In RASNZ Occultation Section Circular CN2002/4 Brian Loader describes a PC program, KIWI.EXE written by Geoff Hitchcox of Christchurch which uses a GPS unit connected to a PC and provides time signals and the ability to accurately record the time of events such as occultations.   To supplement this system Brian also wanted a small low cost 'standalone' portable unit to provide audible time signals which could be recorded in a similar manner to the VNG time signals when observing a graze or a minor planet occultation.

Over the past few months I have been developing an electronic 'Timer Box' to fulfil this requirement.   The box is approximately 10 cm x 8 cm x 4 cm and runs on four AA batteries.   It uses a Piezo buzzer to output 'beeps' at one second intervals.

Features of the timer box include:
     -    Portable and simple to use.
     -    Low cost.
     -    Runs from 4 AA batteries (6 volt).
     -    Provision for alternative power source of 9 - 12 volt batteries.
     -    Audible one second time signals as in the paragraph above.
     -    Electronic synchronization with a KIWI/GPS system.
     -    Electronic calibration with KIWI/GPS system.
     -    Manual reset.
     -    Optional 'mute' operation.
     -    Optional additional output to LED.
     -    Optional connection to video sound systems.
     -    Accurate timing system.

Some of these features are discussed in more detail below.

Portability, simplicity.
The timer box is approximately 10 cm x 8 cm x 4 cm.   With a battery holder for 4 AA batteries it can be run anywhere at any time and used by observers who do not have their own timing equipment available.

Low cost.
All of the components are readily available in retail electronic component shops.   The timer box can be assembled for approximately NZ$50.   The most expensive component is the programmable microchip which costs around NZ$15.

Power source.
The timer box can be run from 4 AA batteries (total 6 volts).   My prototypes also contain provision for an additional 9 - 12 volt DC supply.  I use an external battery holder which can easily be disconnected and allows the batteries to be changed or removed when the timer is not in use.   If required the batteries could be placed inside a [slightly larger] timer box.

Audible time signal.
The audible signal is provided by a small Piezo buzzer mounted inside the box.   The buzzer emits a 'beep' of approximately 50 milliseconds (msec) duration at one second intervals. Every tenth second the 'beep' is slightly longer.   On the 55 - 58th seconds of each minute the 'beep' is very short.  The short beeps start at 51 seconds on the fifth minute.  The 59th second is silent and the minute is marked by an extra long beep (500 msec).  This very closely emulates the former VNG time signals.

Electronic synchronization with KIWI GPS system.
To obtain electronic synchronization with the KIWI GPS system Geoff  very kindly modified the KIWI program so that in addition to the one second time signals, KIWI outputs a separate electronic signal each minute.  The Timer box can be temporarily connected to the KIWI system, synchronized to the minute time signal and then disconnected from KIWI, taken to the graze observing site to provide an
audible time signal for recording on to audio tape.   Note that once the timer has been synchronized with the GPS it should be kept connected to the batteries.   If the batteries are switched off or disconnected the timer operation ceases. When the power is turned back on the timer automatically restarts, but will no longer be synchronized with the GPS system.

Electronic calibration with KIWI GPS system.
In addition to the one second signal the Timer box also provides an electronic signal on each minute.  This permits checking and calibration of the Timer box.  The Timer box can be reconnected to the KIWI system so that the one minute signals activate the KIWI timestamp function.   This records the time to 1 millisecond accuracy so that any errors or drift in the Timer box timings can be measured and taken into account when reducing observations.

Manual reset.
This allows the user to set the timer on a time signal other than a KIWI system.  The manual reset was initially a simple push button switch which could be pressed at a given moment (e.g. radio time signal 'pips').  It is provided so that the timer box can be used even if the user does not have access to a KIWI GPS system but wishes to set the timer reasonably accurately.   I now incorporate two reset push buttons in series.   This reduces possibility of accidentally manually resetting the timer after it has been synchronized with a KIWI system.   Both push buttons now have to be pressed to reset the timer.

Optional 'mute' operation.
A simple on/off switch connected in series with the Piezo buzzer allows the timer to continue operating but without the audible signal.   This is useful when it is desirable to synchronize the time with a KIWI GPS unit at home, then travel to a grazing occultation site.   In such a situation the audible signal is not needed while travelling but can be turned on once the graze site has been reached and the observer is ready to start observing.   The audible signal can then be turned off after the event, allowing the timer to continue operating until such time as its drift rate can be checked.  (See notes below on calibrating the timer unit).

