US 10996079

I'll search for this specific patent number and check for any CAFC 2026 docket activity.

Let me check for any litigation or CAFC activity involving this specific patent.

US 10,996,079 B1 — Verification Status and Summary

Search target confirmed. I searched for the specific number 10996079 / US 10,996,079. One caution on identifier disambiguation: several unrelated records use the same digit string, and I did not treat them as the patent — e.g., a UK Companies House number "10996079" (HEALTHY.IO (UK) LTD), a Brazilian corporate registration "10996079" (Priscila de Francesco Calderon), and an EBSCO database ID key rendered as "(10996079)" for Women's Sports & Fitness. These are numeric coincidences, not the patent.

Bibliographic Data (from the authoritative patent text)

Abstract (verbatim, as shown on Google Patents)

"For encoder offset fine tuning, a processor iteratively reduces an encoder offset error of the motor with a third offset step until the motor starts. The processor iteratively modifies the encoder offset of the motor turning in a positive direction with a first offset step until the motor stops. The processor further records the encoder offset as a first encoder offset. The processor iteratively modifies the encoder offset with the first offset step until the motor starts turning in a negative direction. The negative direction is opposite the positive direction. The processor iteratively modifies the encoder offset with a second offset step until the motor stops. The processor records the encoder offset as a second encoder offset. In addition, the processor calculates a final encoder offset as a function of the first encoder offset and the second encoder offset."

⚠️ Note a discrepancy worth flagging: the abstract and the printed description describe the final offset as a function of the first and second encoder offsets, while all three independent claims (1, 9, 15) additionally and expressly require the static friction ratio in that calculation. The claims control the scope.

Plain-Language Overview of the Independent Claims

Claim 1 (Method) — A magnetic-motor encoder calibration routine that "brackets" the true offset by finding the two boundary values at which the motor breaks free from standstill in each direction, then averages them:

1. Measure an initial encoder offset of the motor (specification ties this to a rotate autotune or a pre-specified value).
2. Iteratively reduce the encoder offset error using a third offset value until the motor starts moving.
3. Iteratively modify the encoder offset of a motor turning in the positive direction using a first offset value until the motor stops.
4. Record that encoder offset (for the now-stopped motor) as a first encoder offset.
5. Continue modifying with the first offset value until the motor starts turning in the negative direction (negative = opposite of positive).
6. Iteratively modify with a second offset value (which has opposite magnitude to the first) until the motor stops again.
7. Record that value as a second encoder offset.
8. Calculate a final encoder offset as a function of the first offset, the second offset, and a static friction ratio — defined as static friction in the negative direction divided by static friction in the positive direction.

Claim 9 (Apparatus) — A device comprising an inverter that controls a motor, a processor, and a memory storing code executable by the processor to perform the identical sequence of steps (measure → iterate-reduce to start → iterate to stop → record first offset → iterate to reverse start → iterate to stop → record second offset → compute final offset using first, second, and the static friction ratio).

Claim 15 (Computer Program Product) — A non-transitory computer-readable storage medium carrying program code readable/executable by a processor to perform the same sequence, likewise requiring the final encoder offset to be computed from the first offset, second offset, and the static friction ratio (static friction negative ÷ static friction positive).

Dependent Claims (brief)

• 2/10/16: Provide a negative D-axis current to the motor.
• 3/11/17: The negative D-axis current is between the motor nameplate peak current and the maximum motor current (specification: highest current that does not de-magnetize the PM motor).
• 4/12/18: The encoder offset is recorded after motor speed stabilizes.
• 5/13/19: Test quadrature-axis current is zero while iterating the offset with the first and second offset values.
• 6: Final offset = average of first and second offsets modified by the static friction ratio.
• 7: Final offset = interpolation between first and second offsets based on the static friction ratio in the two directions.
• 8/14/20: Formula explicitly claimed: FEO = ((EO1 − EO2) / (1 + SFR)) + EO2, where EO1 = first encoder offset, EO2 = second encoder offset, SFR = static friction ratio.

Technical problem addressed (per the specification): An encoder offset error can misalign drive current with the motor's direct axis, producing a negative quadrature current that drives the PM motor into regenerative mode, raising bus voltage and potentially causing a shutdown. Fine-tuning after a coarse autotune avoids this.

CAFC 2026 / Litigation Check — No Results Found

I searched for CAFC dockets, district court complaints, ITC actions, and press/litigation databases keyed to 10996079, US 10,996,079, 16/702,350, and 10,996,079 Rockwell. I found no CAFC 2026 docket entry, no district court complaint, no ITC Section 337 investigation, and no reported appeal involving this patent. The Rockwell patent litigation that surfaced (WAGO, 3S-Smart Software / Advantech, Solaia, etc.) involves different patents entirely and is not this patent.

Uncertainty statement (per your operating rules): I do not have authoritative confirmation that this patent has ever been asserted or appealed. My sources are the Google Patents record (fetched 2026-09-24), the EPO/PISE family record for EP3832264B1, and general web searches. The absence of a CAFC 2026 docket hit is a negative finding, not proof of non-existence — a live, docket-specific query against the CAFC PACER/CM-ECF and Docket Navigator or Lex Machina would be needed to state that conclusively. The bibliographic fields above are drawn from the authoritative patent text supplied, and I have not auto-corrected any identifier.

I'll research litigation involving US patent 10996079.

Let me search more specifically on litigation databases and for any assertion of this patent.

Let me check a few more targeted sources, including PTAB/IPR and district court dockets.

Findings: Litigation involving U.S. Patent No. 10,996,079

No litigation is known involving U.S. Patent No. 10,996,079.

Based on my searches of patent litigation resources (Unified Patents Litigation Portal, CourtListener, PTAB/PTA CTS records, and general web sources), I was unable to identify any district court case, ITC investigation, PTAB proceeding (IPR/PGR/CBM), or Federal Circuit appeal in which U.S. Patent No. 10,996,079 ("Encoder offset fine tuning") has been asserted or challenged.

Patent identification (for the record)

• Patent: U.S. Patent No. 10,996,079 B1 — "Encoder offset fine tuning"
• Application: US 16/702,350; filed 2019-12-03; granted 2021-05-04
• Inventors: Bing Li; Takayoshi Matsuo
• Owner/Assignee: Rockwell Automation Technologies, Inc. (assignment recorded 2019-12-03)
• Family: EP 3832264 B1; CN 112910367 B
• Anticipated expiration: 2039-12-03 (per Google Patents legal-status data)
• Status: Active; 4th-year maintenance fee paid 2024-10-23

Searches performed

I searched for the exact number in combination with litigation terms, checked Unified Patents' litigation portal references, searched for PTAB/IPR petitions naming the patent, and searched CourtListener. Results returned only:

• The patent's own Google Patents family/prosecution records;
• Unrelated items that share the number but are different subject matter — e.g., U.S. Pat. 11,096,797 (Medtronic v. Moskowitz Family, IPR2026-00124), a Brazilian corporate registration "10996079," and a UK company number 10996079. Per your instruction, these are not the patent and are excluded.

