On Aug 31, 11:12 am, VVX wrote:
On 30 Aug, 22:18, Silvar Beitel wrote:



My brother the civil engineer asked me (the supposed family GPS
expert) about the following scenario (to help in developing initial
site plans without immediately having to go to expensive centimeter-
accurate surveying GPSs and/or bringing in and paying real surveyors,
which would come later on in the process):

Take a pair of cheap identical datalogging GPSs. Park one at a
reference point and let it sit there logging. Send cheapo employee
around with the other unit to the points of interest in the site.
Gather the two and dump out their logs. Match the timestamps, check
that both units used the same satellite set at the same times
(probable over a limited area, like a large construction site?), and
calculate all the relative offsets between the points of interest and
the reference. How accurate would those offsets be? Especially
altitudes (important for things like sewer lines)?

I'm guessing pretty accurate, like down to the resolution of the
units. Say 6 inches for a GPS that reported minutes with 5 digits to
the right of the decimal point.

Thoughts?

--
Silvar Beitel

Actually, I did a zero-baseline experiment (two units connected to one
antenna) as a side-product of other (carrier-phase related)
experiments. Those two units were identical receivers using the same
firmware version.
Indicated altitudes diverged by up to 5 meters, horizontally the
difference was around 3 meters approx.

Interesting. Were the units using the same satellite constellation at
the same time? Were the divergences ever zero (or close to it?) for
any length of time?

I have no idea what caused that, as obviously there was no difference
in multipath conditions.

Just as a datapoint, what GPS devices were you using?

--
Silvar Beitel

On 31 Aug, 23:29, Silvar Beitel wrote:
On Aug 31, 11:12 am, VVX wrote:



On 30 Aug, 22:18, Silvar Beitel wrote:

My brother the civil engineer asked me (the supposed family GPS
expert) about the following scenario (to help in developing initial
site plans without immediately having to go to expensive centimeter-
accurate surveying GPSs and/or bringing in and paying real surveyors,
which would come later on in the process):

Take a pair of cheap identical datalogging GPSs. *Park one at a
reference point and let it sit there logging. *Send cheapo employee
around with the other unit to the points of interest in the site.
Gather the two and dump out their logs. *Match the timestamps, check
that both units used the same satellite set at the same times
(probable over a limited area, like a large construction site?), and
calculate all the relative offsets between the points of interest and
the reference. *How accurate would those offsets be? *Especially
altitudes (important for things like sewer lines)?

I'm guessing pretty accurate, like down to the resolution of the
units. *Say 6 inches for a GPS that reported minutes with 5 digits to
the right of the decimal point.

Thoughts?

--
Silvar Beitel

Actually, I did a zero-baseline experiment (two units connected to one
antenna) as a side-product of other (carrier-phase related)
experiments. Those two units were identical receivers using the same
firmware version.
Indicated altitudes diverged by up to 5 meters, horizontally the
difference was around 3 meters approx.

Interesting. *Were the units using the same satellite constellation at
the same time? *Were the divergences ever zero (or close to it?) for
any length of time?

Yes, the constellations were identical (which is quite natural for
zero-baseline setups) if remember correctly.
There were no obvious divergence/convergence (timespan - 15 minutes or
so)

I have no idea what caused that, as obviously there was no difference
in multipath conditions.

Just as a datapoint, what GPS devices were you using?


u-blox LEA-4T OEM receivers

On Sun, 31 Aug 2008 10:29:26 -0700 (PDT), Silvar Beitel
wrote:

On Aug 31, 11:12 am, VVX wrote:
On 30 Aug, 22:18, Silvar Beitel wrote:



My brother the civil engineer asked me (the supposed family GPS
expert) about the following scenario (to help in developing initial
site plans without immediately having to go to expensive centimeter-
accurate surveying GPSs and/or bringing in and paying real surveyors,
which would come later on in the process):

Take a pair of cheap identical datalogging GPSs. Park one at a
reference point and let it sit there logging. Send cheapo employee
around with the other unit to the points of interest in the site.
Gather the two and dump out their logs. Match the timestamps, check
that both units used the same satellite set at the same times
(probable over a limited area, like a large construction site?), and
calculate all the relative offsets between the points of interest and
the reference. How accurate would those offsets be? Especially
altitudes (important for things like sewer lines)?

