I'm testing a TMAG5273 for use in a game controller product, using it as an angle sensor. The IC is in a roughly on-axis configuration with some small amount of offset from true center, but the IC outline is 'within' the diametric magnet's face area, so it's not very much of an offset if anything.
By 'within the face area', I mean, if the face of the diametric magnet was projected to create a longer cylinder, the IC would be contained inside this cylinder. I estimate that the IC is no further than 1.5mm from the shaft's center of rotation. The IC is mounted on a horizontal PCB with a vertical shaft. The shaft's vertical mounting is not perfect, it's a couple of degrees off from 90: maybe 2 degrees off (88 degrees from the PCB).
Rotating the shaft, I'm seeing what appears to be a couple of dead-bands in the angle output from the device. I'm rotating the shaft with my hand and it's a fairly smooth mechanical system so I'm quite certain that the shaft isn't binding and there's no mechanical sticking; the angle is changing smoothly.
Yet, I can find regions where movement of possibly 2-3 degrees yields 0.0 change in angle-sense output. Otherwise, the sensitivity is excellent.
I wasn't expect this based on the setup. Some amount of nonlinearity due to the shaft-axis offset, and the shaft's angle offset, is expected.
Are dead-bands of 3 degrees or more typical for this device? I can't see anything in the data-sheet or application notes when I search 'deadband'.