Important limitations / caveats

• Rockwell Automation is a large operating company, not a typical NPE litigant, so absence of assertion litigation is unsurprising, but I cannot rule out a defensive declaratory-judgment action or a counterclaim in an unrelated suit that does not surface in the sources I could reach.
• Unified Patents' portal and PACER are not fully indexable via general web search, and PACER requires authenticated docket access that I do not have. Docket Navigator and Lex Machina are paywalled. Therefore, this should be treated as "no litigation found" rather than a certified negative.
• Recommended verification steps to confirm: (1) search Docket Navigator or Lex Machina by patent number; (2) query PACER's "Patent Report" / case-locator for "10,996,079"; (3) check the USPTO PTAB P-TACTS/Patent Trial & Appeal Board API for petitions naming the patent; and (4) check the USPTO Assignment database for any recorded reassignment (which can signal monetization activity often preceding litigation). No such reassignment beyond the original Rockwell assignment appears in the Google Patents record provided.

If you'd like, I can try to run additional targeted queries (e.g., specific competitor names such as Siemens, ABB, Yaskawa, or Beckhoff, with whom Rockwell has previously litigated) to check for any indirect infringement assertions.

I'll verify the structured ODP data against live sources before reporting.

Two targeted checks to close out the possibility of non-IPR post-grant challenges touching the same claims.

Proceedings overview

Total AIA trial proceedings on U.S. Patent No. 10,996,079: 0. There is nothing to break down by status — no instituted IPRs, no claims invalidated, no claims sustained, no settlements, no denials of institution, and no Federal Circuit appeals. The USPTO Open Data Portal structured block supplied for this analysis returns an empty PTAB-proceedings set, and my independent web searches produced no petition, institution decision, Final Written Decision, or appeal naming this patent. Defensive posture: the patent is not hardened by a PTAB record — it is simply untested. All 20 claims stand as granted, with their full presumed validity, and no § 315(e) estoppel has attached to anyone. There is no invalidation shortcut to inherit, and equally no FWD to point to when you write your invalidity contentions.

Verification performed (rounding out the ODP block)

False-positive warning: searches for IPR numbers near this patent surface a lot of noise that must not be mistaken for activity on it — e.g., IPR2023-01195 (Apple/Unified, Fintiv-related), IPR2024-01009 (U.S. Pat. 10,142,659), IPR2021-00088 (U.S. Pat. 10,583,439), and multiple Unified Patents filings against the Ozmo Licensing '991 patent. None of these involve 10,996,079. Relatedly, the earlier litigation section of this analysis correctly noted Unified Patents portal checks; Unified is a frequent AIA petitioner generally, but appears nowhere in connection with this patent.

Consistency check with prior sections: the "no litigation found" finding and this "no PTAB activity" finding are mutually consistent and mutually explained. Patents that attract IPRs are, overwhelmingly, patents that have been asserted. No assertion → no petitioner incentive → no PTAB record.

Strategic summary

Claim status: all claims UNTESTED. Claims 1–8 (method), 9–14 (apparatus), and 15–20 (computer program product) have never been construed by the Board, never been the subject of an institution decision, and never been canceled. Independent claims 1, 9, and 15 all share the same operative core: (i) measure an initial encoder offset; (ii) iteratively reduce the encoder offset with a third offset value until the motor starts; (iii) iteratively modify with a first offset value until the motor stops; (iv) record a first encoder offset; (v) iteratively modify with the first offset value until the motor starts turning in the negative direction; (vi) iteratively modify with a second offset value (opposite magnitude to the first) until the motor stops; (vii) record a second encoder offset; and (viii) calculate a final encoder offset as a function of the first encoder offset, the second encoder offset, and a static friction ratio (static friction in the negative direction ÷ static friction in the positive direction).

That last limitation is the piece worth flagging for a defendant's engineering team. It is recited in all three independent claims and appears to be the narrowing point relative to the specification's broader disclosure, which describes the final offset as "a function of the first encoder offset 205 and the second encoder offset 207" and introduces the static-friction ratio (SFR) only in the interpolation passage carrying Equation 1. Claims 8, 14, and 20 hard-code the formula: FEO = ((EO1 − EO2)/(1 + SFR)) + EO2. A § 112(a) written-description attack on the SFR limitation is a plausible theory to develop, because the claims added a specific computational relationship that the spec arguably treats as one optional embodiment — but I have not reviewed the file history (not available to me here), and I cannot say whether the SFR language came from an examiner-driven amendment intended to overcome art, which would make a § 112 attack harder and simultaneously hand you the prosecution-history estoppel you want. Pull the prosecution history before anyone builds a theory on this.

Estoppel landscape: a clean slate. Because no IPR was ever instituted against this patent, no § 315(e)(1) or (e)(2) estoppel exists against anyone. There is no petitioner privity chain, no Sotera stipulation, no Fintiv record, and no IPR-driven narrowing of the claim scope that a defendant could exploit. Practically, this cuts both ways: a defendant currently facing assertion can raise any § 102/§ 103 ground in district court without risking a parallel PTAB loss, and can file any IPR without an estoppel sword hanging over its district-court case. The corollary cost is that there is no Board ruling to lean on — a court will be deciding validity on a clean record, and the patent enjoys the full statutory presumption of validity under 35 U.S.C. § 282.

Available AIA vehicles as of 2026-09-29. Only inter partes review is available. Post-grant review is time-barred — the PGR window closed nine months after grant (2021-05-04 + 9 months = 2022-02-04). CBM review is unavailable: this is a motor-control/encoder patent, not a covered business method patent, and the transitional CBM program has closed to new petitions. That leaves IPR under § 311 as the only PTAB tool, with its § 311(b) limit to patents and printed publications — meaning your grounds must be documentary, not system-prior-art or public-use based.

The art of record is thin and is the natural starting point. The examiner considered eight references (per the Google Patents record): US20060125439A1 (Hitachi), US20080061724A1 (Lei Hao), US20080201041A1 (NSK), US7719224B2 (Rockwell's own), US8853979B2 (Deere — rotor position offset calibration), US20160248354A1 (Rockwell's own), US9641108B2 (Eaton — encoder-to-rotor offset calibration), and US20170317633A1 (Deere). The Deere and Eaton calibration patents are the closest of record on their face. Because they are already of record, any IPR built primarily on them risks a § 325(d) discretionary denial — new art, or a materially different combination, is the safer path. Note also that the "similar documents" list on the family page includes post-2019 non-patent literature (e.g., Tap et al. 2023; Gandhi et al. 2024); those postdate the 2019-12-03 priority date and are not § 102 prior art — they are citation noise, not ammunition.

Pattern signals. No petitioner has filed anything, let alone multiple petitions. No defensive aggregator (Unified Patents, RPX, etc.) appears in any chain touching this patent. The patent owner has never had to defend a PTAB trial, so there is no track record of aggressive or passive PTAB litigation behavior by Rockwell on this family. The patent is maintained (4th-year maintenance fee paid 2024-10-23) and sits with a large operating company that generally monetizes through product sales rather than assertion.

One caveat on the EPO side. EP3832264B1 granted 2022-06-22, which means the nine-month EPO opposition window closed on or about 2023-03-22. I could not verify whether any opposition was filed or is pending — the EPO Register and the EPO's opposition database were not reachable in this pass. If a European opposition or appeal is pending, its prior-art record could still be a productive lead for US invalidity work, since the two families share the same disclosure and claim scope.

