Got a cool new speedometer - basically 100% accurate

[JackConrad]

Subject seems to be drifting from speedometer to GPS, so here is my GPS relative post. We run Delorme and MS Streets and Trips simultaneoulsy on our laptop (split screen S&T on left half of screen and Delorme on right half each usng their proprietary antenna). THe S&T altitude seems to always be inaccurate and considerably different readings on different occasions at the same location. Our drive way has registered as anywhere from -23' to 114' (our actual elevation is 18').  Speeds seem to both be very accurate (checked with portable radar speed displays used on the highways). Anyone have any ideas why the altimeter function does not work properly?  Jack

[belfert]

As far as GPS altitude goes, one friend of mine said that the government purposely doesn't allow accurate altitude readings on civilian GPS.  This is supposedly to prevent use as a guidance system for weapons.  I've never researched this and have no idea if true.

Sean, yes, the GPS speedometer I bought has an external antenna. 

[Sean]

... each usng their proprietary antenna). THe S&T altitude seems to always be inaccurate ...  Speeds seem to both be very accurate ... Anyone have any ideas why the altimeter function does not work properly?  Jack

Yes...

First off, understand that the little box you connect to the computer is not just an antenna, it's an entire GPS receiver.  Your computer itself can't possibly do the sophisticated signal processing required to receive GPS signals.  What the little box sends to the computer is a position report (latitude, longitude, altitude) at frequent intervals.

GPS receivers can have as few as four "channels" and as many as 12 (or more for military and high-precision units).  Unlike channels on your TV, or on a Ham radio or a police scanner, where you can receive only one channel at a time, the number of channels on a GPS receiver represents the number of different signals the unit can be receiving simultaneously.

GPS receivers locate themselves at a unique point in three-dimensional space using simple triangulation.  At any given point in time, the satellites are at known positions, broadcasting a continuous stream of clock pulses.  When the receiver sees a clock pulse, knowing the speed of radio energy, it can determine how far away that satellite is.  As always with triangulation, knowing your distance from one known point places you anywhere on a sphere equidistant from that point.  Two known points puts you anywhere on a circle defined by the intersection of two spheres.  And three known points will place you, usually, at either of two points along that circle, where it intersects with the third sphere.

This is why, in order to get a unique fix, you need to be receiving four satellites simultaneously.  The fourth satellite allows you to determine which of those two points is your actual position.  However, there is a "trick" that receivers use to establish a fix with only three satellites:  They use the sphere defined by the surface of the earth at a given altitude, and assume you must be somewhere on that sphere.  The receiver uses the last known altitude for this trick, and that works nearly perfectly for boats, and almost as well for land vehicles and commercial aircraft.  Things that change altitude rapidly, like fighter jets, space craft, and missiles, can't use this shortcut.

So the ability of any GPS to give you an accurate, continuously updated fix depends on its ability to always have a continuous signal stream from four satellites.  You'd think that could be done with only four channels, but it can't, for several reasons.  First, GPS satellites are constantly "rising" and "setting" above and below the horizon.  So even if you had a perfectly clear view to the horizon on all sides from your antenna, when any of the four satellites sets, you will need to search for, pick up, and synchronize with a different satellite that is still above the horizon.  This takes time, during which you will have lost part of your fix.

Secondly, we seldom have a good view to the horizon all the way around.  So, while there are usually about 12 satellites above the horizon (half the constellation), any given satellite may fade in and out as we pass in front of buildings, under trees, past mountains, etc..  This is particularly an issue if your receiver's antenna is sitting on your dashboard instead of on the roof -- the signal will go through glass, but not anything else like the coach's roof.  So your vehicle itself will generally block your view of several of those satellites.

For this reason, good receivers have 12 channels -- enough to at least try to acquire every bird above the horizon.  And, since the chipsets are now cheaply mass-produced, it's hard to imagine anyone making a unit without 12 channels nowadays.  However, many older receivers have only six, which was a common cost-cutting measure.

So, long buildup, but the reason for the difference in performance between your two receivers is either (1) the more accurate unit has a better antenna, with more "gain" and thus able to receive more marginal signals than the other receiver or (2) the less accurate receiver has fewer channels, and thus is having to switch birds constantly to maintain a fix.  And whenever any receiver drops below four input signals for any period of time, it will switch to a two-dimensional navigation mode as I described above, where it attempts to place you accurately in latitude an longitude (and give you correct speed-over-ground), and punts on the altitude.  When this happens, you will see inaccurate and sometimes rapidly changing altitude displays.

-Sean
http://ourodyssey.blogspot.com

[Sean]

... the government purposely doesn't allow accurate altitude readings on civilian GPS.  This is supposedly to prevent use as a guidance system for weapons.  ...

Well, sort of.  The system does belong to the military, and it is designed so that the accuracy of the system can be lowered on command.  In the event of missiles inbound to the US, the system can be de-tuned or even shut down in a matter of minutes.

