I’ve been reading about z-axis error and GDOP this morning but I’m not sure if this large error is expected or not. I have four anchors on PVC poles outside with clear LOS to each other and the tag. The anchors are arranged in a square about 3m apart from each other. I set the first two anchors (including the initiator) to a height of 1.8m. The second pair are set at 1m.
Inside the square z-axis is fine, but as soon as I step outside the accuracy drops down. If I walk 15m away the tag tells me it is 14m in the air. For reference I am outdoors, on flat ground with no obstacles. I have a clear LOS to all tags. And I see the same data in my software and using the Quorvo Android app.
Is this level of accuracy to be expected? Or is there something I can do to make it better?
You have a 3m square of anchors that you are 15 meters away from and you expect any sort of accuracy? With that sort of geometry the horizontal position is going to be fairly poor too. I don’t know exactly what would be typical for that sort of setup, I’ve never tested it because I’d expect it to be unusably bad.
You say that the anchors are at two different heights. How are these arranged? The two high ones should be on opposite corners for best performance.
Actually the horizontal accuracy is pretty good, just the z-axis was off. The two high ones are in opposite corners. After manually measuring and setting the anchors x,y,z positions it got better. Maybe down to 4m at that distance. Maybe that’s the best it could do but if I could get it to 1m that would be better.
At 14m horizontal distance and a true tag height of 2m the distance from an anchor at 1m to the tag is 14.04m.
If you move the tag 1m vertically the distance to the becomes 14.14m, a 10cm difference.
So using a bit of hand waving approximation rather than any actual calculations that would imply that if you can get range accuracies of around 10cm then vertical accuracies in the 1m region should just about be possible.
If you can get the anchors higher then things improve, a 1 m height change going from 3 meters vertical difference to 4 meters different results in a range change of almost 25 cm rather than 10 cm. Generally putting things 4 m in the air isn’t practical but it illustrates the effect.
And as you discovered, accurate surveys are critical to good results.