1. This forum section is a read-only archive which contains old newsgroup posts. If you wish to post a query, please do so in one of our main forum sections (here). This way you will get a faster, better response from the members on Motherboard Point.

GPS module with speed out 10 times / sec

Discussion in 'Embedded' started by Jyrki Holopainen, Oct 9, 2006.

  1. Are there any GPS-modules from where it is possible to get out speed
    more frequently than once / second?

    - 10 Hz would be the ideal rate.

    - I'd like to measure speeds on vehicles, typically 50 km/h. Speeds less
    than 10 km/h or over 150 km/h are not important.

    - The location information is not important.

    - Low power consumption is important (battery powered unit).
     
    Jyrki Holopainen, Oct 9, 2006
    #1
    1. Advertisements

  2. This sounds like a good job for an accelerometer.

    You get speed from a GPS by looking at where you are at one time, and
    where you are at another, and dividing distance by time. If you wanted
    10 Hz readout at 50 km/h, then you'd be looking at a change of position
    of 14 meters per second, or 1.4 meters per update. If your GPS could
    measure to 14 centimeters in a tenth of a second, then you would be
    able to tell whetehr you were going 50 km/h or 55 km/h. But you can't
    get that good with civilian GPS.

    So you should use a GPS to tell you how fast you are going over one
    second, and an accelerometer to tell you that during the third tenth of
    that second, you were going 2 km/h faster than average...
     
    David M. Palmer, Oct 9, 2006
    #2
    1. Advertisements

  3. Jyrki Holopainen

    larwe Guest

    If you can connect to the OBD connector, I believe the VSS information
    is available at that point (and you can power off the vehicle's
    electric system, too).
     
    larwe, Oct 9, 2006
    #3
  4. That's what I am afraid too, but the device should work without _any_
    installation.
    GPS does not measure speed as a change of position / time, but using
    doppler effect. I do not see why a device could not provide more than
    one measurement / second. A civilian GPS should be able to measure speed
    to 0.1 nautical miles accuracy.
     
    Jyrki Holopainen, Oct 9, 2006
    #4
  5. Not every vehicle has an OBD-connector, so that cannot be used in this case.
     
    Jyrki Holopainen, Oct 9, 2006
    #5
  6. How? This would only give you acceleration, not speed. Of course you could
    integrate acceleration, but this will give you drift errors.
    That's not how a GPS receiver determines speed. A GPS measures the doppler
    shift of the satellite signals to measure speed, so the indicated speed is
    instantly accurate.

    Meindert
     
    Meindert Sprang, Oct 9, 2006
    #6
  7. Jyrki Holopainen

    larwe Guest

    In my area, at least, it is a federal mandate that all vehicles have an
    OBD connector.

    Having exhausted that possibility, how much are you allowed to modify
    the vehicle? A magnet on the wheel and a Hall effect sensor in the
    vehicle body would do the same thing. (But this is quite challenging to
    keep working longterm).

    You can also use a MEMS accelerometer attached to the wheel hub with
    double-sided tape. Count the peaks on one axis...
     
    larwe, Oct 9, 2006
    #7
  8. Jyrki Holopainen

    Al Balmer Guest

    I don't think that's quite correct. GPS calculates speed by using
    doppler shift (sort of). I'm not a GPS expert - just a heads-up that
    you may want to research further.
    http://www.datrontechnology.com/all/site/sections.php?op=viewarticle&artid=14

    Commercial velocity measurement systems are available that claim 7
    updates/second with 0.1 mph accuracy.
     
    Al Balmer, Oct 9, 2006
    #8
  9. To what kind of precision?

    Keep in mind that GPS devices typically use Doppler to measure speed
    --- that basically means to compare the speed of your vehicle to the
    speed of light. That's a dynamic range of 3*10^7 you're talking
    about, and now you want to get 10 readings of that per second, on a
    severely restricted power budget? That's not particularly likely to
    happen. Not for civilian-grade hand-held GPS units running on
    batteries, anyway.
     
    Hans-Bernhard Broeker, Oct 10, 2006
    #9
  10. One clarification, I mean an accelerometer plus a GPS unit. Also, a
    compass would help, if you don't want to work things out from context
    or if the road is extremely twisty.

    You don't need any installation beyond what you need for the GPS unit.
    They can be in the same box, however it is not installed.
    It works out to the same thing, in theory. "By looking at the radio
    signals, I determine that I have moved 20 cm (= 1 wavelength) in the
    past second" is equivalent to "I have detected a doppler of 1
    cycle/second, so I am moving at 20 cm/s".

    In practice, working in the frequency domain uses different techniques
    that working in the time domain. (The difference between theory and
    practice is that, in theory, there is no difference.)
    A device can report many measurements per second. It usually doesn't
    because it gains accuracy by using a low-pass Kalman filter, and so the
    additional reports are little better than inter/extrapo-lations.

    A speed of 0.1 mph measured over 0.1 seconds corresponds to 5 mm, or
    less than 10 degrees of carrier phase. GPS signals are very weak (-130
    dBm, like a microwatt EIRP a mile away) and it is very hard to reliably
    measure such a weak signal to such high phase accuracy in such a short
    time, even without real-world complications such as multi-path.
     
