Skip to main content

Power Sequence

Power Sequence - MacBook Pro Apple Silicon (M-Series Pro / Max)

Status: Family-level repair reference.

Primary scope: MacBook Pro 14-inch and 16-inch logic boards introduced with M1 Pro/Max and their later Pro/Max successors.

Reference designs: A2442/A2485 (M1 Pro/Max), A2779/A2780 (M2 Pro/Max), and later related designs. M3 Pro/Max (A2992) is included on the strength of its shared architecture, but every M3 statement on this page is an extrapolation from M1/M2 boards and has not been confirmed on an M3 board. A2337/A2338 diagrams are useful early Apple-silicon references, but their exact rail names and checkpoint numbering must not be assumed to apply unchanged to Pro/Max boards.

Important: This is a functional sequence, not a time-scaled waveform. Rail names, controller designators, voltages, phase counts, port topology, NAND configuration, and the order of closely grouped rails vary by board number and revision. Always confirm against the schematic or boardview for the board under test.

Board-level detail: The stage timings, designators and rail names in the board-level walkthrough further down apply to 820-02098 / 820-02100 only. They are not family-wide names. See the provenance note at the end of that section.

Evidence labels: [diagram] comes directly from a schematic or power-sequence diagram; [measured] is a working-board observation with the test conditions recorded; [inferred] is a functional interpretation and must not be treated as a guaranteed timing specification; [unverified] is asserted by a source but not confirmed against a schematic, boardview, or our own measurement.

Safety and measurement rules

  • Use a common logic-board ground for ordinary DMM measurements.
  • Use a 10x high-impedance oscilloscope probe with a short ground spring for low-voltage logic signals.
  • Never attach the ground clip of an earth-referenced bench oscilloscope to a switching node, MOSFET source, bootstrap node, or other non-ground point. Use a suitable differential or isolated probe for charger and multiphase switch nodes.
  • Do not probe crystal pins with a normal high-capacitance probe. Prefer a buffered clock output or test point; otherwise use an active low-capacitance probe.
  • Do not probe PCIe differential pairs with a normal passive probe. For power-sequence diagnosis, check the related power rails, reference clock presence where safely accessible, and reset signal instead.
  • Low resistance on SoC, GPU, RAM, and other sub-1 V rails can be normal. Compare diode-mode or resistance readings with the same board family, polarity, temperature, and capacitor charge state.

Architecture at a glance

StageReference checkpointsExpected functional resultPrimary instrument
A1-7Adapter attach, initial VBUS, port-controller startup, charger startup, and establishment of `PPBUS_AON`USB-C power meter and DMM
B8-9`P3V8AON_PWR_EN` enable chain and stable `PP3V8_AON`DMM and oscilloscope
C10-12MPMU/SPMU always-on rails and the SoC low-frequency clockDMM and oscilloscope
D13-14Power-on request accepted; AWAKE, S1/S2, memory, and NAND rail groups begin sequencingOscilloscope and DMM
E15-18SoC reference clock, reset release, SOCROM access, PMU communication, and early firmware self-testOscilloscope or logic analyzer
F19-24Port-controller communication, higher-voltage PD contract, and stable buck-mode system busUSB-C/PD analyzer, DMM, and oscilloscope

The checkpoint numbers are retained as a practical reference to the commonly circulated Apple-silicon sequence diagram. On Pro/Max boards, use the actual schematic net names and PMU sequence for the specific board rather than assuming a one-to-one match.

Stage A - Adapter attach and always-on system bus

  1. Cable/adapter attachment is detected. [diagram]
    One or more USB-C/ACE controllers detect the CC connection. MagSafe-equipped machines use an additional path, but the system still depends on the appropriate port-controller and charger communication.
  2. Initial VBUS appears. [diagram]
    A USB-C source normally begins at approximately 5 V before a higher-voltage PD contract is negotiated.
  3. The port-controller auxiliary supply starts. [diagram]
    The CD3217/ACE controller generates its local supply, commonly a rail such as PP3V3_UPC*_LDO (and, on the M1 generation, an internal PP1V1_UPC*_LDO core rail), and accesses its external ACE/SPI ROM. On the reference boards that ROM is UF260, read over SPI_ACE_MOSI / SPI_ACE_MISO / SPI_ACE_CLK / SPI_ACE_CS_L. [diagram] (board-scoped: 820-02098/02100)
  4. Input power reaches the charger path. [diagram]
    VBUS passes through the applicable input switches, fuse/current-sense elements, and becomes the charger-input rail. Names and designators vary by port and board.
  5. The charger controller becomes biased. [diagram]
    The charger IC receives its analog/digital bias and senses adapter, battery, and system-bus conditions. On the reference boards the charger is U5200 (ISL9240); the input path runs through current-sense resistor R5220 to its P_IN pin, CHGR_AUX_DET asserts above ~1.2 V, and the charger then raises PPCHGR_VDDA and PPCHGR_VDDP (the latter via R5275). [diagram] (board-scoped: 820-02098/02100)
  6. The charger establishes the system bus. [diagram]
    During the initial low-voltage-input state, the bidirectional buck-boost charger operates in boost mode — with U5200 switching Q5230/Q5240 across L5230 — to raise the 5 V input up to the system-bus level. This is established behaviour going back to the Intel CD3215 generation, not an inference from a single board.
  7. PPBUS_AON becomes stable. [diagram]
    The schematic nominal for the reference boards is 13.05 V. In the initial 5 V-input state the charger is still boosting toward that value; once the adapter input rises to 20 V the charger switches to buck mode and regulates the same bus from above. A single-board reading of ~12.3 V in boost and ~12.0 V in buck has been reported but is unconfirmed — the schematic value is the one to trust.

