The Repair LibraryRead · Learn · Master

Board-Level Charging Repair — the M92T36 and BQ24193

The port replacement fixes the mechanical half of the Switch's charging grief; this section fixes the electrical half. When a console arrives dead with a clean, firm port — or comes back from a third-party dock that negotiated it to death — the fault lives past the connector, in the front end's two famous ICs: the M92T36 that speaks power delivery at the port, and the BQ24193 that fills the cell. This is the chapter's deepest board work, and it is built on a discipline before it is built on a technique: the right chip is convicted by measurement, not guessed at by reputation. The section opens with that conviction — diode-mode readings around the port and the ICs compared against known-good values, the failure signatures that separate a dead negotiator from a dead charger, and the honest role of the shotgun approach: when replacing both is a decision, and when it is just a diagnosis skipped. Then the shorted-capacitor hunt, because the M92T36's rails fail short often enough that the cap, not the chip, is sometimes the whole repair — found with diode mode, current-limited injection, and heat, never by pulling parts at random. Then the rework itself, which teaches the volume's next package: the QFN, a leadless chip whose real joint is the center pad underneath — removal against that hidden connection, footprint preparation that must leave the center land flat, and the placement-and-reflow sequence where surface tension does the final alignment if the bench lets it. And then verification in the order the last section made law: diode-mode re-checks against known-good before any power, the current-limited first feed, the meter's negotiation and charge-current classes from the platform map, and the record closed with evidence. By the end, the dock-killed console is a repair you perform, not a story you retell — and the QFN discipline is in hand for every chapter that follows.

AdvancedMedium Risk23 min read

What You Will Learn

  • You will learn to convict the right chip — diode-mode readings against known-good values, the dock-kill and no-charge signatures, and when the shotgun approach is a decision rather than a skipped diagnosis.
  • You will learn the shorted-capacitor hunt — diode mode to find the shorted rail, current-limited injection with heat or freeze to find the part — because sometimes the cap is the whole repair.
  • You will learn QFN removal and footprint preparation — working against the hidden center pad, and leaving every land flat and clean for the rebuild.
  • You will learn QFN placement and reflow — alignment on a leadless footprint, paste or pre-tin, and the surface-tension settle that finishes what tweezers start.
  • You will learn the verification order for board-level work — diode-mode re-checks before any power, the current-limited first feed, and the meter's negotiation and charge classes closing the record.

What You Will Be Able To Do

  • You will be able to convict the failed IC by measurement — diode mode against known-good values and the platform's failure signatures — and defend when a double replacement is justified.
  • You will be able to run the shorted-capacitor hunt — locate the shorted rail in diode mode, then the shorted part with current-limited injection and thermal or freeze evidence.
  • You will be able to remove a QFN and prepare its footprint — center pad included — flat, clean, and inspected under magnification.
  • You will be able to place and reflow a QFN — aligned, pasted or pre-tinned, settled by surface tension — and inspect its fillets and edges for bridges.
  • You will be able to verify board-level charging work in order — diode mode, current-limited power, then the meter's negotiation and charge-current classes — and close the record with evidence.

Required Tools

  • Multimeter with diode mode — the section's primary diagnostic instrument
  • Hot-air station with a small nozzle, fine iron, flux, wick, and low-melting-point alloy
  • Board preheater, kapton and foil masking, board holder
  • Magnification — a microscope strongly preferred for QFN inspection
  • Current-limited bench supply for injection and first power, thermal camera or freeze spray for short localization
  • Known-good board or community diode-mode charts for comparison values
  • Replacement M92T36 and BQ24193 from reputable suppliers, plus solder paste or pre-tinning supplies

When NOT to Attempt This

Do not attempt this section if any of the following apply to you:

  • You are not comfortable working with small surface-mount components.
  • You have not completed the prerequisite sections for this skill.
  • You do not have the required tools in working condition.

