the specifications cover the 1 PPS time output port.  In Australia, there are several manufacturers / wholesalers, including Sigtec Navigation  in Fyshwick, ACT,   Rojone  in Sydney,  and Commlinx Solutions in Tasmania.  It is always wise to also check prices in the very competitive US market (see below).  A useful source of GPS Antennas and associated hardware is   Gilsson Technologies   - also check their home page with further links.
An interesting source of Rockwell Jupiter GPS modules and a complete "Do it Yourself Kit"  can be found at    This site gives full technical information and prices and advises that this module and the kit circuit do have the necessary 1 PPS time output and the NMEA sentences.  These modules are unused, still up-to-date, receivers with respectable specs.

Also the VNG Users Consortium in Canberra, ACT, Australia, is considering some form of 'bulk purchase' to obtain a larger number of current production GPS modules at wholesale cost with low shipping overheads.  However, the fire at Mount Stromlo Observatory has caused considerable delay in this.

Low Cost OEM GPS Receiver Boards

Occasionally, a batch of lower cost 'recovered' (used) or 'overstocked' (new) GPS modules becomes available in the USA, and some of these have highly respectable credentials as time generators (e.g. the Motorola Oncore series).  Prices in this 'bargains' paragraph are in US Dollars.

From mid 2002   BG Micro   has been selling  Trimble SV6  boards (used) with  1 PPS output priced as low as US$ 25 for a complete system including active antenna.  As of late 2002,  small numbers of these SV6 modules with active antenna appear to become available at unpredictable times at BG.   BG provides very prompt service and charges low shipping costs to the South Pacific (e.g.: US$13.50 for one SV6 by airmail to Sydney).  Geoff Hitchcox has developed a special SV6 version of KIWI  software because the Trimble SV6 module works better in the native Trimble TAIP protocol than in 'standard' NMEA.  This "SV6 version" of KIWI is probably also preferred for other Trimble GPS units that use the TAIP protocol.

An alternative source (thanks to Jim Vail for his advice) of these Trimble SV6 modules is  Rodax  whose price of US$50 includes shipping within the US.  Overseas shipping charges and arrangements are not known.

BG Micro   also sell a very low cost Motorola GT Oncore module with 1 PPS timing for US$15, but this is without a matching active antenna, which the user has to source from elsewhere (e.g. see Synergy Systems, below who sell antennas with matching OSX / MCX connectors).  Only very few years ago, this GT Oncore was regarded as a superior GPS module in the industry.  Full details can be found in the     BG Micro GT Quick Start Guide .  This Motorola module will work very well with the Motorola option of the  NMEA version of  KIWI .

A useful US source of Motorola OEM boards is  Synergy Systems   Check out their "Excess Inventory" list from time to time; and their normal prices are also quite good, although their 'quoted' shipping and handling charges appear higher than BG.  Recent 'specials' on VP Oncore boards with 1 PPS are listed for US$ 25 -50 for the receiver plus another US$ 15 - 25 for the active antenna (Motorola or  M/A-COM ), with postage and handling additional.  Synergy also make available an excellent detailed  Motorola Oncore Manual  and the VP Oncore Command Reference (650k ZIP file).  When you consider buying these specials please confirm that the supply voltage and connector type on the antenna & cable match those of the receiver (at these high frequencies, changing a connector is not a trivial matter).  Whatever you do, make sure you can download or borrow the  manual for the particular brand and type of module  (also see below).
An interesting source of Rockwell Jupiter GPS modules and a complete "Do it Yourself Kit"  can be found at     This site gives full technical information and prices and advises that this module and the kit circuit do have the necessary 1 PPS time output and the NMEA sentences.  These modules are unused, still up-to-date, receivers with respectable specs.

