1. Scope Guard: Every Basic Check Done, and the Machine Still Will Not Charge
This guide applies only after every basic check has been completed and the machine still fails to charge, cuts out the moment power is plugged in, or shows an abnormal charging light. The adapter has already been swapped for a known-good unit of the same specification, the cable has been swapped, the charging port has been checked for looseness and debris, the battery has been compared, USB-C Power Delivery negotiation has already been worked through layer by layer in the previous guide, the four-layer intermittent-charging investigation has been run, and the battery-not-detected, embedded-controller failure, and power-button-no-response topics have all been excluded. If any of those steps is still unfinished, do not jump to a board-level conclusion. Go back and finish the basic checks first, because a board-level conclusion reached in the wrong direction will only make every later swap, of battery, adapter, or charger IC, more confused.
The question this guide answers is a single sentence: why does the problem remain at board level when the basic checks are normal? The answer is not the name of one part but a path. Laptop charging is not a straight wire from the adapter into the battery. Power must pass through selection, detection, conversion, and monitoring before it reaches the battery, while also supplying the whole system. A failure in any one function can present as failure to charge, yet the layer behind it, and therefore the inspection direction, is completely different. The guide below uses layer logic throughout and avoids any claim of the form "replace this and it will work."
Remember the boundary rule: only after adapter, cable, port, battery, PD negotiation, and the embedded controller itself have all been excluded should the board charging circuit be discussed. The purpose of that rule is to prevent misdiagnosis, not to encourage skipping basic checks. This guide does not rewrite the basic-check steps. It explains, once those steps are normal, how to narrow the fault down to the charger IC and its surrounding power path, rather than replacing the battery on sight whenever charging fails. The main audience is Windows laptop users. Different brands and models use different charger IC part numbers, so this guide explains mechanisms and layer logic. No pin value taken from one part number may be applied to every machine. Wherever a figure is given, it is given together with its part number, and wherever a light pattern is described, it must be read together with the words "according to that model manual."
| Earlier topic | Content not repeated here | How it is mentioned here |
|---|---|---|
| Adapter, cable, and port basic checks | The full steps for swapping adapters, swapping cables, and cleaning ports | Treated only as an excluded precondition, without repeating the steps |
| Battery failure itself | Battery ageing, swelling, and cycle-count judgement | Mentioned only through comparison results that separate the battery layer from the board layer |
| USB-C PD negotiation | The four negotiation steps, PDO tiers, E-markers, and EPR | Mentioned only as a boundary: once PD has succeeded, a break after the charger IC still prevents charging |
| Four-layer intermittent charging | The full map across supply, battery, port and board, and system settings | Intermittent behaviour is given a second board-level explanation here: intermittent solder-joint contact |
| Battery not detected | Battery communication and system detection layers | Mentioned only at the boundary: a battery communication fault can look like a charger IC fault and must be separated by a known-good battery comparison |
| Embedded controller failure | Multiple EC functions failing together when the EC itself is faulty | Mentioned only at the boundary: a working EC that cannot talk to the charger IC is a different matter |
| Power button no response | Power sequencing and power-button circuitry | Mentioned only at the boundary: booting but not charging, and not booting at all, are two different clues |
2. The Four Stages from Wall Socket to Battery
Using the Texas Instruments BQ family and the Renesas ISL family as examples, the charging path can be understood as four functions. The first is input and path selection. The charger IC together with external MOSFETs selects whether the adapter or the battery powers the system, and the changeover must avoid both supplies being connected at once. The BQ24702 and BQ24703 documentation states explicitly that selector feedback ensures break-before-make switching, meaning one side is disconnected before the other side is connected. The BQ24715 provides N-channel ACFET and RBFET drive for power-path management. The Renesas ISL95521 likewise uses NFET devices for all switching and states support for both HPB and NVDC configurations. When the selection layer fails, the machine may cut out on plug-in, boot on battery but not on adapter, boot on adapter but not on battery, or show an unstable system voltage.
The second function is detecting whether the adapter is valid, and the key signals are ACDET and ACOK. ACDET is the adapter-detect input pin, with a threshold set by a resistor divider, used to judge whether an adapter is present and valid. The BQ24702 and BQ24703 datasheet states that when ACDET falls below its threshold, the chip enters sleep mode and PWM is disabled. The BQ24725A is more specific: ACDET must lie between 2.4 V and 3.15 V, the supply must clear its undervoltage threshold, and the supply must exceed the sense reference by more than 275 mV before the adapter counts as valid and ACOK goes high. If any one condition fails, ACOK is pulled low. The practical meaning is direct: an adapter producing output does not mean the board recognises the adapter. A drifted ACDET divider resistor, a corroded or broken detection trace, or damage to the detection pin itself can make the system believe there is no adapter, so it does not charge and the charging light behaves abnormally. These figures apply to the BQ24725A only; other parts must be judged against the datasheet for the machine under inspection.
The third function is conversion and regulation. Older barrel-adapter machines mostly use a buck architecture, stepping a higher input voltage down to the charging voltage and the system voltage. The BQ24715 is a 2-cell to 3-cell NVDC buck controller with a 6 V to 24 V input range, and the BQ24725A is a 1-cell to 4-cell buck controller. Newer USB-C PD machines mostly use buck-boost, because the input voltage can sit above or below the battery voltage and the chip must switch seamlessly between buck, boost, and buck-boost operation. The BQ25713, BQ25713B, and BQ25720 are NVDC buck-boost architectures supporting 1 to 4 cells, with input ranges of 3.5 V to 24 V and 3.5 V to 26 V respectively. NVDC, or narrow VDC, is the key idea: the system voltage follows the battery voltage but never falls below the minimum system voltage, so the system can boot immediately even when the battery is fully empty or removed. For that reason, failing to boot on adapter power with the battery removed and booting on adapter power with the battery removed but failing to charge are two different clues and must not be confused.
