Conference auss::oz

    re .0
    
    Not that it helps ... in fact it's pretty irrelevant ... but did you
    know that GPS will break before the year 2000? I ASSuME the position
    service will continue to work but the time service will not work unless
    the receiver firmware is modified.
    
    P.S. The position service also gives your altitude.

    
    If I wanna know the time I'll wear me watch dude !!
    
    Strange as this may seem but most people buy GPS to know WHERE they are
    not WHEN they are.............
    
    BTW no mention of the year 2000 breakage in the GPS notes on the net,
    where can I view this info, thanks
    
    CRIM

RE: .1

>    Not that it helps ... in fact it's pretty irrelevant ... but did you
>    know that GPS will break before the year 2000? 

This is almost correct.

An event known as the GPS Week Rollover occurs at Midnight 21-22 Aug 1999 UTC.
The GPS 'week' is maintained in a 10 bit field in the almanac transmitted
by the satellites (and used to determine receiver absolute time). This allows
the time to be determined modulo 1024 weeks. GPS time started at 00:00:00 UTC
on Jan 6th 1980, and will wrap on the above date.

*Some* early GPS systems haven't taken this into account, so the time on their
GPS receivers will rocket back to 00:00:00 UTC 6 January 1980 at Midnight,
21-22 Aug 1999. Exactly which GPS receivers are affected is very difficult to
determine - the manufacturers are being rather quiet - although several have
stated that they will not have a problem, because they have taken it into
account in the software.


> I ASSuME the position
>    service will continue to work but the time service will not work unless
>    the receiver firmware is modified.
Yes - this is true. The satellite locations are determined in terms of
seconds into a week number - so the GPS receiver should still locate
the satellites and calculate a valid navigation solution.

The time service will still 'work', you'll just need to add exactly 
7168 days to the displayed time :-)



This topic is done to death every few months in sci.geo.satellite-nav



RE: .2

A suprisingly large number of GPS systems are sold as 'very accurate clocks',
they are able to generate a time reference signal with precision and accuracy
approaching atomic clocks, but without the cost and calibration problems.
Quite neat really.

re .2
    
>    If I wanna know the time I'll wear me watch dude !!
    
    Except that I am not completely convinced that an error in the time
    will not affect the position determination. The article cited below
    specifically allows this but they may have just been playing it safe.
    
>    BTW no mention of the year 2000 breakage in the GPS notes on the net,
>    where can I view this info, thanks
    
    See for example
    
    http://tycho.usno.navy.mil/gps_week.html
    
    which is pretty authoritative.
    
re .3
    
>The time service will still 'work', you'll just need to add exactly 
>7168 days to the displayed time :-)
    
    At least the day of week is still correct (in case one is seriously
    disoriented).

    It's hard to believe that such recently deployed technology is
    building in Y2K type problems but I think I read somewhere that the GPS
    standard borrowed some pre-existing NASA standard.
    
    Imagine if we (Digital) could get away with calling the Y2K problem
    "year rollover" and leaving it to the users ...

Derek,

After reading the web page you suggested I am now hoping Garmin have got some
s/w upgrade at  a reasonable price (ha HA) to correct for this problem.

I can do without the time but I do need accurate nav info coz I go we no man has
gone before ( at least on the weekends :^(  .....)

We get pretty bad weather off the Gold Coast with storms and what have you and I
use the GPS to get me home safely when I can't see the land.

CRIM

re .5

Garmin will upgrade your F/W for $$. AFAIK, the 12XL has FLASH technology,
others have PROMS.

The good news is that Garmin have stated that all their GPS units
will correctly account for the rollover (At least for this time,
maybe not the next one in 21 years???)

re: .4

This is not really an architectural problem with GPS. As long as your
receiver allows you to set the 'almanac base date' or 'current year'
somewhere, and work as an offset from that. I'd argue that the problem
is really an implementation deficiency.

In processing this field, I ASS-U-ME that the 'system clock'
on the host computer is approximately correct (give or take a few years)
and use this to determine which 1024 week period I am in.

Ar alternative heuristic would be to include a model that used
the difference between GPS time and UTC (which is transmitted by the
satellites as a leap-second offset). This has been monatomically
increasing since GPS started, and is currently 12 seconds. Looking at
this would allow you to make a guess at whether you are in the 1st,
2nd or 3rd count of 1024 weeks.

Also, GPS is not really 'recently deployed', the system was devised in the
mid 70's, when Y2K was still 'a long way off' :-)

> BLAH BLAH BLAH ... currently 12 seconds

Just realised that I'm off by one - the value is
11 seconds, at least until July 1st.

Ref: ftp://tycho.usno.navy.mil/pub/series/ser14.txt

    re .5
    
    I suggest *writing* to the vendor and ask when it will stop returning
    correct navigational info due to week rollover. Also ask the same
    question regarding time info.

    re .6
    
    Perhaps manufacturers of receivers could have included one or more DIP
    switches that set the cycle number. That should extend the lifetime
    enough e.g. beyond warranty, beyond documented lifetime.

DIP Switches disappeared in the dark ages -

If these thingies are handheld, they have a keyboard,
display and non voaltile memory - so you can enter the
data there.

If they are the serial interface/PCMCIA versions,
they have a nice programming interface to use.

    
    I'm buying the PC kit for downloading waypoints so this should be the
    same kit for upgrading the s/w ...? ! unless Garmin marketing can find
    a way of getting us to buy some MORE kit for our toys.....poor ol'
    cynical me, just put me hand back in me pocket for more money....
    
    CRIM

re: .10

I hope you enjoy giving away money.

Before you buy from Garmin, you might like to
consider the ample collection of shareware and
freeware at

ftp://sundae.triumf.ca/pub/peter/index.html

BTW: At least with the last release of this S/W
from Garmin, there was no way to upgrade the F/W

DM

    
    Thanks for the tip David on the URL, now all I have to do is make up a
    cable.
    
    CRIM

To make a cable, 'buy' a connector from Purple Computing.
I don't have the URL, but it should be easy to find.
This guy has a distributor in Australia, and the
connectors are quite reasonable, and good value for
money, because you pay what you think they are worth.

The serial interface should really be buffered, because
The Garmin's NMEA output and the PC's RS232 interface are
'not quite' equivalent.

I have heard of people connecting with wire, but
because I'm a hardware engineer and I care about these
things, I used a buffer to correct the voltage levels.
Of course, you are free not to - your mileage may vary,
your GPS may smoke, etc etc

Garmin are one of the more well-known GPS manufacturers,
especially at the low end of the market. They have many
S/W features compared to other units in the same price
range, and their receivers technology seems reasonable
(although there is some scope for improvement here)