If you are interested in following this route, please monitor developments in the  IOTA Occultations Yahoo! Email Group   with discussion on these matters, where from time to time members alert each other to the availability of low cost GPS modules and their interfacing requirements.
The capable, low cost 'KIWI' GPS to PC Interface

The OEM GPS board described above has to communicate with a computer or microprocessor using (fast) software to extract correct time (and positional) Information, in order to identify the seconds received from the 1 PPS port.  The following web site describes the KIWI  software package developed by Geoff Hitchcox in New Zealand:

On the KIWI web site is a full description of the hardware and software functionality.  For Trimble SV6 users, there is a link to a "mirror" page about the version of KIWI that uses the Trimble TAIP protocol.  Further, there are detailed construction notes, circuit diagrams and circuit lay-out photographs from Don Oliver and Geoff Hitchcox, with references to earlier work by George Silvis.

KIWI is a freeware program that uses a PC and GPS (with 1 PPS) to timestamp an event to millisecond accuracy to UTC, and also generates time signals (sound and sight) in a format similar to WWV.  The program uses the GPS to firstly calibrate the PC timing and then every 5 seconds resynchs the PC timing to the GPS (to track thermal changes in the PC quartz crystal).  This method enables any old PC to be used to timestamp an event over long periods (days or months even) and yet retain the 1 millisecond accuracy to UTC.  The event to be timestamped can trigger KIWI via a logic signal or by a manually operated switch connected to the PC printer port.  For astronomical events using video, a LED (optional) is flashed for 50 ms to ident the video at the time of the "trigger" for subsequent field/frame accuracy determination.  The PC timing acts as a "flywheel" to the GPS 1 pulse per second drumbeat.  Even if the GPS briefly loses synch with the satellites, the software will retain (for some time) its accuracy, although a warning is displayed on the screen.   The software continually analyses the 1 PPS and serial data for Integrity.  Any change to the 1 PPS is detected and an Integrity Fault report is made, giving the date, time, error, and various GPS status conditions.  The Integrity Monitoring ability gives the user confidence of using the GPS "anywhere" as well as the ability to choose the best place to use a GPS to avoid multipath effects, etc.

In addition to giving access to accurate time, KIWI also displays the normal GPS type information: latitude and longitude (in WGS84), altitude, number of satellites tracked, etc. etc.

At the above sites can be seen that the hardware issues are very straightforward:  a small interface circuit needs to be built with two ICs for a total component cost of around $25.  The programs run on a standard IBM compatible PC, which can be any old 486, 386, 286 ('AT') or even the ancient 'PC'.  A number of KIWI systems have been assembled by observers in New Zealand, the USA and Australia.  An excellent example is the SV6 based KIWI Time Receiver system built by Dave Gault in New South Wales.

The KIWI program has been tested by Art Lucas of IOTA (US) against local traceable time sources (see the Review of the Performance of KIWI  on the above KIWI site), where it was found that the program performs well and matches stated specifications.  Again, it was demonstrated in July 2002 by Geoff Hitchcox to a small group of experienced occultation observers near Christchurch, New Zealand.  Used for this were a Motorola GT Oncore module, KIWI interface and a slow 386 computer.  The program ran off a bootable floppy disk so there was no need to modify the PC in any way (but of course it can be installed on a hard disk if preferred).  During this demonstration the equipment worked exactly as planned, so that time (or positional) information was available on demand.  It was also noted and demonstrated by moving or covering the antenna that the software is designed to prevent the generation of erroneous output.  KIWI will not start until a satisfactory signal is received from the GPS, and during a run it alerts the user e.g. when temporarily the number of tracked satellites drops below a minimum set level.  Clearly, a low cost KIWI system is a reliable source of time at the millisecond level  - this was a very convincing 'proof of concept'.


