Fluxmeter 2 drifts more than fluxmeter 1, and additionally is somehow out of phase with fluxmeter 1, causing artefacts seen in the second image.
The solution is to only use one fluxmeter.
Normally
However one could also argue that one compute the gradient with relation to the center of the magnet, which is zero, hence
So in theory we could just use the output of one fluxmeter.
Notice the artifacts before and after flat top.
