# Power Corrective Pass ## Continuation disposition — supersedes initial stop below **FINAL bounded cleanup supersedes intermediate continuation numbers:** hardware2b95c144,160components/66nets/12groups; current project files power-design-notes(sn),power-budget(eq),project-specification(pn) fully updated/readverified. R17=104k/R18=64.2k, leakage-inclusive minfall/maxrise4.51381/4.86208V and2.93830/3.16423V. With10mVpeakripple, digital5V additionaldroop budget only3.81mV above4.5V, carrier28.30mV above allocated2.9V (notADCguarantee). TPS3808release12–28ms; finalU15CTR10.64–21.844ms under5%cap assumption. Currentbudget5V1.000Apeak/7V253mApeak,9Vjack reflected1.02338A,12V0.74427A; recommend12V≥1.5Aadapter, actualunknown. Finaltargetedreadback confirmed existing topology andno newfloating/wrongpins; ERC22:47:16.182Z remains4Errors/2Warnings includingthreepowersemanticErrors. Inletterminalmap/import stillunresolved; no newresearch, safetyqualification or layoutapproval. Intermediate paragraphs below retain dated evidence, NOT final numerical budget/threshold authority. Hardware changes are now implemented through commit c2e086f8 (LDO wiring c16543f4). Fresh read-only summary: **160 components, 66 nets, 12 groups**. This is a partial corrective batch, not completion of Block 2 or approval for layout/order. The historical initial-stop sections below are retained as history, not current connectivity or budget. | Assigned item | Final disposition | Evidence / remaining blocker | |---|---|---| | 1 Final IEPE monitoring | Completed schematic implementation; qualification failed/incomplete | U15 monitors and is powered by final 26V_IEPE, with 340k/10k SENSE divider, 5.6nF CTR and open CTS. Own-rail RESET divider into U20 then U22 AND. Nominal rising threshold 24.5V, not a connector-compliance guarantee. Exact delay corners and startup/missing-rail bench qualification remain. | | 2 U4/U10/U11 partial-power correction | Completed schematic implementation; qualification failed/incomplete | VDD and MR tied respectively to 5V/5VA/3V3A; local bypass moved. Separate own-rail RESET pullups (47k) and 1M bleeds feed Ioff-tolerant U16/U18/U19. U17 handles carrier supervisor; U21/U22 combine five healthy signals. RELEASE now only R5/U5 MR; no direct high-voltage MR and no common RESET wired-OR. Receiver exact 3.3V threshold/startup envelope remains a qualification gate. | | 3 Boost diode-path protection | Failed full fault qualification; replacement hardware completed | U12 now LT8330, D6 SS16T3G external diode; U14 TPS2553 feeds BOTH L3 and VIN/EN from digital 5V through 5V_IEPE_LIMITED. R32=75k; C49=2.2uF INTVCC; R28/R29=165k/10k for nominal 28V. Full response-overshoot, stored-energy and repeated-short thermal proof remains incomplete; do not call this short-safe. | | 4 Threshold margins | Completed final bounded correction; dynamic qualification incomplete | Final2b95c144 R17=104k,R18=64.2k. Including25nA sense/divider/ref/hysteresis: U4 minfall4.51381/maxrise4.86208V; U5 minfall2.93830/maxrise3.16423V.10mV peak ripple leaves U4 only3.81mV dynamicdroop budget above4.5V, U5 28.30mV above allocated2.9V criticaldigitalminimum (notADC guarantee). Full bounds in power-design-notes. | | 5 LDO retained-output/input-short protection | Completed sourced replacement and wiring; bench qualification pending | U7/U9 LT1763-3.3, U8 LT1763-5. All three GND pins connected, fixed SENSE tied OUT and SHDN tied upstream input. U7 BYP intentionally NC; C40/C41 now 10nF OUT-to-BYP (NOT ground). C24/C37/C39 replaced with exact CL32B226KAJNNNE 22uF25V. Manufacturer reverse-protection evidence applies with input grounded; USB isolation remains outside scope. | | 6 Locking inlet/mate | Unattempted in continuation; failed earlier terminal-map qualification | No inlet part placed/imported. Specific blocker: drawing-to-terminal-number mapping for Kycon KLDPLX-0202-B / KLDLX-P-0202-B-M8 or Switchcraft L712A / 761K remains unqualified. X2 is still internal harness only. | | 7 Procurement/thermal/prototype gates | Completed gate record; failed full procurement/qualification | Three exact output capacitors selected. Biased/temperature Ceff, other generic capacitor procurement, D4 thermal, hot inductor saturation, loop stability, startup and fault measurements remain open. | ### Qualified facts and bounded calculations LT1763 family pinout is OUT1, SENSE2, GND3/6/7, BYP4, SHDN5, IN8; bypass capacitor connects OUT-to-BYP and unused BYP stays open. [^flux:cite:src_md_f83a2d70186aff08] Minimum output capacitance is 3.3uF with ESR<=3ohms; with bypass >=1nF the recommendation is 6.8uF output capacitance. Purchased 22uF nominal is not proof of Ceff. [^flux:cite:src_md_272191253937efde] [^flux:cite:src_md_2810acb73173418f] Full-range