Tuesday, July 22, 2014

Low Power Is Bad

I like power when windsurfing. Speedsurfing does not work without power. Turns out this is also true for the GT-31 GPS (yes, this is another geeky post about GPS, so feel free to stop reading).

In my recent post about comparing 5 GPS units,  I had described how one of the two GPS units I used had given substantially worse results than the other one. One of the great names in speedsurfing, Andrew Daff (aka Sailquick), pointed out that this could be due to differences in settings - and he was right. I discovered that my "bad" GT-31 still was set to "low power" mode, while the other unit was in "normal" power mode. It appears that the "low power" mode restricts the GT-31 to tracking at most 6 satellites; in normal power mode, the unit can track 8 or more satellites, which gives much better accuracy. I took both GT-31s I had for a drive to see the effect - here are the tracks from "low power" mode:
Low power mode tracks from two GT-31s
For comparison, here are the tracks in normal power mode from the way back:
Normal power mode tracks from two GT-31s
In normal power mode, the positional accuracy is a lot better - both tracks are right on top of each other, and I'm not driving over houses anymore. The differences are most pronounced when driving in circles. Here are enlargements of the top sections of the tracks:
Low power mode circles
Normal power mode circles
In normal power mode, the tracks actually show the circles and ovals I drove; In low power mode, they are all over the place. In low power mode, the GPS units tracked only 4 satellites; I normal power mode, they tracked 7-8.

Here are the speed and doppler speed graphs for the low power mode:
Speed graphs for low power mode
In low power mode, the speed graph shows a lot of spikes (I actually removed the biggest one to get better scaling). The doppler speed graph does not show any spikes, which is a nice illustration why we doppler speeds are preferable for speedsurfing. However, even the doppler speeds are quite different for the two units at many time points.

For comparison, here are the speed graphs in normal power mode:
Speed graphs for normal power mode
With 7-8 (mostly 8) satellites tracked, the positional data are free of spikes, and very close to the doppler data. Differences between the two units are much smaller, and limited to just a few short regions.

Fortunately, the "normal" power mode is the default setting for the GT-31 (I think). So why did I ever switch my GT-31 to "low power"? I did it when I tried to set a personal best for distance within 24 hours. Somewhere, I picked up a suggestion to switch to low power mode to make sure that the battery would last long enough. Neither the suggester nor I had seen Redsurfbus' comment that low power mode creates spikes...

I did not even sail 24 hours that day. I got too tired after sailing about 12 hours in chop, sailing only 270 km. Apparently, I was still too tired the next day to remember to set the power mode back to normal. This affected my jibe analysis using GPS data for the next two years...

What do we learn? Well, if you have a GT-31, go to "Settings", scroll down to "Power mode", and make sure it is set on "Normal". The other take-home message I took from this is that the number of satellites tracked is a very useful bit of information to have - it is tightly linked to speed accuracy. Although I believe that other factors also come into play when SDoP and HDoP are calculated, the number of satellites could be a good-enough stand-in for recreational (i.e. non-record) windsurfing. With newer GPS chips typically tracking more satellites than the old Sirf3 chip, if should be possible to set the minimum number of satellites higher that the currently suggested cutoff of 5 - at least to 7 or perhaps 8. Why does this matter? Because only very few GPS units can record accuracy values (SDoP and HDoP) directly, but many support the NMEA format that does support satellite information, or similar formats.

I really hope that I'll be done with this geeky stuff for now, and that the promised wind will show up tomorrow and this weekend so that I can spend some time on the water!


  1. Does the canmore 102 have Doppler speed read out in the file? There is a disagreement about this, some user don't see the Doppler.

    1. Not sure why you are asking this in response to an article that is entirely about the GT-31. But yes, the Canmore GP102 records doppler speeds. The speed it records are different from the positional speeds, and virtually identical to the GT-31 doppler speed if the GT-31 is used in "normal" power mode. There is really no disagreement about this. What is missing in the data files are the precision estimates for the doppler speed. Those would be nice to have, but are required only for record attempts. GT-31 data generated with the "low power" setting have been accepted for years, and these are clearly inferior to GP102 data.