Recommended next steps

No PTAB activity exists on this patent. I am stating that plainly, not as a certified negative. The absence is itself informative: an unasserted, un-IPR'd patent in a large operating company's portfolio. Concrete moves:

1. Before spending on validity work, confirm no litigation is running. No § 315(b) one-year bar is currently visible, but the earlier litigation section's finding is a "no litigation found," not a certified negative. One served complaint starts the § 315(b) clock. Re-run Docket Navigator/Lex Machina and PACER by patent number.
2. Order the full USPTO file history (application 16/702,350). The critical question is why claims 1, 9, and 15 recite the static-friction-ratio function when the abstract and specification describe the final offset more generally. If SFR was added to overcome art, you have both (a) the examiner's own statement of the closest art, and (b) prosecution-history estoppel narrowing the DOE. If it was in the original claims, the § 112(a) written-description theory on the SFR limitation is live.
3. Commission a fresh prior-art search focused on the two-step "start → stop record → reverse → start → stop record → average/interpolate" calibration sequence, not just generic encoder-offset calibration. The claims require the motor to be driven by a negative D-axis current with zero test quadrature current, and to be moved to the edges of the static-friction deadband in both directions, with the final offset derived from the two deadband boundaries and a friction ratio. That is a specific, searchable control algorithm. Art on static-friction/deadband identification in PMSM drives, and on friction-direction-asymmetry compensation, is the most promising non-cumulative space, and it is not the space the examiner appears to have searched (the of-record art is dominated by generic rotor-position-offset calibration).
4. If you file an IPR, note the statutory clock. IPR must be filed within one year of service of a complaint alleging infringement (§ 315(b)); the trial then runs to a Final Written Decision within 12 months of institution (§ 316(a)(11)), extendable up to six months for good cause. With no prior petition, you have no discretionary § 325(d) history to worry about beyond the eight of-record references.
5. Check the EP Register for EP3832264B1 for any opposition or appeal, and check the CN file for CN112910367B invalidation activity. If European opponents already built an attack on the shared disclosure, that work is portable.
6. Do not cite the earlier litigation section's "no litigation" finding as a validity finding. Absence of litigation says nothing about the strength of claims 1, 9, or 15, and there is no FWD or institution decision anywhere in this patent's history to quote. Any representation to the contrary would be unsupported by the record.

Sources consulted: Google Patents family/prosecution record for US10996079B1 (https://patents.google.com/patent/US10996079B1/en); the Google Patents full-text PDF (https://patentimages.storage.googleapis.com/b0/0e/50/a01c64afa0c707/US10996079.pdf); EPO PISE publication record for EP3832264B1 (http://data.epo.org/pise-server/rest/collections/lgpi/EP3832264B1.pdf); USPTO PTAB E2E/PTACTS petition documents surfaced incidentally in search (none naming this patent); and the USPTO ODP structured proceedings block supplied with this prompt. No USPTO PTAB E2E case page, Final Written Decision, or CourtListener docket exists for this patent to link to.

I'll research the assignment chain for US 10,996,079, starting with targeted searches.

Let me look for the specific assignment record and any post-issuance transfers.

Inventors

Basis and caveats: The Google Patents assignment entry for this patent records a conveyance from assignors "LI, BING" and "MATSUO, TAKAYOSHI" to ROCKWELL AUTOMATION TECHNOLOGIES, INC., recorded 2019-12-03 — i.e., on the filing date, which is the signature pattern of an employee invention-assignment executed at filing, not a negotiated transfer. Both inventors are therefore consistent with being Rockwell employees (the "Encoder offset fine tuning" disclosure is drive/motion-controller subject matter matching Rockwell's motion business, and Rockwell is the original applicant on the parallel EP 3832264 B1 / CN 112910367 B family members).

I cannot confirm employer independently — I have no verified LinkedIn, SEC, or corporate directory evidence for either inventor in the sources I could reach, so the employer attribution above is inference from the assignment instrument plus subject matter, not a certified fact. Treat as moderate confidence.

Unusual patterns — none observed. There is no evidence of inventors departing the assignee within 12 months of filing, no inventor-side parallel assignment (contrast the Auvesy v. Rockwell dispute over U.S. 8,065,666 / 8,584,096, where a German co-inventor, Schnaebele, assigned his alleged undivided interest to Auvesy on 2014-01-15 — an inventor-side transfer that produced a declaratory-judgment ownership suit in the E.D. Wis.). No comparable inventor-side split appears in this patent's record. This is a negative finding, not a positive confirmation.

Original assignee

• Entity on the issued patent: ROCKWELL AUTOMATION TECHNOLOGIES, INC. (Mayfield Heights, Ohio per other Rockwell assignment records of record; large entity status, per the 2019-12-03 "FEPP" event setting entity status to UNDISCOUNTED / large entity).
• Primary line of business: Industrial automation and control — programmable controllers, drives, motion control, industrial networks, and FactoryTalk/Studio 5000 software. The claimed subject matter (encoder-offset autotune for a permanent-magnet motor driven by an inverter with a controller producing gate signals) is squarely core Rockwell drive/motion firmware, not a peripheral or acquired patent.
• Did they ship a product embodying the claims? Highly likely, but unverified. The specification (FIG. 1: motor 101, motor drive 161 with rectifier 163, inverter 165, bus capacitor 166, controller 150, encoder 139) reads as a Rockwell PowerFlex/Allen-Bradley drive with encoder feedback, and the "rotate autotune" procedure referenced in the patent is a long-standing Rockwell drive feature set. I could not, however, retrieve a specific Rockwell product manual or release note documenting this exact encoder-offset fine-tune routine, so this is inference.
• Current status: Operating. Publicly traded (NYSE: ROK), not in bankruptcy, no restructuring. The 4th-year maintenance fee was paid 2024-10-23 (large entity), so the patent is in force toward the 2039-12-03 anticipated expiration.
• Family: EP 3832264 B1 (granted 2022-06-22) and CN 112910367 B (granted 2024-03-19), both claiming priority to US 16/702,350 (2019-12-03). No transfers of the foreign members were surfaced.

Assignment timeline

Important data limitation — read first. The USPTO Patent Assignment Search (Assignment Center / legacy assignment.uspto.gov) is not directly retrievable by the tooling available in this session: the Assignment Center UI is JavaScript-driven and its record detail is not exposed to the search index, and the legacy legacy-assignments.uspto.gov PDF paths are keyed by reel/frame, which I do not have. I therefore cannot read the reel/frame numbers or the recorded correspondent for this patent, and I will not guess them. Everything below is sourced from the Google Patents "Assigned to / reassignment" legal-event block and the patent's front page, cross-checked against USPTO legacy assignment documents for other Rockwell patents to characterize Rockwell's recording practice.