There is a name for this:  it is called "Selective Availability".  And ten years ago, Selective Availability was in constant use, and civilian receivers could not get a fix better than about 30 meters or so (about 100 feet).  Military receivers are unaffected, because, while they use the same satellites, they use a different and highly encrypted time signal unavailable to civilian receivers.  As more and more of the world grew dependent on the civilian GPS system, including commercial aviation and shipping, pressure mounted to turn off Selective Availability, and I don't remember the date anymore (although I'm sure it can be Googled), but it was in fact turned off sometime during the Clinton administration.

So the rubric about civilian receivers not being accurate for military reasons is now out of date.  Today you can get a fix accurate to about ten meters with the GPS satellites alone, and even better than that, down to 2-3 meters, if you apply corrections to it which are now constantly broadcast from a pair of geosynchronous satellites operated by the FAA, known as WAAS (Wide Area Augmentation System).

In any case, even if Selective Availability was turned back on, the system has no way to control "altitude" versus any other part of the fix -- see my previous post.  Your receiver is doing simple triangulation to determine its position in three dimensions.  It's just that receivers built for navigation prioritize determining latitude and longitude over determining altitude, and so, when not enough data exists for a completely accurate 3-D fix, the altitude number is the one most likely to be wrong.

-Sean
http://ourodyssey.blogspot.com

[JackConrad]

Thanks Sean,
    Since I can't change my altitude in the bus, is not really that important, just informational. And come to think of it, the Delorme (the accurate elevation) GPS did say something about being WAAS compatible.  Jack

[Hartley]

I've used GPS and Loran in airplanes since day one and I found that there are many situations in which neither one worked well or sometimes not at all.

Between tall buildings, in mountain passes, inside anything including tunnels and sometimes just for no reason at all.

How did the airplane in the tunnel part work? I just gotta know?

[jeepme]

I gather from the battery part of your post that this is a portable unit.  Are you using an external antenna, or just the built-in?

Handheld Magellan, no external antenna. Tracks 12 satellites so adding an ext antenna would probably, according to your explanation of how they work, make signal loss an issue only in tunnels.

Jason Whitaker
4104

[captain ron]

THe S&T altitude seems to always be inaccurate and considerably different readings on different occasions at the same location. Our drive way has registered as anywhere from -23' to 114' (our actual elevation is 18').  Speeds seem to both be very accurate (checked with portable radar speed displays used on the highways). Anyone have any ideas why the altimeter function does not work properly?  Jack

According to my calculations You are at 18 ft ASL, Your antenna is located on top of your MCI, You have an 8 inch roof raise, So you should technically be reading 30 feet ASL.  HTH

[Tim Strommen]

I've ben watching this thread with some amusement for som etime now - all for a speedometer  

Sean's description is the closest I've seen to the reality of GPS to date.  Although, there was a missed point...  Specifically, around the problem of tunnels and parking garages and "urban canyons"...

GPS receiver companies have been trying to get auto-makers GPS recivers that were suitable for in-dash navigation in less-than-deal singal quality situations.  Enter dead-reckoning ("DR", via accelerometers).  Once the GPS receiver gets its fix on your trajectory, the DR sub-system begins assisting the actual locations by smoothing errors, and keeping track of changes in your course and speed when the birds are out of sight.

Accelerometers need occational updates to correct for minute errors in the tracking, and GPS receivers need short-term help keeping the heading info coming...  Happy marriage .

I've been in a few vehicles with the new DR+GPS hardware like this and they track very well even in tunnels, multi-story parking garages, and sky-scraper lined cities (like say San Francisco).

Any one can make a speedometer out of an off-the-shelf GPS with a serial output.  The standard "VTG" NMEA 0813 word (which can be set to be continually output from most any GPS receiver) gives Ground Speed and "Track Made Good".  This is easily converted to a simple digital display (what Nordskog has done) for MPH readings.  If the GPS digital dash speedo can keep giving you speed reading when you're in a tunnel, they used a $250 embedded GPS receiver (but if it loses the signal and goes to zero - you probably over-paid for that speedo - the trimble embedded DR+GPS development kit was quoted to me at $650, which included 3 receivers ).

Cheers!

-Tim

[Sean]

...  Although, there was a missed point...  Specifically, around the problem of tunnels and parking garages and "urban canyons"...

Actually, I did say "Without another input source, such as wheel sensors, it's just not possible for a system based solely on GPS to know [how fast you're going all the time]"

GPS receiver companies have been trying to get auto-makers GPS recivers that were suitable for in-dash navigation in less-than-deal singal quality situations.  Enter dead-reckoning ("DR", via accelerometers). ...

You say this like DR is a new discovery.  Come to think of it, several GPS manufacturers seem to think this too.  However, I can assure you that dead-reckoning in-dash navigation systems actually pre-date GPS by some time.

Back in the early 80's, I dealt with a company known as Carlin & Collins.  (Or maybe Karlin was spelled with a "K" -- it's been a long time.)  They had invested hundreds of engineer-years and millions of lines of code, as well as a good deal of hardware development, in exactly such a system, which worked pretty well, that they were trying to persuade the major automakers to adopt.  No accelerometers involved, back then -- the system used a magnetic compass, and sensors in wheels on both sides of the car.  The wheel sensors gave the distance/speed, but the differential between the two sides also gave turn information, which was checked against the compass.