Using a Garmin GPS 18 LVC as NTP stratum-0 on Linux 2.6

In early 2008 we received the following email advising another low cost solution:

Using a Garmin GPS 18 LVC as NTP stratum-0 on Linux 2.6

<>My first foray into this subject lead me to the following pages:

The former page actually refers to the latter.  They both discuss using the Garmin GPS 18 LVC as a source.  I have found that these are available at as low as $68US including shipping within the US if you search the Internet.  The lowest price I found currently is located at Provantage, but shipment down under would un-doubtably take that price up and might allow a more local site to provide a cheaper price.  These units are weather-proof and simple electronics discussed on the above mentioned websites will permit connection to computers for time server functionality.  I don't know much about the electronics you discuss on your site, but this may well work with them as well.  My research has also revealed that Garmin offers another unit, GPS 17 (available in a low or high voltage package) which also provides PPS.  These units are originally designed for marine use, so they are certainly weather resistant.  The lowest cost I have found for delivery to the US is just over $100US.

I plan to experiment with these units over the coming summer and hope to have a design for a unit that is solar powered and connects to a computer to provide a stratum 1 time source.  My hope is to build the unit for about $100US total.   This will likely use the GPS 18 as it has been tried before and is cheaper.    Then again, should I find a GPS 17 unit on EBay cheap . . .    Both of the above noted units list 1 micro-second (+/-) accuracy for the rising edge of the PPS signal.    Hope this information is useful.

The need for GPS Receiver documentation

To use OEM GPS modules we need access to the Specifications and the essential input and output chapters of the Manual.  This information is needed to correctly interface and 'initialise' the module.
You will need to download or borrow or copy the manual for the particular type and firmware version of module.  Some manuals are available on manufacturer's web sites, or from user groups, but others may need to be purchased.  Because the cost of the Manual can be very high (some are US$50), access to essential documentation needs to be considered before placing an order.

For new users: another point that must be kept in mind (also see Art Lucas' notes)  is that it is probably necessary for the OEM GPS module to be 'Initialised' to:
(i)   'Enable' the sending of the 1 PPS data on the 1 PPS port.
(ii)   Set the 'option' to send its serial port output data strings (sentences) in the data format expected by the version of KIWI - often NMEA, but for Trimble modules TAIP.
(iii)   Instruct the receiver to only send short sentences to ensure they can be read and processed in well under a second.  The 'default' sentence may be too long, so that there is a distinct risk that the identification of the supposedly preceding second could be attributed to the wrong second.  To do this you need access to the Manual for this particular type of GPS module, or know someone who is familiar with it.

Some helpful notes from Geoff Hitchcox on this:  "Most GPS receivers can output all sorts of data sentences on standard RS232 serial format (that can be read by any PC).  If your system uses NMEA standard (at 4800 baud) then there are many (or all) sentences that can be requested from the GPS.  If the GPS receiver has been requested to output either all or too many data sentences, then there is not enough time each second for the GPS receiver to transmit all the data requested, so you get data "leaking" from one second to the following second.  This may explain "whole 1 or 2 second" errors in the identification of the second = the time shown.  If the above scenario is what is happening, then the time display "sometimes" may be reasonably correct and at other times a few seconds "slow".  This situation can be easily fixed by requesting a "single" time sentence containing only the required information, and is further improved if the serial "speed" can be increased to the maximum the GPS can use."

Because, "as received, default" the OEM modules may not always do this, a small "instruction" session is required where the computer tells the OEM module what to do, using software that is often available on the manufacturer's web site.  This "instruction" uses the same serial port connection that is later used to output the NMEA data.  These settings need to be retained in the receiver until the next time it is used.  It is therefore important to ensure that the module has continuous (battery) backup power to maintain its on-board memory containing these ('instructed') settings.  If no backup power is provided, the settings will be lost and the initialisation will need to be redone each time the unit is used.

