GPS Location Precision

[Harry]

So the Emprum and the XGPS160 smooth / filter data approximately the same while the VBOX applies a lot less filtering if get it correct. Opposed to the Emprum, I assume the XGPS had an stable lock (many satellites) from the beginning?

- Harry

[gplracerx]

So the Emprum and the XGPS160 smooth / filter data approximately the same while the VBOX applies a lot less filtering if get it correct. Opposed to the Emprum, I assume the XGPS had an stable lock (many satellites) from the beginning?

- Harry

The first fix lists 12 satellites for the XGPS160. Unlike the Emprum, it's a little hard to start the XGPS160 cold. It usually has a satellite lock before it establishes BT communications.

I'm not at all sure why there is so little filtering with the VBOX Sport. They calculate everything in Circuit Tools from the position data. To me, it makes more sense to filter the position data first and then calculate speed, filter some more and calculate acceleration. If you're going to filter the position data, you might as well do it as it's being acquired, rather than later. The question, though, is whether the XGPS160 filter time constant is so low as to lose information. I don't think it is, but I haven't tested enough.

By the way, the VBOX sport can log data on its SD card at the same time it's sending data to HLT. The timestamp on the VBOX Sport data appears to be 0.03 seconds earlier than the timestamp in HLT. When plotted on the same graph, the data appear to be identical. It's a little hard to tell because HLT only uses 6 decimal places for latitude and longitude in decimal degrees while the VBOX Sport uses 7 (before multiplying by 60 to get decimal minutes).

Edit: That also more or less proves that Location Services isn't messing with the data, as it was on at the time.

[Harry]

The Emprum and XGPS need to deliver data that pleases *every* app / application / user. The VBOX usually delivers everything to RL's own systems, allowing to defer all types of post processing to the desktop systems / analysis software. Makes sense from my point of view...

- Harry

[gplracerx]

I need to retest the Emprum with HLT. I was using the Sensor Data app, and that, as I remember, does use Location Services.

[gplracerx]

This is what Circuit Tools did with the data in the .vbo file on the SD card from the test above. It's total acceleration in m/sec^2 rather than g. Changing the smoothing setting had zero effect.



I'm not impressed. But then, I think CT is worth exactly what I paid for it.

[gplracerx]

The Emprum and XGPS need to deliver data that pleases *every* app / application / user. The VBOX usually delivers everything to RL's own systems, allowing to defer all types of post processing to the desktop systems / analysis software. Makes sense from my point of view...

- Harry

I compared simultaneous data acquired with an XGPS160 connected to HLT and a VBOX Sport in stand alone logging mode with an external antenna. The only good data in the .vbo file from the VBOX Sport are the time, latitude, longitude, speed and longitudinal acceleration. The course direction data are heavily filtered and those filtered data are apparently used to calculate lateral acceleration, so that's distorted as well. I can almost exactly match the VBOX Sport course and lateral acceleration data by applying an exponentially weighted moving average filter with an alpha of 0.2. With a sampling time of 0.1 seconds, that's a time constant of 0.4 seconds or a cutoff frequency of 2.5Hz. The iPhone accelerometer was set to 100Hz. A setting of 30Hz produces nearly as much distortion of the acceleration data.