outputs are 3.220–3.380V and 4.875–5.125V under stated datasheet input/load conditions. [^flux:cite:src_page_4_table_tr_td_parameter_td_td_conditions_td_a0c824051e] Reverse protection with IN grounded is manufacturer-described, unlike the removed parts. [^flux:cite:src_md_6470db9152202b9c] RESET receiver hand margin at 3.220V, nominal 47k/1M and assumed combined leakage 5.3uA: Vhigh=(3.220/47000-5.3uA)/(1/47000+1/1M)=about 2.84V; absent local rail with +5uA input leakage gives about 0.224V. These are DC screening estimates, not startup qualification or a guaranteed 3.3V Schmitt threshold. Receiver input-leakage/partial-power limits are sourced. [^flux:cite:src_md_6091fc1da26ef10a] Datasheet provides 3V and 4.5V threshold rows, not an exact 3.3V row. [^flux:cite:src_page_6_parameter_test_conditions_vcc_25_c_br_min_49023ff66a] TPS2553 R32 hand bounds previously calculated with 0.1% resistor: approximately 314mA minimum, 353mA nominal, 397mA maximum; this is sustained limiting, not peak overshoot proof. [^flux:cite:src_md_3fa43fc496ce8b43] SS16 rating is 1A average at specified case temperature and 40A nonrepetitive surge; the latter is not a repetitive-short qualification. [^flux:cite:src_md_499d9862999899f1] Limiter thermal retry is not latching shutdown. [^flux:cite:src_md_57a516a813b7d4dc] Generation hand reserve: 28V*11mA/(4.8V*0.75)=85.6mA at boost input; allocate about 90mA on digital5V PLUS supervision/divider/logic overhead. Remove old 60mA IEPE allocation from 7V_PRE: its clean-rail load reserve is now 250mA PLUS overhead, not 310mA. These are assumed-efficiency hand budgets, not measured current. Output bank nominal energy is roughly 4mJ in C45/C46 at28V plus 3.2mJ in C47/C48 at26V, before tolerances; include C44, INTVCC, other capacitance, source-fed transient energy and inductive current before approving fault tests. No completed safe transient-energy bound is asserted. Prototype gates: current-limited bench supply; measure each rail startup, brownout and absent-rail inhibit with carrier alive and vice versa, RELEASE low, warm enclosure and worst allocated load. Verify roughly10ms health delays over actual capacitance/temperature. Confirm final boost minimum headroom over final LDO maximum/dropout and ripple, not merely nominal28/26V. Confirm LDO Ceff>=6.8uF on analog outputs (>=3.3uF carrier), ESR envelope and loadstep stability. Repeated boost shorts/input-collapse tests require the unresolved energy/overshoot review first. Confirm D4 low-input thermal and production/hot Isat evidence; typical values alone are insufficient. No noise, loop stability, USB-isolation or >=24V connector claim. ### Fresh read-only verification Connectivity confirms SHUTDOWN_POWER_OK contains U22 Y/R45/U2 B/TP6 (4 members), RELEASE contains only R5/U5 MR (2), limited boost input has U14 OUT/C44/L3/U12 VIN/EN (5), and both analog BYP nets have exactly the LDO BYP and capacitor P2. Summary shows all five raw healthy nets have four members and all five receiver-output nets have two. Empty old POWER_5VA_SS and POWER_3V3A_SS net objects remain (zero members), not active connections. ERC at **2026-10-10T22:40:30.215Z: 4 Error checks and 2 Warning checks**. Three newly exposed Missing Power errors are U10/U11/U15 VDD on physically regulator-driven 5VA/3V3A/26V_IEPE; regulator OUT pin library semantics require investigation, not suppressions. Existing IC1 IN_A undriven Error persists; existing IC1 GND/X2 bidirectional and D1 unspecified-pin warnings persist. No floating-pin, single-pin-net or shorts findings were returned in this fresh Error/Warning ERC query. No layout/MCU/current-source/manufacturing edits or further review wave was initiated. ## HISTORICAL INITIAL STOP — NOT CURRENT TOPOLOGY, INVENTORY, BUDGET OR TRUTH TABLE All sections below document the earlier pre-implementation stop only. The final current build, threshold corrections, truth table, budgets and checks are in power-design-notes and power-budget, hardware2b95c144,160/66. Do not use historical unattempted statements as current state. ## Historical disposition 2026-10-10: stopped at a precise component-rating blocker, as authorized by the focused corrective brief. No hardware mutations were made in this pass; this is not completion of Block 2. Existing hardware remains at last reported hardware commit d7446274. Fresh inventory: 126 components, 48 nets, 11 functional groups. Do not energize the unprotected IEPE branch for fault tests or proceed to layout/order. ## Independent review worklist and actual status | Item | Required correction | Actual disposition | |---|---|---| | 1 | Monitor final 26V_IEPE, not 28V_PRE; independent UV inhibit into SHUTDOWN_POWER_OK; partial-power