Stage A failure boundary

  • No initial VBUS: verify the adapter, cable, connector, CC path, and port-controller auxiliary supply before investigating downstream rails.
  • Initial VBUS present but absent after the ingress path: verify input switches, fuse/sense path, corrosion, and the charger-input net.
  • Charger input present but PPBUS_AON missing or collapsing: check for a short or overload on the bus, charger bias/enables, current-sense network, battery path, and charger power stage.

Stage B - 3.8 V always-on rail

  1. The 3.8 V regulator enable chain becomes valid. [diagram]
    On designs using the familiar charger support chain, CHGR_EN_MVR passes through the delay/buffer circuit and becomes P3V8AON_PWR_EN. Confirm the exact logic and designators on the target board (sometimes the Schmitt-Trigger is optimized away and CHGR_EN_MVR is directly attached to P3V8AON_PWR_EN [820-02841 "BATTERY CHARGER SUPPORT", pg 30].
  2. The multiphase regulator generates PP3V8_AON. [diagram][820-02841 "POWER: 3V8 AON", pg 34,35]
    This rail supplies the MPMU/SPMU and other always-on loads. On the reference boards this regulator is U5700, an RAA225501-class 3.8 V buck fed from PPBUS_AON. A stable enable with a missing output points toward the regulator, its bias/input rails, power stages, current-sense network, output short, or OTP/configuration issue.

Stage B failure boundary

  • PPBUS_AON present but P3V8AON_PWR_EN absent: work backward through the charger enable and delay/buffer chain.
  • Enable present but PP3V8_AON absent: check resistance to ground with power removed, then check the controller input/bias rails and switching activity with suitable probes. Also check the detailed PP3V8_AON page for in-depth details.
  • PP3V8_AON pulses or collapses: capture enable, output, and fault behavior on the same trigger instead of relying only on a DMM average.

Stage C - PMU always-on rails and low-frequency clock

  1. MPMU and SPMU always-on outputs appear. [diagram]
    Typical families include PP1V8_AON, PP1V2_AON, PP3V3_AON, early S2 rails, and board-specific PMU LDO outputs. On the reference boards PP3V8_AON arrives at master PMU U8100 as PMU_VDDHI. Use exact schematic names; not every rail exists under the same name on every generation.
  2. The PMU low-frequency oscillator operates. [diagram]
    The PMU oscillator/crystal starts — on the reference boards crystal Y8100, producing MPMU_CLK32K. This must not be confused with the separate SoC 24 MHz reference clock (Y2000) that appears later.
  3. PMU_CLK32K_SOC reaches the SoC. [diagram]
    The approximately 32.768 kHz clock allows the SoC always-on domain to continue sequencing. PMU_CRASH_L sits high (~1.8 V) while the PMU watchdog reports no overvoltage or thermal trip, and the PMUs move into the S2 standby state (board-scoped examples: PP1V2_S2, PP5V_S2_MAIN, PP0V72_S2_VDDLOW). [unverified] (board-scoped net names)

Stage C failure boundary

  • Check the PMU input rails before condemning a missing PMU output.
  • Check PMU crash/fault signals where available, such as PMU_CRASH_L.
  • Use a low-capacitance probe for the clock. A DMM frequency reading is not sufficient evidence of clock integrity.

Stage D - Power-on request and switched rail groups

  1. A valid wake or power-on request is received. [diagram]
    Depending on model and state, the initiating event can involve the power button, lid-open detection, charger attachment, or another PMU wake source. In the early reference diagram, PMU_ONOFF_L is the explicitly numbered request signal.
  2. The PMUs sequence AWAKE, S1/S2, memory, and NAND rails. [diagram]
    These rails form a group rather than one guaranteed simultaneous event. Pro/Max examples use board-specific SoC, GPU, CPU-SRAM, memory, I/O, and NAND rail names. NAND generations may use names such as PP2V58_AWAKE_NAND_VCC/PP1V25_AWAKE_NAND_VCCQ/PP0V88_AWAKE_NAND_VDD or channel-specific rails such as PP0V9_NANDx/PP1V2_NANDx. On the reference boards the family is PP1V2_NAND0 / PP0V9_NAND0 / PP2V5_NAND0 with PMICs UN400/UN480. Never substitute one naming family for another without checking the board schematic.

Between the S2 rails of Stage C and the AWAKE rails above, the reference boards wake a multi-phase controller designated U9300 ("Viper"): P1V8VDDH_PWR_EN asserts, CLVR_RESET_L releases the integrated power stages ("Monaco"), and PP1V8_S1_CLVR_VDDH comes up to feed the CPU, GPU and SRAM awake domains. [unverified]

Stage D failure boundary

  • A missing power-on request is different from a PMU that receives the request but fails to sequence its outputs. Capture both the request and first missing rail.
  • If one NAND rail is missing, check its owning PMU channel, enable, inductor/output path, and resistance to ground before assuming failed NAND.
  • DFU visibility proves that part of the SoC, USB, clock, reset, and firmware path is alive; it does not by itself prove that all NAND or operating-system boot stages are healthy.

