Solar · Storage · Energy Management

Total visibility and control over every watt.

MWatt.io is an end-to-end platform for photovoltaic and battery storage assets — from Modbus registers read at the edge, to live dashboards, automated peak shaving, and the demand-charge savings that follow.

Login to the dashboard See the platform
  • Runs at the edge — keeps logging when the internet doesn't
  • Hardware-agnostic drivers built from manufacturer register maps
  • Monitoring and dispatch control in one system
60 sField polling cadence
0.5 sBattery control loop
64Values logged per device row
5Layers, edge to dashboard

The platform

One system, from the register to the invoice.

Most tools stop at a pretty chart. MWatt.io covers the whole chain — acquisition, transport, storage, visualisation and closed-loop control of the assets themselves.

Edge data acquisition

Modbus RTU and Modbus TCP polling driven by per-device drivers compiled from official manufacturer register maps — never guessed addresses. A bus arbiter gives every application a fair lease on the RS-485 line.

Nothing gets lost

Readings buffer to disk the moment a link drops and upload when it returns. Every row carries its own health header, so an outage is an explicit record in the time-series — not a mysterious gap in a chart.

Live dashboards

Power-flow view, KPI strip, per-device drill-down to any raw register, and energy summaries by day, week or month. Pushed over WebSocket, rendered in the site's own timezone, exportable to CSV or PNG.

Peak shaving & EMS

A safety-gated dispatch loop holds the site under its demand target, with a tariff-aware predictive controller that plans against forecast PV and load — and reports both projected and realised cost avoidance.

PV curtailment

Inverter-agnostic active power, reactive power and power-factor setpoints with a heartbeat re-assert, so a curtailment order survives a comms blip instead of silently reverting.

DER economics

Size a system before it exists. Model PV, storage or both against real interval usage and a versioned utility tariff, then optimise for net present value with IRR, payback and LCOE over system life.

Architecture

Five layers, each isolated from the next.

Field devices are noisy, networks are unreliable and databases should never face the internet. The platform is built around those three facts.

  1. 01

    Edge node

    A hardened single-board computer polls every inverter, meter, battery and weather station on site, and reports its own health as a device in its own right.

  2. 02

    Buffer & upload

    Timestamped, self-describing log files are compressed and pushed over HTTPS. Polling and uploading are fully decoupled loops — neither can stall the other.

  3. 03

    Ingest

    Uploads are authenticated by hardware identity and validated against a driver registry, so a firmware or map change can never quietly corrupt historical decoding.

  4. 04

    Time-series store

    Raw values land in a dedicated TimescaleDB server on a private network. No derived metrics are stored — history stays exactly as the device reported it.

  5. 05

    Application

    A React dashboard and REST API resolve raw positions into named metrics at query time, behind role- and site-scoped access control.

Managed over a private tunnel, not an open port.

Remote access runs across a self-hosted WireGuard VPN, entirely separate from the data upload channel. The two have independent failure domains — losing one never takes out the other, and the time-series database is never internet-facing.

Beyond monitoring

Dashboards report the bill. Control changes it.

Demand charges are set by a handful of minutes each month. MWatt.io watches the grid meter continuously and moves the battery before those minutes happen.

  • Safety gates first. Dispatch is blocked on stale comms, missing grid measurement or an out-of-range battery state — never on optimism.
  • Predictive, not just reactive. Forecast PV and load feed a rate-aware optimiser that plans the day rather than chasing it.
  • Tariff-accurate. Versioned rate schedules with effective dates, so a site keeps the tariff it was actually billed under.
  • Local first. An on-site web UI runs on the edge node itself — commission, configure and observe a site with no cloud connection at all.

Supported equipment

Vendor-neutral by design.

Devices are described by declarative register maps, not hard-coded into the software. Supporting a new model means writing a map — no dashboard rebuild, no API change.

Solar inverterYaskawa Solectria PVI 50/60TLModbus RTU
Solar inverterSolarEdge 3-Phase (SunSpec)Modbus TCP
Battery storageTesla Megapack 2XLModbus TCP
Hybrid inverterSol-Ark 30K / 60KModbus RTU
Revenue & PQ meterSEL-735Modbus RTU / TCP
Power meterAccuenergy Acuvim IIModbus RTU
Charge controllerVictron SmartSolar MPPTModbus TCP
Weather stationRainWise PVMet-150Modbus RTU

Your site is already talking. Start listening.

Existing customers can sign in to view live site performance, history and controls.

Login

Dashboard hosted at mwatt.io/app