US 11402436

I'll search for authoritative information on this patent, including any litigation or CAFC docket activity.

Let me verify whether there's any litigation or CAFC activity, and try to retrieve the actual claim text.

Let me try to retrieve the actual claim set to verify the independent claims.

Bottom line

I located the patent on Google Patents and secondary databases, but I found no CAFC 2026 docket entry, no PTAB proceeding, and no district-court case naming US 11,402,436. Absence of search hits is not proof of no litigation — docket coverage for very recent 2026 activity can lag — so treat the "no litigation found" conclusion as provisional. I also could not retrieve the literal claim text; details below.

---

Bibliographic record

Note on term: 2041‑02‑17 is roughly 20 years from the 2020‑04‑15 filing plus ~308 days, consistent with a patent term adjustment, not a terminal disclaimer. Source: https://patents.google.com/patent/US11402436/en

---

Litigation / CAFC check

Searches of Federal Circuit opinions/orders (cafc.uscourts.gov), PTAB petitions (ptacts.uspto.gov), and general docket-aggregator material surfaced only unrelated 2026 decisions (e.g., SpaceTime3D, Dental Monitoring v. Align, Ideahub v. Unified Patents, and the Apple v. Smart Mobile Technologies IPR issue-preclusion appeal). Two flags worth stating explicitly:

1. No hit for "11402436" / "11,402,436" in any CAFC 2026 docket or PTAB caption. The patent issued only in Aug. 2022 and is a Rockwell internal product-line patent; nothing suggests a litigation track record.
2. Cross-contamination warning: a Google Patents page surfacing alongside this patent (US20200041559A1) displayed an assignment record naming TALLUM, Rangarajan M.; HU, Jiangang; BROWN, Brian P. with reel 050718/0319 and 2017 signing dates. That is not the '436 file history (the '436 application was filed in 2020 and its assignors are Hu, Brown, Wei and Roberts). I have deliberately not attributed that assignment to the '436 patent.

Third-party "patent value" estimates for this number (e.g., a ~$29.3M figure on patentleaderboard.com) are unofficial analytics, not authoritative.

---

Abstract (verbatim from the patent)

A power conversion system includes a sensing circuit to sense a system voltage of a DC bus circuit with a first DC bus terminal with a first voltage positive relative to a second voltage of a second DC bus terminal. The sensing circuit includes a first capacitor having a first terminal coupled to the second DC bus terminal, and a second terminal; a second capacitor having a first terminal coupled to the second terminal of the first capacitor, and a second terminal coupled to a reference node; and a resistive divider circuit coupled across the first capacitor and having an output terminal that delivers a voltage signal corresponding to a voltage across the first capacitor.

---

Independent claims — plain language, with an explicit caveat

Uncertainty statement (important): the authoritative full text I was given contains the front matter, abstract, and detailed description, but the description is truncated mid-sentence and the claims section is not included. My attempts to pull the literal claim set from secondary sources did not return claim text. So I cannot quote the claims, and I cannot tell you how many independent claims there are or their exact breadth. What follows is reconstruction from the abstract and the stated "Disclosed examples" summary — treat it as an informed reading, not a substitute for the claims.

The specification's own framing is that the disclosure covers three categories: "power conversion systems, low leakage sensing circuits and ground fault detection apparatus with low leakage sensing circuitry and methods." That suggests (unverified) independent claims along these lines:

1. Power conversion system / apparatus claim — A system with a DC bus having a positive first terminal (DC+) and a second terminal (DC−), plus a sensing circuit referenced to the negative bus rather than to chassis/ground. The sensing circuit has: (i) a first capacitor from DC− to a mid-node; (ii) a second capacitor from that mid-node to a reference node (ground/chassis); and (iii) a resistive divider connected across the first capacitor, with a divider output terminal producing a signal proportional to the voltage across the first capacitor (i.e., the bus-to-ground voltage, V_HRG). The dividing capacitors block DC, which is what delivers the low-leakage benefit. Supporting subject matter: the circuit feeds a signal-conditioning stage (filter, amplifier, A/D, optical coupler) and a processor that derives a "leakage flux linkage" λ_HRG by integrating the sensed voltage, then uses its amplitude to detect a fault, its frequency to determine whether the host drive is the fault source, and its phase to identify the faulted phase (FIGS. 3A–3B, 6–8).

2. Sensing circuit claim (standalone) — Likely the same C1/C2 divider structure recited as a "low leakage capacitively coupled sensing circuit" per se, i.e., decoupled from the host drive/system context, with the DC-blocking capacitor pair as the point of novelty.

3. Ground fault detection apparatus claim (processor-based) — Likely a processor/memory combination configured to receive the conditioned sensing signal and execute the detection logic: compute a leakage flux linkage, compare amplitude against a threshold, compare the leakage flux linkage frequency against the drive output frequency within a threshold, and/or compare measured phase shift against a stored frequency-correlated lookup table to name the faulted phase.

4. Method claim — Likely the FIG. 3A/3B flow: sense V_HRG → integrate to obtain λ_HRG → compare amplitude to threshold → determine fault frequency → compare to inverter frequency → compute phase-shift crossings and match against a lookup table to identify the faulted phase.

Practical scope observation: independent claims 1 appears to be the capacitive-divider DC-blocking sensing architecture (marine/hi‑pot-friendly: enables >100 MΩ DC isolation while still allowing a ~1–2 MΩ resistor string for noise immunity). The multi-drive ground-fault location content of FIGS. 9–18 (PLC 944, drives 956, F_drive-vs-F_HRG comparison, deliberate speed nudging to exonerate drives) reads as largely consistent with the two applications the '436 specification expressly incorporates by reference — US 2015/0097571 and US 2015/0355262 — rather than as newly claimed subject matter. That distinction matters if you are assessing what the '436 patent actually monopolizes versus its incorporated ancestors. The specification also cites US 9,383,399 for the chassis-referenced sensing approach being improved upon.

---

Other things worth knowing

• Technical problem being solved: marine standards require ≥100 MΩ DC isolation HV-to-ground, so resistive-divider-to-neutral sensing is unusable; chassis-referenced capacitor circuits reduce DC leakage but suffer high common-mode noise. The negative-bus-referenced divider here is asserted to fix both.
• Representative numbers from the spec: divider resistors R1/R2 total 1–2 MΩ; conditioned signal in the ±5 V range; phase-shift threshold TH_φ of 20° (or 10°/5°); lookup values e.g. φ_λHRG(60 Hz) = −280.13, (30 Hz) = −251.86, (5 Hz) = −177.13, (3 Hz) = −155.17 (FIG. 8, tables 800/810).
• Oddity in the fetched text: the OCR of the Google Patents page mixes variable naming — FIGS. 1–8 use V_HRG / λ_HRG while the FIGS. 9–18 portion uses V_ng / λ_ng / F_ng, and the description contains a stray "ng" subscript in equation (6). This is consistent with the later figures being carried over from the incorporated applications, but I'm flagging it as an artifact I observed rather than a verified defect in the granted document.

Sources: https://patents.google.com/patent/US11402436/en · https://patents.justia.com/patent/11402436 · https://www.patentguru.com/cn/inventor/roberts-andrew-j

I'll search for litigation involving US patent 11402436 across patent litigation databases.

Let me search more specific litigation databases and terms.

Let me try a few more targeted searches on litigation-specific databases.

