[time-nuts] GPS Timing Antenna Failure - Long
lmcdavid at lmceng.com
Sat May 12 16:07:57 EDT 2018
I recently had an unexpected failure of a white-conical-dome
HP/Symmetricom 58532A GPS antenna that had been in-place about 5 feet
above the roof of my two-story home in Southern California for about ten
years. I have two similar GPS antennas located about ten feet apart on
this roof, one fed with about 50 feet of Andrews Heliax and the other
with LMR400; the other antenna continued to work ok. The antennas feed
4x and 8x amplified GPS Source (brand name) antenna splitters. I noticed
the failure when several GPSDO units and a GPS Clock failed to sync with
the GNSS. I confirmed the failure was not the antenna splitter and I
replaced the failed GPS antenna one of the same type, after which all
returned to normal.
I removed the conical radome from the failed antenna and was surprised
to find the antenna element was actually a patch, not the quadrafilar I
expected under that conical dome. Subsequently I opened the radomes of
three other similar GPS timing antennas made by various manufacturers
and found that all use patch antennas. I had believed these timing
antennas used a quadrafilar design to benefit from higher low-angle gain.
So, it appears the conical radome shape is really only to prevent snow
accumulation. Well... from my experience here on the flatlands of
Anaheim near Disneyland, that seems to be completely effective as I've
surely had no snow buildup! :) But, I had surely expected the conical
radome covered a quadrafilar antenna. Am I alone in expecting a
Further troubleshooting of this failed antenna revealed many discrete
components on the underside of the round board holding the patch
antenna. The circuit uses a three-stage gain amplifier with three Toko
bandpass filters, numerous bypass capacitors and stripline inductors.
Probing the circuit with a sig gen and spectrum analyzer showed that all
three gain stages were working about as expected. Of course, even with
26-30 dB gain in the antenna, the SA did not have enough gain nor low
enough noise floor to see any GPS signal from the antenna. But, each
gain stage seemed to be working ok. So, what was the failure?
Upon removing the radome, one unexpected thing was seen. The
construction uses a short coax cable up from the N connector, through a
hole in the circuit board, where it is bent over and finally soldered to
circuit board pads for the shield and center conductor. There was a
great deal of very dark flux residue around this coax solder connection.
The appearance was so bad it even looked like a cracked solder joint,
though that proved not to be the case when the flux residue was
thoroughly removed. It did not occur to me to functionally test the
antenna at this point. Later, it was necessary to unsolder this coax so
the board could be removed to access the components on the underside for
detailed testing. But, stage-by-stage RF gain testing did not reveal any
problems, so the antenna was reassembled for actual field testing.
The result? The antenna now works ok; locking sync to the GPS GNSS. I
gotta conclude the flux residue was attenuating the signal out of the
antenna. Careful inspection of that coax solder joint absolutely did not
show any problem after the flux was removed so I believe continuity was ok.
I next removed the radome from one of my (new) Symmetricom antennas to
inspect its coax solder joint and discovered this (perhaps newer)
version has a metal shield-can soldered over the coax solder pads; I am
loathe to remove that shield just to inspect the solder joint flux.
However, there is no flux evident on the solder tabs where the metal
shield-can is soldered to the circuit board so the whole thing must have
been defluxed after soldering. That would be a better process anyway.
To make this very long story into a short one, I learned that the
HP/Symmetricom 58532A GPS Reference (timing) antennas use a simple patch
antenna instead of a quadrafilar antenna and that old solder flux
residue will attenuate the even amplified GPS signal out of this antenna.
I welcome your constructive comments.
Larry McDavid W6FUB
Anaheim, California (SE of Los Angeles, near Disneyland)
More information about the time-nuts