Microprocessor Developments

It was also suggested that for certain types of field use we could investigate the option of not using the full KIWI program suite, but perform only limited essential tasks using a small self contained industrial microprocessor (microcontroller) module.  This is the approach followed by the VNG Users Consortium  and the  Japanese MICON-GHS  group - see the references above.  Numerous varieties of these cheap processors exist (PIC, AVR, Stamp, NEC, etc.) but it is necessary to know (or find someone who knows) how to program and interface them.  One type of microprocessor often used by hobbyists is the 'PIC', a small single chip computer with programmable ROM and I/O on board.   For field GPS based timing units such a chip could (in principle) read the time identification data string and generate the output pip sequences.  The cost of PIC chips (in local currency in New Zealand and Australia) is around $10 - $20, and a 'programming and test' interface board in 'kit' form costs about $50.  In New Zealand and Australia these are advertised in electronics magazines and available from electronic component centres like Jaycar and Dick Smith.  These magazines regularly publish PIC projects and advertise programming manuals.  Once programmed the chip will run on its own (will need regulated power and a local clock crystal).  Making this work will cost very little but involves a PIC learning curve.  Some software is available on the web.
Tom Clark's "Totally Accurate Clock"

An earlier very interesting "kit" on the market, first published in 1995, was Tom Clark's "Totally Accurate Clock"  and you may wish to check out this site for more background on GPS and on how it can be used for timing.  It consists of an OEM GPS module, a specific 'TAC' interface board, using a computer for processing and display.  In some ways an early version of the KIWI approach.  Note that the most of the published documentation is a few years old, although the group is still active.  Some components used in its original version are no longer available, but alternatives can be used by those with the right expertise.  Note that the device is complex, requires specific components, skilled assembly and is rather expensive.  However, at least one version of this put out 'WWV' style audio time pips.  Considering the early start of this project (1995) this was an impressive piece of work.

*  *  *



As many scientific observations are now made using video recordings there is a growing interest in "Video Time Inserters" which add time information to a video signal stream.    A number of active individuals have developed very low cost video time inserters and useful technical information is available from the respective web sites and on-going discussion can be found in the  IOTA Occultations email group:
Some  VERY USEFUL  Video Time Inserter references

<>Geoff Hitchcox'  "KIWI  OSD" Project    This has developed into the highly recommended commercial  KIWI-OSD system, designed in New Zealand, and built in the US.  It time stamps each field in the video stream. It is simply inserted between the video camera and the recorder. For full details see the      PDFsystems   web site

Don McAfee's "VTI" Project   (not sure if still active?)

The BlackBoxCamera Company 'STV Astro'

There was a commercial video time Inserter on the market made by the   BlackBoxCamera Company Ltd. in the UK.   For some years they manufactured a range of Video Time (Overlay) Inserters, one version of which derives its timing from a GPS Time Receiver.  This should address the problem of acquiring time together with the application that requires this time information.  A number of US astronomical observers use these units and discuss their experiences in the  IOTA Occultations Yahoo! Email Group   where these devices are commonly referred to as "STV" or "STV Astro".   However, in 2004 the company web site advised that production was halted (pending an increase in orders) so that this item may no longer be available.  Anyone interested should contact the company direct.  Some technical information on the product is still present at the web site.

There are many things that can generate unexpected (temporary) incorrect time output from GPS based devices.  The problem with (navigational) handheld units was referred to earlier. 

Another recently detected 'hiccup' was   reported by Oliver Kloes in Germany   and relates to the operation of a GPS receiver when stored backup almanac data have been lost, e.g. after an idle period.   Upon re-starting the GPS unit, for an unknown period of time (which can be considerable.....!), some units may display "GPS Time" instead of UTC.   This lasts until such time that the full almanac data have again been downloaded from the satellites to the receiver. 

As of 1 January 2006 (the most recent insertion of a UTC 'Leap Second')  GPS Time-UTC (BIPM) = 13 seconds.  Please note, however, that   the difference between GPS time and UTC will change in the future if further Leap Seconds are needed (click here for authoritative references).

Again, this can only be detected, recognised by using an independent second source of time, such as WWV or in some cases even a good watch. Therefore, a full understanding of your technology is essential and your work should preferably use a second system (based on different principles) as a 'sanity check'.   It is unwise to automatically accept any GPS time output / display (or any other technology, for that matter) as being 'correct'  just because it is derived from the GPS system.  


Click here to return to the Time Home Page