In order for the whole thing to work, however, the system needed to know what the roads looked like, in order to match your movements against them (and correct your position to match).  Thus the company was extremely focussed on making hyper-accurate, machine-readable maps for every major city in the US (the initial target market).  The system looked and worked great, and was very impressive, and I had the sense of being in a James Bond movie (where the "technology" debuted in an Aston Martin, you may remember).  Somewhere in the middle of my dealings with them, they changed their name to what you may know them as today: Navigation Technologies (NAVTEQ), to better reflect what they were producing.

Of course, in the middle of all this, the release of civilian capability for the GPS came down the pike, wiping out virtually all of their investment in the DR measurement portion of the system nearly overnight.  (Ironically, in typical Silicon Valley style, Trimble Navigation opened up shop nearly literally across the street in Sunnyvale.)  Even with GPS, though, which in the early days of civilian receivers was hardly more than a digital display of position, speed, and heading, the demand for the machine-readable maps was greater than ever.  So NAVTEQ ditched their entire navigation system project, and re-focussed strictly on making the maps.  They are now one of the premier electronic cartography companies in the world.

It's taken twenty years for the major consumer GPS players to get back around to installing dead-reckoning inputs to their navigational displays.  Cheap accelerometers and other technology that avoids the problems of installing those pesky sensors out in dirty places like the wheels is one of the reasons -- it's finally becoming cost-effective to add some of the terrestrial-based information back into the system.

And now you know, as Paul Harvey would say, the rest of the story.

-Sean
http://ourodyssey.blogspot.com

[JackConrad]

According to my calculations You are at 18 ft ASL, Your antenna is located on top of your MCI, You have an 8 inch roof raise, So you should technically be reading 30 feet ASL.   HTH

Capn Ron,
   Actually that test was done with the both antennas laying on the drive way. And you are correct, with the WAAS enabled Delorme, raising the antenna to the top of the bus did give us a different reading.  Jack

[maria-n-skip]

The full story......

http://www.navcen.uscg.gov/pubs/gps/gpsuser/gpsuser.pdf

FWIW

Skip

[Tim Strommen]

..You say this like DR is a new discovery.  Come to think of it, several GPS manufacturers seem to think this too.  However, I can assure you that dead-reckoning in-dash navigation systems actually pre-date GPS by some time...

GPS receiver companies have been trying to get auto-makers GPS recivers that were suitable for in-dash navigation in less-than-ideal singal quality situations.  Enter dead-reckoning ("DR", via accelerometers). ...

The "revolution" is the micro-machined, failry accurate, cost-effective accelerometers that has recently become avilable, as well as years of experimentation in the proper use of such devices.  As you mentioned about Navtec, the very mechanically intricate and, hard to calibrate (requiring military/survey-grade mapping and a huge database) is not cost competative in comparison to a few $15 electronic parts out of harms way - and is subject to accuracy based on the reliability of a large group of individual components.  Auto makers are deliberatly trying to get away from this type of complex system, to improve product quality and reduce warranty liability.

Dead-Reckoning is simply knowing one's speed and direction over time and comparing it to a good map.  This "technology" has been in use for more than a century or more - infact it could be said that the first satellite navigation and dead-reckoning was done well before GPS electronic satellites were put in orbit.  One could (for the sake of argument) use a callendar, sextant, star chart, map, hourglass, and compass - along with your boat's speed in knots charted over time to figure out where you are (frequent updates on your position "plot" via sextant improve your positional accuracy like a GPS reading does - Navy guys will recognize this as the classic Quarter Master's job).  I had practice doing this while helping a friend's dad move his sailboat up the pacific coast a few summer's ago.

With GPS systems that have map "datums" (thanks to companies like Navtec), 12-channel Receivers looking at as many "stars" of a known position (through the use of GPS "almanacs") as they can, high-accuracy real-time clocks (updated by GPS time), and now accelerometers and digital compasses built-in, the dead-reckoning process is now mostly digital, self-contained, and automatic - and really accurate.

...Actually, I did say "Without another input source, such as wheel sensors, it's just not possible for a system based solely on GPS to know [how fast you're going all the time]"...

Sorry I missed that...  You are indeed correct.  Not trying to clarify for you necisarily, but rather for others .

...It's taken twenty years for the major consumer GPS players to get back around to installing dead-reckoning inputs to their navigational displays.  Cheap accelerometers and other technology that avoids the problems of installing those pesky sensors out in dirty places like the wheels is one of the reasons -- it's finally becoming cost-effective to add some of the terrestrial-based information back into the system...

This was the point I probably didn't make very well, but you definately filled in the blanks .


Cheers!

-Tim

[belfert]

My GPS speedometer works great since I plugged it back in.  I couldn't get it to work initially and left it unplugged.  I called the company and they said it needed to to be plugged in for 20 minutes the first time before it would acquire lock.

The only issue is losing lock twice due to the antenna turning sideways.  I need to make a mount for the antenna.