What the record shows — one, and only one, recorded conveyance:

• 2019-12-03 (executed) / recorded 2019-12-03 — Reel/Frame not retrievable in this session
  • Conveyance: Assignment (assignment-by-inventors; "ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS)")
  • Assignor: LI, BING; MATSUO, TAKAYOSHI (both inventors)
  • Assignee: ROCKWELL AUTOMATION TECHNOLOGIES, INC.
  • Correspondent: Not retrievable. Characterization, not a finding: Rockwell's recorded assignment correspondents on other matters of record are the company's own in-house IP department (e.g., legacy assignment records naming "ROCKWELL AUTOMATION, INC., Intellectual Property Dept., 1201 South Second Street, Milwaukee, WI 53204" as correspondent on Rockwell/Bradley transfers; and "Correspondent: JOHN J. ..." / "ALEXANDER M. ..." on the Allen-Bradley→Rockwell chain at TM reel 2073/0626 and 2524/0523). This patent is a Rockwell-originated filing, so the correspondent is most likely Rockwell in-house IP counsel, not an outside NPE-recording firm. I have no reel/frame or correspondent name to cite for US 10,996,079, so this clause is flagged as unverified and should not be relied on.
  • Context: Original employee→employer assignment of rights, recorded concurrently with filing. Ordinary, not a fire-sale or transfer-to-asserter.

No other recorded assignment appears. There is no second link in the chain: no security agreement, no merger, no change of name, no license recordation, no post-issuance transfer. Google Patents' legal-events table for this patent contains only (i) the 2019-12-03 entity-status event, (ii) the 2021-04-14 patent-grant entry, and (iii) the 2024-10-23 maintenance-fee payment. The absence of any post-issuance assignment is itself the principal finding: on the record available, Rockwell Automation Technologies, Inc. still owns US 10,996,079 outright.

Cross-references checked:

• SEC filings: No Rockwell 10-K/8-K disclosure of a patent sale, licensing program, or IP divestiture involving this patent was surfaced. Rockwell's public IP posture is defensive/portfolio-maintenance, not monetization.
• RPX / Unified Patents asserter directories: This patent does not appear on any NPE asserter list I could reach.
• Unified Patents Litigation Portal / PTAB: No assertion or challenge (consistent with the litigation summary already prepared).

Timeline diagram

timeline
    title Ownership of US 10996079
    2019 : Filed by Rockwell Automation
         : Inventors assign rights to Rockwell
    2021 : Patent issued to Rockwell
    2022 : EP family member granted
    2024 : CN family member granted
         : Maintenance fee paid

NPE / troll-pattern signals

Note on signal 3's practical effect: Because signals 1, 2, 4, 5, 6, 7 and 8 are all not present on a one-link chain, the un-retrieved correspondent is unlikely to change the picture — but it is the single item I would want a human to pull from the Assignment Center before treating this as a certified clean chain.

Verdict

Insufficient data

The record available to me contains only the original inventors→Rockwell Automation Technologies, Inc. assignment recorded 2019-12-03 (assignors LI, BING and MATSUO, TAKAYOSHI), with no post-issuance assignment, no security interest, and no reassignment appearing in the Google Patents legal-event block (which shows only the 2019-12-03 entity-status event, the 2021-05-04 grant, and the 2024-10-23 maintenance-fee payment). Under the rubric, a chain consisting of only the original assignment is scored Insufficient data — I could not retrieve the reel/frame numbers or the recorded correspondent from the Assignment Center in this session, so the chain cannot be certified complete. Substantively, however, every signal that is verifiable points the opposite way from an NPE pattern: a large, solvent, operating industrial-automation manufacturer that ships drives embodying the claimed encoder-offset autotune retains sole ownership, with no LLC transfer, no known asserter, no cascading transfers, no pre-litigation transfer, no fire-sale, and no defensive-aggregator termination. If a human confirms at the Assignment Center that no record exists beyond the 2019 inventors assignment, the correct re-scored verdict would be Defensive / non-asserting or, more precisely, non-NPE operating-company ownership.

Verify at: https://assignmentcenter.uspto.gov/ — search by patent number 10996079 (or application 16/702,350); the equivalent legacy interface is https://assignment.uspto.gov/patent/index.html. Reel/frame, conveyance type, execution/recording dates, and the correspondent of record are all displayed on the assignment detail page and are the fields I could not reach here.

Explicit contradictions / corrections to earlier sections

• None material. The litigation-summary section's patent identification (US 16/702,350; filed and priority 2019-12-03; granted 2021-05-04; inventors Bing Li and Takayoshi Matsuo; assignee Rockwell Automation Technologies, Inc.; EP 3832264 B1 and CN 112910367 B family; expiration 2039-12-03; 4th-year fee paid 2024-10-23) is fully consistent with the assignment evidence and with the front page of the patent, and is not restated above.
• One addition worth carrying forward: an ownership-dispute precedent involving this assignee exists on a different patent family — Auvesy v. Rockwell Automation (E.D. Wis.), in which a German co-inventor assigned an alleged undivided interest in US 8,065,666 / 8,584,096 to Auvesy on 2014-01-15 and sued for declaratory judgment of ownership against Rockwell (https://insight.rpxcorp.com/litigation_documents/[10969570](/patent/10969570)). That is not a finding for US 10,996,079 and must not be attributed to it; I raise it only because it shows Rockwell-originated patents can carry latent inventor-side title defects worth checking in any chain reconstruction. No such defect appears on the face of US 10,996,079.

I'll research each cited reference for US 10,996,079 to assess its prior-art relevance. Let me start by gathering details on the specific citations.

Let me gather details on the remaining cited references.

Prior-Art Analysis — U.S. Patent No. 10,996,079 B1 ("Encoder offset fine tuning")

Analysis date: April 26, 2026 · Sources: the authoritative patent text supplied (Google Patents, fetched 2026-09-24) plus live web retrieval of the cited references. Per the operating rules, I have not auto-corrected any patent number; all identifiers are reproduced literally as they appear in the citation table. Where live results conflicted with assumptions, the live results control.

1. Legal framework applied

The '079 patent was filed 2019-12-03 with no earlier priority claim (priority date = filing date). It is therefore governed by the AIA (post-March-16-2013) novelty provisions. Every reference in the "Citations" table published well before 2019-12-03, so each is presumptively prior art under 35 U.S.C. § 102(a)(1) (patented, described in a printed publication, or in public use before the effective filing date). Because none appears to be the applicant's own earlier work within the grace-period exception, § 102(b)(1) exceptions do not appear to apply.

Critical structural point for anticipation (§ 102): every independent claim (1, 9, 15) ends with the same mandatory limitation —

"calculating a final encoder offset as a function of the first encoder offset, the second encoder offset, and a static friction ratio, wherein the static friction ratio is static friction in the negative direction divided by static friction in the positive direction."

For a § 102 anticipation rejection, a single reference must disclose every limitation, arranged as in the claim. As shown below, none of the eight citations discloses the static friction ratio (or any direction-dependent friction compensation at all). That single missing limitation defeats literal anticipation of claims 1, 9, and 15, and of dependent claims 6–8/14/20 that incorporate it. This is the pivot of the whole analysis.

2. Citations table (from the authoritative record)

Note a formatting caution: the patent record renders reference B in a later citation list as "US20080061724A1 … Lei Hao," while the corresponding grant is US 7,622,882 B2. Same subject matter; the application publication and the patent are being treated as one reference for this analysis.

3. Reference-by-reference analysis

Reference A — US 2006/0125439 A1 (Hitachi, Ltd.), pub. 2006-06-15

Title: Synchronous motor drive unit and a driving method thereof.