Stage E - SoC clock, reset release, and early firmware

  1. The SoC reference oscillator starts. [diagram]
    A reference design shows a separate 24.000 MHz oscillator — Y2000 on the reference boards. Confirm the frequency and test point for the actual board.
  2. SoC reset is released. [diagram]
    PMU_RESET_L or the board-equivalent active-low reset transitions high and releases the SoC COLD_RESET* input. On the reference boards PMU_ACTIVE_READY asserts first.
  3. The SoC accesses SOCROM/boot storage. [diagram]
    Look for activity on the applicable SPI_SOCROM_* or equivalent external boot-ROM interface — on the reference boards flash U2010, with SPI_SOCROM_1V8_CS_L going low as the SoC reads its initial-stage bootloader. Activity proves bus access, not valid contents.
  4. Early firmware initialization and self-test proceed. [diagram]
    The reference sequence identifies internal initialization, RAM verification, SPMI/I2C device checks, NAND access, EEPROM/configuration checks, and later trackpad/Touch ID communication. The exact order and visibility vary by model and firmware. The embedded SMC and Secure Enclave cores come up in this window, after which the SoC opens the management buses: SPMI to the PMUs, and I2C to the NAND PMICs, the charger, and the port controllers.

Stage E failure boundary

  • Reference clock present but reset held low: diagnose PMU prerequisites, fault inputs, missing rail groups, and PMU communication.
  • Reset released but no SOCROM activity: confirm ROM power, chip select, clock, signal integrity, and the SoC-side prerequisites.
  • SOCROM activity followed by a repeating reset: capture the period and identify which rail, reset, or PMU fault changes first.

Stage F - PD negotiation and stable buck-mode operation

  1. The system communicates with the port controller. [diagram]
    The reference design shows I2C/interrupt communication between the system controller and USB-C port controller. The request for a higher-voltage contract originates in the embedded SMC, which commands the CD3217 over I2C — the port controller then runs the PD messaging itself.
  2. The port controller processes CC/PD communication. [diagram]
    USB-C Power Delivery messaging occurs on the CC line. A proper PD analyzer is more informative than a simple voltage/current display.
  3. VBUS rises to the negotiated higher voltage. [diagram]
    Common USB-C operation is approximately 20 V. A 28 V EPR/MagSafe-capable result applies only to models, adapters, cables, and ports that support it; it is not a universal expectation.
  4. The charger detects the elevated input. [diagram]
    The adapter-detect path, shown as AUX_DET on a reference design, confirms that the higher input is available.
  5. The charger operates in buck mode. [diagram]
    With adapter input above the required system-bus voltage, the charger regulates down to the always-on bus and manages the battery path as commanded. What triggers this mode change on U5200 is not settled — see Open questions.
  6. PPBUS_AON remains stable in buck mode. [diagram]
    The system bus should remain near its board-specific nominal value (13.05 V on the reference boards) while adapter VBUS is higher.

Beyond checkpoint 24 the reference boards continue into display and peripheral bring-up: the SoC asserts LCD_PWR_EN to the T-CON connector (J8500), verifies the panel over I2C_SEEPROM_*, checks the Lid Angle Sensor (IPD_LID_OPEN must read high, ~1.8 V), then ramps the mini-LED backlight boost to 40–44.5 V on PPVOUT_S0_LCDBKLT and asserts WLBT_PWR_EN for the wireless modules. Display, audio, PCIe, Thunderbolt and macOS boot are later consumers and must not be treated as the definition of checkpoint 24. [unverified]

Stage F failure boundary

  • Initial 5 V present but no higher-voltage contract: check CC integrity, port-controller power/ROM, system-to-port-controller communication, adapter/cable capability, and firmware state.
  • Higher-voltage contract appears but the board resets: capture adapter VBUS, charger input, PPBUS_AON, PP3V8_AON, and reset together to find the first collapsing node.
  • Do not assume that successful 20 V negotiation means the complete boot sequence succeeded.

Board-level walkthrough - 820-02098 / 820-02100

Board scope. Every designator, rail name, timing and current figure in this section applies to 820-02098 (M1 Pro 14") and 820-02100 (M1 Pro/Max 16") only. Do not lift these names onto another board number without checking its schematic.

Provenance. The charger path (Stage 1) is schematic-confirmed. Everything else in this walkthrough comes from a technician's annotated sequence and has not been confirmed against a schematic, boardview, or our own measurement — hence [unverified]. The timings in particular are single-board observations, and the page's standing rule (see Interpretation limits and T01) still applies: a current value is not a fault code.

Mapping to the family stages: walkthrough Stage 1 → A; Stage 2 → C; Stage 3 → C→D; Stage 4 → D→E; Stage 5 → D/E→F; Stage 6 → beyond F.

Walkthrough stage Reference checkpoints (approx.) Timing [unverified]
1 - 5 V input, ACE-ROM handshake, initial PPBUS_AON boost 1-7 (A) 0 - ~80 ms
2 - MPMU/SPMU and AON/S2 rails 10-12 (C) ~80 - 350 ms
3 - Viper/Monaco core awake rails 13-14 (D) ~350 ms - 1.1 s
4 - Embedded SMC, 24 MHz, SOCROM execution 15-18 (E) ~1.1 - 2.8 s
5 - NAND power sequence and 20 V PD contract 19-23 (F) ~2.8 - 4.5 s
6 - Display, backlight and peripheral enables beyond 24 ~4.5 - 6.2 s

Stage 1 - 5 V input, ACE-ROM handshake and initial boost

Schematic-confirmed ([diagram], 820-02098/02100): PPVBUS_USBC* arrives; the attached CD3217 enables PP3V3_UPC*_LDO and PP1V1_UPC*_LDO; it reads UF260 over the SPI_ACE_* bus; PPDCIN_AON appears as the input FETs close; the path runs through R5220 to U5200's P_IN; CHGR_AUX_DET asserts; PPCHGR_VDDA / PPCHGR_VDDP come up; U5200 boosts through L5230 / Q5230 / Q5240 to raise PPBUS_AON; P3V8AON_PWR_EN (EN_MVR) enables U5700 to produce PP3V8_AON.

