The phrase “in a non-steady system” in a converter title is doing real work. US10916991B2, granted to Mainspring Energy on February 9, 2021, is a DC-DC converter patent, but claim 1 binds it to a context where the input is not a steady DC bus — it is the pulsing output of a reciprocating generator.

The CPC tags confirm the coupling. Alongside the converter classes (H02M 3/06, H02M 3/1582) sit H02K 7/1884 (generator coupled to a reciprocating engine), F02B 71/04 (free-piston engines), and H02P 9/04 (generator control). This is a power-electronics claim that only makes sense attached to Mainspring’s linear-generator engine.

“Multiphase electromagnetic machines, such as free-piston engines or compressors, may require, or supply, a pulsed power profile from or to a DC bus, respectively. The pulsed power profile may include relatively large fluctuations in instantaneous power.”— U.S. Patent No. 10,916,991 source

Claim 1 itself is not a bare converter claim — it is a power management system, and reading its limitations shows why the “non-steady” framing is structural rather than cosmetic. The claim recites a split DC bus (“a first bus at a first voltage and a second bus at a second voltage lower than the first voltage”); an inverter containing a first DC-DC converter that “output[s] a smoothed power output relative to a power output from a free-piston power generator”; at least one energy-storage device on the DC bus that stores and releases energy; and a power-electronics module containing a second DC-DC converter coupled both to the DC bus and to the free-piston generator, configured to receive the generator’s output. The architecture is purpose-built to absorb a pulsing source: storage buffers the energy swings, the second converter interfaces the generator, and the inverter delivers a smoothed result.

The dependent claims make the source explicit and the buffering concrete. Claim 3 specifies that the free-piston generator “comprises at least one linear electromagnetic machine (LEM)” — Mainspring’s actual hardware. Claim 2 adds circuitry that “control[s] a plurality of currents in a plurality of respective phases” of that generator, i.e. multiphase commutation of a linear machine. Claim 5 makes the energy-storage device a capacitor (fast, high-power buffering for instantaneous swings rather than bulk energy). Claims 6 and 8 have the inverter regulate the DC-bus voltage — the explicit job of keeping the bus stiff while the source pulses. Claims 9 and 10 add a control system that reads status, voltage, current, and temperature off the converter. There is also a mirror-image independent claim 11: the same topology run in reverse, where the power-electronics module “provide[s] a pulsed power input to the free-piston machine” — motoring the piston rather than generating from it. A free-piston engine has no crankshaft to spin it up, so the converter that smooths its output is also the converter that has to drive it; claim 11 fences both directions.

Claim construction matters here because the “non-steady” limitation both narrows and strengthens the patent. A general DC-DC converter claim would drown in prior art; a two-converter, storage-buffered, bus-regulating power-management system specifically adapted to source from — and motor — a multiphase free-piston linear generator is novel precisely because the context is unusual. The fence is small but well-defended, and the bidirectional claim 11 closes the obvious escape of describing the same hardware as a starter drive.

Why a grid-IP desk cares: Mainspring’s linear generator is pitched as a flexible, fuel-agnostic distributed-generation asset, and the power electronics that turn its non-steady output into grid-usable power — and that crank it to life — are core IP. The split-bus, capacitor-buffered, inverter-regulated topology is the moat around the architecture, not a standalone converter business.

The caution, as always: claim 1 fences off a power-management system for this generator, not converters generally and not the engine itself. Anyone building steady-state DC-DC converters, or buffering a conventional rotating generator, is untouched. The patent’s value is entirely a function of how essential the free-piston-plus-storage topology proves to be — the IP is bolted to the engine it serves.

It is worth being precise about why a free-piston machine forces this topology, because that is what gives the claim its grounding. A conventional rotating generator delivers a roughly constant torque and a smooth electrical output; a free-piston linear generator does not — the translator reverses direction at the end of every stroke, so its electromotive force passes through zero twice per cycle and its instantaneous power pulses hard, exactly the “relatively large fluctuations in instantaneous power” the abstract names. Feeding that directly to a grid inverter would push those pulsations onto the bus and, ultimately, the grid. The claimed architecture intercepts the problem in two places: the second DC-DC converter (claim 1) in the power-electronics module shapes the generator-side exchange, while the capacitor energy-storage element (claim 5) absorbs the stroke-to-stroke energy swing, and the bus-regulating inverter (claims 6 and 8) holds the DC bus stiff so the output is “smoothed.” The multiphase current control of claim 2, applied to the linear electromagnetic machine of claim 3, is what lets the converter commute the machine cleanly through those reversals. And the mirror-image motoring path of claim 11 is not an afterthought — a free-piston engine cannot bootstrap itself, so the same electronics must be able to drive the translator to start combustion. Reading the claim set as a whole, it is less a converter patent than a complete bidirectional power-management architecture for a machine that, by construction, refuses to produce steady power. That specificity is the moat, and also its limit: it protects Mainspring’s architecture precisely because almost nobody else builds a generator that needs it.