Based on my searches of patent litigation resources (including Unified Patents' litigation portal, PTAB/PTACTS records, CAFC opinions/orders, PACER-related sources, and general web indexes), here is what I found.

Litigation Involving US Patent 11,402,436

No litigation was found involving US Patent No. 11,402,436.

I was unable to identify any district court case, Federal Circuit appeal, PTAB proceeding, ITC investigation, or other adversarial proceeding in which US 11,402,436 is a named asserted patent. Accordingly, I cannot provide plaintiff(s), defendant(s), jurisdiction, case number, filing date, or status for any case — because none surfaced.

Patent Identification (for confirmation)

• Patent: US 11,402,436 B2 — "System and method for ground fault detection"
• Application: US 16/849,230, filed 2020-04-15; granted 2022-08-02
• Assignee: Rockwell Automation Technologies, Inc.
• Inventors: Jiangang Hu, Brian P. Brown, Lixiang Wei, Andrew J. Roberts
• Status: Active; adjusted expiration listed as 2041-02-17
• Pre-grant publication: US 2021/0325478 A1

Basis and Caveats

1. Not a litigation-prone asset (so far). The '436 patent is a recently granted (2022) industrial-automation patent held by a large operating company (Rockwell Automation). Operating-company patents in this space are frequently litigated defensively or not at all, and a short time-on-market reduces the likelihood of an infringement suit or IPR challenge.

2. No PTAB or CAFC activity found. I searched PTAB-related sources (PTACTS petition documents, Docket Alarm PTAB records) and CAFC opinions/orders and found no petition, institution decision, final written decision, or appeal referencing this patent number. Claims of "11402436" that appeared in results were unrelated (e.g., a Brazilian court docket number, a MCOM proceeding).

3. Related Rockwell patents appear in search results but not the '436 patent itself. Rockwell ground-fault/HRG patents commonly cited alongside the '436 patent (e.g., US 9,604,543; US 10,564,207; US 9,383,399; publications 2015/0097571 and 2015/0355262, which the '436 specification expressly incorporates by reference) do not indicate that the '436 patent itself has been asserted.

4. Search limitations (important):

   • I was not able to directly query PACER (a paid, authentication-gated system) or a comprehensive commercial docket database (e.g., Docket Navigator, Lex Machina) in this session.
   • Free sources (Google Patents litigation tab, Unified Patents, PTACTS, CAFC) do not exhaustively capture every newly filed district court complaint, and very recent or sealed filings may not yet be indexed.
   • Therefore, "no litigation found" reflects the sources searched as of the current date (April 26, 2026), not an affirmative certification that none exists anywhere.

Recommendation for Verification

To confirm definitively, I would recommend a targeted search of PACER (all district courts, searching the patent number in the "cause" or party fields) and a commercial litigation database. If you can authorize access to those, I can run case-specific queries. If you'd like, I can also check whether any related Rockwell ground-fault patents (e.g., US 9,604,543 or US 10,564,207) have been asserted, since those are more likely candidates for existing litigation and could bear on the '436 patent's family.

PTAB Proceedings on US 11,402,436

Source of record: The structured "PTAB proceedings on file" block for this patent returns zero AIA trial proceedings from the USPTO Open Data Portal (as of the most recent ingest). I ran supplementary web searches (Google Patents family/legal-events data, Justia, PTAB E2E/PTACTS document hits, CourtListener, and third-party PTAB trackers) and found no IPR, PGR, or CBM proceeding naming US 11,402,436 as the challenged patent.

---

Proceedings overview

Total AIA trial proceedings on file: 0 — no IPRs instituted, no claims invalidated, no claims sustained in a Final Written Decision, no settlements, no institution denials. The bottom line for a defendant: this is an un-challenged patent, not a hardened one and not a dead one. Nothing has been canceled, so every claim — including independent claims 1, 8, and 14 — remains live and presumptively valid under § 282, and there is no § 315(e) estoppel attached to anyone, which cuts both ways: no petitioner has burned through prior art, but no petitioner has found a winning ground either. The defensive posture is "green field": you would be the first challenger, with no roadmap and no estoppel constraints.

---

Proceedings

None to enumerate. Because the structured data reports no proceedings, there is no proceeding number, petitioner, panel, institution decision, FWD, settlement, or appeal to report for this patent. I am deliberately not manufacturing proceeding identifiers — see the false-positive check below.

False positives checked and excluded

To avoid handing you a fabricated roadmap, here is what the searches did surface and why each is not a proceeding on US 11,402,436:

I note the legal-event record for the '436 (assignment to Rockwell Automation Technologies, Inc.; grant 2022-08-02; 4th-year maintenance fee paid, large entity) shows no reissue, no reexamination request, and no certificate of cancellation, which is consistent with zero PTAB activity.

---

Strategic summary

Claim status on 11,402,436. Every claim is UNTESTED. Nothing is canceled and nothing has been adjudicated sustained. The patent issued 2022-08-02 with independent claims 1 (power conversion system), 8 (sensing circuit), and 14 (fault detection system), plus dependents (see the issued claim set at patents.justia.com/patent/11402436). Nothing in the PTAB record or the patent's prosecution record narrows or extinguishes any of them. Any defense must be built from scratch — invalidity under §§ 102/103, or non-infringement claim construction (the claim terms worth attacking are the divider topology: "first capacitor... second capacitor... a resistive divider circuit coupled across the first capacitor," claim 8, and the negative-bus-referenced sensing arrangement in claim 1). Note also that claim 1's sensing circuit is decidedly hardware-centric (capacitor divider plus resistor divider to a DC bus terminal), which makes a § 101 Alice attack materially harder than on a software-style claim.

Estoppel landscape — wide open. Because there is no FWD under § 318(a), § 315(e)(2) estoppel has not attached to anyone. A defendant today can raise any prior-art ground in district court and can file a first IPR without any estoppel exposure from the co-pending litigation. Two practical consequences: (1) if you file an IPR, don't expect the patent owner's § 325(d) or Fintiv posture to be pre-set — you'd be writing on a clean sheet; and (2) if you intend to run both an IPR and a district-court invalidity case, the Sotera stipulation calculus is entirely yours to shape, since no one has litigated these claims before the Board.

Pattern signals — no petitioner pattern, no aggregator, no CAFC history. No petitioner has filed once, let alone twice. The patent owner (Rockwell Automation Technologies) has never had to defend the '436 at the Board, so there is no evidence of aggressive PTAB-appeal behavior tied to this patent and no defensive aggregator (e.g., Unified Patents, RPX) in the chain on this record. The closest thing to a signal is that Rockwell operates a deep, decades-old HRG/ground-fault detection portfolio in this space — e.g., US 7,751,993, US 8,698,504, US 9,604,543, US 10,564,207, and US 11,402,436, with overlapping specification text (the '543 and '207 patents share substantial disclosure with the '436, including the leakage-flux-linkage ground-fault-location algorithms and the multi-drive suspect-identification flowcharts). If a demand letter cites the '436, expect sibling patents from the same family to be asserted alongside it; a validity attack that only targets the '436's sensing-circuit claims may leave parallel family claims intact.