Reported input signature [unverified]: 5.0-5.2 V, 0.00-0.02 A with the bus not yet established.

Stage 2 - MPMU, SPMU and AON/S2 rails

[unverified] PP3V8_AON feeds U8100 as PMU_VDDHI; U8100 brings up PP1V8_AON, PP1V2_AON and PP3V3_AON; Y8100 starts the 32.768 kHz MPMU_CLK32K; PMU_CRASH_L reads high; the PMUs enter S2 (PP1V2_S2, PP5V_S2_MAIN, PP0V72_S2_VDDLOW); PMU_CLK32K_SOC is handed to the SoC.

Reported signature: ~0.05 A static at 5 V. This is the classic "bus established but the PMU state machine is held" plateau — see [[m3-spmu]] and [[m1pro-soft-short]] for two different rail faults that produce it.

Stage 3 - Viper, Monaco and the core awake rails

[unverified] U8100 releases P1V8VDDH_SLP_L and asserts P1V8VDDH_PWR_EN to wake U9300; CLVR_RESET_L releases the Monaco stages; PP1V8_S1_CLVR_VDDH appears and the awake domains follow — PPVDD_CPU_E_AWAKE (~0.65-0.85 V dynamic), PPVDD_CPU_P_AWAKE (~0.70-0.90 V), PPVDD_GPU_AWAKE (~0.60-0.80 V), the SRAM rails, and the fabric supplies.

These rails transition within milliseconds, so a cracked switch-node solder joint fails here under dynamic load rather than showing up at rest — see [[dfu-inductor]] for the L5800 / PP1V2_AWAKE case.

Stage 4 - Embedded SMC, 24 MHz crystal and SOCROM

[unverified] PMU_ACTIVE_READY asserts and PMU_RESET_L is released; Y2000 starts the 24 MHz reference; the SoC Boot ROM runs and pulls SPI_SOCROM_1V8_CS_L low to read the LLB from U2010; the Secure Enclave and embedded SMC cores initialise and PMU_SYS_ALIVE asserts; the SoC opens SPMI and the I2C management buses.

Stage 5 - NAND power sequence and 20 V contract

[unverified] The SMC sequences the NAND PMICs UN400/UN480: PP1V2_NAND0 and PP0V9_NAND0 appear, P2V5_NAND0_EN enables PP2V5_NAND0, NAND0_RESET_L de-asserts, and NAND0_CLK24M_R starts. With the storage rails stable the SMC commands CD3217 over I2C to request the 20 V profile; PPDCIN_AON steps to 20 V; U5200 transitions from boost to buck and holds PPBUS_AON at its nominal from above.

Reported signature: current ramps from ~0.25 A at 5 V to 0.65-1.10 A at 20 V.

Stage 6 - Display, backlight and peripherals

[unverified] See the Stage F note above for the chain (LCD_PWR_EN, J8500, I2C_SEEPROM_*, IPD_LID_OPEN, backlight boost to 40-44.5 V on PPVOUT_S0_LCDBKLT, WLBT_PWR_EN). Two wiki case pages sit directly on this stage: [[m1-las-backlight]] (a degraded LAS asserting "lid closed" and killing the backlight) and [[miniled-arc]] (the 40 V+ boost arcing through moisture at the hinge flex).

Open questions

  • What triggers the charger's boost → buck change on U5200? It is not yet established whether the mode change follows a dedicated detect pin (PBUS_PWR), part of U5200's own I2C protocol with the SMC, or both. Until this is settled, describe the transition by its observable effect (bus regulated from above once VBUS is high) rather than by a named trigger. This supersedes the older claim F04 wording.
  • M3 (A2992) coverage. The M3 boards share this architecture, but no statement on this page has been checked on an M3 board. Confirm before relying on any M3-specific figure.

Working-board measurement checklist

Purpose: Capture a reproducible baseline from a known-good board of the same board number, revision, memory/NAND configuration, and test state. Do not publish a value as universal unless it has been confirmed across multiple boards.

Record the test conditions first

  • ☐ Model and board number/revision:
  • ☐ SoC and RAM configuration:
  • ☐ NAND configuration:
  • ☐ Battery connected/disconnected and battery state of charge:
  • ☐ Display, keyboard/trackpad, fans, speakers, and other assemblies connected:
  • ☐ Adapter wattage, PD/EPR capability, cable type, and selected port:
  • ☐ Trigger used: adapter insertion, power button, lid open, DFU command, or other:
  • ☐ Board temperature and starting state:
  • ☐ Oscilloscope model, probe type, attenuation, bandwidth limit, sample rate, and trigger:
  • ☐ DMM model and measurement mode:

Powered checks

DoneStageInstrumentMeasure/verifyExpected result on a working boardRecord actual result
☐AUSB-C power meter or PD analyzerInitial adapter contractInitial approximately 5 V contract; current is dynamic and test-condition dependent
☐ADMMVBUS at connector and both sides of the input protection/sense pathNo unexplained voltage loss across the path
☐ADMMPort-controller auxiliary LDO, such as `PP3V3_UPC*_LDO`Approximately 3.3 V and stable
☐AOscilloscope or logic analyzerACE/SPI ROM `CS`, `CLK`, `MOSI`, and `MISO`Boot-time transaction present; valid logic levels with no stuck line
☐ADMMCharger bias and input-detect railsMatch the schematic values and remain stable
☐ADMM; oscilloscope for transients`PPBUS_AON`Stable at the board-specific nominal value — 13.05 V on the reference boards
☐BOscilloscope`CHGR_EN_MVR` and `P3V8AON_PWR_EN`, where usedClean enable transition without repeated pulsing; delay matches the board design
☐BDMM and oscilloscope`PP3V8_AON`Approximately 3.8 V, stable; record startup ramp and ripple
☐CDMMAll schematic-listed MPMU/SPMU AON and early S2 railsEach rail reaches its specified nominal value and remains stable
☐COscilloscope with low-capacitance probe`PMU_CLK32K_SOC`Approximately 32.768 kHz, continuous and with valid logic amplitude
☐CDMM or oscilloscope`PMU_CRASH_L` or board-equivalent fault outputInactive/healthy state according to signal polarity
☐DOscilloscope`PMU_ONOFF_L`, lid-open input, or board-specific wake requestA valid request transition followed by rail sequencing
☐DDMM and oscilloscopeAWAKE, S1/S2, CPU/GPU, RAM, I/O, and NAND rail groupsCorrect nominal voltages; no collapse or repeated restart
☐EActive or low-capacitance oscilloscope probeSoC reference clock at a safe buffered pointBoard-specified frequency, commonly 24 MHz in the reference sequence
☐EOscilloscope`PMU_RESET_L`/`COLD_RESET*`Held low during prerequisites, then transitions high once without cycling
☐EOscilloscope or logic analyzer`SPI_SOCROM_*` or equivalent boot-ROM busBoot-time access on chip-select, clock, and data lines
☐EOscilloscope or logic analyzerSPMI and relevant I2C busesTransactions present; no rail-clamped or permanently stuck clock/data line
☐D/EDMM and oscilloscopeNAND supplies and NAND resetCorrect board-specific voltages; reset releases after supplies stabilize
☐FUSB-C/PD analyzerPD negotiation and final contractHigher-voltage contract appropriate to model and adapter, commonly 20 V; 28 V only where supported
☐FDMM and oscilloscopeAdapter input, `PPBUS_AON`, and `PP3V8_AON` during contract changeNo collapse or reset while VBUS changes; system rails remain regulated
☐FUSB-C power meterCurrent profile from attach through boot and idleRecord the complete profile; do not reduce it to a universal single current value

Powered-off baseline checks

Disconnect the adapter and battery, discharge the board safely, and document the same meter polarity for every measurement.

DoneInstrumentMeasureRecord
☐DMM resistance mode`PPBUS_AON` to groundInitial and settled resistance, polarity, temperature, and settling time
☐DMM diode mode`PPBUS_AON`, `PP3V8_AON`, PMU AON rails, and NAND railsForward/reverse polarity readings using the same meter and board state
☐DMM resistance or diode modeEach low-voltage CPU/GPU/RAM railComparison value only; low readings are not automatically shorts

Quick diagnostic flow

Adapter connected
|
+-- No initial VBUS
|   +-- Check adapter, cable, connector, CC path, and port-controller auxiliary power
|
+-- Initial VBUS present, no port-controller LDO/ROM activity
|   +-- Check controller bias, local LDO, reset, ROM power, and SPI bus
|
+-- Charger input present, PPBUS_AON missing/collapsing
|   +-- Check bus short/overload, charger bias, sense network, power stage, and battery path
|
+-- PPBUS_AON good, PP3V8_AON missing
|   +-- Check CHGR_EN_MVR -> delay/buffer -> P3V8AON_PWR_EN -> U5700 (RAA225501-class)
|
+-- PP3V8_AON good, PMU AON/S2 rails or 32 kHz clock missing
|   +-- Check MPMU/SPMU inputs, fault signals, oscillator, and exact missing output channel
|
+-- Rail groups good, reset never releases
|   +-- Check wake request, PMU prerequisites, PMU communication, and fault inputs
|
+-- Reset releases, no SOCROM/firmware activity
|   +-- Check reference clock, SOCROM power/bus, reset integrity, and early SoC prerequisites
|
+-- Early boot activity present, no higher-voltage PD contract
|   +-- Check system-to-port-controller communication, CC/PD path, adapter/cable, and firmware state
|
+-- Higher-voltage contract appears, then board resets
    +-- Trigger on reset and capture VBUS, charger input, PPBUS_AON, PP3V8_AON, and the first collapsing rail

Interpretation limits

  • A voltage being present does not prove that its ripple, startup ramp, sequencing, or current capability is correct.
  • Bus activity does not prove that the transmitted data or firmware contents are valid.
  • DFU or Recovery visibility does not prove NAND integrity or a complete operating-system boot.
  • A successful 20 V or 28 V contract does not prove that all SoC, display, storage, or peripheral stages completed.
  • Current draw alone is not a unique fault code. Always correlate it with rails, reset, clocks, and bus activity.

Editor Verification Checklist

Purpose: This section tracks statements that are inferred, generalized across models, based on workshop experience, or not yet tied to primary evidence. It is an editorial checklist, not part of the electrical power sequence itself.

Rule: A plausible explanation is not the same as a verified fact. Until a claim passes the checks below, retain the [inferred]/[unverified] label, narrow its wording, or remove it.

Status legend

  • ☐ Unverified - claim currently has no adequate supporting record.
  • ☐ Diagram verified - confirmed directly in a schematic, boardview, or manufacturer document.
  • ☐ Measured once - observed on one known-working board with recorded conditions.
  • ☐ Cross-board verified - confirmed on multiple applicable board numbers or revisions.
  • ☐ Inferred but retained - useful interpretation, explicitly labeled and not presented as fact.
  • ☐ Reword or remove - evidence does not support the current wording.

