[time-nuts] Difference in GPS antennas

[Lux, Jim (337C)]

On 8/14/09 7:47 AM, "Graham / KE9H" <TimeNut at austin.rr.com> wrote:

> John Green wrote:
>> Mark Sims wrote:The quality of the results match the quality of the antenna.
>>  A geodetic/survey grade antenna gives results to within a few inches.  A
>> conical timing antenna within 8 inches.  A cheap patch antenna around a
>> foot.
>> I took one of those conical timing antennas apart only to find that it has
>> the same ceramic patch as most cheap GPS antennas. How is a survey grade
>> antenna different? I would like to be able to know one when I see it.
>> 
>>  
> John:
> 
> The "survey grade" antenna usually incorporates "choke rings", which are
> ring structure surrounding the antenna, designed to suppress signals arriving
> 
> But when the antenna is mounted with a clear view of the sky, and
> nothing much above
> it for signals to bounce off of, then the survey antenna will (mostly)
> ignore signals arriving
> horizontally or from below the antenna, which are by definition, bounce
> or distorted.
>
Having recently discussed this with some folks at work in a couple other
contexts, I was surprised how low the multipath needs to be.  BUT, you don't
have to worry about multipath from things "far" away.  At this level, you're
doing carrier phase measurements, so multipath signals more than a chip away
(at P code rates of 10.23) don't really contribute to phase errors (although
they "raise the noise floor") because they're decorrelated.  That's on the
order of 100 ns or 30 meters path length.

Within that 1 chip time kind of span, though, if you want mm precision,
that's a phase shift on the order of 1 degree (e.g. Lambda is 20cm, 1 degree
is 1mm, roughly), which requires that the interfering signal be at least 35
dB down (=20log10(sin(1 degree)).  Keeping side and back lobes down to -40dB
is challenging, because you've got things like creeping evanescent waves
propagating over a surface and stuff like that.


The real problem with multipath is that if it's from a distant reflector
(but not so distant it decorrelates), the phase of the reflected signal is
always changing relative to the direct signal. It's like having a
interferometer with a long (in wavelengths) baseline, so as the spacecraft
moves across the sky, the apparent path length (e.g. The carrier phase your
receiver sees) keeps changing.  If you don't have the multipath signal, the
antennas themselves have very small changes in apparent phase center with
respect to look angle, and that's where fancy radome design comes in
(because it's a dielectric, and even though thin, it has an effect on the
propagation of the wave)