(Nearly) All about the time inside your GPS
The time that
comes out of the NMEA port when the unit is locked on is always UTC (even
Garmin got this right otherwise they would have had problems interfacing with
other navigation equipment).
The time
displayed on the screen is UTC (with the user set time zone correction
in whole hours) when the unit is locked on (otherwise it drifts off slowly).
However this may not be the time recorded on the track log. The time
recorded on the track log may be GPS time (currently 13 seconds later).
The time recorded
in the track log should be UTC. However Garmin screwed up and on some of
their units the track log time stamps are in GPS time. It could be only the
Garmin units that have this error. GPS time has no business escaping from
inside the GPS onto the track log and it should never have been there.
GPS time is for
spacemen, Satellites, and the inside of GPSs. UTC is for us earthlings (no
matter how high the thermal goes). Because the earth doesn't take exactly 24
hrs to rotate us earthlings agreed to average our clocks so everywhere we go on
earth we can use UTC and have it exactly the same. However orbiting objects
need to make precise calculations affected by the earths rotation and therefore
use GPS time.
Big Brains based
in the USA keep the time used in orbit a whole number of seconds offset from
ground based UTC and decide only a month or so in advance when they will take a
second out or add a second on to the difference between UTC and GPS time. The
messages from the Satellite to the GPS contain this offset data so all the GPSs
on the ground can correct their internal GPS time to UTC.
Now the offset
between GPS time and UTC is quite a lot in terms of the separation of gliders
crossing a goal line in a hang gliding competition. Currently 13 seconds.
You can see an
illustration of the difference at:
http://www.leapsecond.com/java/gpsclock.htm
Note this
displays the relative differences based on your PC clock and not the true
reference times.
Nobody realised
that some GPS units have the wrong time on the track log when they are flying
at the start of the course even though they are carefully timing their exit to
the start circle or sector. This is because their track log is correct or 13
seconds (at the moment) later. So when they are timing their start to the 15
minute intervals they have to leave 13 seconds too early before they get
clocked by 15 minutes. Setting off a few seconds early doesn’t help much so
nobody realised this.
At the finish the
checking programs do some maths to try an pin point when the pilot crossed the
line by using an average speed between track log points, or projecting the
average speed toward the line (or goal circle). Finishing times didn’t seem to
be accurate but this was put down to the unpredictable glide path beyond the
line and the complicated sums in the checking program. Recently though we
realised something was wrong with the track logs.
Christian Quest
reports that Garmin are reluctant to come up with a list of which units are
affected.
The only mention
of the leap second correction on Garmin’s update page for the common units (12,
12XL, 12Map) is this:
“GPS12MAP
Changes made from version 2.02 to 2.03:
Latest version is 2.05
I know that one day in Australia (at
Denilequin using GPS goal circles) I found I could get a score about 13 seconds
better on my MLR GPS than with my GPS12XL running version 3.53. (this is an
early unit and this is the latest software it can handle).
I suspect that all the “12” series have
this problem except for the 12MAP which definitely doesn’t if running 2.03 or
later and all these 12MAP units can run 2.05.
One day at Denilequin I entered the goal
circle simultaneously with two pilots, at least one of whom was French and
probably using an MLR. I was recorded as arriving later than these two guys.
This could have been the 13 seconds.
Maybe Christian Quest or Ivan Twose can
come up with a way of reading the track log times when we load the units with
waypoints at the beginning of the comp. At the moment the only way I can think
of is by simulating a start or finish by running the comp checking program on a
number of GPSs that have been driven round a course.
Maybe I’ll do that if the weather stays
like it is now! I could cover 12XL 2.53, 4.57 and another version on a 12 and
compare that with the 12MAP on 2.05.
We could build up a database pretty
quickly probably!
In CAT2 and CAT1 internationals there is
often a sprint finish and there are a lot of points separating pilots who are
less than 13 seconds apart. Until we have a way of correcting for the leap
minutes we must use a goal marshal to correct the finishing order. We can use
the GPS times as a guide and at least we know that we should be able to get to
the right times by adding or subtracting 13 seconds!
If you have a GPS 12MAP make sure you are
running software 2.05 and you know you have the right time, (and make sure you don’t cut the start too
fine) (same for MLR users). You might like to be sporting and ask for a
correction if you see pilots who were ahead of you scored as slower.