[time-nuts] u-blox NEO-M8T GPS initial tracking test

David Witten wittend at wwrinc.com
Thu Feb 9 06:12:25 EST 2017


I am doing similar tests, seeing similar results.  I have 5 x NEO-M8T's, 3
M6T', 4 Skytrack Venus 838Lp-T's.  Mostly have Toaglas triple band antennas
on PCB groundplanes with LNA-SAW-LNA conditioning in line.  I have built a
three arm test rig (arm < =1.5 M) for testing RTK measurments, etc.  Hope
to deploy it at local USGS office RTK calibration/test range.

So far the performance of the M8T's seems impressive.  Usually 29 sats in
view from basement windowsill (inside) with woods 15 M away.  Facing south.

Dave, KD0EAG

Message: 5
> Date: Wed, 8 Feb 2017 21:11:51 -0500
> From: MLewis <mlewis000 at rogers.com>
> To: Discussion of precise time and frequency measurement
>         <time-nuts at febo.com>
> Subject: [time-nuts] u-blox NEO-M8T GPS initial tracking test
> Message-ID: <589BCFE7.4010704 at rogers.com>
> Content-Type: text/plain; charset="utf-8"; Format="flowed"
> Anticipating issues with multipath, signal levels and a limited skyview,
> along with discovering that the tallest building in my skyview with a
> military computer centre has various stealth antennas around its roof, I
> went with:
> - u-blox NEO-M8T, mainly for its sensitivity, multi-GNSS (GPS, GLO & GAL
> (and Beidou)), and multipath handling, and
> - Tallysman TW4722 "Accutenna" wideband (GPS/GLO/Bei/GAL) active antenna
> with LNA-SAW-LNA filter, using a dual-feed patch element, for 'enhanced
> multipath rejection' & 'excellent out of band signal rejection'.
> Powered up for the first time, my u-blox NEO-M8T connected through USB
> to the u-blox u-center V8.24. It reported two satellites, with the
> antenna sitting inside a concrete/metal/brick building, behind two panes
> of glass, closed metal blinds, two feed below grade. So before
> installing it outside, I figured lets see what it does on the window
> sill behind glass.
> Setup: facing the SE, only ~45% of a full skyview with a bank of
> buildings opposite and parked cars between, antenna on 2" of wood with
> the 100 mm circular stainless ground plane almost touching the inside
> pane of glass, approximately 18" above grade, backed by the closed metal
> blinds.
> With GPS, GLO, SBAS & WAAS enabled, it picked up twenty satellites, most
> with a very usable signal. But around 2/3 were north of my building, not
> line-of-sight (LOS) in my skyview. So as expected, there's a lot of
> multipath (MP) going on. Reported location was initially within an 2 m x
> 8 m oval, but as the various satellites travelled and their multipath
> changed, the footprint expanded to 20 m x 120 m, then a 200 m trip to
> the south and back, before roaming a 4 m by 20 m oval. Altitude started
> +/- 5 m, but then ranged over 80 m. Apparently doable for upper micro
> second accuracy, but not for nano second accuracy.
> With the M8T set to test for the antenna connection of open circuit,
> this removed power from the active antenna. Of course all of the signal
> levels dropped dramatically. To my surprise, I observed that the signal
> levels from both the LOS satellites and non-LOS satellites were not only
> low, but very similar in amplitude, with only a few even lower. With the
> amp back on, I observed that the LOS satellite signals and the MP
> satellite signals were distinctly separate in amplitude. Now I
> understood what a multipath-filtering active antenna would do for me! An
> arbitrary 30db signal level threshold set in the M8T promptly had it
> ignoring most of the satellites not LOS in my skyview.
> The u-center SkyView showed those remaining were some satellites at the
> horizon to my NW, with u-center WorldView placing them over the Bearing
> Strait down through the Aleutian chain. Their signals had to be coming
> in over the building and reflecting back at the window, most likely
> straight back off of the tall building to the SE. I set the M8T's
> elevation exclusion at 15 degrees and that filtered those sources out.
> Fifteen degrees also excludes LOS satellites while at a low elevation,
> thereby ignoring signals prone to low-angle multipath off the buildings,
> parked cars & ground. As seen on the u-center Skyview: closer/higher
> satellites on that NW heading didn't get that same reflection; signals
> on the equivalent heading on the other side of north couldn't reflect
> off the park to the SW. The elevation exclusion may also be of benefit
> as it excludes those satellites low to the horizon, hence with the
> longest signal paths through the atmosphere?
> The above settings combined with the TW4722 antenna leave the M8T with
> only LOS signals from within the actual skyview to seek its timing
> solutions from.
> I did a quick test with the glass removed. LOS satellites gained 3 to 8
> db. MP satellites dropped 5 to 8 db. In any event, it wasn't enough for
> a meaningfully change. It just meant a satellite coming into range did
> so a few seconds sooner, or dropped out a few seconds sooner, with no
> material change in coverage. And not relevant, as I've got a fixed
> survey and a minimum of three satellites for timing solutions.
> I was going to be doing a stealth antenna install outside, but given how
> well it's done, I leaving it inside behind glass:
> Good:
> I've got a survey dialed in within 1/3 metre and fixed.
> Between GPS & GLO, I get between three to eight LOS satellites providing
> their timing solutions to the M8T. And that is before adding in GAL
> coverage.
> This setup with the M8T & TW4722 also receives a SBAS sat and/or a WAAS
> sat, usually both, along with an EGNOS available.
> The u-center tools are largely intuitive, the visuals meaningful & useful.
> The antenna is secure indoors, and I won't have to brush the snow off
> the antenna!
> Bad:
> I don't get to play with RF absorbing foam to custom shield the low- and
> high-angle multipath off of the surrounding buildings, parked cars &
> ground.
> Next:
> A com cable to the interface on the NEO-M8T breakout board.
> A dedicated NTP server (well, a PC).
> Learning a setup for Lady Heather, using the USB to COM driver, just
> because.
>
> Michael


More information about the time-nuts mailing list