On Sun, 20 Nov 2016 14:13:58 -0800
Hal Murray hmurray@megapathdsl.net wrote:

There are multiple issues. As already mentioned, SNR is only a part of the
picture. What you are looking for is an uniform gain pattern over most
of the hemisphere, with a sharp decrease at low elevations. Then the
left vs right polarization ratio should be as high as possible over
the whole hemisphere (most antennas only have good polarization ratio
at the zenith and behave like a linear polarized antenna at low elevations).

Additionally to this comes the phase center stability. Ie that the phase
center is a fixed location, independent of azimuth and elevation. And this
is probably the hardest to measure.

Absolute (and probably the most precise) measures of these properties are
done in an anechoic chambers with a rotating antenna mount.

The second way to do it, is to use a "known good" reference antenna and
use this as a comparison with a short (3-15m) baseline between reference
and antenna under test. For additional fancyness and to get better results
one can add the antenna onto robotic arm (like on the picture in [1]) and
get a more complete picture of the antenna. In this setup you want to have
an as fancy receiver as possible. At the minimum it needs to support carrier
phase data. The better receivers allow you to connect two antennas to the
same receiver and do a direct phase/amplitude comparison of the signals.

For the equipment hobbyists usually have, the phase center is not that
important. Most antennas have a variation <5mm. Even 10mm would lead to
just a ~33ps variation. Ie for the normal GPSDO that has a loop time constant
in the 100s to 1000s seconds and is using "normal" receivers, this is
completely drowned in the noise of the receiver's PPS output. Having good
sky view and as little multipath as possible is much more important.

			Attila Kinali

[1] https://www.ife.uni-hannover.de/aoa-dm-t_absolute.html

--
It is upon moral qualities that a society is ultimately founded. All
the prosperity and technological sophistication in the world is of no
use without that foundation.
-- Miss Matheson, The Diamond Age, Neil Stephenson

On 11/21/16 6:54 AM, Attila Kinali wrote:

I'm not sure about whether an anechoic (which is really "hypoechoic")
chamber is going to get you the data you need.  Calibrating the chamber
to the needed level of accuracy might be harder than doing field
measurements.

It might just be because there's a ton of analysis software out there,
but the folks who really, really care about 0.1 mm shifts in phase
center seem to use field data in a well characterized site, and
accumulate it for a number of days.

The GPS antenna folks at JPL, when they're testing a spacecraft antenna
for things like precision orbit determination (a basic choke ring sort
of thing) go out with the antenna and a test receiver on a cart in a
parking lot.

Looking at it in terms of numbers:
1mm is 1/150 wavelength, or about 2-3 degrees of phase.

sin(2 degrees) is 0.034, or -30dB.  So a spurious reflection that is 3
cm different path length (modulo wavelength) and 30 dB down will give
you a 1mm phase center error.  0.1 mm is -50dB.

Now, it's true that if you had a good spherical near field range, with
time gating, you can probably get rid of the reflections from the
chamber (and, in fact, you can do the measurements in a regular lab, or
your garage). But even there, it's tricky, because the probe calibration
has to be very good, and the structure supporting the scanning probe
also has to be accounted for.  You might be able to do it by doing
transmit/receive measurements on something like a spherical target of
appropriate size.

I've done measurements on what was essentially an interferometer with a
2 meter baseline, in a conventional chamber on a conventional pedestal
(JPL Mesa 60 ft chamber  http://mesa.jpl.nasa.gov/60_Foot_Chamber/).
You could easily see -40dB specular reflections as the array rotated.
(and you could also see things like the ladder on the positioner behind
the antenna we accidentally left in there, even though it was behind the
horn antennas in the array)

I think a good test using satellites and a very well characterized
comparison antenna in a open air test site is probably the easiest, and
most accurate, way to do it.
Arranging your test on a post well above the terrain, and making sure
that the surface is flat is easy.

Hi

If “sum the S/N” gives you a difference you should immediately ask why.

Normal receiving antennas are a gain = directivity sort of beast. There is
not a lot you can do about that. For a GPS antenna, you want to be able to
receive over a hemisphere. You don’t really know in advance where the antenna
will be used, so that’s how it’s done.

Early on the designs had more gain straight up than at the horizon. That’s a
bad thing. If anything, you want more gain at the horizon. Signals are strong
from straight overhead (short path, less atmosphere) and weak(er) at the
horizon. They could easily give you a better “sum the S/N” number while
actually performing worse in a location with sat’s mostly overhead. A “straight up”
antenna might be wonderful at a location on the equator. It’s probably a disaster
at a location on the arctic circle.

The real answer for signal to noise will always be location dependent. If I’m in
an urban canyon the only “sky” may be straight up. If I have a lot of terrestrial
broadband noise close to the horizon, again straight up might be the answer.
If my antenna is on top of a pole and I have a clean view 360 degrees around and
down to < 5 degrees elevation, a straight up antenna is very much what I do
not want.