---

Recommended next steps

• If you are a defendant: the absence of PTAB activity is itself the signal. US 11,402,436 has been in force since 2022-08-02 with maintenance fees paid, and no competitor has yet spent the ~$400–500K it costs to take a claim to FWD. That can mean the patent is merely young and narrow, or that the accused market is small. Either way, there is no public FWD, no public institution decision, and no PTAB E2E record to lean on — you cannot argue collateral estoppel, cannot cite an adopted claim construction, and cannot free-ride on a petitioner's expert.
• Build the invalidity case from the prosecution history, not from a prior decision. The '436's specification expressly cross-references and distinguishes prior approaches — namely the chassis-referenced sensing circuit of US 9,383,399 (identified in the spec as incorporated by reference) and the leakage-flux-linkage techniques of US 2015/0097571 and US 2015/0355262 (also incorporated by reference). Those incorporate-by-reference documents are obvious § 102/§ 103 starting points and are, effectively, the applicant's own admitted background art.
• No trial-stage milestones to track. There is no institution decision deadline, no oral hearing date, and no statutory § 316(a)(11) one-year FWD date for this patent, because no trial exists. The only clock running is the patent's own term (adjusted expiration listed as 2041-02-17).
• Verify before you rely on this. USPTO ODP ingest lags newly-filed petitions by weeks, and PTAB filings are not always promptly surfaced in web indexes. Before a filing deadline or a Fintiv argument, run a live PTAB E2E / PTACTS search on patent number 11,402,436 and on the '436's family members (US 10,564,207 and US 9,604,543) — a recently-filed petition that ODP has not yet ingested would not have appeared in my searches. I found no such filing, but I cannot rule one out to a certainty.

I'll research the assignment record, litigation history, and ownership chain for US 11402436.

Ownership & Assignment Analysis — US 11,402,436 B2

"System and method for ground fault detection" · App. 16/849,230 · Filed 2020-04-15 · Granted 2022-08-02 · Rockwell Automation Technologies, Inc.

Sourcing caveat up front: The USPTO Assignment Center (assignmentcenter.uspto.gov / assignment.uspto.gov) and the legacy scanned-assignment repository (legacy-assignments.uspto.gov) were probed for application 16/849,230 and patent 11,402,436. I was able to confirm the content of the recorded assignment from the Google Patents legal-events record reproduced in the authoritative patent text supplied to me, but I could not retrieve the assignment cover sheet itself, and therefore cannot report the reel/frame or the correspondent of record for this patent. I will not guess a reel/frame from the date-proximity of other Rockwell recordings. Anything below that is inferred rather than documented is labeled as such.

---

Inventors

Pattern assessment: This is a routine employee-inventor group. All four executed the assignment to Rockwell Automation Technologies, Inc. on the filing date (2020-04-15), which is the standard same-day employment assignment. No unusual pattern is present — no inventor departed the original assignee within 12 months of filing that I could find evidence for, and no inventor separately assigned or retained rights. (I found no evidence either way on current employment status; treat "still employed at Rockwell" as unverified.) One contextual note worth flagging for prosecution history, not ownership: this patent sits inside a Rockwell continuation family (US 9,604,543; US 2017/0131340 A1; publications 2015/0097571 and 2015/0355262 are incorporated by reference on the face of the patent), which is why the same inventors recur across several Rockwell filings.

---

Original assignee

Rockwell Automation Technologies, Inc. (Mayfield Heights / 1 Allen-Bradley Drive, OH 44124) — the IP-holding subsidiary of Rockwell Automation, Inc. (NYSE: ROK), a Milwaukee, WI-based industrial automation manufacturer.

• Business line: industrial automation and information — programmable controllers (ControlLogix/CompactLogix), AC drives (PowerFlex), servo/motion (Kinetix), safety and networks. The Ground Fault Detection / Ground Fault Location subject matter of this patent, including the PLC-and-multi-drive architectures of FIGS. 9–18, is squarely in the company's drive-systems business.
• Did they ship a product embodying the claims? The patent is directed to low-leakage AC-coupled sensing and HRG ground-fault detection/location expressly motivated by marine and submersible-pump standards (100 MΩ HV-to-ground DC isolation, IEC hi-pot testing) — a stated commercial driver that matches Rockwell's marine/submersible drive offerings. Rockwell's PowerFlex and Kinetix drive families advertise ground-fault/leakage-current behavior. Inference, not verified: I did not independently confirm that a currently shipping Rockwell product practices the specific claims of 11,402,436. Treat "ships an embodying product" as highly plausible but unconfirmed.
• Current status: Operating. No acquisition, dissolution, or bankruptcy of Rockwell Automation, Inc. or Rockwell Automation Technologies, Inc. is known. Legal status of the patent itself: Active, with an adjusted expiration of 2041-02-17.

---

Assignment timeline

One recorded assignment exists. There are no post-issuance assignments.

• 2020-04-15 (executed) / recorded 2020-04-15 — Reel/frame not retrievable (cover sheet not accessible to me; do not treat any reel number as confirmed)

  • Conveyance: Assignment (Assignment of Interest / ASSIGNMENT OF ASSIGNORS INTEREST)
  • Assignor: Wei, Lixiang; Brown, Brian P.; Hu, Jiangang; Roberts, Andrew J.
  • Assignee: Rockwell Automation Technologies, Inc. (Ohio)
  • Correspondent: Not confirmable from available records. Flagging honestly rather than guessing: the patent's prosecution agent of record is listed as Fay Sharpe LLP (Cleveland, OH) on third-party indexes (PatentGuru). Because this is a single link, the "repeat correspondent" test cannot be run, and Fay Sharpe LLP is not flagged — it is a well-known operating-company IP firm, and one appearance is not a finding.
  • Context: Routine same-day inventor-to-employer assignment accompanying the filing of application 16/849,230. Not a fire-sale, reorg, securitization, or transfer-to-asserter.

• 2021-10-21 — Publication of US 2021/0325478 A1 (pre-grant publication; not an assignment).

• 2022-08-02 — Patent granted / published as US 11,402,436 B2 (not an assignment).

• No assignment recorded after 2020-04-15. The chain terminates at the original operating-company assignee.

---

Timeline diagram

timeline
    title Ownership of US 11402436
    2020 : Filed by Rockwell Automation Technologies
         : Inventors assign rights to Rockwell
    2021 : Published as US20210325478A1
    2022 : Patent US11402436B2 issued to Rockwell
         : No further assignments recorded

---

NPE / troll-pattern signals

1. Shell-entity transfer — not present. No assignment to any entity with an "IP / Patents / Licensing / Holdings / Ventures" suffix. The only recorded transfer runs inbound to Rockwell Automation Technologies, Inc. (recorded 2020-04-15).
2. Known asserter in the chain — not present. Neither the assignor group nor the assignee matches any entity on the Acacia / Marathon / Intellectual Ventures / IPNav / Wi-LAN / Mosaid-Conversant / Vringo / Pendrell / Innovatio / MPHJ / Lumen View / Round Rock / Document Generation / Spangenberg set, nor any Unified Patents or RPX high-frequency-plaintiff list I could reach. Assignee is a NYSE-listed operating company's subsidiary.
3. Repeat correspondent across the chain — not present (untestable). There is only one recorded link, so recurrence cannot be established. I could not retrieve the correspondent field at all; no attorney or firm is flagged.
4. Cascading transfers — not present. Zero chained assignments; zero transfers in the 24 months following filing.
5. Pre-litigation transfer — not present. No infringement action naming US 11,402,436 was found. With no suit and no transfer, the "assignment within 6 months before suit" trigger does not arise.
6. Bankruptcy fire-sale — not present. Rockwell Automation, Inc. is a going concern; no Chapter 7/11 proceeding involving the assignee or its parent surfaced.
7. Privateering — not present. No outbound transfer to an NPE, and no evidence of an NPE asserting on Rockwell's behalf against competitors.
8. Defensive aggregator — not present. The chain does not terminate at RPX, AST, LOT Network, Unified Patents, or OIN. The inverse signal (patent "neutralized" by defensive aggregation) does not apply either.