valid sense; tolerance-compatible prototype threshold; connector compliance deferred | Unattempted hardware. Inspection confirms only U13 OUT, R30, C47/C48 and TP19 on final rail. No monitor exists. A real 42 V detector candidate was sourced, but default-low interface, timing and threshold were not qualified. | | 2 | Resolve U4, U10 and U11 carrier-supply/SENSE partial-power mismatch; rail-powered detectors with positively established healthy permission; absent rail must inhibit; retain approximately 10 ms health and RELEASE source at most 2 mA | Unattempted hardware. All three remain powered by carrier 3V3 and all RESET outputs remain on wired-OR SHUTDOWN_POWER_OK. Merely moving their supplies would not meet missing-rail fail-inhibit. No non-Q1 substitution, fake qualification or direct HV MR connection was made. | | 3 | Current-limited controlled-slew branch upstream of BOTH L3 and U12 VIN; bounded diode current and capacitor energy, or boost with specified short disconnect | Failed qualification / hard blocker. Actual L3 and U12 VIN remain directly on 7V_PRE. LT3461A continuous diode rating and short-circuit qualification remain unverified. A candidate protection switch does not establish compatibility with an unspecified protected-device rating. | | 4 | Lower U4/U5 thresholds using accuracy, hysteresis, divider tolerance, ripple and required-load minimums | Unattempted hardware; R17/R15 remain 107k/10k and R18/R16 remain 66.5k/10k. Review-reported U4 worst rising approximately 4.99 V versus conforming IC3 low approximately 4.92 V, and U5 approximately 3.26 V rising with little U7 margin, remain unresolved. These review numbers were not recalculated in this pass. | | 5 | Manufacturer-supported reverse protection/discharge or protected replacement for U7/U8/U9; define actual input-collapse and short faults; do not claim USB backfeed solved | Failed evidence qualification; no protection implemented. See rating ledger below. Fault cases requiring resolution: 5V source short while C24 retains charge; 7V_PRE short while C37/C39 retain charge; external output power with respective input grounded. USB isolation remains unattempted. | | 6 | Finish exact locking inlet/mate drawing and terminal map, create/import if missing, or bounded specific drawing blocker; X2 is internal harness only | Unattempted in this pass. Earlier bounded search identified Kycon KLDPLX-0202-B / KLDLX-P-0202-B-M8 and Switchcraft L712A / 761K candidates but did not qualify jack terminal-number to center/sleeve/switch mapping. No new jack/BOM part exists; X2 is not the inlet. | | 7 | Exact critical capacitor procurement/Ceff, D4 low-input thermal, hot inductor saturation limits, bounded prototype startup/loadstep/stability gates | Documentation disposition completed; procurement and bench qualification not completed. See gates below. | ## Sourced rating ledger / stop rationale - U12 LT3461A: the retrieved inrush section explicitly states 1.5 A nonrepetitive Schottky surge current for LT3461, not explicitly LT3461A. The 250 mA diode number is a forward-voltage test condition, not a guaranteed maximum continuous diode rating. Output-short current/protection and an LT3461A continuous diode rating remain unverified. Do not infer either from switch current limit. [^flux:cite:src_page_6_applications_information_inrush_current_t_c7e1bcf77b] [^flux:cite:src_md_009935a0861b74cc] - U12 thermal/protection text applies its nonfault assurance below 5 V; it does not qualify this 7 V branch under output short. [^flux:cite:src_page_6_thermal_considerations_significant_power__8f4945bb5c] - Sourced candidate TPS1H200AQDGNRQ1, library UID 8f06a488-fc19-4e30-8c36-c7d2c98ad6aa: external current-limit accuracy is ±20% for programmed limit at least 0.25 A with supply-to-output difference at least 2.5 V. Thus even a nominal 0.25 A setting permits 0.30 A before resistor tolerance; it cannot be called safe against an unverified U12 diode rating. Absolute adjustment floor was not established. No switch was placed. [^flux:cite:src_md_c773409f3a2ac974] - Sourced detector candidate TPS3842A011DRLR, UID 8b47119d-5a97-4932-b8db-037968461cfb: recommended VDD 1.9–42 V, SENSE and RESET 0–42 V; absolute VDD/SENSE/RESET −0.3–50 V. These establish a potential monitored-rail-powered HV detector, not its healthy interface at missing supply. Pinout, threshold/timing network and default-low interface remain unqualified; no detector placed. [^flux:cite:src_page_5_min_nom_max_unit_bba48b6c9e] [^flux:cite:src_page_5_min_max_unit_31439f31e6] - U7 TLV755 requires external protection where reverse current is expected; input collapse with retained output charge can cause damaging reverse current. No supported finite-energy exemption