Peak to floor is 100 ppmWhat is the scale of the variance on your small graph? Is it even meaningful? You also show a sharp drop before a slow rise. That seams more significant.
The sharp drop was when I did a water change with IO RC salt and it is low on mag, which is the part that makes sense. It is the steady increase since that doesn't. I am wondering if it could be bound up somewhere, but testing error seems to be the prevailing theory right now.