Batboard

Does anyone do any avl gps subsystem testing to verify navigation accuracy, sanity, and TTFF? I've had to put together a suite of tests, and some of them take quite a bit of time to complete. Anyone noticed a correlation between 802.11b/g activity and gps errors? Snap to position errors, or rubberbanding?

Ouch. So, nobody services gps in their shop? It is a niche market. Well, for your entertainment - this is what a plain old 2.4g consumer quality wifi access point can do to a gps board navigation accuracy. I am not a physicist. Can't explain the how or the why though I suspect it is repeating the L1 signal with a teensy bit of latentcy causing position errors.

Here is a different way of looking at the boards - line chart over time where you can correlate dop to position. You can also check whether two boards are calculating the same psuedo-range solution.

Neat softwares...we deal with AVL stuff...Airlink/Sierra Wireless & Bluetree mostly. We've never had to deal with your level of analyzing though...the darned things just plain work. Good to know that 802.11 & 2.4GHz wireless systems can cause issues though.

Both of those charts are roll your own using Windmill. Similar in concept to Wonderware. Multi i/o, multi interface, parsing engine, graphics engine, lib files, readme files, wonder files, worry files, pull your hair out files, sit up nights files, finally get a clue configurations, and then it all comes together kinda software. You can monitor industrial processes, fresh water stream turbidity, velocity versus fuel consumption of your favorite rocket, or sanity tests of suspected failed gps boards.

802.11b/g only seems to affect boards that do not employ raim, or who don't employ it well even though the OEM claims "it's in there". I've also found issues with faulty flash where the almanac is written. GIGO and even augmented garbage is still just garbage.

When differential goes bad (dramatic music) ...

Bill_G wrote:Ouch. So, nobody services gps in their shop? It is a niche market. Well, for your entertainment - this is what a plain old 2.4g consumer quality wifi access point can do to a gps board navigation accuracy. I am not a physicist. Can't explain the how or the why though I suspect it is repeating the L1 signal with a teensy bit of latentcy causing position errors.

Do what? I was inclined to believe you up until this point.

GPS L1 is at 1575MHz... for most devices to receive it, they employ a directional antenna with a 25-ish dB LNA. An 802.11b/g AP runs on 2.4GHz with omnidirectional antennas, not to mention being half-duplex. But it's repeating the L1 signal and interfering with the GPS? Nope.

AP interfering with a component on the GPS board? Interfering with the LNA? Little green spacemen? Maybe I can buy these, however unlikely they might be. An 802.11 AP rebroadcasting GPS? Umm, no.

In both images, you're showing one satellite at low signal quality which is unlocked in the "bad" image and locked in the "good" image (although the first "good" image shows one zig that looks like the "bad" image). I'd be far more inclined to believe that the GPS is attempting to use that poor-quality satellite which is inducing the error. RAIM/FDE would address this issue by ignoring that particular satellite assuming enough satellites were in view.

Ha! I totally understand your sentiment. It's not the first time I've heard it. I said it myself a long time ago. The good news is I have taken an origami class so I can make really cool hats out of aluminum foil. They're awesome!

Honest. In the presence of wifi, there are gps products that go insane. Even through a L1 pass filter. Turn off the ap, they normal up. Turn it back on, start passing traffic, the gps weirds out. The more traffic there is, the worse their pseudo-ranging is. Drive them through an area of numerous ap's (as identified by NetStumbler), and they will rubberband and/or wander. Continue driving out of the area and they will suddenly correct. Get out into the country and they keep up for mile after mile. Drive back into the city and they happily report a 3D fix with the wrong position. Scary.

While I cannot prove the source of the errors, or fix it, or even reliably recreate it, I can identify an individual failed gps, take it out of service, and replace it with one that is immune. That's the bottom line. That's what's important. In these days with more and more "stuff" being thrown on the air with no oversight, and no remedy, the best we can do is rub two brain cells together and hope for heat.

I dealt once with someone using a little Microsoft Streets & Trips USB GPS unit on his Toughbook in his car that one day, started to go sour. What was happening was that it would be always be about 100 to 200 feet off location to the left of where he really was and what concerned me is that when stopped, it was jumping all over the place up to like 2 miles away at random.

I just replaced it with a different USB GPS unit I had laying around for him, but I determined once I got the puppy open (which I had to use a dremel) that the SiRF GPS just flat out failed very slowly from some werid voltage regulation issue and didn't handle interference real easily.

I've had odd things happen with USB gps units on laptops too. The cursor would move wildly, windows would be selected and deselected randomly, and trying to type in screen was impossible. Turned out to be the laptops didn't like booting with an active input from the gps. You had to remove the gps cable, wait for the laptop to wake up, and then reinsert it.