    David M. Palmer, Oct 10, 2006
    #10
  11. Jyrki Holopainen

    rickman Guest

    I read all 10 replies before I responded. I think a lot of what they
    say is correct in that the higher update rate will not buy you a lot
    due to the filtering done on the values returned, but certainly there
    may be applications that still need the frequent updates. I have
    recently worked with two brands of GPS modules. By modules I mean very
    small surface mount devices that are used much like a chip. You
    provide an antenna connection, power and an LVTTL level serial
    interface to a processor. The module will spit out a variety of
    reports including speed and will do so at 4 updates per second. These
    modules are designed for embedding in a cell phone (aren't they all at
    this point?) and so the size and power are very low. I belive they are
    about 100 mW.

    The one we picked for our project was from Fastrax called the iTrax03S.
    The second runner up (and a very *close* second it was) is the LEA-4S
    from uBlox. uBlox is the company that is putting up the money for the
    Antaris devices from Atmel. I believe Atmel is the fab house and they
    have an agreement to sell the chips to markets that uBlox is not
    pursuing. uBlox seems to be selling Antaris 4 based modules while
    promoting the next gen uBlox 5 "all on one" chip at 8x8 mm. Add the
    crystals, the SAW filter, antenna and 50 mW of power and you have a
    working GPS unit!
     
    rickman, Oct 10, 2006
    #11
  12. On rally-cars, motorcycles, snowscooters, bicycles, go-karts, trucks and
    busses too? I have understood that OBD-connector is mandatory on
    registered passanger cars only.
     
    Jyrki Holopainen, Oct 10, 2006
    #12
  13. It seems that the 10 Hz is not easily possible, so anything better than
    the typical 1 Hz would help. The typical speed precision of 0.1 nautical
    miles / h would be ideal, but on worst case I could accept precision of
    1 km/h.
     
    Jyrki Holopainen, Oct 10, 2006
    #13
  14. Thanks, I'll check these ones.
     
    Jyrki Holopainen, Oct 10, 2006
    #14
  15. Without _any_ installation I mean the the device could be "thorwn in".
    The device can be in any angle, which means that simple accelerometers
    cannot be used.

    I could accept lower update rate and/or lower accuracy. The typical
    update rate (1Hz) is however too low.
     
    Jyrki Holopainen, Oct 10, 2006
    #15
  16. Jyrki Holopainen

    SamSvL Guest

    Some low cost modules output raw data (pseudoranges, integrated carrier
    ranges and instantaneous Doppler) at 10Hz (e.g. the u-blox Antaris based
    receivers). I do not know of modules that are able to output position and
    velocity with 10Hz, Probably their processor lacks power to do the
    calculations at this rate. But the math is not too difficult, so if you
    add
    your own processor you could calculate position and velocity at 10Hz
    using the raw data. I have done some tests with the u-blox TIM-LP (see
    http://home-2.worldonline.nl/~samsvl/TIM-LP.htm) and found a velocity
    accuracy of 0.15 m/sec 95%.

    Cheers,

    Sam
     
    SamSvL, Oct 10, 2006
    #16
  17. Jyrki Holopainen

    dalai lamah Guest

    Un bel giorno Jyrki Holopainen digitò:
    If you just need the speed, you could try to use the so called "raw data".
    Most GPS modules are able to output some intermediate values (basically
    pseudoranges, ephemeris and doppler shift) at a higher speed than the
    normal update rate; the really heavy calculations (triangulation and
    coordinate transformation) are made after.

    You need to check on your GPS manual/datasheet if it has the ability to
    output raw data (and perhaps to disable the triangulation/trasnformation
    algorithms to save power). Then you need some very strong basics on GPS,
    since you have to convert the doppler shift of each satellite signal into a
    speed value.
     
    dalai lamah, Oct 10, 2006
    #17
  18. A lot of the UBLOX modules will update at 4Hz. I've found that
    adequate for navigation at speeds up to about 20m/second (~45mph).

    Mark Borgerson
     
    Mark Borgerson, Oct 10, 2006
    #18
  19. That's gonna be *very* hard to find. You're asking for 1e-9 precison
    in an uncontrolled environment.

    A precision of 1 km/h during 0.1 seconds corresponds to less than 0.03
    meters of position tolerance. You're asking for a GPS that's
    principally accurate to an inch, but just refuses to tell you about
    it. Or, to put it differently: you're requiring the frequency of a
    very faint GHz wave to be counted to less than one cycle time.

    And as if that wasn't enough, the unit will be getting different
    Doppler-shifted frequencies from different satellites, all mixed into
    a single antenna signal, so it also has to read the data contained in
    those streams to know where the satellite with the dominant signal is,
    so you can know its Doppler signal.
     
    Hans-Bernhard Broeker, Oct 10, 2006
    #19
  20. If you have 3 accelerometers at right angles, you can have the device
    at any angle, as long as it doesn't slide around (which would also be a
    problem in the GPS-only case). It will have to teach itself which
    direction corresponds to forward acceleration, and depending on the
    vehicle type (e.g. how much sideslipping it does, how fast it turns...)
    a compass would be useful for converting among north, direction of
    motion, and the derived nominal forward.
     
    David M. Palmer, Oct 11, 2006
    #20
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.