Section Overview

The port job fixed the mechanical half of the Switch's charging grief; this section fixes the electrical half — the front end's two famous ICs, repaired under a conviction discipline (replacing-the-switchs-usb-c-charging-port). The right chip is convicted, not guessed. Diode-mode readings around the port and the ICs, compared against known-good values, separate a dead negotiator from a dead charger — and the shotgun approach of replacing both is legitimate only as a costed decision, never as a skipped diagnosis (diode-mode-the-repair-technicians-best-friend). Sometimes the cap is the whole repair. The M92T36's rails fail short often enough that the shorted-capacitor hunt runs first — the rail found in diode mode, the part found with current-limited injection and thermal or freeze evidence (low-ohms-and-voltage-drop-short-localization). The rework teaches the volume's next package. Both ICs are QFN parts — leadless, with the real joint hidden underneath: the center pad that carries ground, heat, and most of the grip — so removal fights it, cleanup must flatten it, and reflow must wet it (component-placement-with-hot-air). Placement ends in a settle. Aligned by eye and boardview, pasted or pre-tinned, then floated on molten solder while surface tension finishes the alignment tweezers started. And verification keeps the last section's law. Diode-mode re-checks before any power, the current-limited first feed, then the meter's negotiation and charge classes closing the record. Convict, hunt, remove, rebuild, verify — the dock-killed console, repaired by evidence.

Why This Matters

This is the repair that separates benches from parts-swappers — and the package it teaches carries the rest of the volume (replacing-the-switchs-usb-c-charging-port). This matters because the dock-kill queue is real: a generation of consoles died to bad negotiation, they still arrive weekly, and the bench that can convict and replace the M92T36 turns a community legend into routine work. This matters because wrong convictions cost twice: a BQ24193 replaced for an M92T36 fault is a QFN job done for nothing, and the console still returns dead — the diode-mode hour is cheaper than any wrong chip (diode-mode-the-repair-technicians-best-friend). It matters because the cap fools everyone: a shorted capacitor on a negotiator rail mimics a dead chip perfectly, and the bench that hunts shorts before pulling ICs sometimes closes the whole job for the price of one passive (low-ohms-and-voltage-drop-short-localization). It matters because the QFN is the volume's package: charge controllers, power-delivery chips, and audio ICs across the coming chapters are leadless parts with hidden center joints — learned here, on the best-documented board available, with community values to check against (component-placement-with-hot-air). And it matters because board-level verification is where reputations are made: the console that leaves with restored negotiation, proven charge current, and a before-and-after record is the one whose owner sends the next three. Convict by measurement once, and every front end after this one reads like a worked example.

Required Prerequisites

  • Flux, wick, and low-melting-point alloy — to run the removal and the center-pad cleanup the job turns on (component-placement-with-hot-air)
  • Solder paste in a fine-tip syringe, or fine solder for pre-tinning — to rebuild the QFN's lands with the right amount in the right places
  • Kapton tape and foil — to keep the blast radius honest on a front end dense with survivors (replacing-the-switchs-usb-c-charging-port)
  • Isopropyl alcohol and lint-free wipes — to clean between every stage, because flux residue lies to microscopes and meters alike
  • Replacement ICs from reputable suppliers, one spare each — to survive the counterfeit and the corner-crushed part the cheap listing ships
  • A donor Switch board — even a dead one — to practise the whole sequence where failure costs nothing, and to harvest known-good diode-mode values first (diode-mode-the-repair-technicians-best-friend)
  • Any scrap board with QFN packages — to drill leadless removal and reflow cheaply before the real footprint
  • A current-limited bench supply — to practise injection at safe limits and the current-limited first feed (low-ohms-and-voltage-drop-short-localization)
  • A thermal camera or freeze spray — to watch a short reveal itself instead of imagining it
  • A microscope — to inspect QFN edges the way the package demands
  • Community diode-mode charts for the platform — to compare readings when no known-good board is on the bench