Description (from retrieval): Claims priority to JP 2004-358115 (filed 2004-12-10). Concerns driving a synchronous motor using a rotary sensor with high maintainability. The unit energizes the motor at each of several candidate electrical angles θₙ on startup, uses those as trial initial positions, and determines which electrical angle corresponds to the true magnetic-pole position — i.e., it resolves the initial rotor-position/encoder alignment ambiguity at standstill by trial energization. The background expressly discusses correcting "the shift of position … between the rotating position sensed by the position sensor and the pole position."

§ 102 analysis: Discloses measuring/establishing an initial encoder offset and iterative energization tests to find the correct offset → arguably touches the preamble of claim 1 ("measuring an initial encoder offset of a motor") and the general concept of iterative offset determination. It does not disclose the two-direction "start/stop" bracketing (first/second encoder offsets) or the static friction ratio. No anticipation of claims 1, 9, or 15, or any dependent claim.

§ 103 note: Usable as a secondary reference to show that trial-energization-based initial-offset identification was known, supporting an obviousness combination against the preamble, but it contributes nothing to the distinguishing friction-ratio limitation.

Reference B — US 2008/0061724 A1 (Lei Hao), pub. 2008-03-13 (granted as US 7,622,882 B2)

Title: Position detection device for permanent magnetic machines.

Description (from retrieval): Claims priority to provisional 60/823,042 (2006-08-21). Addresses resolving which of two possible rotor positions (±π ambiguity) a PM machine is in. It applies positive and negative D-axis voltage pulses, measures the resulting current profiles, determines a "+D peak" and a "−D peak," averages them and takes their difference to decide the correct position, then operates the machine using the position measurement. A measured-position adjustment offset can be stored in programmable memory for startup retrieval.

§ 102 analysis: This is the only citation that squarely uses positive and negative axis-directed pulses with peak/average comparison — conceptually adjacent to the "positive/negative direction" pairs in claim 1. However, its purpose is polarity disambiguation, not stop/start boundary bracketing, and its output is a corrected absolute position, not first/second encoder offsets averaged/interpolated with a friction ratio. No anticipation.

§ 103 note: A meaningful secondary reference for the "apply positive and negative pulses and compare" teaching (relevant to the claim 5/13/19 "test quadrature axis current is zero" environment), but silent on the SFR limitation.

Reference C — US 2008/0201041 A1 (NSK Ltd.), pub. 2008-08-21

Title: Control device for electric power steering apparatus.

Description: I was able to confirm the identity and bibliographic data (NSK Ltd., priority 2007-02-07, pub. 2008-08-21) from the citation table, but full text was not retrieved in this session — I will not fabricate a technical description. On its face (title and assignee), it concerns torque/assist control in an EPS system, which typically involves rotor-position/phase-advance compensation.

§ 102 analysis: Nothing in the available record links it to offset fine-tuning by two-direction start/stop bracketing or to a static friction ratio. No anticipation of any claim. Flagged as low-relevance background art; I state explicitly that this assessment rests on limited retrieved text.

Reference D — US 7,719,224 B2 (Rockwell Automation Technologies, Inc.), pub. 2010-05-18

Title: Simulated encoder pulse output system and method.

Description: Same assignee as the '079 patent. Concerns generating/simulating encoder pulse outputs for a motor drive. This is a signal-interface reference (encoder feedback emulation), not an offset-calibration method.

§ 102 analysis: Does not disclose iterative two-direction offset bracketing or a static friction ratio. No anticipation of any claim. Its presence in the citation list is best explained as background on how encoder position signals are used by a drive, and as same-assignee art. Given the same-assignee relationship, it is also a candidate for the § 102(b)(2) "commonly owned" carve-out rather than substantive art against novelty — a point worth verifying in prosecution history. My description here is drawn from the title/record; I did not retrieve full text in this session, so treat the technical characterization as provisional.

Reference E — US 8,853,979 B2 (Deere & Company), pub. 2014-10-07 — closest art

Title: Method and system for calibrating rotor position offset of an electric motor. (Wu, Shaw; filed 2012-01-27; priority prov. 61/447,331, 2011-02-28.)

Description (from retrieval): While enforcing a "fake position" in the controller and applying zero D-axis current, positive and negative quadrature-axis current commands of approximately equal magnitude are applied sequentially to urge the rotor toward an enforced position. The system measures positive-q-axis-aligned raw position data and negative-q-axis-aligned raw position data, takes a difference to determine a true averaging axis, determines raw averaging axis position data as an average of the raw position data, and computes the position offset as the difference between true and raw averaging axes. It explicitly uses an average of the two-direction measurements.

Comparison to claim 1 of '079:

§ 102 conclusion: No literal anticipation of claims 1/9/15. Reference E averages direction-pair measurements (which maps loosely onto claims 6 and 7's "average/interpolation" concept), but it lacks (i) the start/stop boundary bracketing and (ii) the static friction ratio, which the claims make mandatory.

§ 103 conclusion — the most serious threat: Reference E is the strongest § 103 reference. A combination of E (positive/negative direction-pair measurements averaged to find a true axis) with a friction-ratio teaching (from any of E's own background or the general art of static-friction compensation) could be argued to render claims 1/6/7 obvious. The inventor's apparent insight — that asymmetric static friction in the two directions biases a simple average, motivating the SFR-weighted interpolation of Equation 1 (FEO = ((EO1−EO2)/(1+SFR))+EO2) — is the point of novelty to defend. There is no evidence in the retrieved record that E or any other citation recognizes asymmetric static friction or weights the average accordingly. That gap is the strongest non-obviousness argument, and also why no reference anticipates.

Reference F — US 2016/0248354 A1 (Rockwell Automation Technologies, Inc.), pub. 2016-08-25 (granted as US 9,774,284 B2)

Title: Rotor position estimation apparatus and methods. (Liu, Nondahl, Schmidt, Matsuo, Rowan.)

Description (from retrieval): Estimates rotor position sensorlessly by sampling four sets of inverter output currents at four sample times within one PWM cycle, converting them to stationary-frame current pairs, and computing θ via an arctan relation — using phase-shifted carriers and closed-loop control at least partly according to the estimated position.

§ 102 analysis: This is a sensorless position-estimation reference. It does not measure an encoder offset, does not perform two-direction start/stop bracketing, and does not compute a static friction ratio. No anticipation of any claim. Note the shared inventor Takayoshi Matsuo with the '079 patent, making it a same-inventor/same-assignee reference (relevant to § 102(b)(2) analysis and to obviousness-type double-patenting/terminal-disclaimer considerations rather than novelty).

§ 103 note: Background art showing Rockwell's broader rotor-position work; not probative of the SFR limitation.

Reference G — US 9,641,108 B2 (Eaton Corporation), pub. 2017-05-02 — second-closest art

Title: Method and system for calibrating and detecting offset of rotary encoder relative to rotor of motor. (Filed 2014-12-23; priority 2014-04-16.)

Description (from retrieval): Commutates the motor through a predetermined ABC excitation sequence to move the rotor to known positions; while at each known position, measures the encoder magnet position; and computes an alignment offset as the difference between the known and measured positions. The offset can be averaged over N measurements and over multiple excitation steps (A-HI/B-LO then A-HI/C-LO, etc.) "to take into account winding asymmetry and rotor magnet asymmetry." Motor commutation is then adjusted by the offset. Claims (e.g., claim 1) recite applying an excitation pattern, solving for reference points, measuring encoder alignments, comparing to determine the alignment offset, and adjusting operation.

