Sunday, May 2, 2010


I had to send my Garmin 305 back for a warranty repair. I decided to buy a Garmin 500 as sort of an alternate, so if and when my 305 is in the shop, I'll still be able to anally retentively collect data.
I always figured GPS must be pretty accurate. No more wheel calculations, wires, magnets, easy transfers from one bike to the other.
Well Doug and Mookie shattered my impressions with stories of Garmins inaccuracies in both mileage and elevation gain.
So since I have two units now I decided to do a little test. I used them both simultaneously. I had the 500 on my stem and the 305 in my jersey pocket.
The 305 recorded a further distance, but less elevation gain and a crazy high calorie burned.
You would think the unit on the stem would have the higher mileage count? Or that two units would somewhat closer on elevation? Or what kind of whacked out algorithm are the using to calculate calories, and why the 1300 C difference?
They keep pretty good time though.


Hill Junkie said...

Not just Garmin's. Any GPS will have similar shortcomings. I keep a wired computer on all my bikes. When I ride rode, the GPS will stay within 1% of wired computer for both measured distance and moving time. When I ride the FOMBA trails (extremely tight and twisty), the GPS will typically lag the wired computer by 20% in distance but stay pretty close in moving time. At least a GPS is good at telling when you move or not. When you look at the track file, you get a hint at why it falls short. The GPS records data only periodically, even though it nearly coninously makes measurements from the satellites. The recorded data cuts off many of the tight turns, as if you were cheating and cut through instead of taking the switchback all the way around. I can only assume Garmin is applying a similar kind of smoothing algorithm to the odometer measurement.

As for altitude measurements, barometric altimeters can work well in certain situations. Elevation data from satellites is bogus. It is maybe good to calibrate an approximate starting altitude for a baro altimeter, but that is it. Baro's work well where you have relatively few large monotonic climbs and descents. Where they fall short is in up-down singletrack, which is where we mountain bikers care most. Again, they apply a rigorous smoothing algorithm to the raw data to eliminate the wind noise effects. Without this, barometric altimeters would over measure climbing. So what happens is small hills don't even register, as the small pressure change is treated like wind noise.

What bugs me about this is the only way I can measure distance and vertical XC skiing is with the GPS. I know in certain terrain it will undermeasure both by quite a bit. It will probably do so consistently, so at least I can compare workout to workout. That is probably all that really matters anyway.

Mookie said...

Yeah, what Doug said.

Mark said...

You know, I noticed that the 305 always recorded insanely high C burn but the 205 and 705 seemed to be pretty consistent with each other.

sujal said...

I'd also expect more variability if the device is tucked into your pocket vs. being out on the stem. It's not the jacket per se, but the fact that it's closer to your body. That will cut down on a satellite fix in one or more directions.

I keep mine attached to my pack. and it doesn't do as well as when it's free in my hand during a walk or whatever.

The other thing I've noticed is that trees are another source of interference. When I'm going on narrower roads with a lot of dense tree cover, the GPS can only lock on within 10-30m or so, which introduces a lot of error compared to my wired computer.

Just my observations.