GPS Technical Discussion

[BVCBR]

I know that there have been a few threads discussing various GPS devices, but I thought I'd like to start another one that discusses more of the technical details and specifications of some of the units that are being used.

There was a mention of using GPS for speed certifications in the North American Eagle 2012 plans thread and instead of hijacking their thread for this discussion, I would like to move it here.

Here is the last post regarding the GPS from that thread;

I'll have to get the model number for you but the manufacturer is Topcon. We designed our network so we transmit from the nose of
the vehicle to three antenna's located 2.3 miles apart. The wireless ethernet changes nodes as we travel the distance of the course.
The Topcon system gets it's location triangulated from as many as 14 to 19 satellites. The data that is transmitted via the wireless
ethernet is captured to a mobile ground station and is simultaniously captured on the onboard racecar computer as a backup. We
actually have to pick two points in space along the track and extapolate that data to come up with the actual speed measurement.
It is very accurate and will pull 20 samples per second with an accuracy of .020 of an inch. 

Here is my follow-up question;

With an update rate of 20 samples per second at 200mph the vehicle will travel 13.968... feet. How does the Topcon GPS keep the .020" accuracy in position in that 13+ feet? Does it sample from the ground station and if so, do you know what the sample rate is at or does it have something like on-board accelerometers that extrapolate a position based on the last satellite readings?

I have been looking at GPS technology for a few years and haven't found the right one for a project I've been working on.

[TD]

I won't speak for F104A but I think there are several topics here

* Measurement of position
* Transmission of position (and other operating parameters) from the moving vehicle to fixed receivers
* Calculation of speed

Having some background in this area I think the mention of "Ethernet" indicates the AE team are using a wireless LAN to transmit data acquired from the vehicle back to the pits.  Remarkable that WiFi works at those ranges and speeds.  Of course other forms of radio telemetry have been around motor racing for a long time.

Measurement of position could be a rapid sampling of the GPS-derived coordinates and the time at which each position was captured.  GPS is really good at providing precision time.

Choosing two {position, time} values 1 km and/or 1 mi apart would allow you to calculate average "kilo" and "mile" speeds.  An advantage here would be that you, rather than the timing or sanctioning body, would get to choose where the speed traps start  That is not to suggest the AE team are doing it, only that it is possible.

Interesting paper on high-precision fixed positioning surveys using GPS is here http://www.gpstk.org/pub/Documentation/GPSTkPublications/precision-gps-hhstt-ion-gnss-2006.pdf

These guys go really fast.

I should add that I don't know how an onboard GPS could maintain 0.020 accuracy while moving, at least not without internal (inertial nav) or external augmentation.  The Topcon web site describes millimeter-accuracy positioning capabilities for some of their products (intended for use in paving applications), but it isn't immediately obvious how that's accomplished.  Again, not doubting what the AE team are doing in any way, just saying I don't understand how it works yet.

[TD]

Well, read and learn... The Topcon product described here http://www.topconpositioning.com/products/gps/receivers/hiper-ii has a specification sheet that suggests it can report "fast static" position information at 20 samples / sec with O(10mm) accuracy.

I don't know how it works, but it might be using a combination of GPS and GLONASS (Russian GNSS).

[F104A]

Our data aquisition engineer, Steve Wallace, is better suited to give finite details of how the GNS system works. I do know that we use
a combination of US and Russian satellites. When tested at Bonneville against the SCTA clocks, we were as accurate as they were
and went three digits beyond. I'll see if we can get Steve onto this thread and he can explain it a little better.

[BVCBR]

TD, thanks for the link to the paper.

I was involved in some testing of a triangulated GPS system in 2000 that had stated it was accurate to +/- 0.5". To be fair, it was very accurate in a static position. It did not do so well on a moving vehicle.

Another system we tested used a 3 axis accelerometer to calculate position along with the GPS signal. It worked pretty well as long as the acceleration was fairly linear or stable. it did have some issues with rapid deceleration though.

F104A, any information Mr. Wallace could provide would be appreciated.

[maj]

Finding this interesting too guys, we use gps for steering in agriculture, 2cm accuracy no probs, but our systems have a problem if you loose the signal from interferance , it can take a while to pick it up again esp at speed (10-14mph)
currently looking at cors based and using cell towers as a step up from the radiowave tower
one of our units is topcon  . they have a pretty good reputation   

[TD]

Keep reading, keep learning...

Some introductory info on GPS Real-Time Kinematic measurement and Continuously Operating Reference Station (CORS) at
http://en.wikipedia.org/wiki/Real_Time_Kinematic

So it would appear to be relatively simple (easy for me to write while sitting here) to measure out a mile or a km trap with very good accuracy.  How one could record vehicle position rapidly, repeatedly, and accurately while moving at >= 600 MPH is less clear but I now think it is possible or even relatively straightforward.

My $200 Garmin handheld receiver will report speeds in the 600 MPH range while carried on board a commercial airliner.  I have no idea how accurate that is or how frequently it updates position information.

[Jon]

No GPS units measure speed, they work out their position by calculating distances from satellites using transmitted times.

Then to calculate speed they use position and time, very similar to the way LSR mile times are done.

There is always an error in the calculated position due to clock errors, multipath, satellite orbit errors etc.

RTK (Real Time Kinematic) systems use a base station at a  known reference point to work out the error, by comparing its's known position to its internally calculated position.
The base then transmits this correction to the rover units, the rovers add the correction to their calculated position and use that position.