I noticed something of interest pertaining to this thread the other day. While looking over the specs for some GPS antennas, I found one manufacturer (PC Tel) who listed certain models with the following blurb:

"...provides consistent, clear GPS signal reception while minimizing loss-of-lock in high-RF fields"

So, it would appear to be a known phenomenon.

Yes. We've run into that many times at busy sites where the customer equipment uses a gps disciplined master oscillator. Some manufacturers use an LNA antenna through coax to a receiver indoors in the equipment. Some place the receiver in the antenna and run serial data down a signal cable to the equipment. The latter seems to be the prevailing type since signal cable is cheaper. While they put a lot of effort into keeping stray energy induced on the line from reaching their equipment, the receiver itself is buck naked and subject to being beaten to death by everything else at the site. Placing the antennas as low on the tower as possible doesn't always work because they may become fully obscured by the surrounding forest. Putting them a little bit higher may put them close to a xmitter antenna that either kills them, or farbles them. We recently had to move one to the top of the tower at the 250ft receive level to make it work. It becomes especially difficult with cell sites where numerous towers are colocated on the same property.

I have recently run into GPS errors when this touchsmart lapotop has the wireless card active.

My position error goes up and my direction indicator(course) does not steady, but constantly rotates, attempting to display direction.

I shut off the wireless card, and the GPS receiver settled down and began reporting a far more normalized position.

My Delorme Street Atlas showed dithering when stopped, so the display showed me wandering my parking space while placing a green dot at every position
point the receiver downloaded.

So much for WAAS and DOP...my WiFi does that on its own.

As for the GPS receiver and WiFi interacting/interfering, I can see how, with GPS being weak signal and most HH receivers are not supplied with active antennas, a small, nearby OOB signal can affect reception of the L1 signal, but I doubt the L2 would be affected nearly as much.

I tried both my eTrex Vista HCx and my (Globalstar MR350) trunk mounted units, and both are affeccted, but the MR350 is affected more even though it is outside, on the rear trunk lid. But I can assume the control/data cable is the real culprit, whereas my eTrex is powered/controlled by a far shorter length of cable. I have to test these theories more to be certain though.

I only install GPS receivers in vehicles, and all to date are consumer items used only for positional reporting, with a few used to call home as the company spy.
so,no, we have not yet had a need to test any GPS device for strange anomolies to date, and I doubt we ever will.

I loved the idea of testing, I am always interested in such things, and thanks for writing about this as well.

You're welcome Earnest, and thank you. Our local bus transit system has over a thousand vehicles. I maintain the bus data system backbone, and assist their techs in the day to day maintenance of the buses with training and guidance. The gps is an integral subsystem for route and schedule adherence, as well as event recording. And we've had odd problems with the gps since the system went on line. At first we noticed many buses, but not all, would report off route in specific parts of town. We also got reports of impossible speeds, as well as numerous unscheduled stops. Outside of those areas the same buses would perform normally. So, we knew early on that it was not the operators. You can't be doing 70 on a busy street. You can't be driving through a grocery store. You can't be stopping anyplace you want near busy intersections. You can't keep having these odd things happening over and over again enough to plot them on a map to show clusters, and have it be the operator's poor driving, or bad gis data, or poor schedule programming. Things really blew up when a city wide public access wifi went in. So, I had to dev a way to test the boards that was a simple pass/fail so we could get the bad boards out of the system. Thankfully there are plenty of resources on the web to create your own charting programs.

GPS is a funny creature, easily misled when you do not have 3D coverage and are running under 2D, which no longer provides proper speed/position reporting, nor will it indicate properly when overhead obstructions increase, like tall buildings, elevated rail systems and of course my personal favorite; HV lines of the power company.

I noted a decrease in receiver sens. when my WiFi is active, and it gets worse as I move down the road at speeds of 30-45Mph.

I get direction errors, wide swings on the display in NAV mode, and once stopped, slow reporting once the vehicle is accellerating.

I've seen my car remain stopped at a light I took off from a minute ago, and reached city V2 speed shortly after, but my reported speed was 10.

Plus I was headed in the wrong direction....the map(Delorme) indicatedI was eastbound instead of north(compass).

I have had the GPS Rcvr. lose tracking when I was using 900 as well(902.4125), but this error was only resulting in a 3-6Mph error for brief periods while talking.

I am going to test the RF interference proiblem this week with a puck right next to my 900 gain antenna to see what the actual error is.