I'm guessing pretty accurate, like down to the resolution of the
units. Say 6 inches for a GPS that reported minutes with 5 digits to
the right of the decimal point.

Thoughts?

--
Silvar Beitel

Actually, I did a zero-baseline experiment (two units connected to one
antenna) as a side-product of other (carrier-phase related)
experiments. Those two units were identical receivers using the same
firmware version.
Indicated altitudes diverged by up to 5 meters, horizontally the
difference was around 3 meters approx.

Sam Storm van Leeuwen has done some interesting experiments, worth a look, see:

http://home-2.worldonline.nl/~samsvl/TIM-LP.htm

Lots of interesting stuff, start:

http://home-2.worldonline.nl/~samsvl/index.htm

Interesting. Were the units using the same satellite constellation at
the same time? Were the divergences ever zero (or close to it?) for
any length of time?

I have no idea what caused that, as obviously there was no difference
in multipath conditions.

Just as a datapoint, what GPS devices were you using?

On Sun, 31 Aug 2008 09:40:44 -0500, Happy Trails wrote:

I'm am having fun with it - your sense of humour is obviously
different from mine.

I'll just picture a sort of Dick Cheney sneer then, LOL.
--
Mike Russell - http://www.curvemeister.com

On Sun, 31 Aug 2008 20:08:47 GMT, Sam Wormley
wrote:

Identical satellite configuration is not likely.... The old
Rockwell boards could be programmed to use satellites A, B, C and D,
but can you guarantee your two receivers are using the same combinations?
Probably not.


One of the satellite selection modes of the old Motorola Oncore oem
boards was "highest in the sky". Using this mode on the original 6
channel boards, I always got the same satellites used by both
receivers (5 to 10 miles apart), and got excellent dgps results, but
that was real dgps - precisely time-stamped range & range
rate-of-change corrections - not some half-assed delta xy(and maybe z)
arrangement like this guy is talking about.

Sometimes one or the other of the two receivers would switch sats a
few seconds sooner than the other - but only a few seconds. When
running magnetometer or sidescan lines 20 or 100 meters apart, it
didn't screw things up at all.

I did set up 2 receivers in a non-differential mode at either end of
my VW microbus once, and got a pretty good gps compass if I averaged
the readings over short times, but I also had good electronic
compasses at the time, so did not want to dedicate the use of an
onboard PC for this.

"Sam Wormley" wrote in message
news:h_Cuk.322295$yE1.122612@attbi_s21...
VVX wrote:
On 30 Aug, 22:18, Silvar Beitel wrote:
My brother the civil engineer asked me (the supposed family GPS
expert) about the following scenario (to help in developing initial
site plans without immediately having to go to expensive centimeter-
accurate surveying GPSs and/or bringing in and paying real surveyors,
which would come later on in the process):

Take a pair of cheap identical datalogging GPSs. Park one at a
reference point and let it sit there logging. Send cheapo employee
around with the other unit to the points of interest in the site.
Gather the two and dump out their logs. Match the timestamps, check
that both units used the same satellite set at the same times
(probable over a limited area, like a large construction site?), and
calculate all the relative offsets between the points of interest and
the reference. How accurate would those offsets be? Especially
altitudes (important for things like sewer lines)?

I'm guessing pretty accurate, like down to the resolution of the
units. Say 6 inches for a GPS that reported minutes with 5 digits to
the right of the decimal point.

Thoughts?

--
Silvar Beitel

Actually, I did a zero-baseline experiment (two units connected to one
antenna) as a side-product of other (carrier-phase related)
experiments. Those two units were identical receivers using the same
firmware version.
Indicated altitudes diverged by up to 5 meters, horizontally the
difference was around 3 meters approx.

