What works today

The limits are on the node

Hard limits live on the Tank Node itself. The router drops, the Pod is unplugged, you pause everything else — the limits are still being enforced, and a beacon on every screen tells you which state you are in without hunting through menus.

The node also switches its own twelve volt ports, so a doser, the auto top-off and a CO₂ solenoid all hold with the network down. Being able to stop those is worth more than being able to start anything. None of it is a fallback mode. It is where the decisions are made on a good night too.

Sump level is a number with a history

The node reads the distance to the surface, continuously, from above the water, so there is nothing submerged to foul or calcify. Level becomes a parameter with a trend behind it, the same as temperature or pH.

The shape of that trend is the information. A steady fall across the day is evaporation. A fall that keeps going after the top-off has run is a leak or a pump. A fall that runs past the top-off's own trigger without stopping is a top-off that has quietly stopped working. None of those are visible to a switch. Still being tested: whether it reads reliably under a closed lid with condensation forming on it.

The float, and the thing that watches the float

The float is wired so healthy is closed, so a broken wire, a pulled connector and a dead switch all report the same way — as the alarm. A float stuck closed does not, and that is what the next check is for.

And because the level sensor reads the same water by a different method, the controller knows what the float should be saying. When the two stop agreeing, that is a fault and you are told. That is the honest answer to the only real objection to a float, which is that floats stick.

Two supplies, and no transfer between them

The node has a second DC input on a barrel socket and takes a pack you already own — any 12 volt pack, a lithium jump box, a sealed lead-acid brick, a USB-C bank through a trigger cable.

Both supplies sit on the same rail behind ideal diodes and the higher one simply wins. The transfer time is zero because there is no transfer. Nothing switches, nothing detects a failure and reacts to it. The node is a load and never a charger, enforced in the wiring rather than in software, and the reserve inside the node is a capacitor rather than a cell — the part that has to outlast everything has nothing in it that wears out.

Eight isolated channels, all of them yours

Every node is the same board. There is no version with fewer channels and nothing on the node to unlock. Isolation keeps a probe in the water from sharing a ground with the rest of the controller, and it costs less to build every channel that way than to decide later which ones deserved it.

Temperature and level arrive on their own inputs, so all eight are free and all eight are visible from the first boot, unlit, waiting. You buy probes on your own schedule, because that is how probes actually get used.

Calibration that tells you the probe is finished

Two points for pH, one for ORP and salinity. The screen walks it — rinse, place, wait for the reading to settle, capture — and the buffer value is editable, because the number on your bottle is the one that matters.

At the end it works out the slope and says how far off ideal it is. A probe reading 70% of ideal slope is not miscalibrated, it is worn out, and that is the number that tells you to buy a new one. Every probe carries the date it was last calibrated.

A node’s record is bound to its board

Each node carries a fixed hardware ID alongside the name you give it, taken from the radio MAC burned into the board at the factory. Rename it, reset it, reflash it, restore a config — the history stays with it.

Replace the board and you have a different unit with a different ID, so it arrives empty. The controller sees it has taken the old one’s place and offers to move the name, the calibration history and the alert rules across, so the trend does not restart at the swap. And nothing joins from a screen: adoption is a three-second touch on the node itself, inside a thirty-second window.

Your hand tests go in the same record

A syringe and a vial do not drift, do not clog and have never needed a firmware update. So the first release is built around the test kit you already own: a hand result goes into the same record as the probe data and carries the same history, instead of living in a notebook.

There is no automated alkalinity testing, and that is a decision rather than a gap. A controller acting on a wrong number is worse than one with no number at all.

A screen that can sit in a room you live in

The Display Pod goes where you do — at the tank, on a table beside it, or across the room where you actually sit. Nothing gets screwed to a wall. After an idle timeout you set, it falls back to a clock or your own tank photos, and a tap brings the dashboard back. A PIN on that tap is optional, and so is a PIN on individual tabs.

Three palettes ship with it, and under those are eight color roles you can set yourself. The preview updates live on a working dashboard before you apply anything, and a theme exports as a file. From anywhere else you can look in and see the same parameters on your phone. The screen in the room is the point, not the limit.