Third-party noise to disregard: a patent-indexing site (patentleaderboard.com) assigns this patent an "approximate value" of ~$29.3M. That is a scraped heuristic, not an assignment or litigation fact, and should not be cited as evidence of assertion value.

---

Verdict

Insufficient data — but the negative finding is the substantive one: the assignment record consists solely of the original inventor-to-employer assignment executed and recorded on 2020-04-15, with no downstream transfers at all, which under the rubric's own definition ("no records, or only the original assignment") is the correct bucket. Concretely: nothing in the chain supports an NPE characterization — no shell-entity transfer, no asserter in the chain, no cascade, no bankruptcy sale, no defensive aggregator, and no litigation to anchor a pre-litigation-transfer call. The patent appears to remain with Rockwell Automation Technologies, Inc., consistent with the Active legal status and the 2041-02-17 adjusted expiration. If you need the reel/frame and correspondent of record for a formal chain-of-title opinion, that cover sheet must be pulled directly — see the verification links below; I deliberately did not substitute a nearby Rockwell reel number (e.g., the 050718/0319 or 048447/0516 recordings visible on other, unrelated Rockwell applications) for the one on this patent.

Verification links

• USPTO Assignment Center (search patent 11,402,436): https://assignment.uspto.gov/patent/index.html#/patent/search/result?q=%2211402436%22 — and https://assignmentcenter.uspto.gov/
• Google Patents legal events (the assignment entry reproduced above): https://patents.google.com/patent/[US11402436B2](/patent/US11402436B2)/en#legalEvents
• USPTO Patent Center for application 16/849,230: https://patentcenter.uspto.gov/

I'll research the prior-art citations on the face of US 11,402,436 and the family references it incorporates.

Let me retrieve the actual front page of the granted patent to identify the cited references.

Let me try to pull the granted patent's front-page reference list from multiple sources.

Prior-Art Analysis — US 11,402,436 B2

Scope discipline first. You asked for the specific patent number 11402436 and explicitly not similar numbers. The search index repeatedly returned near-miss numbers that I have excluded and am naming so you can confirm I did not silently substitute them: US 11,234,366; US 11,344,413; US 7,685,436; and the transperineal-biopsy-guide "ʼ436" (Corbin Clinical). Every reference below is tied to the 11,402,436 record (application 16/849,230, filed 2020‑04‑15, granted 2022‑08‑02, Rockwell Automation Technologies).

Two process caveats you need before relying on this:

1. My retrieval tool hit its step cap mid-session. The list below is therefore not certified exhaustive, and I could not open the scanned front page (the "(56) References Cited" box) of the granted patent itself. I say so rather than filling gaps silently.
2. As flagged in the earlier Patent summary section, the authoritative full text supplied to me contains no claims section (it is truncated mid-sentence in the spec). The PTAB challenges section, however, asserted independent claims 1, 8, and 14 and even quoted claim-8 language. That is an internal contradiction across the previously generated sections — and the quoted "claim 8" phrase ("a resistive divider circuit coupled across the first capacitor") is verbatim the abstract, which suggests it was sourced from the abstract rather than from an issued claim set. I therefore treat claim 1 / 8 / 14 as unverified claim buckets and map art to subject matter, not to confirmed claim numbers.

---

A. The two Rockwell references the '436 specification itself relies on (Tier 1 — verified in the authoritative text)

These are the highest-confidence § 102 references because the authoritative text states them on its face — and critically, the '436 says they are "incorporated by reference," not that the '436 claims § 120 benefit of them.

Why the "own patent family" argument does not rescue these: Both published in 2015 — more than one year before the 2020‑04‑15 filing date. The AIA § 102(b)(1)(A) grace-period exception is capped at one year, and the § 102(b)(2)(C) "commonly owned" exception applies only to § 102(a)(2) art (patents/published applications), not to § 102(a)(1) printed publications. So if the '436's effective filing date really is 2020‑04‑15 (as the Google Patents record shows, with no § 119/§ 120 benefit claim), A1 and A2 are unconditional prior art.

Contradiction flagged: the earlier Patent summary section concluded the FIGS. 9–18 content "reads as largely consistent with the two applications the '436 specification expressly incorporates by reference… rather than as newly claimed subject matter." Correct as to specification provenance — but note the legal consequence that section did not draw: because the '436 appears not to claim § 120 benefit of those applications, incorporating them by reference makes them admissions and prior art, not ancestors that shield the '436. The '436's likely novelty therefore has to live in the sensing circuit, not in the multi-drive location logic.

---

B. The reference the '436 distinguishes as the problem being solved (Tier 1)

---

C. Art appearing in the '436's own citation record (Tier 2 — retrieved, with a sourcing flag)

The indexing I could reach shows a short examiner-style citation block associated with the US 11,402,436 record:

Sourcing flag (important): this two-reference block was retrieved from an adjacent Google Patents page (US 2020/0041559 A1) whose snippet interleaved the '436's title, a "Cited By"-style entry (US 11,362,507 B2, ABB) and a "Citations (2)" table. I could not confirm to a certainty that the "Citations (2)" table belongs to the '436 rather than to US 2020/0041559 A1. Treat C1 and C2 as strongly suspected but unconfirmed. Verify on the '436's own "Citations" tab.

---

D. Non-prior-art items that will otherwise contaminate the analysis (explicitly excluded)

These surfaced near the '436 in indexing and are post-dating documents — they may be cited-by references but can never be § 102 art against a 2020‑04‑15 filing:

• US 11,362,507 B2 (ABB Schweiz AG), "Ground fault protection in a high resistance grounding system" — filed 2020‑08‑06, granted 2022‑06‑14. After the '436's filing date. Excluded.
• WO 2024/061473 A1 (Hitachi Energy Ltd), "Method for detecting earth faults and frequency converter system with earth fault detection" — published 2024‑03‑28. Excluded.

One genuine § 102(a)(2) nuance worth preserving: a reference can be prior art under § 102(a)(2) (patent/published application, "effectively filed" before the critical date, naming another inventor) even if it issued after 2020‑04‑15. Two candidates meet that test:

• US 11,175,349 B2 (Eaton Intelligent Power Limited), "Electrical system" — effectively filed 2018‑06‑12, granted 2021‑11‑16.
• US 11,456,660 B2 (General Electric), "Grounding scheme for power conversion system" — effectively filed 2015‑12‑25, granted 2022‑09‑27.

Neither is a better reference than A1/A2, but they are available if you need § 103 combinations on grounding topology.

---

E. Family-record art — high-value candidates, but not confirmed on the '436's face (Tier 2/3)

The following appeared in the citation/cited-by records of the sibling Rockwell family (US 7,751,993 B2; US 10,564,207 B2 pages), not necessarily in the '436's own (56) box. I list them because they are the most technically probative § 102/§ 103 references for the sensing claims, and one of them (E1) is a third-party reference squarely on the multi-drive location point.