Real-World Applications

The conviction discipline runs every front-end intake. A bench with a dock-killed console reads the port pins in diode mode against the community chart, finds the M92T36's signature dead-short where a junction should be, and books the negotiator replacement with the evidence on the record (diode-mode-the-repair-technicians-best-friend). A technician with a console that negotiates 15 V but never gains charge reads the signature the other way — the negotiator is doing its job, the cell never fills — and convicts the BQ24193 side instead of shotgunning both (replacing-the-switchs-usb-c-charging-port). Someone hunting a dead-short rail injects current-limited voltage and watches the thermal camera paint one capacitor warm — the cap comes off, the rail clears, and the console lives without either IC leaving the board (low-ohms-and-voltage-drop-short-localization). A shop replacing its first M92T36 spends the extra minutes flattening the center pad and pasting sparingly — and the chip settles square under surface tension instead of skating on a solder hill (component-placement-with-hot-air). And a bench closing the job re-checks diode mode against its known-good chart before any power, feeds the board current-limited, and files the meter's restored negotiation beside the intake's near-zero — the evidence pair that quotes the next dock-kill with confidence. The failures this prevents: the wrong chip replaced on reputation, a shorted cap misread as a dead IC, a QFN skating on an unflattened center pad, and power applied to an unverified footprint.

Common Challenges

  • The two ICs' symptoms overlap at the edges. A dead negotiator starves the charger; a shorted charger rail can silence negotiation — the difficulty is refusing to conclude from one reading: the conviction is a pattern of diode-mode values against known-good, not a single number (diode-mode-the-repair-technicians-best-friend).
  • The center pad is invisible at every stage. It cannot be seen during removal, cleanup, placement, or reflow — the difficulty is working by evidence instead of sight: the release you feel, the flatness you check with a straightedge of light, the settle you watch for, the fillets that prove wetting afterward (component-placement-with-hot-air).
  • Injection tempts recklessness. More volts finds shorts faster — and cooks traces, or the eMMC's neighborhood — the difficulty is holding Volume 5's limits: low voltage, current-limited, on the identified rail only, with the platform's assessment sheet in force (low-ohms-and-voltage-drop-short-localization).

Safety Notes

Risk Level: Medium. This is hot-air and injection work on a battery-powered consumer board — the port job's rules apply whole, and injection adds its own.

Professional Tips Before Starting

  • Harvest known-good values before you need them. Ten minutes of diode-mode charting on a working board — the comparison library that convicts every future front end (diode-mode-the-repair-technicians-best-friend).
  • Buy the ICs from suppliers you would name aloud. Counterfeit M92T36s exist and fail convincingly — reputable sources, one spare each, graded on arrival.
  • Photograph the chip's orientation before removal. The dot, the corner, the silkscreen — a QFN placed rotated solders beautifully and works never (component-placement-with-hot-air).

The Front End by Evidence — Convict, Hunt, Remove, Rebuild, Verify

Recap and Frame

Section 2.1 mapped the front end and 2.2 repaired its connector; this section goes to the silicon behind it — under a rule the whole handbook has been building toward (replacing-the-switchs-usb-c-charging-port). The rule is conviction by measurement. The M92T36's dock-kill reputation is real, but reputation is a hypothesis, not a verdict — the chip that comes off this board comes off because diode-mode evidence said so, the same standard Volume 5 set for every fault (diode-mode-the-repair-technicians-best-friend). The two ICs divide the labor cleanly. The negotiator speaks power delivery at the port and routes what arrives; the charger manages the cell — so their failure signatures differ in exactly the ways the platform map predicts, and the meter classes from intake already lean one way before a probe lands. The hunt comes before the conviction. Shorted capacitors on the front end's rails mimic dead silicon, cost pennies, and are found by method — which is why the shorted-cap hunt is a stage of this job, not a footnote (low-ohms-and-voltage-drop-short-localization). And the package is the point. Both chips are leadless QFN parts — the removal, cleanup, placement, and reflow discipline this section teaches is the same one the volume's remaining charge controllers and power chips will reuse (component-placement-with-hot-air). Hold the frame — evidence before parts, the hunt before the conviction, the package as a transferable skill — and the chapter's deepest work becomes its most methodical.

