Edge-to-converter pulse survivability
The pulse is faster than any switch. So we stopped trying to catch it.
Keraunophylax protects grid-tied power-electronic converters from the early-time (E1) high-altitude electromagnetic pulse — the transient whose leading edge outruns every clamp, relay, and controller built to stop it.
E1 rise ≤ ~1 ns · the grid is now electronics · defend the converter, not just the transformer
The grid became electronics. Its protection did not.
Decades of pulse hardening went into the bulk transformer. Meanwhile, generation, storage, and load moved behind inverters — and the inverter is exactly what E1 destroys first.
The edge outruns the clamp
E1 rises in under a nanosecond. Varistors react in tens of nanoseconds; spark gaps in microseconds. Most of the energy is past before any active device conducts.
Built for the transformer
Neutral-blocking devices guard high-voltage transformers against the slow E3 component. They do nothing for the gate drivers, controllers, and sensors inside a converter.
The converter kills itself
A pulse-corrupted controller can command a false turn-on. Two series switches conduct at once — shoot-through — and the bridge destroys itself from the inside. Terminal surge protection never sees it.
Three pulses, ten orders of magnitude, one weak point.
The figures below are not ours — they are the public HEMP environment, characterized in IEC 61000-2-9 and the MIL-STD-188-125 protection standards. They explain why the converter, not the transformer, fails first.
| Component | Time scale | Field strength | Induced cable current* | First to fail |
|---|---|---|---|---|
| E1 · early | ~2.5 ns rise · ~23 ns wide | ~50 kV/m radiated | ~2.5 kA | Semiconductors, gate drivers, controllers, sensing |
| E2 · intermediate | ~1 µs – 1 s | ~100 V/m | ~250 A | Lightning-class — conventional arresters mostly cope |
| E3 · late | ~1 – 100+ s | ~40 V/km | ~1 kA | Long lines; transformer-core saturation (GMD-like) |
*Representative MIL-STD-188-125 pulsed-current-injection values; real coupling depends on geometry. E1 is the one that kills power electronics: it arrives and is gone before a lightning arrester begins to conduct.
The exposed surface is growing faster than the protection for it.
The grid is electronifying. Solar, storage, EV charging, and grid-forming inverters put an ever-larger share of the network behind semiconductor switches rather than electromechanical apparatus.
Power electronics are the soft target. Gate drivers, digital controllers, and sensing front ends fail at field strengths far below what a transformer shrugs off — and they sit at the distribution edge, in the open, by the thousands.
The standards lag the hardware. Mandated hardening centers on bulk transmission transformers. There is no settled, converter-level protection requirement, and therefore no incumbent owning the method.
Hardening is cheaper than replacement. A converter parked safe and re-synchronized is back in minutes; a destroyed bridge — or a fleet of them — is a supply-chain event.
A book-to-physical grid that still ties at the substation can hide a distribution edge that no longer exists.— the exposure Keraunophylax addresses
One event. Three components. Five moves.
A high-altitude pulse is not one transient but three, spread across ten orders of magnitude in time. Conventional protection arrives after the part that matters. Keraunophylax divides the labor across the whole event.
Time axis logarithmic, illustrative. The leading edge is conceded to passive means at the boundary; everything an active system can reach is handled in sequence.
One protected region. One way in.
The converter’s vulnerable electronics live inside a continuous shield. Every conductor enters at a single bonded point, treated by a staged passive front end — and the interlock reaches the gate drivers on a hardware path the controller cannot touch.
Gold path = controller-independent. The safety command never traverses firmware.
A division of labor across the event.
No active element can intercept a sub-nanosecond edge — so we don’t ask one to. Each part of the pulse is met by the means suited to its time scale.
Concede the edge
The converter’s electronics sit inside a continuous shield. Every conductor enters at a single bonded point through a staged passive front end — bulk diverter, lossy magnetic, low-capacitance fast clamp — that attenuates the leading edge with no trigger and no latency.
See it before the controller does
A field-derivative sensor reads the pulse’s rate of change and discriminates a HEMP-class event from lightning and switching transients by rise time and spectral content — so it fires for the real thing and ignores the rest.
Force the bridge safe core claim
On detection, a hardware interlock drives the gate drivers to a non-conducting state — overriding the digital controller, in hardware, on a path that never touches firmware. Even if the pulse has corrupted the controller, the converter cannot be commanded into shoot-through.
Hold through the storm
The communication interface is galvanically isolated and the converter rides through on DC-link or reserve energy while the E2 component and the seconds-long E3 tail pass.
Come back, or stay down
Once the slow component subsides, a validated self-diagnostic runs. The converter re-synchronizes to the grid under anti-islanding rules only if it passes — otherwise it holds safe and reports the fault.
Watch a corrupted controller try to destroy the bridge.
A half-bridge has two switches in series across the DC link. If both ever conduct at once, the link shorts through them — shoot-through. An E1 pulse can corrupt the controller into commanding exactly that. Inject the pulse, then try it with the interlock on.
Interlock off: the corrupted controller commands both switches on → shoot-through. Interlock on: detection forces the gates off in hardware before the controller can act.
Survivability, not just suppression.
Protection scoped to the gate-drive, control, sensing, and communication stack — the part of the modern grid the transformer-era corpus leaves exposed.
The sub-nanosecond edge is handled by elements that need no triggering, instead of competing on clamp speed against a transient nothing can outrun.
A controller-independent path forces the bridge off, so a disrupted controller can’t drive the converter into shoot-through. No firmware in the safety loop.
Rise-time and spectral discrimination avoid nuisance trips; a validated self-diagnostic governs automatic, standards-compliant reconnection.