E‑group 1 — Multi-drive / frequency-based ground-fault location (third party — the strongest non-Rockwell art)

E‑group 2 — HRG / DC-bus ground-fault detection (the § 102 attack surface for C1/C8/C14)

E‑group 3 — Rockwell sibling patents (same title, same 2013‑10‑08 priority)

---

F. Bottom line for the § 102 attack

Where § 102 is realistic:

1. Multi-drive location claims (FIGS. 9–18). A1 (US 2015/0097571 A1), A2 (US 2015/0355262 A1 / US 9,604,543 B2), E1 (US 10,585,134 B2), and E18 (US 9,684,026 B2) each plausibly disclose the full "F_drive vs. F_ng suspect-set + speed-nudge exoneration" logic. Because A1/A2 published >1 year before filing and were incorporated by reference rather than chained by a § 120 benefit claim, they function as § 102(a)(1) anticipation art, self-owned or not. This is the single highest-value finding in this analysis, and it cuts against the '436's breadth — exactly the distinction the earlier Patent summary section raised but did not resolve.
2. Inverter/DC-bus ground-fault detection preambles. E13 (US 2005/0099743 A1) and E4 (US 4,589,048) are clean pre-2020 publications hitting the C1/C14 preamble.
3. The sensing-circuit claims (C1/C8). § 102 is only realistic if C1 (Ford, US 2014/0132278) or a sibling (E16/E17) discloses the three-element topology — C1 from the second (DC−) bus terminal to a mid-node, C2 from that node to the reference node, and the resistive divider across C1 with the divider output delivering a signal corresponding to the voltage across C1. Neither C1 nor C2 (NXP) is confirmed to show that exact arrangement. If they do not, § 102 fails on C1/C8 and you are in § 103 territory.

Where § 103 is the realistic route (and the better one): the '436 admittedly characterizes two things as known art in its own background — (i) chassis-referenced capacitor circuits "having capacitors that create an AC neutral point with connections to positive and negative bus terminals" with "reduced DC leakage current … but … high common mode noise," and (ii) resistive divider sensing with 1–2 MΩ strings (leakage problem). Combine that admitted art with B1 (US 9,383,399 B2, chassis-referenced sensing) and any of E3 / E8 / E13 (DC-bus-to-ground sensing), with a marine/hi-pot ≥100 MΩ isolation motivation supplying the reason to modify (KSR rationale: known problem, finite number of identified solutions, predictable result). The negative-bus referencing and the resistive divider coupled across the first capacitor only are the only real battlegrounds — attack those with an explicit "no unexpected result" argument.

Do not bother with: C2 (NXP) as a standalone § 102 reference, and E7/E12/E20/E21 as standalone anticipation. They are § 103 support at best.

---

G. Contradictions and unresolved issues you must reconcile before relying on any of this

1. Claims are missing. The authoritative text has no claims. The earlier PTAB section's "independent claims 1, 8, 14" is unverified, and the one quoted claim phrase is verbatim the abstract. Every claim mapping above is therefore subject-matter mapping, not claim-charting. Get the issued claims from the USPTO Patent Public Search / Patent Center for 16/849,230.
2. Effective filing date is the pivot of this whole analysis. The Google Patents record shows priority 2020‑04‑15 with no § 119/§ 120 benefit claim — yet the spec text is essentially identical to US 2017/0131340 A1 (Tallam, Hu, Brown; filed 2017‑01‑19; published 2017‑05‑11; itself a CIP of 14/048,156 filed 2013‑10‑08 and 14/296,502 filed 2014‑06‑05 → US 9,604,543). If a § 120 chain exists in the file wrapper, the 2015–2017 Rockwell publications stop being prior art and my A1/A2 § 102 conclusions flip. Confirm the benefit claim on the face of the granted patent.
   • This also resolves the earlier Assignment section's "cross-contamination warning." The reel 050718/0319 assignment (TALLUM, RANGARAJAN M.; HU, JIANGANG; BROWN, BRIAN P.; signed 2017‑01‑17/18) is the assignment for US 2017/0131340 A1 — the same-family, same-specification application — which is why it surfaces on pages adjacent to the '436. That section was right to refuse to attribute it to the '436, and the reason it appeared is now identified.
3. The V_ng / λ_ng vs. V_HRG / λ_HRG naming "oddity" is now explained, not a defect. The FIGS. 1–8 text was re-drafted for the '436 using "V_HRG," while the FIGS. 9–18 text was carried over from A1/A2 with leftover "Fng / INTNng / ng" subscripts (e.g., the '436 spec's "determines the fault frequency Fng as a frequency of the integral INTNng at 1408"). It is a partial find‑and‑replace artifact consistent with the specification being assembled from the incorporated documents — which is itself corroboration of point 2.
4. B1 (US 9,383,399 B2) is cited in the authoritative text but I could not verify its dates, title, or assignee. I have deliberately left those fields blank rather than guess. Pull it before using it.
5. Sourcing flag on Section C. The C1/C2 "Citations (2)" table came off an adjacent page; verify on the '436's own citations tab.
6. My search was step-capped. A live PTAB E2E/PATACTS run and a full front-page (56) pull were not completed. A recently indexed citation or a recently filed petition would not have appeared.

Verification links

• Google Patents, US 11,402,436 B2 (Citations / Cited By / Legal Events tabs): https://patents.google.com/patent/[US11402436B2](/patent/US11402436B2)/en
• USPTO Patent Public Search (application 16/849,230 — pull the full issued claim set and the front page): https://ppubs.uspto.gov/pubwebapp/
• USPTO Patent Center (file wrapper, to confirm § 120/§ 119 benefit claims): https://patentcenter.uspto.gov/
• A2's grant for cross-check: https://patents.google.com/patent/[US9604543B2](/patent/US9604543B2)/en
• The same-specification family member: https://patents.google.com/patent/US20170131340A1/en

I'll research prior art relevant to the '436 patent's claimed subject matter before performing the § 103 analysis.

I now have the actual claim text. Let me pull the full claims and the key family/prior-art references.

Obviousness Analysis — US 11,402,436 B2 under 35 U.S.C. § 103

0. Corrections and contradictions to flag before the analysis

Three things that affect how much weight to put on the prior sections:

1. Claim text is now available — the earlier "claims not retrievable" caveat is superseded. The prior "Patent summary" section stated the claim set could not be obtained and reconstructed the claims from the abstract. A live search of the Justia copy (https://patents.justia.com/patent/[11402436](/patent/11402436)) returns the literal issued claim text, including all three independent claims (1, 8, 14) and dependents 2–7 and 9–13. My § 103 analysis below is therefore run against the actual claims, not the reconstruction. A verbatim copy of the independent claims (as retrieved) appears in § 3.

2. Internal contradiction in the prior sections, now resolved. The "Litigation summary" said claims could not be retrieved; the "PTAB challenges" section simultaneously asserted "independent claims 1 (power conversion system), 8 (sensing circuit), and 14 (fault detection system)." The PTAB section was right — those are the three independent claims. The litigation section's caveat was over-conservative.

3. Date inconsistency. The system header gives today as 2026‑09‑29; the live fetch of Google Patents is stamped 2026‑09‑24; the task block says "Current Date: April 26, 2026," and the prior sections say April 26, 2026. I use 2026‑09‑29 as "today" for prior-art-on-sale/public-use purposes, but this does not move the analysis, because every reference below is a printed publication dated well before the critical date. Flagging it because it will matter to anyone relying on the "no litigation" conclusion in the other brick.