Convicting the Right Chip

The front end offers three suspects — negotiator, charger, or a shorted passive — and diode mode sorts them before any heat exists (diode-mode-the-repair-technicians-best-friend). The readings happen at the port first. Probing the port's pins in diode mode against ground — CC lines, VBUS path — and comparing against known-good values: a junction where the chart shows a junction is health; a dead short or an open where the chart shows a number is the first conviction evidence. The dock-kill signature points at the negotiator. A console dead after third-party docking, near-zero at the meter, and CC-line readings collapsed to shorts or opens — the M92T36's classic failure, often accompanied by shorted capacitors on its rails that must be found regardless of the chip's fate. The no-charge signature points at the charger. A console that negotiates — the meter shows the 15 V contract — but never pushes current into the cell, or never charges while running: the negotiator is speaking, the BQ24193's side is not delivering, and its rails and thermals get the next probes. Known-good comparison is the standard of proof. A donor board's values or the community's charts turn 'this reading feels wrong' into 'this reading is 0.3 volts below every documented good board' — the difference between suspicion and conviction (replacing-the-switchs-usb-c-charging-port). The double replacement is a decision, not a reflex. Dock kills sometimes take both chips; replacing both in one heat cycle is legitimate when the evidence implicates both or when a costed judgment says the second chip's price is cheaper than a second teardown — what is never legitimate is both-because-unsure, which is a diagnosis skipped and billed. Port pins first, signatures matched, known-good as the standard, doubles by decision — the right chip comes off for the right reason. Convict with the meter, and the hot air never works for nothing.

The Shorted-Capacitor Hunt

Before any IC leaves the board, the rails get cleared — because a shorted cap mimics dead silicon and costs pennies to fix (low-ohms-and-voltage-drop-short-localization). Diode mode finds the rail. Probing the front end's supply rails against ground: a healthy rail reads as a slow capacitive climb or a documented junction; a dead short reads near zero and names the rail — but not yet the part. The rail's caps are all suspects equally. A dozen capacitors can share the shorted rail, and every one reads shorted because the rail is shorted — the classic trap that pulls good parts off boards; the short is somewhere, not everywhere. Injection localizes by heat. A current-limited bench supply set to the rail's safe voltage and a conservative ceiling feeds the short; the shorted part dissipates the power and warms; the thermal camera paints it — or freeze spray frosts the board and the short thaws its own outline first (diode-mode-the-repair-technicians-best-friend). Voltage-drop walks where cameras cannot. Millivolt readings along the rail's copper fall toward the short — Volume 5's gradient technique, at home on a board this dense. The suspect comes off to testify. The convicted cap is removed and the rail re-read: short cleared, the cap was the fault — sometimes the whole repair; short remains, the hunt continues honestly toward the silicon (replacing-the-switchs-usb-c-charging-port). Rail by diode mode, part by heat or frost, confirmation by removal — the hunt in order. Clear the rails before convicting the chips, because pennies fix what panic replaces.

The QFN — Removal and the Center Pad

The package changes here: both ICs are leadless, and the joint that matters is the one nobody can see (component-placement-with-hot-air). The QFN's grip is mostly underneath. The perimeter lands are the visible minority of the connection; the center pad — the thermal and ground slab under the die — carries most of the solder, most of the heat path, and most of the mechanical hold, which is why a QFN resists removal long after its edges glisten. The removal is the port job's rule, smaller. Masking on, preheat under the board, flux at the edges, low-melt alloy worked around the perimeter, and even heat over the whole package — with the lift a weightless tweezer test, because the center pad releases last and prying against it rips the land off the board (replacing-the-switchs-usb-c-charging-port). The footprint cleanup has a hidden half. Wick flattens the perimeter lands — and then the center pad, which must end flat: a leftover solder hill under the new chip becomes a pivot that lifts every perimeter joint off its land. Flatness is checked with light. Under the microscope, a low-angle look across the center pad shows hills as shadows — the straightedge-of-light check that stands in for the sight the package denies. The gate applies as always. Magnified inspection before any placement: lands attached, mask intact, the center pad flat and bright — the port job's footprint gate, inherited whole (diode-mode-the-repair-technicians-best-friend). Mostly-hidden grip, patient removal, a cleanup with an invisible half, flatness proven by light — the QFN's removal discipline. Respect the pad you cannot see, and the lands you can see survive.