Minimum evidence required for every claim

  • ☐ Exact Mac model, EMC/model identifier, board number, and PCB revision are recorded.
  • ☐ SoC variant, RAM capacity, and NAND configuration are recorded where relevant.
  • ☐ Source type is recorded: schematic, boardview, datasheet, oscilloscope capture, DMM measurement, USB-C/PD log, repair result, or firmware log.
  • ☐ A schematic claim includes the page/sheet, component designator, and exact net name.
  • ☐ A measured claim includes battery state, connected assemblies, adapter/cable, selected port, trigger event, board temperature, and instrument configuration.
  • ☐ An oscilloscope claim includes probe type, attenuation, bandwidth limit, sample rate, time base, trigger source, and ground/reference method.
  • ☐ A DMM claim includes meter model, measurement mode, polarity, and whether the board was powered or fully discharged.
  • ☐ The raw evidence is attached or linked instead of only recording the author's conclusion.
  • ☐ The claim states whether it applies to one board, one generation, or the complete Pro/Max family.
  • ☐ Conflicting evidence and known exceptions are recorded.

Claims requiring verification

IDClaim to auditWhy it is not automatically provenAcceptable verificationEditor decision
A01Every supported model initially receives approximately 5 V before negotiating a higher-voltage contract.USB-C, MagSafe, EPR, battery-only startup, and already-negotiated reconnect behavior may differ.Capture PD logs and VBUS waveforms for each input method and supported generation from a fully off state.☐ Verified ☐ Add exceptions ☐ Narrow to USB-C
A02The CD3217/ACE controller always generates a local `PP3V3_UPC*_LDO` and loads an external SPI ROM before the remaining sequence.Controller generation, ROM topology, rail name, and port arrangement vary.Confirm controller part number, local LDO, ROM device, supply, reset, and bus activity for each board family.☐ Verified boards: \_\_\_\_\_\_ ☐ Generalize wording ☐ Remove exact IC
A03The charger initially operates in boost mode to create a 12-13 V `PPBUS_AON` from 5 V.**Resolved:** the boost mechanism is established practice (CD3215 generation onward) and is now stated as such in Stage A. The remaining uncertainty is the trigger for the later boost→buck change (see F04).Scope the switch node with a differential probe to document the mode boundary and the bus value in each mode.☑ Mechanism accepted ✗ Old "12-13 V from 5 V" figure superseded by the 13.05 V schematic nominal
A04`PPBUS_AON` is normally 12.0-13.1 V across the supported family.**Partly resolved:** the schematic nominal for the reference boards is 13.05 V. Whether other board numbers share it is still open.Cite the schematic nominal per board; measure the boost-mode and buck-mode values separately on a known-good board.☑ Schematic nominal 13.05 V recorded ☐ Per-board table still required
B01The universal enable path is `CHGR_EN_MVR` → delay/buffer → `P3V8AON_PWR_EN`.This path is verified on known designs but may change with charger or PMIC generation.Trace the enable path and logic polarity in each supported schematic; measure both sides of the buffer on a working board.☐ Verified boards: \_\_\_\_\_\_ ☐ Add alternate path
B02`PP3V8_AON` directly powers the primary MPMU/SPMU and specific audio/wireless circuits.Downstream loads and intermediate regulators vary; a net name alone does not prove every listed consumer.Use schematic net references or boardview connectivity to enumerate confirmed consumers by board.☐ Consumer list verified ☐ Remove unverified consumers
B03Missing `PP3V8_AON` with enable present implicates the RAA225501 regulator.**Partly resolved:** on the reference boards this is `U5700`, an RAA225501-class part, so designator and part number are now cross-referenced. The fault can still be input/bias loss, a failed external MOSFET, current-sense fault, output short, open inductor/trace, or configuration mismatch.Require input, bias, enable, fault, phase activity, resistance, and output-path checks before naming the controller.☑ Designator/part reconciled (U5700 = RAA225501-class) ☐ Diagnostic wording is non-exclusive ☐ Rewrite
C01Generic names such as `PP1V2_AON` or `PP0V8_AON` accurately represent every board's PMU outputs.Some boards divide these domains between MPMU/SPMU or use more specific names and different nominal voltages.Create a per-board AON/S2 rail table copied from the applicable schematic.☐ Table added ☐ Generic names marked as examples
C02The PMU crystal in the early stage is 24 MHz.The reference drawing separates the PMU low-frequency oscillator from the later SoC 24 MHz reference clock. The board-level walkthrough agrees: `Y8100` (32.768 kHz) is early, `Y2000` (24 MHz) is later.Read the crystal value/BOM and measure a buffered output. Do not infer frequency from component placement.☑ Claim already contradicted by the walkthrough ☐ 24 MHz claim removed
C03`PMU_CLK32K_SOC` is continuously present at 32.768 kHz on all supported boards.Clock gating, state, amplitude, and routing may differ.Capture frequency, amplitude, duty cycle, and availability across off, attach, DFU, boot, sleep, and running states.☐ Verified conditions: \_\_\_\_\_\_ ☐ Add state qualifier
D01Power button, lid open, and charger insertion all produce the same effective PMU wake event.These are different inputs and may enter different firmware or power states.Capture the relevant PMU inputs and first rail/reset response for each trigger independently.☐ Separate sequences documented ☐ Generalize wording
D02The listed AWAKE/S1/S2, CPU/GPU, RAM, and NAND rails start together at one checkpoint.A diagram may group outputs that actually have measurable internal ordering and dependencies.Perform a multi-channel scope capture or repeatable single-channel captures using one common trigger. Record first-rise time and stable time for every rail.☐ Timing captured ☐ Keep grouped without order claim
D03NAND rail names and nominal values can be generalized across generations.NAND count, vendor, interface, PMU ownership, and voltage domains vary. A third naming family (`PP1V2_NAND0`…) has now been added, board-scoped.Build a per-board NAND power table including controller/PMU channel, inductor, nominal voltage, enable, power-good/reset, and NAND devices supplied.☐ Table added ☑ Naming families kept separate, each scoped
D04A particular USB-C current plateau indicates failed NAND or DFU state.Current is affected by battery, port, adapter, connected assemblies, temperature, firmware state, and many unrelated faults.Record complete current/time profiles from multiple known-good and failed boards, then correlate with confirmed rails, reset, DFU visibility, and final repair result.☐ Evidence set attached ☐ Reword as weak clue ☐ Remove
E01The SoC reference clock is universally 24.000 MHz and starts at the same point.Clock source, frequency, enable ownership, and accessibility can change.Confirm BOM/schematic value and capture a safe buffered clock point for each generation.☐ Verified boards: \_\_\_\_\_\_ ☐ State as reference example