Bottom line up front: I found a credible § 103 invalidity case against independent claims 1 and 8, and a good-but-messier case against claim 14, built on Rockwell's own pre‑2020 family (which the '436 specification itself incorporates by reference) combined with Caterpillar's capacitor‑divider/resistor‑divider leakage-detection circuit and a PCT publication on capacitively coupled DC-bus ground-fault detection. The single most serious attack is not obviousness but anticipation: Rockwell's own US 2017/0131340 A1 appears to disclose the same "capacitively coupled sensing circuit 115" drawn in the '436's FIG. 1. That is a § 102(a)(1) question, and it needs to be run down before anyone spends money on an IPR.

---

1. Legal standard applied

• Graham v. John Deere factors: scope and content of the prior art; differences between the prior art and the claims; level of ordinary skill; secondary considerations.
• KSR Int'l v. Teleflex: a claimed combination is obvious where the improvement is a predictable use of prior-art elements according to their established functions; an express teaching, suggestion, or motivation is not required. Where the problem is known and the solution is one of a finite number of identified, predictable approaches, the claim is obvious.
• AIA § 102(a)(1)/(a)(2): applicable here because the '436's effective filing date is 2020‑04‑15, with no earlier priority or benefit claim of record (the front matter retrieved from Google Patents shows priority date 2020‑04‑15, and no "(60) Provisional" or "(63) Continuation" data). That is important: it makes Rockwell's own 2016/2017 filings fair game as prior art.

Level of ordinary skill in the art (PHOSITA). The '436 and the prior art are directed to sensing and diagnosing ground faults in industrial motor drives (AC/DC/AC voltage-source converters with a DC link, high-resistance grounding). I would define the PHOSITA as a person with a B.S. in electrical engineering (or equivalent) and 2–5 years of experience in power-conversion/motor-drive design, including familiarity with DC-link sensing, insulation/ground-fault detection, and drive control. This is not an unusually high level; a journeyman drives engineer. Nearly all the references below come from that same group of ordinary skill (Rockwell, ABB, General Electric, Caterpillar, Siemens, Mitsubishi, Fanuc), which materially helps a § 103 case — the references are analogous art by definition.

Claim terms worth construing (dispositive terms in italics):

• "reference node" — the '436 spec maps this to "a ground or other reference node 104," optionally the chassis, optionally a Y-source neutral through R_HRG. Broadest reasonable reading: a node treated as the system reference/ground.
• "low leakage sensing circuit" — a functional label; the claim itself supplies the structure that produces the low leakage (the series DC-blocking capacitor path to the reference node). Should not be read to impose a numeric leakage limit absent a definition.
• "coupled across the first capacitor" — the '436's own spec defines the resistive divider as R1 from DC− to a tap 117, R2 from the tap to C1's second terminal, i.e., R1+R2 in electrical parallel with C1, with the output taken at the R1/R2 tap. That reading should control.
• "a voltage signal corresponding to a voltage across the first capacitor" — the divider tap voltage is a scaled version of the C1 voltage; this is the V_HRG signal.

---

2. The actual claims (independent claims, as retrieved)

Claim 1 — a power conversion system comprising: a rectifier (AC input, DC output); an inverter (DC input, AC output); a DC bus circuit including (i) first and second DC bus terminals individually coupled between the rectifier DC output and the inverter DC input and (ii) a DC bus capacitor coupled between the first and second DC bus terminals, the rectifier configured to provide a DC output voltage with the first DC bus terminal positive relative to the second; and a low leakage sensing circuit to sense a voltage between the second DC bus terminal and a reference node, including:

• a first capacitor having a first terminal coupled to the second DC bus terminal, and a second terminal;
• a second capacitor having a first terminal coupled to the second terminal of the first capacitor, and a second terminal coupled to the reference node;
• a first resistor having a first terminal coupled to the second DC bus terminal, and a second terminal; and
• a second resistor having a first terminal coupled to the second terminal of the first resistor, and a second terminal coupled to the second terminal of the first capacitor.

Claim 8 — a sensing circuit per se, for a power conversion system having a first DC bus terminal with a first voltage and a second DC bus terminal with a second voltage, the first voltage positive relative to the second, comprising: a first capacitor (first terminal coupled to the second DC bus terminal, and a second terminal); a second capacitor (first terminal coupled to the second terminal of the first capacitor, second terminal coupled to a reference node); and a resistive divider circuit coupled across the first capacitor and having an output terminal that delivers a voltage signal corresponding to a voltage across the first capacitor.

Claim 14 — a fault detection system, comprising: a sensing circuit to sense a voltage between the second DC bus terminal and a reference node, the sensing circuit comprising a capacitor circuit coupled between the second DC bus terminal and a reference node and a resistor circuit coupled to the capacitor circuit, the resistor circuit having an output terminal that delivers a voltage signal corresponding to a voltage between the second DC bus terminal and the reference node; and a signal conditioning circuit having a first input terminal coupled to the output terminal of the resistor circuit and an output configured to provide an output voltage signal representing a voltage signal from the output terminal of … (retrieved text truncates here; the remainder of claim 14 — and dependent claims 6 and 7, which add the A/D converter / optical coupler / processor chain — should be verified against the printed patent).

Structurally, then, the '436 monopolizes three things that are worth separating:

• (A) the architecture of a drive's low-leakage sensing front end: a DC-blocking capacitor pair from DC− to ground, with a resistive divider sitting across only the bus-side capacitor, tapped to produce the sensed voltage (claims 1, 8);
• (B) the box — a fault-detection system with that sensor plus signal conditioning (claim 14);
• (C) the software — leakage-flux-linkage amplitude/frequency/phase-shift logic (in the spec, largely carried over from the incorporated publications and possibly claimed in the 2017-family, not visibly recited in claims 1/8/14 as retrieved).

The validity fight is about (A). (C) is essentially the subject matter of documents the '436 expressly incorporates by reference — a point the patentee is going to have to live with.

---

3. Prior art references

Caveat on element-by-element confidence. I have verified the text of the claims and the text of each reference's disclosure, but I have not been able to pull the figures of US 2017/0131340 A1, US 2015/0097571, or US 9,383,399 into this session. The figures are where the exact divider topology lives. The § 102 question about 2017/0131340 in particular cannot be closed without its FIG. 1 and FIG. 2. I say so rather than assert anticipation.