Placement and Reflow — The Settle

A leadless chip cannot be soldered pin by pin; it is placed, fed, and floated — and surface tension finishes what tweezers start (component-placement-with-hot-air). The solder goes on first, sparingly. Paste from a fine syringe on the perimeter lands and a modest print on the center pad — or a thin pre-tin leveled with wick — because excess on the center pad is the hill the chip will skate on, and excess at the perimeter becomes bridges under the package's skirt. Orientation is a one-way door. The pin-one dot matched to the silkscreen and the boardview before any heat — a rotated QFN reflows beautifully, inspects cleanly at the edges, and works never (replacing-the-switchs-usb-c-charging-port). The alignment is coarse, then chemical. Tweezers set the chip within a fraction of a pad; hot air brings the joint to reflow; and the molten solder's surface tension pulls the package square onto its lands — the settle, a visible small snap into place that is the package telling the bench it found home. The heat is even and brief. The whole package reflows together — center and perimeter — with the same circling patience as removal; dwelling long enough to wet, not long enough to cook the die or the neighbors. Inspection reads the edges. QFNs solder underneath, but they confess at the sides: the microscope pass looks for side-wetting fillets climbing each visible land — though sawn-edge terminations do not always wet up the side, so one absent fillet prompts closer inspection rather than automatic condemnation — no bridges under the skirt, and the package sitting flat and square, plus the gentlest push test: a seated QFN does not rock (diode-mode-the-repair-technicians-best-friend). Sparing solder, orientation checked, the settle watched for, edges read under magnification — placement as a sequence. Feed it right and let the solder align it — surface tension is the steadiest hand on the bench.

Verification — Diode Mode Before Volts, Volts Before Chargers

Board-level work earns power in stages, and each stage is a checkpoint the next one depends on (replacing-the-switchs-usb-c-charging-port). Diode mode re-testifies first. The same port-pin and rail readings that convicted the old chip, re-taken and compared against known-good: junctions where junctions belong, the shorted rail now climbing, the collapsed CC lines restored — the footprint proves itself to the meter before it meets a single volt (diode-mode-the-repair-technicians-best-friend). First power is current-limited. The bench supply, not a charger, feeds the board at a conservative ceiling while fingers and the thermal camera watch for heat — a fault that survived the rework announces itself in milliamps here instead of magic smoke later (low-ohms-and-voltage-drop-short-localization). The meter classes close the loop. On the known-good USB-PD setup: negotiation restored where intake read near-zero, charge current flowing where the cell starved — Section 2.1's classes, now reading healthy, filed beside the intake numbers. The charge test runs to proof. The reassembled console charges a partially drained cell while the bench watches percentage climb and the case warm normally — the customer's actual complaint, actually resolved (component-placement-with-hot-air). The record closes with the pair. Intake readings and photos beside the after set — conviction evidence, the removed part's photo, the restored classes — the file that quotes the next dock-kill in thirty confident seconds. Meter, limited volts, restored classes, a real charge, the evidence pair — verification in stages. Let the board earn each level of power, and no repair leaves as a gamble.

Common Mistakes

  • Replacing the famous chip on reputation. The M92T36's legend convicts it before the meter does — diode mode against known-good first: sometimes the negotiator is innocent and a cap is the fault (diode-mode-the-repair-technicians-best-friend).
  • Pulling every capacitor on a shorted rail. They all read shorted because the rail is shorted — find the part with injection and heat, not with a parts bin (low-ohms-and-voltage-drop-short-localization).
  • Leaving a solder hill on the center pad. The new chip pivots on it and every perimeter joint lifts — wick it flat and prove it with the low-angle light check (component-placement-with-hot-air).
  • Skipping the orientation check. A rotated QFN solders cleanly and works never — pin-one dot against silkscreen and boardview before heat.
  • Feeding the board a charger first. A surviving fault meets unlimited current — diode mode re-checks, then the current-limited supply, then the meter's classes (replacing-the-switchs-usb-c-charging-port).