Optional output to a LED.
If a video system is being used to record an event (or multiple events) an accurate timestamp facility on the video is needed to allow accurate time analysis of the events recorded.   The Timer box also has provision for output to a LED.   This provides light flashes with exactly the same timing characteristics as the Piezo buzzer.   The LED can be mounted in the telescope field of view so that the one second flashes are recorded on the video system.   When playing back the video tape in order to analyze the event(s) the one second flashes are visible on the tape.   These can be used to determine the exact time for each recorded event.   Using the slow playback or frame stepping facility of the video player it is possible to determine the time of each event to within  0.02 sec depending on the characteristics of the video player.   Alternatively the flashing LED could be used to accurately calibrate commercially obtained video time stamping systems.

Optional output to a video sound system.
Another way of using the output from the timer would be to connect it directly to a video recorder's sound system.   The signal from the time will then create a 'double click' on the video tape so that when you play back the tape for analysis of the data the time signal is audible.

Accuracy.
Initial tests of my first prototypes indicated that the timers were operating with close to 1 part per million (ppm) at around 25° Centigrade and about 20 ppm at 15°C. As the Timers were required for use on grazing occultations where night time temperatures could be expected to be around 5°C I modified the program to slow the Timer slightly.   They are now operating within 5 ppm at 5°C.   See the technical notes below for more details about this.

Conclusion.
This portable system could be used for recording the times of grazing occultations, minor planet occultations and other events accurately.  Combined with the KIWI GPS system this overcomes the gap left by the closure of VNG.  Furthermore the system is not subject to 'fade out' or noisy interference from local power lines.
Anyone who is interested in creating their own timer boxes is invited to contact me for further information.   I am willing to program suitable model PICs and to provide detailed instructions and circuit diagrams to anyone wishing to create their own timer boxes.

Some additional technical notes.
The main electronic components of the Timer box are a 4 MHz crystal driving a Programmable Integrated Circuit (PIC).   The PIC is programmed to output electronic signals to a Piezo buzzer, LED and KIWI as required.   The connection to KIWI for synchronization and calibration is via a separate lead.
The assembly language program for the PIC was done on an ordinary PC running Windows 98 and a free programming development system downloaded from the Microchip website.   The compiled program is loaded on to the PIC using a programming circuit board connected to the printer port of the PC before inserting the PIC into the Timer circuit.

The accuracy of the Timer depends on the accuracy of the 4 MHz crystal.   This can vary depending both on how the crystal is manufactured and the temperature at which it is operating..
To test the accuracy of the Timer box I used an electronic temperature sensor and another program (also written by Geoff Hitchcox) which automatically logs the temperature (with 0.5°C accuracy) at one minute intervals.   After initial tests the program was modified so the timer boxes now have an error rate of 1 ppm and 5 ppm at 5°C respectively.   An error rate of 5 ppm is equivalent to about 0.018 seconds per hour.  The attached graphs show the results of some of my timing tests after modifying the program.
At 10°C the error rate is approximately 8 ppm different to the error rate at 5°C.   The differences between the two timers is mainly due to variation in the crystals driving the PIC.   This means that each timer box should be calibrated so that its own characteristics are known.   Because of temperature variations each timer should be checked against the Kiwi system as soon as possible after each graze so that any drift can be taken into account when reducing observations.

Future improvements.
I am working on two related improvements to the Timer boxes.   The first is to enable the program on the chip to take into account differences between crystals driving the microchip.   Ordinary commercially available 4 MHz crystals vary depending on how they have been manufactured.  Typically the manufacturers claim approximately 50 ppm accuracy.   The timing tests I carried out above were sufficient to show up differences between the two I am using for my prototypes.   The first improvement I am working on is to modify the PIC program to allow these differences to be taken into account more easily.   The second improvement will then be to add facility to automatically calibrate each individual timer.   This second improvement will involve both additional changes to the  PIC program and changes to the connections for the KIWI GPS system to allow the modified timer to use the KIWI one second signals for the calibration.   Once these modifications have been incorporated it will then be possible to quickly calibrate each timer to work with 1 ppm accuracy at a given temperature.   Timers could then be recalibrated at different times of the year.

Acknowledgements.
I am indebted to Alfred Kruijshoop (Melbourne) for his original idea for the timer, his suggestions and feedback on the first prototype.  Also many thanks to Geoff Hitchcox of Christchurch for advice on how to program integrated circuits, and ideas on how to interrelate with KIWI for synchronization and calibration.   And thanks to both of them and Brian for their encouragement in developing the project.

Tests on timer 2.   This shows the raw data from a typical test run over a period of 24 hours.  The timer was placed in a insulated chilled container and the temperature and timer rate monitored electronically.   The thick black line shows the temperature (scale on left of graph) and the grey line shows the corresponding timer error rate (scale on right of graph).

Comparison of the error rates of two timers against temperature showing clear difference in error rate