GNS systems use the Russian and American satellite systems so give more satellites for the system to use, doesn't generally make the system more accurate but makes it more reliable where part of the sky is obscured by trees, mountains etc.

Maj, are you using a portable or fixed base?
We had much better reliability after we logged our permenant bases and portable base points for 7+ days and sent the log to Auspos, used they're corrected locations for base co-ordinates.

Cheers
jon

[maj]

Jon there is a fixed base ,and a portable we use for more remote locations
biggest issue we have is the transmitter power , supposed to be limited to 1 watt  ,but about to get an upgrade

and the cors has hi powered transmitters well spaced around us when we adopt it later in the yr

Using a garmin x18 NEMA device on my bike logger, refresh is too slow at 1 hz but the info looks good so far , no dropouts
speeds match the timing slips , shows a very consistant rate of accelleration and i used that info to shorten up the runs at the Shootout
so i wasn't beating the bike up as much

[NAE_DAG]

OK you guys, before too much misinformation snowballs out of control, please let me set the record straight.

We have a GNSS receiver mounted in the NAE and a Static GNSS receiver operating back at home base. Both these units receive GPS signals from the same 20 satellites simultaneously at 0.05 second update rate. The combined data is processed using software from Topcon Inc. to calculate the vehicle's position every .05 seconds with a positional accuracy WITHIN 5mm horizontally, and 3mm vertically. The Topcon unit in the NAE is controlled via a link the on-board solid state PC. Access to this PC is accomplished via an Ethernet system furnished by Tropos Inc. I do not use the Ethernet to transmit data, just to initialize data acquisition and trigger data gathering prior to each run.

The Ethernet link is not reliable at high speed, but heals itself quickly as soon as it is able and I can reset the equipment and transfer data to removable media. I ran the same Topcon system at Bonneville in 2010 to compare official timing with the GNSS.  The speeds were dead accurate.

Yes, even at 20Hz some extrapolation would be required to hit start and finish lines dead nuts. This is the fundamental issue: (wrap this one around your head!) Traditionally, we have been measuring "speed" by measuring time between two points in space. With GNSS one calculates speed by measuring the distance traveled between two points in time.  The position and speed is extremely accurate and the data (not just the results) can be saved for scrutiny at any later date by anyone with sophisticated analysis software, should there be any questions. We will have to agree upon a uniform method of determining the time traps and correlating it with the customary positions of the timing lights. Just using the first and second measurement points beyond our customary distance marks would work well.  Error, in the GNSS case is virtually the same regardless of speed (down to 1X10^-14 or a one hundred trillionth of a second)) with the distance measured expanding proportional to speed.  With timing lights, any error in measuring the distance between lights is applied against the time between traps... making measurement accuracy inversely proportional to speed.

There really isn't any reason GNSS cannot be used, other than prohibitive cost of the few systems one would need to pass out to each racer prior to the run (they'd have to give it back!).

[BVCBR]

Yes, even at 20Hz some extrapolation would be required to hit start and finish lines dead nuts. This is the fundamental issue: (wrap this one around your head!) Traditionally, we have been measuring "speed" by measuring time between two points in space. With GNSS one calculates speed by measuring the distance traveled between two points in time.  The position and speed is extremely accurate and the data (not just the results) can be saved for scrutiny at any later date by anyone with sophisticated analysis software, should there be any questions. We will have to agree upon a uniform method of determining the time traps and correlating it with the customary positions of the timing lights. Just using the first and second measurement points beyond our customary distance marks would work well.  Error, in the GNSS case is virtually the same regardless of speed (down to 1X10^-14 or a one hundred trillionth of a second)) with the distance measured expanding proportional to speed.  With timing lights, any error in measuring the distance between lights is applied against the time between traps... making measurement accuracy inversely proportional to speed.

Thanks for the info. I'll have to sit and think about this for a little while and ask a couple more questions once I think I know what I'm talking about.

JW

[Dean Los Angeles]

I took my PT Cruiser out tonight and ran the speed you see below.

What? You doubt the speed on the very reliable GPS?

And that is the problem. Timing lights are surveyed and certified by an independent company.

The GPS is very accurate with a fixed base station. Where is the measured mile? Who verifies that the data hasn't been "adjusted".

[maj]

With gps technology should you have to use a set piece of real estate for the mile ?

as long as you can show a full mile averaged speed and time taken to run that mile , and if required the same coordinates used for a return run

new technologies can open doors to view things differently
 

[Seldom Seen Slim]

My turn, okay?  To the question of using a predetermined set of start and end points -- that'd be necessary if running at an event where the same points have been used in previous events.  That would mean that every vehicle has the same amount of space to run up before getting into the lights.  A new sort of measured mile using the "floating" (for lack of a better word) points would work, of course -- but couldn't very well be compared to speeds with the fixed points.

As for higher accuracy -- well, take this idea:
"RTK (Real Time Kinematic) systems use a base station at a  known reference point to work out the error, by comparing its's known position to its internally calculated position.
The base then transmits this correction to the rover units, the rovers add the correction to their calculated position and use that position"

and use it -- just like I've suggested many times in the past few years.  I'd given up on trying to explain it to the unwashed, but go ahead and work out the details and at least mention that I've been thinking about it for a long time.

[Dean Los Angeles]

Records have been claimed by many with no way to independently verify the claim. Many never approach anything close to the claimed speed ever again.

Racing organizations exist for that reason.