Troubleshooting Guidance

The job runs convict, hunt, remove, rebuild, verify — and its stalls map to its stages. If the readings do not match either signature: stop and widen the chart — compare more pins against known-good before any conviction, because the third suspect is a passive (diode-mode-the-repair-technicians-best-friend). If a rail reads dead short: the hunt comes first — injection at limits, thermal or freeze, voltage-drop gradients — and the cap comes off to testify before any IC does (low-ohms-and-voltage-drop-short-localization). If the QFN will not release: the center pad is holding — more preheat, more flux, alloy at the perimeter, wider circling heat, and the weightless lift; never a pry against the hidden joint (component-placement-with-hot-air). If the new chip sits proud or rocks: the center pad has a hill — remove the part, re-wick to flat, re-check with low-angle light, and paste more sparingly. If the settle never comes: the joint is not reaching reflow together — raise preheat, slow down, and let the whole package arrive at temperature as one. If post-reflow diode mode still reads wrong: inspect the skirt for bridges and re-reflow before suspecting the new part — then suspect the new part, because counterfeits exist (replacing-the-switchs-usb-c-charging-port). If the current-limited feed draws high: power off and return to the hunt — something still shorts, and the limited supply just saved the board from proving it violently. If negotiation returns but charging does not: the other IC's side now stands accused — the signatures divide the labor, and the method walks to the second conviction with the same standard. The throughline: every stall answers to more evidence, more flux, or more patience — and the board meets each level of power only after the previous level testified.

Verification & Testing Methods

Confirm the discipline holds before the next front end arrives:

  • [ ] I convict chips by measurement — diode-mode values against known-good boards or community charts, signatures matched to the platform map — and I can defend every conviction aloud, including why the famous chip is sometimes innocent.
  • [ ] I treat the shotgun approach as a costed decision with evidence implicating both chips — never as both-because-unsure — and I run the shorted-cap hunt before any IC leaves the board: rail by diode mode, part by current-limited injection with thermal or freeze evidence, confirmation by removal.
  • [ ] I work the QFN by its hidden joint: removal with preheat, flux, alloy, and the weightless lift against the underside grip; the footprint gated under magnification before placement.
  • [ ] I leave the center pad flat and prove it with the low-angle light check; I paste sparingly, match the pin-one dot before heat, watch for the settle, and read the edges — fillets, no bridges, no rock — under the microscope.
  • [ ] I verify in stages — diode-mode re-checks before any power, the current-limited first feed watched for heat, the meter's restored negotiation and charge classes, the real charge test — and I close the record with the before-and-after pair.

Then try the practice exercises below — bench work on donor and scrap boards; scenarios differ from the quiz.

Practice Exercises

  1. Chart and convict (6 minutes, donor board and meter). Harvest a known-good chart first — diode-mode values at the port pins and front-end rails of a working or documented board — then read a faulty or scrap front end against it and write the conviction: which readings differ, which signature they match, and which chip or passive they accuse — and a one-line verdict on the double: whether the evidence justifies one chip, a costed both, or neither, defended aloud (diode-mode-the-repair-technicians-best-friend).
  2. Hunt a short by heat (7 minutes, scrap board, supply, camera or freeze). On a scrap board with a real or introduced rail short, find the rail in diode mode, set the bench supply's voltage and current ceiling out loud, inject, and localize with the thermal camera or freeze spray; confirm by removing the convicted part and re-reading the rail (low-ohms-and-voltage-drop-short-localization).
  3. Remove and rebuild a QFN (7 minutes, scrap board with any QFN). Remove a leadless package against its center pad — preheat, flux, alloy, weightless lift — then flatten the footprint including the center land, prove it with the low-angle light check, and reflow the same or a replacement part: sparing paste, pin-one matched, the settle watched for, edges inspected (component-placement-with-hot-air).
  4. Run the staged verification (5 minutes, the rebuilt board). In order and out loud: diode-mode re-checks against the chart, the current-limited first feed with a hand and camera watching for heat, and — where the hardware allows — the meter's negotiation and charge classes; close with the written evidence pair: intake values, conviction, after values (replacing-the-switchs-usb-c-charging-port).

These core steps — conviction by known-good comparison, the shorted-cap hunt, the center-pad discipline, the settle, and staged verification — are tested in the Chapter Quiz at the end of this chapter, where a score of 80% is required to continue.

