L76K GPS Precision: T-Echo vs T-Beam Supreme


I’ve been measuring the GPS data of the two modules, and the results are a bit surprising to me. In both cases, the GPS chip is the L76K, but the difference in precision between the two is brutal. The T-Echo (in red) has measurements that are very far from the average and the real position. The T-Beam results are surprisingly good. They were both in the same position (center of the green circle). Any explanation, except the antenna?

Perhaps explain what the plots you have posted are actually meant to show, all I see is a red line bouncing up and down, with no indication of what it means.

Above is a map. With samples of two devices fixed in the same position. I think every minute. In red, T-Echo, with the points very dispersed. In green, T-Beam, with the points very located.

Below is a graph with time on the abscissa and latitude of the samples from the two devices (same color) on the Y axis, and below, another with longitude.

(It is a screenshot of https://casium2.uma.es/mapaArt/ in which the default date and time corresponds to samples of a pair of t-beams with neo m8n)

Could just be differences in the antennas.

Are both GPSs outside with a clear view of the sky and horizon ?

Yes, they are both on a terrace with more than 7 eighths of visible sky, on a day with clean, clear skies. T-Echo was upright. I will look at the internal location of the antenna and do tests by placing it on the top. In any case, I think the casing itself can also influence it. Thank you!

The graph is presumably the change in position over time for a stationary node, I have seen similar behavior with my node indoors, outdoors it is not nearly as bad, I believe there are several reasons:

Cheap antenna with only single frequency
L76K has no Galileo support
Antenna is in a really bad location, it is right next to the board and right next to the Lora antenna, and it is set down in the body so there is a large signal gap in the direction of the Lora antenna, and when the device is held there is a signal gap from you.

What actual distance variation, from the known fixed point, were you getting ?

Oh I’m sorry. I forgot to clarify that the radius of the circles is 4 meters around the mean, so the T-Beam error is around one meter, and the T-Echo error is several meters. I have not had time to make more measurements, nor to calculate the actual standard deviations