For more information about Geoff Hitchcox's GPS timing systems refer to his page:
http://www.geocities.com/kiwi_36_nz/kea/kea.htm

To be added to the email discussion group on this topic please email Graham.Blow@actrix.gen.nz

Geoff Hitchcox, 8 August 2001:

Occultations using GPS

I am currently doing R&D on a new gravimeter and so I have lots of
software code interfacing to GPS equipment. So I thought I would 
throw together some code to enable my PC to timestamp a contact closure 
to about a microsecond accuracy (if only my reaction time was as good). 
But before I do this, I thought I would double check to see if you knew 
of any IBM PC program that already does this? (No point in reinventing 
the wheel). The only similar project I could find related to a Japanese 
circuit that interfaces to a GPS to enable accurate timing, but that 
seems a very expensive way to go when all the hardware is in the PC.

[Return to Top]


Geoff Hitchcox, 9 August 2001:

Occultations using GPS

> How accurate can you get timewise with GPS? Bearing in mind that I have
> no experience, I'd still be a bit surprised if you could get absolute
> timing accuracies down to about 0.1 sec, which is the aim with
> occultations.

That is a good question Graham. If you go down to Dick Smith and buy a
consumer GPS, the time "displayed" is up to 1 second late (although the
Rx knows the time internally to a few hundred nano seconds). Time is
considered a low priority on the LCD in a consumer GPS.

In fact I have achieved better timing using a $5 digital watch set to Nat
Radio time pips than what a consumer GPS can give!

However if a GPS has available 1PPS (one pulse per second) digital
signal, then one has digital "pips" aligned to UTC to better than 1 
microsecond (300 nS on my Rx for example).

It is the 1PPS that I am feeding into the PC. By shutting down all
interrupts and using a fast polling technique (faster than interrupts)
one can achieve accuracies of 1 microsecond to UTC  - even on an old 
20MHz AT PC!

This means I will "date/time"stamp a contact closure to an accuracy of 1
millionth of a second to UTC (as well as give the Lat and Lon) at time of
measurement.

It is a pity my reaction time (at best) will be about 100,000 times worse
!!!!!!!

> I do know that some people have tried setting their PC's clock using
> wesbites like timeanddate.com, but when I checked this myself I found
> that because of internet propagation delays errors in setting the time
> were quite significant.

It is propagation delays that often limit this. If one has access to a
"time server" on ones own network, it then comes down to (variable) network
delays and operating system.

> Failing that, there's always the IRL talking clock on 0900-456578 -
> assuming you can find some way to get your PC to sync to its time pips.

One has to be careful with time pip source ;-)

I confess to writing all the software for the BBC World Service  audio
equipment in Auckland. I take the BBC off a satellite dish (already
delayed by 2 or 3 geostationary satellite hops) digitise the audio and save 
into memory. By playing back the audio 5% faster (but shifting the audio
frequencies down 5%) I can squeeze in 3 minutes of advertisements into
the hour without losing any of the BBC program (the whole process is
automatic). Consequently I replay the BBC pips about 13 seconds (+/- 3secs) 
late each hour!!

[Return to Top]


Geoff Hitchcox, 12 August 2001:

Experiment on using GPS and PC for Occultation Timing:

As promised (and requested) here are the results of my little
software "project" to develop a PC stopwatch for Occultation
work.

It has been a "cheap" stopwatch for me to make. I already had
all the parts apart from the (25 Metre) cable to the switch,
so a nice stopwatch for $8!

Sunday Morning 12th August 2001:

I set myself up outside with a great view of the moon, reclining
on a sun lounger, with bino's and the switch to the computer
inside the house!

Last looked at my watch at 07:06:30 (local time) then started to
look at Moon with my 10 X 50 Bino's.

The Moon looked wonderful, no cloud what so ever in the viewing
portion of the sky!

The Moon was in perfect focus, but the rest of the sky was (with
the sun at -4 degrees) a faint milky blue. I could not perceive
any stars anywhere near the Moon. Continuing to look I clicked
the "switch" at what I perceived as a small "flash" of light.
However I could NOT perceive any star after this event, the sky
was just too bright. A few minutes later I came inside the house
and looked at the computer screen it read:

Keypress at 19:07:10.378471 UTC, time to next reference second = 0.621529
$GPGGA,190711.00,4331.1916,S,17242.1316,E,1,03,5.1,33.6,M,,M,, RIGHT CRC
$GPZDA,190711.00,11,08,2001,00,00 RIGHT CRC

That was the only reading. It bears a strong similarity to the
correct time for my location, but as my first attempt at
"occulting" I don't wish to claim it as the reappearence of the
star!