E02`PMU_RESET_L`/`COLD_RESET*` transitions high only after all named rail groups are valid.The PMU may monitor only selected rails and internal conditions; some rails may start after reset release.Capture reset with each candidate prerequisite or identify the dependency explicitly in the schematic/PMU documentation.☐ Dependency proven ☐ Reword to observed order
E03SOCROM SPI activity proves successful Secure Enclave or boot-ROM verification.Electrical activity only proves an attempted transaction, not valid content, authentication, or successful execution.Correlate bus activity with reset behavior, DFU/recovery status, firmware logs, and a known-good ROM comparison where legally and technically appropriate.☐ Wording limited to bus access ☐ Additional proof attached
E04The SoC self-test order is RAM → SPMI/I2C → NAND → EEPROM → trackpad/Touch ID.The reference annotation may be simplified and firmware versions can parallelize or reorder operations.Use synchronized bus captures, reset/rail captures, firmware logs, or authoritative documentation. Otherwise call this a functional list, not a strict order.☐ Strict order proven ☐ Reword as unordered checks
F01Who requests the higher-voltage PD contract at checkpoint 19?**Resolved:** the embedded SMC commands the `CD3217` over I2C; the port controller then runs the PD messaging. Adopted as established, not inferred.Decode the SMC→port-controller transaction if a formal record is ever wanted.☑ Adopted as established
F02Every supported board should negotiate 20 V, or 28 V when connected through MagSafe 3.Contract depends on model, adapter, cable, battery, port, firmware, and USB-PD/EPR capability.Record PD analyzer logs for every claimed model/input combination and list expected PDO/EPR behavior separately.☐ Compatibility table added ☐ Remove universal expectation
F03`AUX_DET` proves that the charger has accepted and is using the elevated adapter input.A detect voltage may prove only threshold detection, not correct regulation, current capability, or selected power path.Measure `AUX_DET`, charger input, system bus, battery current, and charger switching behavior together.☐ Functional meaning proven ☐ Wording narrowed
F04The charger switches from boost to buck mode once the adapter input is elevated.**Partly resolved:** the switch itself is established behaviour. What is *not* settled is the trigger — a dedicated detect pin (`PBUS_PWR`), `U5200`'s own I2C protocol, or both. The old "clean switch at checkpoint 23" wording is withdrawn.Identify the trigger signal in the schematic and confirm on a board; capture the switch node with a differential probe during the transition.☑ Effect accepted ✗ Old "clean switch at chk 23" wording withdrawn ☐ Trigger still open
F05Successful higher-voltage negotiation means the full power sequence completed.PD success can occur before NAND, display, peripheral, or operating-system boot succeeds.Keep this claim rejected unless all later rails, reset, buses, display output, and OS state are independently verified.☐ Claim removed ☐ Explicit limitation retained
T01A specific USB-C current value uniquely identifies a failed stage.Current draw is non-unique and strongly dependent on test configuration and time resolution. The board-level walkthrough quotes current figures; they are observations only.Use current only as supporting evidence. Require at least one rail, clock, reset, bus, or thermal observation before assigning a stage.☐ Current values labeled as observations ☐ Universal thresholds removed
T02A component or fault is "common" or "very common."Workshop memory and online reports are subject to selection bias.State the sample size, date range, board families, confirmed repairs, and denominator. Otherwise use "possible" instead of "common."☐ Dataset cited ☐ Wording changed to possible
T03DFU restore is the correct action whenever firmware corruption is suspected.Restore can erase data and cannot repair missing rails, unstable clocks, damaged NAND, or hardware resets.Require stable prerequisite rails, clocks, reset, DFU detection, customer authorization/data-risk assessment, and a documented restore error/result.☐ Preconditions added ☐ Data-loss warning added
G01The board-level stage timings (0-80 ms through 4.5-6.2 s).Reported from one annotated single-board capture; no instrument setup, trigger or conditions are recorded, and the page is explicitly not a time-scaled reference.Re-capture with a multi-channel scope on a known-good board, or at minimum record the trigger, probe setup and board state for the original capture.☐ Measured once ☐ Downgrade to "order of magnitude" ☐ Remove
G02The per-stage current signatures (0.00-0.02 A, 0.05 A, 0.08-0.15 A, 0.25 A, 0.65-1.10 A, 1.2-2.5 A).Same as T01: a single-board observation, and current depends on battery, port, adapter, attached assemblies and firmware state.Record complete profiles across multiple boards; treat as a rough clue only.☐ Evidence set attached ☐ Reword as weak clue ☐ Remove
G03The AON/S2 net names `PMU_VDDHI`, `PP1V2_S2`, `PP5V_S2_MAIN`, `PP0V72_S2_VDDLOW`.Board-scoped claim, not confirmed against a schematic by us.Check each net in the 820-02098/02100 schematic and correct the names where they differ.☐ Nets confirmed ☐ Board scope tightened ☐ Remove
G04The `U9300` ("Viper") / Monaco core-awake stage and `PP1V8_S1_CLVR_VDDH`, `CLVR_RESET_L`.Unconfirmed designators and net names from a secondary source.Confirm in the schematic and, if confirmed, record the phase count and the awake rails it feeds.☐ Confirmed ☐ Remove
G05The CPU/GPU awake rail voltage ranges (~~0.60-0.90 V dynamic).Dynamic operating ranges depend on load, DVFS state and firmware; a voltmeter figure at one instant proves little.Capture the rails across boot, idle and load; state the working range observed rather than a nominal.☐ Measured range recorded ☐ Remove
G06The NAND PMIC sequencing via `UN400`/`UN480` with `P2V5_NAND0_EN`, `NAND0_RESET_L`, `NAND0_CLK24M_R`.Board-scoped; sequencing order and signal names unconfirmed.Confirm designators and the enable/reset order in the schematic.☐ Confirmed ☐ Remove
G07The Stage 6 display chain (`J8500`, `LCD_PWR_EN`, `I2C_SEEPROM_*`, `IPD_LID_OPEN` ~1.8 V, `PPVOUT_S0_LCDBKLT` 40-44.5 V, `WLBT_PWR_EN`).Beyond checkpoint 24; board-scoped and unconfirmed. The backlight voltage is corroborated in kind by a case page but not by a schematic here.Confirm the net names in the schematic; measure the backlight rail on a working panel.☐ Confirmed ☐ Remove
G08M3 Pro/Max (A2992) shares this sequence unchanged.Architectural similarity across generations is not proof that rail names, controllers or ordering carry over.Repeat the checkpoint walkthrough on an M3 board and record the differences.☐ Verified on M3 ☐ Keep flagged as extrapolation ☐ Drop M3 from scope