Against what protects the grid today.
Existing protection is real and effective — for what it was built for. The gap is the converter, the sub-nanosecond edge, and self-inflicted destruction.
| Surge suppressor (MOV / TVSS) | Transformer neutral blocker | Keraunophylax | |
|---|---|---|---|
| Primarily protects | Terminal equipment | HV transformer core | Converter control & gate stack |
| Component addressed | E2 / lightning | E3 / GMD (slow) | E1 (fast) + E2 + E3 tail |
| Sub-nanosecond edge | ✗ too slow | ✗ n/a | ✓ conceded to passive |
| Prevents shoot-through self-destruction | ✗ | ✗ | ✓ hardware interlock |
| Automatic safe restoration | ✗ | – stays in place | ✓ validated re-sync |
| Nuisance-trip discrimination | – clamps all | – | ✓ rise-time + spectral |
Complementary, not competitive: a neutral blocker still guards the bulk transformer; Keraunophylax guards the converters the blocker never saw.
Every standard protects something. None of them the converter.
HEMP protection is well-standardized for what it was written to defend: hardened facilities, command-and-control, long-haul comms, and bulk transformers. The distributed grid-tied converter falls in the gap between them.
Defines the radiated HEMP environment — the E1/E2/E3 waveforms themselves. It describes the threat; it protects nothing.
HEMP hardening for fixed and transportable ground facilities, with the pulsed-current-injection test method. Facility-scale Faraday and filter practice — not an edge inverter.
HEMP immunity test methods for equipment. A framework Keraunophylax can qualify against — not a converter-protection mandate.
EMI emissions and radiated susceptibility (RS103). Broad equipment EMC — silent on the shoot-through self-destruction mode.
HEMP vulnerability of telecom systems. Closest in spirit, but aimed at communications plant, not power conversion.
GMD (E3-like) vulnerability for the bulk system — the transformer-neutral world. Says nothing about E1 at the converter.
The gap is the point: no settled standard mandates E1 protection at the grid-tied converter — so no incumbent owns the method.
Wherever the grid runs on switches.
From claim to qualified hardware.
An honest path, not a promise. Each stage is gated by the one before it, and performance becomes real only at the testing stages — everything earlier is design intent.
- Now
Provisional filing
Thirty-nine claims drafted; provisional in preparation. Establishes priority on the controller-independent interlock.
- +12 months
Non-provisional & FTO
Full utility application with formal drawings, after a freedom-to-operate scan against gate-driver fault-blanking and EMP-detection art.
- Prototype
Interlock board
A bench build of the detector plus hardware interlock, tapped to a representative gate-drive stage. The first measured response times.
- Qualify
Pulsed-current injection
Inject MIL-STD-188-125-class E1 pulses; measure residual let-through and confirm the bridge holds off under a corrupted-controller stimulus.
- Pilot
Converter in the loop
A single PV, storage, or EV-charging converter retrofitted and evaluated with a partner where survivability matters.
Four independent claims, one anchor.
Why this files cleanly
The invention is hardware. §101 eligibility is not at issue — there is no abstract-idea exposure, and the design deliberately uses no machine learning. The work the patent does is on novelty.
The anchor is the controller-independent hardware interlock: the distinction from gate-driver fault-blanking art is that the trigger is an external pulse detector and the path overrides the controller specifically when the controller is the thing being disrupted.
- IND 1Protective systemShielded region, single-point passive front end, sub-nanosecond detector, and controller-independent safe-state command.
- IND 2Interlock apparatusThe hardware interlock as a standalone, retrofittable unit tapped to a gate-enable path.
- IND 3MethodConcede the edge passively, detect, force safe in hardware, isolate, ride through, validate, re-sync.
- IND 4Fleet systemShared detection with a fail-safe all-clear heartbeat across a plurality of converters.
What people ask first.
Does it use machine learning?
Won’t it trip on every lightning strike?
Does it protect the power semiconductors from direct damage too?
What about the slow E3 / geomagnetic component?
Retrofit or built in?
Is it patent-pending?
The vocabulary, briefly.
- HEMP
- High-altitude electromagnetic pulse — the radiated field from a high-altitude nuclear detonation, split into E1, E2, and E3.
- E1
- The early-time component: a sub-nanosecond-rise, ~50 kV/m field that couples into conductors and destroys electronics. The one this protects against.
- Shoot-through
- Two series switches in a bridge conducting at once, shorting the DC link through them. The self-destruction mode the interlock blocks.
- Gate driver
- The circuit that switches a power semiconductor on command from the controller — the path a corrupted controller would use to cause shoot-through.
- DC link
- The capacitive energy store between a converter’s input and its bridge; also the ride-through reserve.
- GIC / GMD
- Geomagnetically induced current / disturbance — the slow, E3-like effect that saturates large transformers.
- PCI
- Pulsed-current injection — the standardized way to test HEMP protection by driving a defined pulse into the front end and measuring let-through.
- Anti-islanding
- The interconnection rule that a converter must not energize a dead grid; it governs how the unit may re-synchronize after an event.
Patent-stage, and open to the right conversations.
Keraunophylax is a patent-stage technology. A provisional application is in preparation; we are scoping licensing discussions and bench-stage pilots now.
Two ways to start.
License the technology
Building in the inverter, storage-conversion, EV-charging, or grid-protection space? The interlock and protective-interface method are available to discuss.
Open a licensing discussionRun a pilot
Operate or test grid-tied converters where pulse survivability matters? Let’s scope a bench-stage evaluation on one converter.
Start a pilot conversation