I could see that had the "event" been an hour earlier, I would
have had perfect "seeing" and would have more clearly "seen" the
actual event. A little dissapointing for my "first" observation,
but encouraged enough to give it a go next time a "bright" star
plays hide and seek with the Moon!

The night before, I did the following experiment:

I listened to National Radio a minute before the "time pips"
with my eyes closed (so I could not see any clocks). I clicked
the "switch" attached to the computer when I first heard the
first of the six time pips. You can see that my reaction time
was about 1/3 of a second!

Keypress at 07:59:55.350895 UTC,
Keypress at 08:59:55.329320 UTC,
Keypress at 09:59:55.309317 UTC,


A little more (techie) info for those that are interested.
How it works:

The "manual switch" cable and GPS 1PPS (1 Pulse Per Second of
UTC to 300 ns) plug into the Printer Port of the PC (IBM clone).

The GPS RS232 data (that contains the alpha/numeric data) plugs
into the PC serial port (comport).

The 1PPS from the GPS is first used to auto calibrate the
internal PC timing.

The Software then measures the time from the manual switch
closure to the next UTC reference second. The "time/datestamp"
of that "reference second" is then read off the RS232 serial
line. All the data for that event is then saved to an ASCII disk
file and also displayed on PC screen.

Containing:

UTC time of switch closure (to micro second precision)
Latitude
Longitude
Number of GPS satellites used in location fix.
Horizontal Dilution of Precision (of satellite positions) (HDOP).
Antenna height (metres).
Result of Checksum calculation, on the serial data from GPS.

Comments:

A "1 click does all" (giving time/location and saving to
disk)

Ideal for laptops and remote use.

GPS has worldwide coverage.

Easy future expansion, like producing marker data for video
recorders, using the same (and improved) concept as in the
Japanaese system (very expensive) as per:

http://www2.synapse.ne.jp/uchukan/data/occult/gpsradio/ghsclock.html

(NOTE: The Japanese system requires manual setup each time used)

[Return to Top]


David Dunham, 20 January 2002:

The least expensive digital tuning (that's important for
quickly finding a signal in the dark, especially away from
home) short-wave receiver that we can get now is the
Radio Shack DX-375 AM/SW/FM receiver.  On the back it says
"custom manufactured in China for Radio Shack, a division
of Tandy Corporation, Ft. Worth, Texas" and it costs $100,
although sometimes it's on sale for $70 US.  They sell an
analog shortwave receiver for $40, but as noted in the first
sentence, those are harder to operate.  Also, South Africa
is a long way from Ft. Collins, Colorado, so I think
reception will be quite difficult there; you would need a
good antenna.

>For instance, is there a GPS unit that can output an audio 
>signal? 

GPS may be the best solution for your timings.  But most
commercial units aren't good at timing; you need to get
one with a 1PPS output signal.  But then you need to do
something to use it to generate an audio signal.  Observers
in New Zealand and Japan have worked on this problem and
can give you some help.  In Japan, the GHS clock has been
developed for this purpose, working through a PC.  For 
video observations, Blackbox video in England has developed 
an STVASTRO system that uses a 1PPS GPS receiver to generate 
an accurate time (and coordinates) display; it plus the GPS 
unit you need cost about $250 US; others, such as Hellmuth 
Cuno in Germany, are working on less-expensive options.  

[Return to Top]


Art Lucas, 20 January 2002:

Re: [IOTAoccultations] Help needed for timing occ'ns in South Africa

Steve Preston has provided the web site for Geoff in NZ 

http://www.geocities.com/kiwi_36_nz/kiwi/kiwi.htm 

Where the GPS system is described.  I have built that system, which is 
basically computer based. However, I built my battery box with a flashing 
light on it that flashes when the GPS triggers.  This makes it independent 
of a laptop computer if you photograph the light immediately before and after 
the event with the same camera and VCR that is taping the occultation.  
Frankly, its a bit awkward.   

I built the battery box above to interface to the Garmin 35.  It was then 
only a solder joint or two away from driving the BlackBox STVASTRO.  As 
I have all them working together, I usually use the STVASTRO as it is 
immediate and precise.  In the end, with a group of people, much will depend on
price.  If its between buying a laptop computer and a STVASTRO, the latter 
is way ahead. 

[Return to Top]


Alan Gilmore, 12 May 2002:

(803) Picka occ'n seen at Mt John but...