---

4. Motivation to combine

The '436 specification itself supplies the motivation, in the applicant's own words (this is unusually favorable to a challenger, because the motivation comes from the patent's own background section):

1. Stated problem. Marine standards "require DC isolation of 100 MΩ resistance or more from high-voltage to ground." Therefore "a neutral or other system voltage cannot be sensed with respect to ground using resistive dividers." Ergo, the whole field had a recognized, articulated need to sense bus-to-ground voltage while blocking DC. That is the definition of a known problem.
2. Stated prior solution and its stated deficiency. "Chassis referenced capacitor circuits can be used, having capacitors that create an AC neutral point with connections to positive and negative bus terminals. These circuits have reduced DC leakage current … but may suffer from high common mode noise." The applicant thus concedes that capacitive AC neutral points across the DC bus were known and practiced — and identifies the residual problem (common-mode noise) that the '436 purports to solve.
3. Stated design trade-off that motivates the claimed structure. "The use of the blocking capacitor(s) facilitates use of a resistor string of sufficiently low impedance [1–2 MΩ] to facilitate high noise immunity for voltage feedback." A PHOSITA reading this would immediately recognize the claimed arrangement as the obvious way to get both: block DC to satisfy insulation/hi-pot, keep a low-impedance divider for noise immunity.
4. Caterpillar's own motivation. US 2009/0001993 A1 motivates the exact combination from the other direction: it teaches that an HVDC drive can be balanced/referenced to frame ground by equal capacitors plus equal resistors, that a divider tap (V_offset across R6) can be measured by a controller, and that the measurement identifies leakage/unbalance worth alarming on. Combining Caterpillar's divider front end with Rockwell's drive GFD processor is combining two references in the same field, directed to the same purpose, with a predictable result.
5. WO 2021/028039's motivation. It expressly frames the design goal as detecting ground faults in a DC bus without compromising insulation resistance (claim 9's R ≥ U/1 mV/Ω constraint) and filters the measured voltage to improve detection (claim 8) — i.e., the same problem/solution pair, addressed with the same class of components.

There is also an efficient "same-spec" motivation that is hard to rebut: the '436's own specification says the sensing circuit 115 "can be used for a variety of applications, including without limitation ground fault detection, ground fault location identification" and expressly incorporates US 2015/0097571 and US 2015/0355262 wholesale. A PHOSITA wishing to make the incorporated GFL algorithms work on a marine-compliant drive would be led, as a matter of ordinary engineering, to swap the admitted chassis-referenced sensing string (US 9,383,399) for the admitted chassis-referenced capacitor circuit, and then to the negative-bus-referenced variant to cure the admitted common-mode-noise defect.

Reasonable expectation of success: High. Both halves of the combination are passive linear networks whose AC/DC behavior is fully predictable by elementary circuit analysis. Nothing in the combination is unpredictable; the claimed structure is a two-capacitor DC-block plus a two-resistor divider. KSR forecloses any "but nobody had thought to put them together in exactly this arrangement" argument.

---

5. Claim-by-claim § 103 analysis

5.1 Claim 1 (power conversion system)

Conclusion on claim 1: Obvious under § 103 over US 2017/0131340 A1 (or US 2015/0097571 / US 9,684,026) in view of US 2009/0001993 A1, and further in view of WO 2021/028039 and US 9,383,399 (the applicant's admitted chassis-referenced sensing circuit). The only element not squarely met by any single reference I found — and it is a genus-of-two selection — is "reference the divider to the negative rail instead of to chassis." When the two options are "chassis" (the admitted prior art) and "negative bus," and the specification itself tells you the chassis option has a common-mode-noise problem, the selection of the negative bus is an obvious design choice with a predictable, articulated benefit. That is a KSR case, not a § 102 case.

Anticipation alternative (flag, do not assume): if US 2017/0131340 A1's FIG. 1/FIG. 2 depict the sensing circuit as negative-bus referenced with the divider across C1 — which is plausible given that the '436 specification appears to be the same disclosure with the figure caption changed to add "with a negative bus referenced circuit" — then claim 1 is invalid under § 102(a)(1), and § 103 need not be reached. This is the highest-value item to verify, and it is cheap to verify (pull the '340 publication's FIG. 1).

5.2 Claim 8 (sensing circuit per se)

Claim 8 is claim 1's sensor with the drive context stripped out. That makes it weaker, not stronger, because the drive context was the only thing tying it to a specific host architecture:

• US 2009/0001993 A1 discloses a capacitor pair spanning the DC bus and a resistor divider producing a measured tap voltage indicative of bus-to-frame voltage/leakage, with a controller and thresholds — i.e., a "sensing circuit to sense a system voltage in a [DC-bus] system having a first DC bus terminal with a first voltage and a second DC bus terminal with a second voltage, the first positive relative to the second."
• WO 2021/028039 independently discloses the capacitor arrangement between the two bus bars and a middle point tied to ground potential through a high resistance, with a voltage-measuring arrangement at the middle point.
• DE 10148740/US 6,747,458 discloses divider-based ground-potential monitoring in a rectifier drive.
• The final limitation — "a resistive divider circuit coupled across the first capacitor and having an output terminal that delivers a voltage signal corresponding to a voltage across the first capacitor" — is met by any two-resistor series network shunted across C1 with a tap; US 2009/0001993's R3/R4/R6 divider and DE 10148740's parallel-to-divider-resistor arrangement are the two closest.

Conclusion on claim 8: Obvious. The claim is a bare electrical topology with no host, no algorithm, and no numerical constraint. Under KSR, a claim that recites the predictable combination of a known capacitive voltage divider (US 2009/0001993; WO 2021/028039; the '436's admitted chassis-referenced capacitor circuit) with a known resistive voltage divider (US 2009/0001993; US 6,747,458) — where both references are in the same field and both are used for the same purpose, measuring a bus-to-reference voltage — does not survive § 103. Claim 8 is probably the single most vulnerable claim in the patent, precisely because it is drafted as an apparatus-per-se with no tie to the drive's operation and no narrow numerical limitation (unlike, say, a claim reciting the 1–2 MΩ string, which would at least create a § 103 argument about criticality).

5.3 Claim 14 (fault detection system)

Claim 14 adds signal conditioning to the sensing circuit and (per the truncated text and dependent claims 6/7, which the same family recites elsewhere) the A/D → optical coupler → processor chain and a processor configured to detect the ground fault.

• US 2013/0218491 A1 (incorporated by reference in the '436) and US 2015/0097571 / US 9,684,026 disclose on-board GFD systems with HPF/LPF filter components, an integrator, and threshold comparison — i.e., the signal-conditioning layer.
• US 2009/0001993 A1 discloses the divider output feeding a controller / op-amp / microprocessor that measures the tap voltage and compares against a threshold.
• US 2017/0131340 A1 discloses the whole "AC coupling → sense system voltage → determine leakage flux linkage → detect ground fault" chain with A/D and processor.
• A/D conversion, opto-isolation (iso-coupler between the high-side sensing front end and the processor), and filtering are the textbook way to bring a divided high-side signal into a control processor — a routine combination of known elements each performing its known function, with a predictable result. The Applicant's own signal-conditioning chain (C3, C4, R5 filter; amplifier 200; A/D 202; optocoupler 201; processor 122) is the conventional pattern.

Conclusion on claim 14: Obvious over US 2017/0131340 A1 in view of US 2009/0001993 A1, further in view of US 2013/0218491 A1 / US 9,684,026 B2 for the conditioning-and-threshold layer. Slightly weaker than claims 1/8 because claim 14 is broader in one respect (it says only "a capacitor circuit" and "a resistor circuit coupled to the capacitor circuit," not the specific C1/C2/R1/R2 topology) — breadth cuts against the patentee here too, because the broader recitation is easier to meet squarely with US 2009/0001993 alone. I would predict the same outcome: obvious.

