This means reef-pi is reading it always on... I think its highly unlikely that we have a driver bug in reef-pi, and we are missing something else here, likely configuration related. Did you already shared your configuration screenshot? If not please do, connectors (inlet GPIO 8 ?? ) and ato details (check interval, control pump etc). I just tested ATO again with master (rc-2) release, and I could not find a bug. I am doing a whole lot of field testing now, and will hate to ship reef-pi with bugs like this (if it turns out) :-( . If your issue does not resolve, then I'll create a dedicated issue on this in github and I'll try to reproduce this with actual circuit. I have spare float switches, so no issue on getting this physically verified.