Once I note the amount of error introduced with standard RF, I am going to place the puck right on the laptop's WiFi antenna and see what the response is when both are in close proximity, and the WiFi actually sending/receving data.

I can barely look at a WiFi signal, so I have to 'gain up' on the receiver and pad it as well so I can see what's going on.

If the GPS signal drops by 6dB, I can understand the loss of precision in reporting correct data, but if close to negligible, there are other 'forces' at work that need to addressed as well.

Maybe the claim of -157dBm isn't all that accurate, given the fact that such a low level signal can easily be wiped out by nearfield RF of just about any type emitter, especially if pulsed in the microsecond range and nearby.

You can make a 1 watt active device pulse at 100 watts for 20us, so why not some other devices in the ISM band?

I love tests that create their own set of empirical data

You experienced rubberbanding Earnest. Kewl! Wait til you experience snap to position - the instantaneous change from one position to another. Usually only a few hundred feet, but I have seen thousands of feet making for fantastic reported speeds. I've also seen a parked bus wandering through the neighborhood on the CAD. The urban canyon has always been a trouble spot, but the system is augmented by dead reckoning. However, it requires the DOP to rise high enough to declare gps confidence bad before employing it. That happens easily between tall buildings downtown. But, when it's on an open street and there are plenty of birds in view, the gps can be totally insane, and perfectly happy with it.

One thing no city has that we have here; tall mountains about 7,000 feet and pretty deep canyons mixed in.

Good luck walking to a waypoint when your GPS is not seeing anything skyward, then all of a sudden, some birds come into view and hose your position fix, as well as altitude.

Too bad nobody can tell where the exact coordinates are when you are at the same location, but at different elevations.

I have had this happen on several occasions, where a single coorcinate could be obtained from two different altitudes, now mix that into trying to locate a cache, a hidden transmitter, or a body in the wilderness and you have a great mystery story about real life enigmas.

I love the duality of that statement: A mystery wrapped up in an enigma.

Tall buildings are one problem, as are mountains, canyons and forests...All in the same Geo terrain!

The next gen of GPS is supposed to be a vast imrovement over the current offering, so time will tell if the accuracy will increase, or will DOP be played on
even more.

WAAS has its problems, as does the differing versions of current WGS84 nav standards.
NGS is still offering conversions from WGS27.

Why can't we all play on the same sheet of music when it comes to civil standards for the parameters used?

I also prefer the degrees, minutes and seconds breakdown(ddd.mm.ss.ss).

But then again, I also use maidenhead for satellite work..

Hmm the "drunken wandering" of the signal on the left looks more like ye olde GPS selective availability being activated than anything else...

MattSR wrote:Hmm the "drunken wandering" of the signal on the left looks more like ye olde GPS selective availability being activated than anything else...

Interesting. I've wondered the same myself though it was 2009. For this test I use a common antenna through a splitter and an L1 filter with the assumption that an error displayed by one board would be seen in the other. Sometimes a common deviation shows up on the chart recorder. Most often they are seen around dawn and dusk. But they are hard to see in the 2D plot since the dots can overlap each other. I have swapped correlator clock xtals hoping that dithering was occurring because of xtal variations. But, I saw no difference in performance.

Here is a board that got turned in for "off route". I dunno. You be the judge.

This project, like so many others, is on and then off again, based on the progress of the techs. There is only so much mud you can throw on the wall hoping for a stain before individuals reach brain saturation. But, we made some steps forward.



In the above image from my original post, the plot on the left shows a snap to position error vectoring about 330 degrees. It snaps out almost 200m to the NW, paints a few points as it attempts to track back to center, then snaps south a few degrees, and finally marches home over many, many seconds back into the civilian 10m circle. The snap out indicates a loss of lock. In this particular dual plot, the two boards share an antenna and line through a splitter. The error is not duplicated in the board on the right. So, the physicality of the cable is not suspect. That is, the test bench antenna and line are probably okay. But, aboard a bus it is possible the line and/or antenna are intermittently opening. And, in fact we found some spread bnc female connectors on the equipment housings that could be hand wiggled to mimic this behaviour. We also found some poorly applied bnc male connectors on the coax with loose braid, and loose sma connectors at the antennas that did not take much tickling to fail intermittently.

Normal vehicle rattling could easily reproduce these physical failures. I could not duplicate a high rf field loss of lock condition on the rooftop by keying the mobile, or by keying a portable near the gps antenna. However, I suspect snap to position errors are indicators of an individual weak gps receiver. As with all things, workmanship plays a large role in reliability. Once those issues are removed, there are still equipment failures that need to be identified properly leaving us with a very small amount of unexplained behaviour.