Skip to main content

M2 Air Battery Not Detected — Sheared SMBus Pull-Up

M2 Air Battery Not Detected — Sheared SMBus Pull-Up

Status: repaired successfully · Channel: Alexey Lavrov · Published: 2025-05-29 · Runtime: 12:45 Model: MacBook Air 13" M2 (A2681) · Board: 820-02536 Source: YouTube Fault class: Sheared SMBus pull-up resistor → battery detection failure + CPU throttling

Summary

After a screen replacement the machine powered on from AC but reported the battery as Not Installed, and the CPU throttled hard (kernel_task ~100 %, clock clamped near 800 MHz). The narrow PCB peninsula near the display/webcam connectors — the so-called "giraffe neck" — had lost the 0201 pull-up on SMBUS_BATT_SDA (reported in the video as R5220; see the refdes note below), which pulls that line to 3.3 V. It was sheared clean off its pads. A replacement 10 kΩ 0201 restored battery telemetry, fast charging and normal clock speed.

Timeline

Time Evidence Observation
01:20 [measured] Powers on from AC; battery menu shows Not Installed; CPU throttled (kernel_task ~100 %, ~800 MHz).
03:40 [stated] Alexey: the PCB layout near the display and webcam connectors forms a narrow peninsula ("giraffe neck"); 0201 passives here are frequently sheared off with metal tweezers during cable disconnection.
06:10 [measured] Microscope: the 0201 pull-up on SMBUS_BATT_SDA (reported as R5220) sheared clean off its pads next to the camera flex bracket.
08:35 [stated] That pull-up holds SMBUS_BATT_SDA at 3.3 V; with it missing the battery bus floats low, halting BMS telemetry.
10:50 [measured] Replacement 10 kΩ 0201 soldered on; battery recognised instantly, fast charging at 20 V / 2.5 A, CPU clock normal.

Measurements

Time Measurement Value Context
01:20 CPU clock ~800 MHz Throttled state
06:10 Resistance OL Pads across the missing pull-up
10:50 Resistance / voltage 10.0 kΩ / 3.30 V Pull-up fitted; bus pulled high
11:30 Voltage / current 20.0 V / 2.52 A Normal charging profile

Diagnostic reasoning

  • Observation — loss of battery recognition and severe CPU throttling immediately following screen service.
  • Stated mechanism — the A2681 layout leaves the battery I2C pull-up exposed to shearing during screen replacements.
  • Resolution — confirming the missing pull-up and fitting a 10 kΩ part restored battery communication and eliminated the throttling.

Refdes note

The video identifies the sheared part as R5220. On the Pro/Max reference boards (820-02098 / 820-02100) R5220 is the charger current-sense resistor in the PPDCIN_AON path into U5200, not a battery-bus pull-up — see [[power-sequence]]. This case is a different board (820-02536, M2 Air), so the two can both be real refdes, but the collision means the designator in this case should be treated as unconfirmed until it is read off the 820-02536 boardview. The electrical finding (a 0201 pull-up on SMBUS_BATT_SDA sheared by display service) does not depend on the designator.

Repair procedure

  1. Inspect the display-connector perimeter under a stereomicroscope.
  2. Clean the ripped copper pads with flux and a micro-iron tip.
  3. Position and solder a 10 kΩ 0201 resistor with a micro-pencil iron.
  4. Seal with UV conformal coating for mechanical durability.
  5. Boot and verify battery status and CPU performance.

Claims audit

ID Claim Evidence Note
C1 Loss of the battery SMBus pull-up forces the Apple Silicon PMU into a thermal safe mode, throttling the CPU to minimum. [inferred] The correlation is demonstrated; the exact throttle trigger (bus fault → safe mode) is the technician's explanation.
C2 The sheared part is R5220. [unverified] Refdes as stated in the video. It collides with the charger current-sense R5220 on 820-02098/02100; confirm on the 820-02536 boardview before quoting the designator.
  • [[m1max-smbus-tvs]] and [[m1pro-scl-open]] — the same bus failing electrically rather than mechanically.
  • [[m1-las-backlight]] — another "damage during display service" failure mode.
  • [[power-sequence]] — board-scoped reference that carries the other R5220 (charger current sense).

Provenance: condensed from the video's audio transcript (Google AI Studio); timestamps and values are as stated in the video and are not independently verified. The R5220 designator is flagged unconfirmed — see the refdes note above.