Even more complex: If I have a bunch of transmitters at a wide range of frequencies running at the
same site as the GPS, I may want really good filtering ahead of the preamp. Those
filters likely will have a temperature sensitivity.The filters create  loss ahead
of the preamp so the noise figure (and thus S/N) take a hit. I get something I desperately need
and trade it off against degraded performance in other areas.

Lots of variables.

Bob

I agree. And besides, for those of us here in Oregon/Washington, the very ground is moving northwest at several inches per year (plate tectonics).

http://leapsecond.com/pages/quake/

/tvb

When I was doing VHF and UHF direction finding antenna design, I would
drive out to the highest readily accessible hilltop for testing.  Once
I came up with a low sidelobe design, I started picking up things like
lamp posts, trees, and bushes in the parking lot, aircraft over LAX
and John Wayne airports 50+ miles away, etc. which limited testing
performance.

While a perfect test environment is handy for design, a GPS antenna is
going to be subject to all kinds of environmental limitations so I
would accept field testing which includes considerations like
multipath, temperature variation, and a generally hostile RF
environment.

On Mon, 21 Nov 2016 08:22:50 -0800, you wrote:

On 11/21/16 8:38 AM, Bob Camp wrote:

the satellites have a pattern that is "edge weighted" so that the
received power on the ground is roughly constant into an isotropic
antenna - that is, extra gain when the satellite is on the horizon is
built into the transmit side of the link.

http://www.lockheedmartin.com/content/dam/lockheed/data/space/photo/gps/gpspubs/GPS%20Block%20IIR%20and%20IIR-M%20Antenna%20Panel%20Pattern,%20Marquis,%20Feb2014%20-%20publically%20releasable%20data.pdf

actually shows that the "peak" of the pattern is at around 45 degrees
elevation (from the user)

And slide 32 shows that the variation (on the newer satellites) is on
the order of 1-2 dB.

Signals are strong

That might be. Also, there's a difference between "optimum combining for
timing" vs "optimum timing for position/velocity".  I don't know that
the geometry makes as much difference for timing, but it sure makes a
difference for position.  That might actually be a rationale for higher
transmit antenna gain at 45 degree user elevation: those are the
satellites you want for a good fix: the one overhead isn't as useful.

A lot of GPS design decisions were based on use cases and requirements
from 30 or more years ago: where were users likely to be, how
sophisticated receivers were (e.g. a "pick the 4 strongest signals and
hope they give good DOP geometry), what kind of antennas were easy (quad
helix)

This is very much the case.. we had a high performance survey quality
antenna/preamp that in one location was unusable because of interference
from close by signals (cellphones, I think).

The receiver had very little front end filtering, presumably for the
stability reasons you mention above.

In message CALg-KtM9NsHTt0RGmB3rxbsxM=hCXY7WGCtgO=xUctVbCXxN=g@mail.gmail.com, Scott Stobbe writes:

ps ?

No way - ever

ns ?

Probably, but it depends a lot on the exact materials and manufacturing.

--
Poul-Henning Kamp      | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG        | TCP/IP since RFC 956
FreeBSD committer      | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.

On 11/21/16 6:38 AM, Scott Stobbe wrote:

Figure it's copper, so 16 ppm/deg C.  velocity factor is about 2/3, so
30 ft is about 45 nanoseconds.  about 1ps/degree

Really, you also need to look at the effect on the propagation velocity
of the radial expansion of the coax, too, and the change in epsilon.

You can look up a "phase vs frequency" curve for most coax which factors
all of this in.

Several years ago I measured the delay of about 80 feet of LMR400
feeding a GPS antenna, much of which was lying on a black shingle roof
in the Georgia sun.  I checked in early afternoon when the sun was
beating, and in the wee hours of the morning, to get the greatest
temperature delta.  My recollection is that the tempco was surprisingly
small -- maybe a couple of nanoseconds.  It was much less than the other
elements of the GPS timing error budget.

I can't find the data right now, but will keep digging.  There's also a
short paper from the early 2000s from Haystack on their measurement of
LMR400 in an environmental chamber.  They came to the same conclusion,
but I can't find that paper either. :-(

I don't know how much different RG58 results would be.

John

On 11/21/2016 09:38 AM, Scott Stobbe wrote:

When I first took a look at some of the coax datasheets I couldn't find
anything. I was able to find the following paper "phase stability of
typical navy radio frequency coaxial cables"
http://www.dtic.mil/dtic/tr/fulltext/u2/628682.pdf I attached the table
from the last page. They estimate RG59 to have a tempCo of -330 PPM/degC
for electrical length. They also estimated RG-58 at -480 PPM/degC.

On Mon, Nov 21, 2016 at 2:44 PM, John Ackermann N8UR jra@febo.com wrote:

John, many thanks for the Haystack tip! That is a wonderful paper, I
believe the one you are quoting is "Dispersion and temperature effects in
coax cables" http://www.haystack.mit.edu/tech/vlbi/mark5/mark5_memos/067.pdf