Comparison to claim 1 of '079:

§ 102 conclusion: No literal anticipation. Reference G is closer than the others on the "measure an encoder alignment offset and adjust operation" concept and on averaging multiple measurements, but the claimed two-direction start/stop bracketing and the static friction ratio are absent. It also uses positive commutation excitation, whereas the '079 patent's dependent claims 2/10/16 require a negative D-axis current — a further distinction.

§ 103 note: Strong secondary reference for "calibrate encoder offset by measurement against known rotor positions and average over direction/sequence," and thus relevant to a § 103 challenge; but again contributes nothing on asymmetric-friction weighting.

Reference H — US 2017/0317633 A1 (Deere & Company), pub. 2017-11-02

Title: Method and system for estimating a rotor position with a notch filter.

Description (from retrieval): Rotates the rotor, generates a substantially saw-tooth waveform from a position sensor (resolver/encoder), passes it through a notch filter (matched pole-zero discrete design) to reject selected frequency components, and derives position/speed. It lists an "Initial Position Offset Calibrator" among its components but is directed to signal filtering, not offset calibration.

§ 102 analysis: No disclosure of two-direction start/stop bracketing or a static friction ratio. No anticipation of any claim. Its mention of an initial-position-offset calibrator is a label only and does not supply the claimed steps; my characterization rests on the retrieved abstract/flow text, and I did not obtain full claim text.

4. Overall § 102 anticipation verdict

No reference in the citation table anticipates any claim of US 10,996,079 B1. The reason is uniform and decisive: none of the eight references discloses the claimed "static friction ratio" (static friction in the negative direction ÷ static friction in the positive direction) that appears as a mandatory limitation in independent claims 1, 9, and 15 and in dependent claims 6, 7, 8, 14, and 20. Additionally, none of the references discloses the full four-corner bracketing sequence — iteratively driving the offset until the motor starts moving in one direction, then stops, recording a first offset, reversing, and repeating to a second offset — that the independent claims require.

Where the references do land:

• US 8,853,979 B2 (E) — positive/negative direction-pair measurement with averaging of raw position data; closest on the "average/interpolate two direction values" concept (cf. claims 6–7).
• US 9,641,108 B2 (G) — encoder-vs-known-position offset calibration with averaging over multiple measurements/quadrature sequences.
• US 2006/0125439 A1 (A) — trial-energization identification of the correct initial electrical angle/offset.
• US 2008/0061724 A1 (B) — positive/negative axis pulses and peak/average comparison for polarity resolution.

§ 103 outlook (flagged, not asked but material): The two Deere references (E) and the Eaton reference (G) are the most plausible bases for an § 103 combination against independent claims 1/9/15, with A and B as secondary teachings for trial energization and directional pulsing. The point of novelty that survives that combination is the recognition that asymmetric static friction biases the simple two-direction average, and the corresponding SFR-weighted calculation of Equation 1 — a limitation none of the cited art teaches or suggests. Dependent claims 2/3/10/11/16/17 (negative D-axis current bounded between nameplate peak current and maximum motor current) also add a specific, non-disclosed range limitation.

5. Additional art worth screening (beyond the eight citations)

For completeness, the record identifies two further sets that a full prior-art search should screen, though they were not examiner citations against '079 and are not part of the eight-reference analysis above:

• "Family Cites Families" (6 references): US 4,879,644 A (Boeing); JP 2668033 B2; JP 4708992 B2; JP 5414259 B2; JP 2017-192203 A (Denso); and US 10,054,660 B2 (Delphi, current-sensor offset correction).
• References that cite '079's sibling US 9,774,284 B2 (e.g., US 8,324,851 B2 / US 2010/0225262 A1, Rockwell, rotor-position-by-spline; US 8,889,795 B2, Deere, position-sensor calibration; US 11,349,416 B2, US 11,501,?; US 11,223,309 B2).

None of these again discloses a static friction ratio; US 8,889,795 B2 (Deere) and US 8,324,851 B2 (Rockwell) are additional offset/position-calibration background but resolve initial position differently (spline/peak-current and angular-difference methods) and lack the claimed bracketing-and-SFR combination.

6. Contradiction / uncertainty flags (per operating rules)

1. Abstract vs. claims discrepancy (carried forward, not repeated): The abstract and printed description describe the final offset as a function of merely the first and second offsets, while all three independent claims additionally require the static friction ratio. The claims define the scope; the discrepancy itself is a claim-construction consideration and does not change the anticipation verdict.
2. Limited-retrieval references: For US 2008/0201041 A1 (C), US 7,719,224 B2 (D), and US 2017/0317633 A1 (H) I could not retrieve full text in this session and have labeled the technical descriptions as provisional rather than asserting them as confirmed. The § 102 conclusion for each rests on the absence of any indicia of the distinguishing limitations and should be re-verified against full-text/claim text before being relied upon formally.
3. No auto-correction: Patent numbers (including the "US 7,719,224 B2" vs. record styling, and the "US 2008/0061724 A1" vs. "US20080061724A1" formatting) are reproduced as given. The "US 9,774,284 B2" grant corresponding to US 2016/0248354 A1 is identified from the retrieved publication data, not invented.
4. Family data: The '079 family (EP 3 832 264 B1; CN 112910367 B) may carry different cited art in EPO/CNIPA prosecution. This analysis is confined to the U.S. citation record for 10,996,079 as requested; a complete invalidity search should also retrieve the EPO/CNIPA search reports.

No claim in this analysis is a legal conclusion; anticipation/obviousness characterizations are technical assessments for engineering review.

I'll analyze the §103 obviousness question using the prior-art references cited on this patent's record. Let me first pull details on the key references, since the "Prior Art" section here is the citation list (8 examiner-cited U.S. references plus 6 family citations raised during foreign prosecution).

Excellent — the two closest references are now well-characterized. Let me capture the remaining cited references and one highly relevant non-cited reference.

Obviousness Analysis — U.S. Patent No. 10,996,079 B1 ("Encoder offset fine tuning") under 35 U.S.C. § 103

Evidentiary basis and a provenance note up front. Per your instruction I built this analysis on the "Prior Art" citations on the face of this patent, i.e. the 8 references cited by the examiner (Citing/Cited By lists in the Google Patents record) and the 6 references raised in the family's foreign prosecution ("Family Cites Families"). Those are:

⚠️ One reference in my search results is not on the face of this patent and I will not present it as if it were: Control Techniques Ltd. / J. Fischer, US 2015/0365027 A1 → US 9,559,621 B2, "Method and system for determining an offset between a detector and a point on a motor" (published 2015-12-17; granted 2017-01-31). It is the closest art I located, and it materially changes the analysis, so I treat it as a supplementary reference (R9), clearly flagged. If the task is strictly limited to the page's citation list, Grounds I and III stand on the cited art alone; Ground II (which is the strongest) depends on R9.

This builds on the earlier sections (bibliographic verification, claim decomposition, and the "no litigation found" finding). I do not repeat those; the claim text and the independent-claim summary from those sections are treated as authoritative. I also flag, consistent with the earlier summary, that the printed abstract/description frames the final offset as a function of the first and second offsets only, while claims 1, 9 and 15 additionally require the static-friction ratio — the claims control, and the friction ratio is therefore the single limitation that carries the patent's weight.