5.4 Dependent claims 2–7, 9–13

• Claims 2, 3, 10–13 (signal conditioning; filter + amplifier) — meet by US 2013/0218491 A1 / US 9,684,026 / US 2015/0097571 (all in the '436's own incorporated disclosure), and by US 2009/0001993's op-amp/controller.
• Claims 4, 5, 6, 7 (A/D → optical coupler → processor; processor detects the ground fault) — meet by the '436's own incorporated disclosure (processor 122 / memory 124 with the FIG. 3A/3B algorithm), which is in US 2017/0131340 A1 and the incorporated publications. An "optical coupler" is nothing more than an electrical-isolation element whose use in high-side sensing is entirely conventional.
• Claim 9 (divider = first resistor with a second terminal coupled to the output terminal) — this is just a definition of a tapped divider, met by US 2009/0001993's R3/R4/R6 tap and by every two-resistor divider in the art.

None of the dependents supplies a limitation that plausibly rescues the independent claims. The dependents are, if anything, more exposed, because they add only textbook components (filters, amplifiers, A/D converters, optocouplers).

---

6. Secondary considerations (Graham factor 4)

I have no evidence of any of these, and the record actively cuts against most:

• Commercial success with nexus — none presented; the only "value" figure in the record (~$29.3M on patentleaderboard.com) is a scraped heuristic, not evidence, and in any event commercial success requires a nexus to the claimed structure, not to Rockwell's drive business generally.
• Long-felt need — partially supported by the '436's own background (100 MΩ marine isolation), but that need was already being addressed by the references (WO 2021/028039's insulation-resistance constraint; Caterpillar's balanced HVDC referencing). A long-felt need that the prior art already addresses is not a saving grace.
• Failure of others — the references show the opposite: Rockwell itself iterated chassis-referenced (US 9,383,399) → capacitor-circuit → negative-bus-referenced sensing circuits, which reads as routine engineering progression, and Caterpillar and the PCT applicant reached variants of the same solution.
• Copying / licensing — no litigation, no licenses of record; the prior brick found zero assignments after the original 2020‑04‑15 inventor-to-employer assignment and zero PTAB or district-court activity, so there is no secondary-considerations record to work from.

Practical effect: the patentee's best non-obviousness arguments will be attorney argument about the negative-bus referencing and the specific divider tap, not evidence.

---

7. The patentee's best rebuttals — and why they likely fail

1. "Resistive dividers are taught away because they leak current to ground." The '436 criticizes resistive dividers "connected to a neutral or other high voltage circuit node." But claim 1's R1/R2 do not connect to ground at all — R2's second terminal goes to C1's second terminal, not to the reference node. The DC path to ground is blocked by C2. So the very criticism the patentee relies on is a criticism of a different circuit (a divider that references ground/neutral through a resistor); the claimed circuit is precisely the fix for that criticism. No teaching away.
2. "Chassis-referenced capacitor circuits suffer high common-mode noise." This is a criticism of the reference node choice, not of the capacitor-divider-plus-resistor-divider architecture. With only two possible reference nodes for a bus-referenced sensor (chassis/frame vs. a bus rail), and the specification itself identifying the chassis option's defect, the negative-rail option is a two-choice design selection — the classic KSR "finite number of identified, predictable solutions."
3. "The phase-shift lookup table and faulted-phase identification are novel." True as to the algorithm, but (i) the algorithm is not what claims 1/8 recite, and (ii) the algorithm and its FIG. 3A/3B flow appear in US 2017/0131340 A1 and in the references the '436 expressly incorporates — so the patentee's own specification forecloses treating that material as the inventive contribution.
4. "The resistor string of 1–2 MΩ is critical to noise immunity." That value appears in the specification but, as retrieved, not in claims 1, 8, or 14. Unclaimed subject matter cannot supply patentability. If the patentee wants to narrow to 1–2 MΩ, that raises a § 112 written-description/enablement-against-criticality fight rather than a § 103 escape.
5. "Unexpected results / better common-mode noise immunity." Argument-only, and the mechanism (referencing the sensor to the negative rail rather than the chassis) is predictable, not unexpected; the specification's own explanation ("significantly better common mode noise immunity") is a statement of expected behavior, not a surprising one.

---

8. Where the obviousness case is weakest

Be candid about this, because it determines strategy:

1. The negative-bus-referenced limitation is the linchpin, and I do not have a single reference that discloses it explicitly. Every structural reference I retrieved references its divider to frame/chassis ground (Caterpillar), to a middle point grounded through a resistor (WO 2021/028039), or to a chassis (US 9,383,399). The '436's negative-rail referencing is the element that distinguishes it, and the § 103 case rests on characterizing that as an obvious design choice. It is a strong KSR argument but not a slam dunk — a patentee could argue that chassis-referenced sensing was the entrenched convention for good reasons (operator safety, chassis continuity) and that rail-referencing sacrificed a benefit.
2. The exact divider-across-C1 topology (R1+R2 in parallel with C1, output at the R1/R2 junction). Caterpillar's divider (R3/R4/R6) is a mid-point divider, not a bus-cap-shunted divider; DE 10148740's is closer but I only have its abstract/claim language, not its figure. A clean § 103 chart on claim 1 needs one more reference — ideally a DC-link voltage-sensing circuit that shunts a resistive divider across a DC-link capacitor and taps it. Rockwell's own family may contain it. This is a 30-minute search, and I did not run it to ground in this session.
3. The anticipation question on US 2017/0131340 A1 is unresolved. If resolved in the challenger's favor, it moots the whole § 103 debate for claims 1/8/14. If resolved against, the obviousness case carries the load and item 1 above becomes decisive.
4. Claim 14 is drafted more broadly than claims 1/8 ("a capacitor circuit," "a resistor circuit coupled to the capacitor circuit"), which actually helps the challenger. Do not treat claim 14 as the hard claim; claim 8 is the hard claim, and claim 1's negative-bus limitation is the hard limitation.

---

9. Recommended next steps (in order of cost-to-information)

1. Pull FIG. 1 and FIG. 2 of US 2017/0131340 A1. This is the single highest-value action. If the sensing circuit 115 there matches the '436's, the correct invalidity theory is § 102(a)(1), not § 103.
2. Pull FIG. 1/2 of US 9,383,399 B2 (the admitted chassis-referenced sensing circuit) and of US 7,751,993 B2 for the DC-bus-sensing lineage.
3. Search Rockwell/Nikola/ABB/Siemens DC-link voltage-sensing arts for a resistor divider shunted across a DC-link capacitor with a rail-referenced tap — needed to close the topology gap in § 8.2.
4. Confirm WO 2021/028039's US designation so § 102(a)(2) is available. Its "effectively filed" date of 2019‑08‑13 precedes the '436's 2020‑04‑15 filing date; a PCT publication is § 102(a)(2) art only if it designates the United States.
5. Verify the truncated claim text (claim 14's trailing elements; dependent claims 6 and 7) against the issued patent at USPTO Patent Public Search or the granted PDF, since my source is a third-party reproduction.

Confidence: substantiated-obviousness as presented for claims 1 and 8 is a strong but not airtight case (I would put it at "more likely than not," i.e., ~60–70% in an IPR), rising to "probable" if either (a) US 2017/0131340 A1's figures are confirmed to disclose the topology (making it § 102), or (b) a DC-link divider-shunt reference is found to close the § 8.2 gap. For claim 14 I would put it modestly higher because the claim is broader. I am explicitly not calling this an anticipation case on the current record, notwithstanding the family overlap, because I have not seen the figures. This is a technical validity analysis, not legal advice, and any filing decision should be confirmed by counsel after reviewing the prosecution history and the actual claim charts against the figures.