Key Takeaways

  • The right chip is convicted by measurement — diode-mode values against known-good boards or community charts, the dock-kill and no-charge signatures dividing the suspects — and the shotgun approach is legitimate only as a costed decision with evidence implicating both chips, never as a diagnosis skipped (diode-mode-the-repair-technicians-best-friend).
  • The shorted-capacitor hunt runs before any IC leaves the board: the rail found in diode mode, the part found by current-limited injection with thermal or freeze evidence, the conviction confirmed by removal — because a shorted cap mimics dead silicon and sometimes the cap is the whole repair (low-ohms-and-voltage-drop-short-localization).
  • The QFN grips mostly from underneath: removal negotiates the hidden joint with preheat, flux, alloy, and the weightless lift, and the footprint gate inherits whole — lands attached, mask intact, and the hidden land proven flat (replacing-the-switchs-usb-c-charging-port).
  • The center pad rules the rebuild: pasted sparingly so the chip cannot skate, with the pin-one dot matched before heat, the surface-tension settle watched for, and the edges read under magnification — fillets climbing, no bridges, no rock (component-placement-with-hot-air).
  • Board-level work earns power in stages — diode-mode re-checks before any volts, the current-limited first feed before any charger, the meter's restored negotiation and charge classes before reassembly — and the record closes with the conviction and the before-and-after pair.

Skills Learned

  • You can now convict the failed IC by measurement — diode mode against known-good values and the platform's failure signatures — and defend when a double replacement is justified.
  • You can now run the shorted-capacitor hunt — locate the shorted rail in diode mode, then the shorted part with current-limited injection and thermal or freeze evidence.
  • You can now remove a QFN and prepare its footprint — center pad included — flat, clean, and inspected under magnification.
  • You can now place and reflow a QFN — aligned, pasted or pre-tinned, settled by surface tension — and inspect its fillets and edges for bridges.
  • You can now verify board-level charging work in order — diode mode, current-limited power, then the meter's negotiation and charge-current classes — and close the record with evidence.

Glossary Additions

  • QFN — quad flat no-lead: a surface-mount IC package with no protruding pins — its connections are flat lands on the package's underside edges, plus a large exposed pad at the center that carries ground, heat, and most of the mechanical grip. The package dominates modern power and interface silicon (the Switch's M92T36 and BQ24193 both wear it), and it dictates its own rework rules: removal must negotiate the hidden center joint rather than the visible edges, footprint cleanup must leave the center land flat, placement relies on surface tension to finish alignment, and inspection reads the package's side-wetting fillets because the real joints cannot be seen. Learned once, the QFN discipline transfers to charge controllers, power-delivery chips, and audio ICs across consumer boards.
  • center pad — the exposed metal slab on a QFN's underside, soldered to a matching land on the board: the package's thermal path, its ground connection, and the majority of its solder area and mechanical hold. The pad is invisible at every stage of rework, so it is managed by evidence instead of sight — it releases last during removal (which is why prying rips lands), it must be wicked dead flat during cleanup (a leftover solder hill becomes a pivot that lifts every perimeter joint), its flatness is proven with a low-angle light check under magnification, and it is fed sparingly at rebuild because excess solder makes the chip skate. A QFN job succeeds or fails on the joint nobody sees.
  • shotgun approach — replacing multiple suspect components in one repair rather than isolating the faulty one: on the Switch's front end, pulling both the M92T36 and BQ24193 because 'one of them did it.' The approach has a legitimate form — a costed decision, made when evidence genuinely implicates both parts or when the second chip's price is cheaper than the risk of a second teardown, and recorded as a decision on the case record. Its illegitimate form is both-because-unsure: a diagnosis skipped and billed, which doubles the rework risk on a dense board, hides which part actually failed from the bench's own case history, and teaches the bench nothing it can reuse on the next console.

Suggested Next Sections

Must read next:

  • Game Boy and Game Boy Advance Repair — Section 2.4 turns from the modern front end to the vintage bench: leaked-electrolyte cleanup and trace repair on the Game Boy family, the screen-line fix that resurrects original DMG displays, and the forgiving boards that make these consoles the perfect board-repair classroom.

Recommended: