garmin sensors

cyclosteve
cyclosteve Posts: 150
edited February 2013 in Workshop
ive got a garmin edge 500. when out riding ive noticed speed seems to flit up and down by a few mph each way and yet my speed has not changed well not that much anyway.is this normal. i dont use the cadence sensor as got 4 bikes and never use turbo so run it on gps all the time.is this the norm or should i run it eith sensors

Comments

  • Hi there, I would run it with sensors if I was you. I had problems using an iphone before to record rides as it was only using GPS. I found that the signal can come and go and the max speed recorded was crazy. If you fit the speed/casence sensor for the Garmin it gives you actual speed and not GPS speeed, and you can use it on the turbo/ rollers as well.
  • sungod
    sungod Posts: 17,434
    the edge 500, like most small gps units, doesn't have doppler circuitry, so it's not reliable for measuring instantaneous speed

    if the route is curvy and/or hilly the distance data will also be iffy, usually less than the true distance

    you'd get better figures from the on-bike sensors, especially if you did a roll-out to measure the wheel circumference and set it in the head unit

    thing is, do you really *need* more accurate speed info? it doesn't tell you anything useful in training terms

    instead of sensors, i'd use the money for more chocolate :-)
    my bike - faster than god's and twice as shiny
  • rolf_f
    rolf_f Posts: 16,015
    sungod wrote:
    the edge 500, like most small gps units, doesn't have doppler circuitry, so it's not reliable for measuring instantaneous speed

    Interestingly, my Bryton doesn't seem to have a significant fluctuation in instantaneous speed. What it does do is slightly lag though. I assume there is something going on to dampen the fluctuations (which I know are there becasue my average recorded speed drops too much if I stop - ie I'm not being recorded as fully static).
    Faster than a tent.......
  • Bozman
    Bozman Posts: 2,518
    sungod wrote:
    the edge 500, like most small gps units, doesn't have doppler circuitry, so it's not reliable for measuring instantaneous speed

    if the route is curvy and/or hilly the distance data will also be iffy, usually less than the true distance

    you'd get better figures from the on-bike sensors, especially if you did a roll-out to measure the wheel circumference and set it in the head unit

    thing is, do you really *need* more accurate speed info? it doesn't tell you anything useful in training terms

    instead of sensors, i'd use the money for more chocolate :-)

    Slight query, I use a cateye strada and have my iPhone running Cyclemeter, on a bad day there's a tenth of a mile difference over a 3 hr ride, the average speed varies by up to .2, one GPS the other not.
    I've just bought a 500 and I wasn't going to fit the cadence yet because of the lack of difference between the two, but you lot are putting doubt into my mind.
  • sungod
    sungod Posts: 17,434
    it depends on the route, your speed, the signal conditions, and the device implementation

    the gps gets the raw data by logging it's position, typically every second

    it then uses a 2d model (ignoring altitude) and calculates the straight line distance between adjacent points - a 2d rather than 3d model is preferred because gps altitude accuracy can be very poor

    some devices have a barometric altimeter to improve altitude recording, but i don't know that any use this to create a 3d distance model, i doubt it as the barometer usually lags many seconds behind altitude changes

    anyway, speed can then be calculated from distance/time, but the distance is subject to those position errors

    aside from the uncertainty caused by the error in each position measurement, the straight line approximation to the true path causes distance to be under read on curvres and the 2d model ignores inclines

    at low speed, error on curves is not so bad, as many samples will be taken around the curve, at high speeds the error increases as the straght line approximation lacks enough points to get close to the true path, on steady inclines the error is not affected by speed, but error increases with gradient

    to avoid displayed speed jumping around, the data are filtered, algorithms will vary by make and model, devices may have an option to change the smoothing period

    whilst instantaneous speed can be off, the speed accuracy over a longer period is far better, the individual position errors tend to average out

    unless the route is fast and twisty, or has a lot of steep inclines, the overall distance and average speed will be pretty accurate, even on a 'bad' route i'm sure they'll be more than sufficient for training purposes
    my bike - faster than god's and twice as shiny
  • sungod wrote:
    it depends on the route, your speed, the signal conditions, and the device implementation

    the gps gets the raw data by logging it's position, typically every second

    it then uses a 2d model (ignoring altitude) and calculates the straight line distance between adjacent points - a 2d rather than 3d model is preferred because gps altitude accuracy can be very poor

    some devices have a barometric altimeter to improve altitude recording, but i don't know that any use this to create a 3d distance model, i doubt it as the barometer usually lags many seconds behind altitude changes

    anyway, speed can then be calculated from distance/time, but the distance is subject to those position errors

    aside from the uncertainty caused by the error in each position measurement, the straight line approximation to the true path causes distance to be under read on curvres and the 2d model ignores inclines

    at low speed, error on curves is not so bad, as many samples will be taken around the curve, at high speeds the error increases as the straght line approximation lacks enough points to get close to the true path, on steady inclines the error is not affected by speed, but error increases with gradient

    to avoid displayed speed jumping around, the data are filtered, algorithms will vary by make and model, devices may have an option to change the smoothing period

    whilst instantaneous speed can be off, the speed accuracy over a longer period is far better, the individual position errors tend to average out

    unless the route is fast and twisty, or has a lot of steep inclines, the overall distance and average speed will be pretty accurate, even on a 'bad' route i'm sure they'll be more than sufficient for training purposes

    So that's that sorted out then.