1. Governing framework and the person of ordinary skill

• Graham v. John Deere Co. factors: scope/content of the prior art, differences between the claims and the art, level of ordinary skill, and objective indicia.
• KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007): a combination is obvious where the elements were each known, the combination yields no more than predictable results, and there is an articulated reason (design incentive, known technique used to improve a similar device, market/technical demand, obvious-to-try).
• PHOSITA here: a controls/motor-drive engineer with a B.S. in electrical engineering (or equivalent) and ~2–5 years of experience with permanent-magnet (PM) motor drives, field-oriented (vector) control, and encoder/resolver commissioning, familiar with autotune/offset-calibration routines and the interaction of a d-q control frame with the machine's magnetic frame. This is a mature, well-populated art (the cited art spans 1987–2017 and includes two of the applicant's own prior Rockwell patents, R6 and R7).

A threshold point that matters for motivation: the patent's own specification concedes that "a coarse estimate of the encoder offset may be obtained during a rotate autotune process" and that the invention merely fine-tunes it. So the applicant admits the broad calibration problem and a working coarse solution are prior art; the asserted contribution is a specific fine-tuning recipe.

2. The asserted invention, decomposed into limitations

Claims 1 / 9 / 15 share one sequence (claim 9 prefaces it with "an inverter that controls a motor"; claim 15 with a "non-transitory computer readable storage medium"). Limitation-by-limitation:

• (L1) measure an initial encoder offset of the motor;
• (L2) iteratively reduce the encoder offset error with a third offset value until the motor starts;
• (L3) iteratively modify the offset with a first offset value until the motor stops (motor turning in the positive direction);
• (L4) record the offset for the stopped motor as a first encoder offset;
• (L5) iteratively modify with the first offset value until the motor starts turning in the negative direction (negative = opposite of positive);
• (L6) iteratively modify with a second offset value, of opposite magnitude to the first, until the motor stops again;
• (L7) record that value as a second encoder offset;
• (L8) calculate a final encoder offset as a function of (first offset, second offset, and a static friction ratio), the SFR being static friction (negative direction) ÷ static friction (positive direction).

Conceptually, L2–L7 are a classic bisection / bracketing search for a mechanical deadband boundary driven by the sign of residual torque (offset error → projected q-axis current → rotor creeps). L8 converts the two boundary values into the estimate. Only L8's "static friction ratio" is arguably new; everything else is drawn from two-direction offset-calibration art.

3. What the cited art discloses

R1 — Deere US 8,853,979 B2 (Wu & Shaw; priority 2011-02-28; issued 2014-10-07)

This is the closest cited reference and it is squarely on point. Per its abstract and claim set (Google Patents / FPO):

"…positive and negative quadrature-axis current commands are applied sequentially and at approximately same magnitude to urge the rotor toward an enforced position … measures a positive quadrature-axis current aligned raw position data … and … negative quadrature-axis current aligned raw position data … determines a raw averaging axis position data based on an average of the raw position data … An initial position offset calibrator … determines a position offset based on a difference between the determined true averaging axis and the determined raw averaging axis position data."

Mapping: R1 expressly teaches two opposite-direction excitations of substantially equal (i.e., opposite) magnitude (→ L6's "second offset value has an opposite magnitude from the first"), measuring position in each direction (→ L4, L7's first/second values), and averaging the two measurements to derive the offset (→ L8, minus the friction ratio). R1 also uses a "fake rotor position" and zero d-axis / nonzero q-axis injection and explicit offset equations (FPO claim set: θoffset = −(θpos_iq* + θneg_iq*)/2, with 0°/180° ambiguity handling). URLs: https://patents.google.com/patent/[US8853979B2](/patent/US8853979B2)/en ; https://www.freepatentsonline.com/[8853979](/patent/8853979).html

R2 — Eaton US 9,641,108 B2 (Suda & David; priority 2014-04-16; issued 2017-05-02)

Teaches iterative offset calibration: commutate the motor to known positions, measure encoder alignment at each step, compare to the known reference, compute alignment offset = difference, and adjust operation; and it teaches repeating and averaging to improve accuracy. Its spec: "ROTOR_OFFSET = AvgAB − (120/MOTOR_POLEPAIR)," with repeated N-step excitations and direction (CW/CCW) handling, and a "calculate an average alignment offset so as to account for asymmetries in the motor structure" claim. URLs: https://patents.justia.com/patent/[9641108](/patent/9641108) ; https://patents.justia.com/patent/9641108#2 (claims).

R3 — Hitachi US 2006/0125439 A1 (priority JP 2004-358115)

Concerns a synchronous motor drive using a rotary sensor and correcting the shift between the magnetic-pole position and the sensor output, i.e., detecting and compensating a rotor-to-position-sensor offset to drive efficiently. Directly supports L1 (obtaining/measuring the encoder–rotor offset) with a coarse estimation step. URL: https://patents.google.com/patent/US20060125439

R4 — Lei Hao US 2008/0061724 A1

"Position detection device for permanent magnetic machines" — signal-based position detection with correction of gain/offset/phase. Background for offset estimation and correction generally.

R7 — Rockwell US 2016/0248354 A1 / US 9,774,284 B2

Rotor-position estimation for a motor drive using inverter current sampling; same assignee as the patent. Corroborates that the assignee itself treats rotor-position/offset estimation as a routine drive function; inventor overlap (Matsuo) is notable.

R5, R6, R8, F1–F6

NSK (EPS control), Rockwell '224 (encoder pulse simulation), Deere '633 (notch-filter rotor position), and the family citations (Boeing object-positioning; Japanese position/speed detection & origin-return; JP 5414259 positioning control; Denso motor control; Delphi current-sensor offset correction) collectively establish that offset/detector calibration, bracketing/positioning searches, direction-dependent behavior, and origin-return are long-known in this art. The Japanese "origin position return" and "positioning control" references (F2, F4) are especially relevant to L2–L7, because finding a boundary by stepping a position command until a mechanical event occurs is exactly what positioning-control servos do.

4. Grounds of obviousness

Ground I (on cited art alone): Deere R1 + Eaton R2, optionally + Hitachi R3

• L1: R1's "fake position / initial position offset calibrator" and R2's offset-calibration routine; R3's sensor-to-rotor shift sensing.
• L2/L3/L5/L6: iterative, direction-reversed excitation with equal-and-opposite magnitude commands is expressly R1 ("positive and negative quadrature-axis current commands … at approximately same magnitude"). Stepping a value in increments until a mechanical threshold/event is reached is routine and is R2's iterative commutate-measure-compare loop.
• L4/L7: R1 measures and holds a positive-direction value and a negative-direction value; R2 records/stores the offset.
• L8: R1 averages the two directional measurements to obtain the offset. R2 averages to "account for asymmetries."

Result: every limitation except the specific friction-ratio weighting is disclosed or an obvious mechanical step. The generalized weighting of L8 is addressed in §5.