Using the single channel photometer on the 0.6-m f/16 Optical 
Craftsmen telescope I recorded the Picka occultation.  The duration 
of the occultation seems fairly well determined but I've got a 
problem with the absolute time.

I set the photometer's (DOS) computer clock using the 
GPS display. (We don't have any direct electronic method for setting 
the computer's clock.)   Tests soon showed that this was running some 
2s slow on the radio time signal coming through on 5 MHz.  The radio 
signal was very poor -- absent on other WWV and VNG frequencies -- so 
I couldn't properly hear the time announcements. However there was no 
doubt that the minute pip was about 2s early compared  with the GPS 
clock.  Does anyone have an explanation for this? It thus appears 
that the above times need to be corrected by -2 seconds.  More 
refined corrections are possible if I base the time entirely on the 
time signals, but I will leave them till later.

This is the first MP occultation I have recorded.  Pity about the 
time uncertainty.

[Return to Top]


Steve Preston, 11 May 2002:

Great work picking up the (803) Picka event.  You asked about the
discrepancy between the GPS display and WWV.  I would not trust the GPS
display to be accurate.  Unless the specifications on your GPS
explicitly state the display's accuracy, there is room for significant
error.  Most consumer grade GPS units (e.g. Garmin), can output the
current time on their RS-232 serial port.  When the GPS is setup to
output the time/position fix on its serial every second then you can
assume an accuracy of less than 1 second with this data.  However,
Garmin sells an OEM unit (less than $200), with no display, that
provides a 1pps output with an accuracy in the microsecond range in
addition to the serial port output.  One could feed this 1pps signal
into the PC parallel port and write a program that uses the serial port
data and the 1pps signal to set the DOS clock accurately.  Of course, PC
clocks are not known for accuracy or stability, but this would still be
more accurate that setting the clock by eye/ear.  You might also find it
interesting to talk with a fellow New Zealander with some knowledge of
accurate timing with GPS and DOS
[http://www.geocities.com/kiwi_36_nz/kiwi/kiwi.htm ].

Clear skies,
Steve
[Return to Top]


Alan Gilmore, 12 May 2002:

Steve

Many thanks for your most helpful comments.  I got the GPS unit built 
by the electronics people in Canterbury University's Dept of Physics 
& Astronomy.  They are very good on the electronics but apparently 
don't know all the caveats in GPS receivers and displays.  I have 
passed on your email to them for comment. 

 I was expecting the GPS display to be good to a fraction of a 
second. Apart from the occasional MP ocultation, I need such accuracy 
for routine CCD exposures of fast NEOs. There time errors can 
contribute more than position measuring errors to the final 
uncertainty. 

Thanks again.
Alan
[Return to Top]


Geoff Hitchcox, 13 May 2002:

I am not sure how the "custom made" project was implemented at Mt John,
but there may be a very easy "fix" for your problem, that will not give
"perfect" results but reduce the error to something approaching "Shortwave
Errors".

I assume the "custom" project uses the serial data stream from a commercial
GPS receiver.

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) then there are many (or all) sentences that can be
requested from the GPS. What "may" be happening at Mt John, is that the GPS
receiver has been requested to output either all or too many data
sentences.
When this happens, 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 your 2 second error of
the GPS.

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 either requesting a "single" time
sentence, and even better if the serial "speed" can be increased to the
maximum the GPS can use.

[Return to Top]


Simon Lyall, 8 July 2002:

Re: ntp based timing.

On Wed, 3 Jul 2002, Graham Blow wrote:

> The comment was based on my limited experience of using an internet time
> server, in which I found that the time I was getting was up to a couple
> of seconds late. Mind you, the server was in the US so one would expect
> there to be delays due to path differences, packet handling protocols,
> etc. I guess the problems would be less with servers closer by, but I'm
> not aware of these. Indeed, my knowledge in this area is quite limited.
> Where would you suggest I try?

You ISP actrix runs a couple of ntp servers ( ntp1.actrix.co.nz and
ntp2.actrix.co.nz ) and there are various others listed for New Zealand
(see http://www.eecis.udel.edu/~mills/ntp/clock2.htm entries 58 and
onwards).

Lists of software are at the following sites and (I think) windows also
has some built in software for it.

http://www.eecis.udel.edu/~ntp/software/win9x.html
http://tucows.ihug.co.nz/sync95.html

One think to make sure is that you are using the NTP protocol and not the
older rdate one, a good ntp client will keep talking to the remote servers
as long as you are online and get more accurate as it works out how much
your clock drifts and what the delays are to the remote servers.

[Return to Top]

[Top of Page][Return to Home Page]