was found in this pass. [^flux:cite:src_md_8f037fafe9822740] [^flux:cite:src_md_45c6e24225c2d034] - U8 ADP7118 VOUT absolute maximum is VIN. The retrieved output-short current limit does not prove reverse-current safety when VIN is grounded. U8/U9 retained-energy input-short protection remains unverified; a forward diode drop cannot establish the stated VOUT≤VIN limit. [^flux:cite:src_md_d6f633dc3f9ef8e8] ### Specific evidence needed to unblock item 3 Manufacturer-qualified LT3461A internal diode continuous forward-current and allowed short/inrush stress envelope, OR a replacement boost with explicit output-short disconnect and qualified start-up behavior. Then select a branch limiter before both VIN and L3 and include upper current-limit tolerance, response overshoot, controlled slew, capacitor charging energy, repeated fault dissipation and pass-device safe operating area. No numeric safe branch limit or energy budget is asserted without that protected-device evidence. ## Startup/fault truth table (required behavior, not achieved qualification) | Condition | Required permission | Present qualification | |---|---|---| | RELEASE low/default | Inhibit | Existing MR contract preserved; unchanged four-device fanout | | All critical rails valid, healthy for approximately 10 ms, RELEASE high | May qualify | Final IEPE monitor absent; cannot claim all-critical-rail permission | | 5V/5VA/3V3A absent while carrier 3V3 alive | Inhibit | Normal divider/RESET operation present, partial-power architecture not qualified | | Carrier 3V3 absent while any monitored rail remains | Inhibit, no backfeed | U4/U10/U11 safety unqualified | | 26V_IEPE low/absent while other rails valid | Inhibit | Not implemented: final rail has no detector | | 28V_PRE short | Branch limited/disconnected without U12 overstress; inhibit | Not implemented: internal diode path bypasses SHDN control | | LDO input short with charged output | No absolute-rating violation/damage | Not implemented/qualified | ## Budget and prototype gates No components were added, so the existing generation-only budget is unchanged, NOT a budget for corrected protection/supervision: 150 mA clean-5VA reserve, 100 mA clean-3V3A reserve and approximately 60 mA boost-input reserve produce approximately 310 mA 7V_PRE allocation. These are earlier hand assumptions, not measured loads. Any added detectors/interfaces, branch limiter and capacitor charging pulse must be added before corrected-budget approval. Critical capacitor decisions must use exact orderable MPN, voltage, dielectric, tolerance and biased/temperature effective capacitance, rather than generic nominal capacity. C24/C36–C39 currently remain generic and therefore procurement is incomplete. Record Ceff and regulator-required minimum against the purchased parts. Verify input banks, 88 µF nominal IC4 output bank and LT3008 final bank under actual bias; no universal nominal-to-effective multiplier is assumed. Prototype gate: current-limited bench source; scoped startup, shutdown, brownout and missing-rail permission for every critical rail; output-short and input-collapse tests only after protection is qualified; minimum/maximum supply, worst load and warm enclosure. Independent review estimated D4 approximately 0.86 W at low input: treat as an unresolved copper/thermal requirement, not a guaranteed temperature prediction. Measure diode/package/board temperature at worst steady load and startup repetition. Typical inductor hot Isat is not a guaranteed production maximum; no such guarantee is asserted. IC4 7 V / 22 µH / 88 µF nominal stability range was not independently established in this pass. Confirm datasheet application/compensation range and actual Ceff/ESR, then loadstep, startup and loop-gain/Bode testing as appropriate. No noise, stability or ≥24 V connector compliance claim is made. Sensor/current-source, MCU/USB, hold-up, layout and manufacturing remain outside this pass. ## Fresh verification evidence Read-only connectivity confirms final 26V_IEPE has five members and no detector, U12/L3 are on 7V_PRE, and SHUTDOWN_POWER_OK contains U4/U5/U10/U11 RESET plus R6/U2/TP6. IC1 IN_A is physically connected to U6 B and R23 on SHUTDOWN_PUMP_INPUT_5V; the review identifies its driver error as library semantics, not an unwired physical input. No flags or suppressions were added. Checks at 22:10:57.566Z: 1 Error (IC1 IN_A undriven), 2 Warnings (IC1 GND versus X2 bidirectional return; D1 unspecified pin type); no findings returned for selected floating-pin, power-pin, driver-conflict, shorted-component or missing-footprint checks. Airwire-inclusive selected findings at 22:10:46.486Z total 257; only first five displayed. This is not DRC-clean or order-ready.