More error when moving

Hi everyone. I’m using 4 anchors (SMA antenna) and 1 tag (antenna on-board) with SS-TWR method. When Tag doesn’t move, the accuracy is 10-30cm (without any filter). But when Tag is moving, the accuracy is so bad. How to improve accuracy without using filter? I have calibrated antenna delay and range bias.

What method are you using to calculate the position? Is this your own position calculation, PANS, or some other 3rd party system?

When you move do you see a zig-zag or sawtooth pattern in the position?
How fast are you moving and how fast / frequently are you measuring the ranges?

Position calculation generally involves making the assumption that all the ranges were measured simultaneously. There is also the assumption that the range is the same at the start and the end of each individual measurement process. For slow movements and high range update rates these assumptions are close enough to true. The faster you move and the slower you update the measurements the less true this becomes and so the higher the error in the position measurement.

The simple way to improve this is to measure things quicker. Our early prototypes used off the shelf systems which only managed ~50 measurements per second, the results were unusable even at walking speed. We upped this to 2400 DS-TWR measurements, that combined with some sneaky radio protocol tricks allow us to track things at up to 200km/h with fairly low noise levels, typically under 5 cm.

Hi @AndyA. I’m using TDMA with SS-TWR method. Frequency is 5 Hz and speed is about 5km/h. When the Tag starts to move, its coordinates are pretty much wrong, like it’s jumping around. But when moving at a steady speed, the coordinates are more stable.

How is the position calculated?
When you say 5Hz do you mean a 5Hz position output with each range measured once during that 200ms update period?
Are the measurements equally spaced over the time period or are they taken in a burst at the start and then idle for most of the time? Walking will give you very roughly 1 m/s of speed, that means up to 20 cm of range change over your update period. If the measurements are spread out over that time then it’s going to result in some sizable position errors.

Two things to look at: Plot the ranges to each anchor over time and look for issues with those. You should get nice smooth curves, if you don’t then look at antenna obstructions or other signal integrity issues. I’ve seen some wonderfully weird issues using SMA antennas on cables where the range would have jumps of 1/4 or 1/2 of the antenna cable length as it sometimes got picked up directly by the components due to poor shielding.

If the raw ranges look good then it’s a time synchronisation or position calculation issue.