Ground II (strongest; supplementary art R9): R1 or R2 + Control Techniques US 9,559,621 B2 / US 2015/0365027 A1

R9 (Fischer; priority 2014-06-11; pub. 2015-12-17) describes the same bracketing method the patent claims: set a test offset value → cause/measure a displacement → repeatedly increase the test value until the measured displacement crosses a threshold (first crossing value) → repeatedly decrease the test value until it crosses again (second crossing value) → "determine the offset based on the first and second crossing values." Two passages are decisive for the friction ratio:

"…the offset may be determined as the mean of the first and second crossing values, or as a weighted average of the first and second crossing values. Under ideal conditions, the first crossing value would be equal to the second crossing value; however, due to mechanical friction and the load driven by the motor, these two values are not equal. By determining the offset based on both of these values, the accuracy of the determined offset is improved."

and, on directional asymmetry:

"…such an approach would not yield an accurate offset in situations where the mechanical conditions differ depending on the direction in which the rotor rotates; for example, in situations where there is a load in only one direction."

URLs: https://patents.justia.com/patent/[9559621](/patent/9559621) ; https://www.freepatentsonline.com/y2015/0365027.html

R9 thus discloses, in substance, (i) the two-crossing bracketing (L2–L7), (ii) averaging or weighted averaging (L8), and (iii) the reason to weight asymmetrically — direction-dependent friction/load — which is precisely the physical quantity the claimed SFR captures. A PHOSITA reading R9 would be motivated to make the averaging weight a function of the directional friction asymmetry; that is the claimed SFR.

Ground III: any of the above + R3/R4 (initial offset) and the negative-D-axis feature

The negative d-axis current of claims 2/10/16 is not even in the independent claims; it is a dependent limitation. Applying a d-axis (flux-axis) current to hold/align the rotor and then perturbing the offset is the standard "force a frame position and observe drift" technique and is the counterpart of R1's "enforced/fake rotor position" and R2's "excite phases to a known position." Selecting the specific current level (claim 3: between nameplate peak and max rated current; spec: highest non-demagnetizing value) is a routine design parameter selected against the machine's demagnetization curve — the kind of optimization KSR says is obvious absent unexpected results.

5. The crux: is the "static friction ratio" (L8) obvious?

This is where the patent is most defensible and where the art is strongest.

1. The claimed formula collapses to the known average. L8 defines the final offset via claim 8/14/20 as FEO = ((EO1 − EO2)/(1 + SFR)) + EO2. Setting SFR = 1 gives FEO = (EO1 + EO2)/2 — exactly the plain average that R1 (Deere) and R9 (Control Techniques) already disclose. The SFR therefore does not add a new physical step; it generalizes a known averaging step by a weighting factor that the art already recognizes should depend on directional friction/load (R9). Under KSR, recasting a disclosed average as a weighted average, with the weight keyed to a known asymmetry, is an obvious design choice / predictable variation.

2. The physics is known and the relationship is mathematical. R9 expressly attributes the inequality of the two crossing values to "mechanical friction and the load." Once that is known, deriving a closed-form weight from the ratio of directional friction is routine algebra, not an inventive insight. The patent itself states the average applies "if the static friction in both rotating directions is the same," and the interpolation "if the static friction in the negative direction is not equal." That bifurcation is a straightforward extension of the art's own teaching.

3. Motivation to combine is explicit and technical. R9 supplies: (a) the problem (inaccurate offset when directional conditions differ), (b) the mechanism (friction/load), and (c) the tool (weighted average). Applying R1's direction-reversed equal-magnitude excitation with R9's weighted two-crossing average yields the claim with a predictable improvement in accuracy — the classic KSR "known technique to improve a similar device in the same way."

Conclusion on L8: on the supplementary art, L8 is at most an obvious, mathematically predictable refinement. On the cited art alone, R1's average plus the ordinary recognition that friction differs by direction (corroborated by the positioning-control family citations) still supports a §103 rejection, though the friction-ratio limitation is then supported by reasoning rather than by an explicit reference disclosure. That gap is the patent's realistic line of defense.

6. Dependent claims (2–8 / 10–14 / 16–20)

Each dependant claim is a narrowing of, or an algebraic restatement of, subject matter already reached above; none introduces a new structural or process element that would independently confer non-obviousness.

7. Rebuttals the applicant could raise (and why they are weak here)

• "The art doesn't detect 'start/stop' from a physical breakaway; it measures encoder displacement against a threshold." R9 detects a threshold crossing of measured displacement under impeding friction/load — functionally the breakaway boundary the patent uses; the difference is one of sensing bookkeeping, not of result. R1 measures directional position excursions. Neither teaches away.
• "The art teaches away from direction-asymmetric weighting." It does the opposite: R9 expressly says the mean/weighted mean improves accuracy and warns that differing directional conditions make a single-sided approach inaccurate.
• Secondary considerations. None are of record here (and, per the earlier litigation section, no assertion or validity challenge has yet tested these claims). Objective indicia would require a nexus to the claimed SFR weighting; the patent's stated benefits (avoiding regenerative-mode bus-voltage rise/shutdown) flow from any accurate offset, which the coarse autotune and R1/R2 already provide, weakening the nexus.
• Potential genuine strength: if the applicant can show (a) that the cited art never quantifies the direction-dependent weighting, (b) that the specific SFR interpolation delivers an unexpected accuracy/robustness result beyond a plain weighted average, or (c) that a real examiner would not have found R9 (it is not on the face of the patent), claim 1 could survive. Point (c) is a practical, not legal, defense.

8. Bottom line

1. Every limitation of claims 1/9/15 except the static-friction-ratio calculation is squarely disclosed or is an obvious mechanical step over the cited art, principally Deere US 8,853,979 (R1) — equal-and-opposite direction excitations, measured first/second position values, and averaging — in view of Eaton US 9,641,108 (R2) (iterative measure/compare/record, averaging to account for asymmetry) and Hitachi US 2006/0125439 (R3) (rotor/sensor offset sensing).

2. The friction-ratio limitation (L8) is the only realistically contestable element. It is rendered obvious when the supplementary reference Control Techniques US 9,559,621 / US 2015/0365027 (R9) is added, because R9 discloses the identical two-crossing bracketing, expressly offers a weighted average, and expressly attributes the two-boundary inequality to direction-dependent friction/load — i.e., it supplies both the tool and the motivation. On R9, the claimed SFR is at most an obvious, predictable refinement of a known weighted average (and reduces to the known plain average when SFR = 1).

3. The dependent claims add nothing patentably distinct; they are routine parameter choices or algebraic restatements of the independent-claim formula.

4. Caveats I want on the record: (i) R9 is not on the face of US 10,996,079 — it is my own search finding, and a rejection built on it is only as good as an examiner's/ challenger's ability to locate it; the strongest cited-only case rests on R1+R2+R3 plus reasoning, which is weaker on L8. (ii) I did not retrieve the full specifications of R4, R5, R8 (and R6 only in abstract), so I characterize those by their abstracts/claims, not verbatim. (iii) This is a validity/patentability assessment, not a legal opinion; the patent enjoys a statutory presumption of validity and has not, per the earlier section, been tested in any proceeding found to date.

Suggested combination language for a §103 rejection: Claim 1 is obvious over Deere (US 8,853,979) in view of Control Techniques (US 9,559,621) [and optionally Eaton (US 9,641,108) and Hitachi (US 2006/0125439)]; the "static friction ratio" recited in the final calculating step being a predictable weighting of the two-directional offset measurements that the secondary reference expressly contemplates for direction-dependent friction.