I have no idea what caused that, as obviously there was no difference
in multipath conditions.

VVX


I watch many receiver combinations (side-by-side) over the years
report with in specifications, but giving alarmingly different
positions.

You want the same answer -- one needs to sync the PVT timing,
ensure identical compliments of satellites... eliminate interference
between the two receivers... and, yes, the multipath can be different.
Spatial separation increases differences in error sources.


Even if you could eliminate the items mentioned above by Sam , the receiver
noise in consumer gps units might be enough to give the deviations seen by
VVX.

--
Tom
http://home.att.net/~tbharvey/

On Sat, 30 Aug 2008 09:18:28 -0700 (PDT), Silvar Beitel
wrote:

My brother the civil engineer asked me (the supposed family GPS
expert) about the following scenario (to help in developing initial
site plans without immediately having to go to expensive centimeter-
accurate surveying GPSs and/or bringing in and paying real surveyors,
which would come later on in the process):

Take a pair of cheap identical datalogging GPSs. Park one at a
reference point and let it sit there logging. Send cheapo employee
around with the other unit to the points of interest in the site.
Gather the two and dump out their logs. Match the timestamps, check
that both units used the same satellite set at the same times
(probable over a limited area, like a large construction site?), and
calculate all the relative offsets between the points of interest and
the reference. How accurate would those offsets be? Especially
altitudes (important for things like sewer lines)?

I'm guessing pretty accurate, like down to the resolution of the
units. Say 6 inches for a GPS that reported minutes with 5 digits to
the right of the decimal point.

Thoughts?

Sounds like a wast of time, you should get an El Cheapo post-procesing
solution like Delorme's PostPro/Xmap and you're sure to get sub-metric
accuracy.

On Aug 31, 4:08 pm, Sam Wormley wrote:

Identical satellite configuration is not likely.... The old
Rockwell boards could be programmed to use satellites A, B, C and D,
but can you guarantee your two receivers are using the same combinations?

Not with consumer-grade sensors, of course. But the idea is to
compare logs and do deltas only on those points that report using the
same satellite set.

Probably not.

I don't know. That was the basis of this comment:

Then there's this problem: What is the probability of having matching
configurations at all POIs? Obviously, any POI measurement when the
sats don't match would be particularly suspect. But I suspect that
over an area the size of a construction site (even something large
like an airport), the probability of having matching sets of
satellites would be high, especially if you left the "slave" unit at
each POI for a few minutes or longer.

Also, "benchmark" and POIs being in areas that differ substantially in
signal blocking characteristics (benchmark on flat hill, POI in grove
of trees) might make getting data with identical satellite sets
impossible.

--
Silvar Beitel

Sorry. I didn't finish responding to Sam's post.

On Aug 31, 4:16 pm, Sam Wormley wrote:
I watch many receiver combinations (side-by-side) over the years
report with in specifications, but giving alarmingly different
positions.

You want the same answer -- one needs to sync the PVT timing,
ensure identical compliments of satellites... eliminate interference
between the two receivers... and, yes, the multipath can be different.
Spatial separation increases differences in error sources.

I suspect that, even with identical units and identical satellite
sets, their guts aren't capturing and calculating at the same time
(multi-threaded simple RTOS), very likely introducing the kinds of
offsets being reported here, even if they are putatively reporting for
the same instant. Another strike against this scheme.

--
Silvar Beitel

On Aug 31, 4:17 pm, Scotty wrote:

Sam Storm van Leeuwen has done some interesting experiments, worth a look, see:

http://home-2.worldonline.nl/~samsvl/TIM-LP.htm

Lots of interesting stuff, start:

http://home-2.worldonline.nl/~samsvl/index.htm

Very interesting! Thanks for the pointer. (And to Sam Storm van
Leeuwen for the work.)

--
Silvar Beitel