GridSentinel
A battery-free, clamp-on sensor that turns the electricity grid edge into a live stream of outage and fault data. It harvests its own power from the line it watches.
Validated concept, specified through Phase 0. A battery-free clamp-on grid sensor, re-scoped during technical discovery from theft detection (the physics don't close on medium voltage) to outage detection and fault localization, which utilities actually want. ~CAD $95 per node.
The problem
Utilities find out about outages from phone calls
The distribution grid edge (the actual conductors carrying power to homes and businesses) is almost entirely unmonitored. When a section goes down, utilities typically learn about it from customer calls, not sensors, and crews patrol blind to find the fault.
- No plug, no accessible power source at most points on the line, so any sensor needs to power itself.
- Storm restoration is slow and risky without knowing which spans are damaged or safe to re-energize.
- Electricity theft is a real cost, but the physics of proving it only close at certain points on the network.
Repositioning
The headline feature doesn't close the math on this line
Electricity-theft detection was the original pitch, but on a medium-voltage feeder, the energy-balance equation needed to prove theft doesn't close; it only works at the low-voltage transformer secondary. Rather than ship a headline feature that can't work where it was pitched, I re-scoped the first product around what a utility ops team actively wants and has no incentive to refuse: automatic outage detection and fault localization. Theft detection moves to a planned LV variant, where the physics actually work.
Discovery
Who has to say yes
Utility operations · Primary
Wants faster outage detection and fault localization without new truck-rolls or infrastructure to maintain: the MVN's entire value case.
Revenue protection · Later
Cares about theft detection, which is served by the LV/transformer-secondary variant once the MVN has proven itself in the field.
Backlog
User stories that shaped the wedge
Requirements
Spec'd as testable acceptance criteria
My role
Finding the wedge the physics actually support
I pressure-tested the original theft-detection pitch against the underlying electrical engineering, found it didn't close on a medium-voltage line, and re-sequenced the roadmap around a wedge utilities would adopt immediately. From there I specified the full hardware architecture, bill of materials, and phased validation plan.
What I designed
A node that powers, senses, and reports itself
Energy harvesting
Split-core CT → surge protection → harvest PMIC → supercapacitor + Li/SOCl₂ backup, feeding a regulated 3.3V rail with no plug.
Sub-GHz mesh comms
Wi-SUN self-healing mesh, pole-to-pole, backhauled to a gateway over cellular or fiber.
Self-orienting physical design
Clamps on as a gravity pendulum: it self-orients vertically, keeps the antenna clear of the wire, hot-stick installable.
Detection logic
Outage, fault-localization, and storm-restoration logic, all running on the same MVN hardware, with no separate device for each capability.
Under the hood
From current collapse to a crew dispatched to the right pole
Outcome
A de-risked platform, staged for a bench prototype
GridSentinel is fully specified through Phase 0: architecture, BOM (~CAD $95/node electronics), and a de-risked product sequence that earns a place on the line with a wedge utilities want, rather than leading with the unsolved theft-detection problem. The dominant cost and risk, correctly identified upfront, is the custom HV enclosure and installation engineering, not the electronics.
Roadmap
Bench → pilot → platform
Bench prototype: prove harvesting, last-gasp, mesh, and fault-surge detection.
Single-feeder pilot: validate survival, outage detection, fault localization, and install time in the field.
LV theft-detection node and storm restoration, then arc/wildfire, dynamic line rating, and power quality as firmware on the installed base.
More case studies
Want a discovery sprint or a prototype like this one for your product? Book a call.