Hybrid RF and powerline communications for smart metering
Neither RF nor powerline reaches every meter. The interesting engineering is in carrying both on one board without either ruining the other.
Outcome A single board carrying both communication paths, with the mains coupling and the radio front-end designed so neither degrades the other.
Diagram: dual-path topology, RF and powerline on one board, with the isolation boundary marked. Original vector artwork.
Every deployed meter that cannot be reached is a truck roll, and a truck roll costs more than the meter. That single economic fact drives the whole design.
Why both links, on one board
Neither path is reliable everywhere.
Powerline carrier works where radio is blocked — deep in a building, in a basement, behind reinforced concrete. It fails where the line is electrically noisy, where the meter sits on a different phase from the concentrator, or where a transformer stands in the way.
Sub-GHz RF works where the line is hostile. It fails where the structure is.
Deployed at scale, both failure modes appear in the same estate. A hybrid gives each meter a fallback, which turns a communications problem into an installed-base problem you can actually manage remotely.
Where the difficulty is
Two front ends that dislike each other. A switching mains coupling network and a sensitive sub-GHz receiver on one board is a self-interference problem by construction. Physical separation, careful grounding, filtering on every boundary crossing, and a floorplan that treats the RF section as a keep-out — decided in the block diagram, not discovered at pre-compliance.
The mains side is a safety problem before it is a signal problem. The coupling network sits across live conductors. Isolation, creepage and clearance, surge withstand and fusing come first; the coupling behaviour comes second.
Measurement you can trust. Hall-effect current sensing gives isolation by construction, and its usefulness depends entirely on the analogue path around it — reference stability, layout away from switching nodes, and a sampling strategy that survives a noisy line.
Losing power gracefully. A meter that dies silently is a meter that has to be visited. Enough hold-up energy plus an early warning signal lets the firmware send a last message and close its filesystem cleanly. Dying-gasp is a hardware feature with an energy budget, not a firmware flag.
| Subsystem | Constraint it imposes |
|---|---|
| Powerline coupling | Isolation and clearance to mains; surge withstand; filtering |
| Sub-GHz RF | 50 Ω feed, keep-outs on every layer, ground stitching, duty cycle |
| Coexistence | Physical separation and floorplan, decided at architecture |
| Current sensing | Analogue path away from switching nodes, stable reference |
| Dying-gasp | Hold-up energy sized against the longest realistic write |
| EMC | An emitter and a receiver on one board, in a chamber, together |
The part that is easy to underestimate
EMC. A board that deliberately injects signal onto the mains and simultaneously listens on a sensitive radio has to pass a chamber session with both active.
That is not a test you tune your way through afterwards. Return paths, plane strategy, filtering at every boundary and the physical separation between the two front ends all have to be right in the first layout — which is why the EMC work on a design like this starts at the block diagram.
Further images to produce
These slots are laid out and waiting for assets. Each carries its own
brief — see shotList in this project’s source file.
Coverage illustration: where RF fails (building structure) versus where PLC fails (noisy or wrongly-phased line), showing why the hybrid exists. This is the argument in one image.
Board photograph or render, with the RF section and the mains coupling network visibly separated.