Working-board verification record

Duplicate this block for every captured reference board.

Record ID Date
Model Board number/revision
SoC/RAM NAND configuration
Battery state Connected assemblies
Adapter/cable/port Startup trigger
Ambient/board temperature Firmware/macOS state
Instrument and probe setup
Raw captures/attachments
Claims verified
Exceptions/conflicts found
Reviewer Second reviewer

Publication gate

  • ☐ Every [diagram] statement has an exact board, page/sheet, component, and net reference.
  • ☐ Every [measured] statement links to raw data and includes complete test conditions.
  • ☐ Every [inferred]/[unverified] statement is phrased as an interpretation, not a guaranteed event.
  • ☐ Board-specific net names are not silently presented as universal family names.
  • ☐ Voltage ranges state the tested boards and operating conditions.
  • ☐ Current readings are observations, not standalone fault codes.
  • ☐ "Common fault" language includes a dataset or has been changed to "possible fault."
  • ☐ Correlation is not presented as causation.
  • ☐ No bus activity is described as proof of valid data without decoding or corroborating evidence.
  • ☐ Destructive or data-erasing actions include prerequisites and an explicit warning.
  • ☐ At least one second reviewer has checked high-impact diagnostic statements.
  • ☐ The page revision history summarizes which claims changed and why.

Sources

The board-level walkthrough was assembled from secondary sources. None of them is a schematic, and the two that claim to be measured captures do not record their instrument setup — which is why the material is labelled [unverified] rather than [measured].

  • Oscilloscope-measured reference (primary source of the timing table): 14-inch MacBook Pro A2442 (820-02098) & 16-inch A2485 (820-02100) 60-Checkpoint Master Power Sequence — IT-Tech Online / RepairWiki.
  • Lecture and circuit analysis: A2442/A2485 M1 Pro and M1 Max Power-On Sequence and Repair Methods — Master Liu Technology.
  • Video reference: MacBook Pro 16" A2485 M1 Max Stuck At 5V | Only Works On Battery | Oscilloscope Diagnosis Fix — YouTube.
  • Schematic-confirmed material: the Stage 1 charger path and the reference-board net names are cross-checked against the 820-02098/02100 schematics; the citations in the family stages (820-02841, 802-02841) are page references to those documents.

To promote any statement above to [measured], re-capture it and record the trigger, probe setup and board state — the checklist at the top of this page lists exactly what is required.

Revision notes

  • v7 (this revision). Merged a board-level checkpoint walkthrough for 820-02098/02100 into the family-level sequence. Changes to previously stated facts: PPBUS_AON nominal corrected to the 13.05 V schematic value (was "12-13 V"); the boost→buck mechanism is now stated as established with the trigger left open (old F04 wording withdrawn); the 3.8 V regulator is cross-referenced as U5700 = RAA225501-class; the SMC→CD3217 I2C request for the 20 V contract is adopted (F01 resolved, was "not proven"); M3 (A2992) added to scope, flagged as extrapolation; a third NAND naming family added, board-scoped. New claims G01-G08 opened for the board-level material.