The fourth function is monitoring and protection. The tiny voltage difference across a sense resistor is amplified into a current-monitoring signal. On the BQ24725A, IOUT is 20 times the sense-resistor voltage difference, and the host can use it to monitor input current or charging current. The BQ25720 likewise uses 5 milliohm or 10 milliohm sense resistors as its charging-current and input-current reference. The BQ24715 lists overcurrent, overvoltage, and MOSFET short-circuit protection, and the BQ25713 datasheet contents list system overvoltage protection, battery overvoltage protection, battery short-circuit protection, system short-circuit hiccup mode, and thermal shutdown. The practical meaning is that a burned-open sense resistor leaves the chip blind to current and stops the charging loop, while a shorted MOSFET can short the input rail directly to ground, so the adapter cuts out on protection or its light goes out the moment it is plugged in.
| Layer | Component or signal | What it does when normal | Typical presentation when faulty or inaccurate | How it is separated from the previous layer |
|---|---|---|---|---|
| L0 External power | Adapter, cable, socket | Supplies voltage and current to specification | No output, unstable output, loose plug | Comparison with a known-good adapter and cable of the same specification; assumed excluded in this guide |
| L1 Detection layer | ACDET divider, ACOK, REGN | Recognises a valid adapter before permitting charging | Power reaches the board but the system treats it as no adapter; charging light off or flashing abnormally; ACOK stays low | Adapter output normal but the system does not recognise the adapter; board-level measurement is required and readers must not open the board to measure |
| L2 Path-selection layer | ACFET, RBFET, BATFET | Break-before-make switching between adapter and battery; battery supplements the adapter at full load | Cuts out on plug-in; boots on battery but not on adapter, or the reverse; unstable system voltage | Compare the three states of adapter only, battery only, and both together; different results point to the selection layer |
| L3 Conversion layer | Charger IC itself, inductor, high-side and low-side MOSFETs, output capacitors | Buck or buck-boost conversion with stable charging voltage and current | No charging at all, extremely slow charging, cut-out under load, abnormal heating at the IC or its surroundings | No charging current after battery, PD, and EC communication have been excluded; board-level inspection is required before any conclusion |
| L4 Monitoring layer | Sense resistor, IOUT, temperature and current monitoring | Measures input and charging current and triggers limiting, load shedding, and protection | Frozen battery percentage, charging that stops and starts, protection triggering for no apparent reason, false overcurrent or overtemperature reports | Battery health report normal but charging behaviour erratic; a burned sense resistor is board-level evidence and must not be judged by appearance |
| L5 Communication layer | SMBus or I2C, EC and host controller | Host sets charging voltage, current limit, and minimum system voltage and reads status back | Settings cannot be written, status cannot be read, charging disabled by a host command | This guide assumes the EC itself is excluded; the point is that a working EC that cannot talk to the charger IC is different from a faulty EC |
| L6 Power-rail layer | Main system rail, 3.3 V and 5 V downstream rails | Distributes converted power to the processor, memory, and input and output circuits | Adapter light goes out on plug-in, no response at all, one shorted rail dragging the whole board down | Adapter protection action and downstream shorts must be separated from a charger IC short itself; only inspection can identify the layer, and repeated plug-in attempts must stop |
3. The Detection Layer: Adapter Output Is Not the Same as Adapter Recognition
The detection layer is the easiest layer to misdiagnose, because the evidence in the reader hands looks decisive: the adapter outputs power, so the board must be receiving power. Board logic does not work that way. The board charges only after it recognises a valid adapter, and the ACDET divider circuit is the entry point for that judgement. Two resistors divide the input voltage and feed the result into the charger IC, where it is compared with an internal threshold. When a divider resistor drifts in value, the threshold moves. When a detection trace corrodes or breaks, the signal distorts. When the detection pin itself is damaged, the judgement stays wrong. The outward result is identical in all three cases: the adapter clearly has power, yet the board acts as if there is no adapter.
The BQ24725A conditions deserve to be taken apart one by one, because they show that a valid adapter actually has three requirements, and all three are necessary. First, the ACDET voltage must sit inside its specified window. Too low can mean the adapter is not properly connected or the input voltage is insufficient, while too high does not prove health either and may mean the divider circuit itself is faulty. Second, the supply must clear its undervoltage threshold, because a chip that has not established its own working voltage is in no position to judge an adapter. Third, the supply must exceed the sense reference by a specified difference, which confirms that the current direction and the supply path are sensible rather than merely checking an absolute voltage figure. Only when all three are met does ACOK go high and tell the host that the adapter is valid. If any one fails, ACOK is pulled low and charging never starts. That mechanism explains a deeply confusing presentation: the adapter light is on and a system light is on, yet the machine never charges. A light on proves only that power reached the board. ACOK proves that the system accepted the adapter.
REGN is another detection-layer clue. It is an internal low-dropout regulator output in the BQ family that supplies the charger IC internal circuits and the external MOSFET drive. Repair-community write-ups note that a missing REGN can mean the IC has no input voltage, or that the IC itself no longer responds. That statement must be handled carefully. A missing REGN is an inspection concept, not a reader measurement task, because internal supply pin names and voltages differ by part, and readers must not open the board to measure them. At inspection, the technician follows the path of whether input reaches the board, whether ACDET and ACOK recognise the adapter, and whether REGN is established. The reader evidence that makes that path start at the detection layer is simple: adapter output normal, system recognition absent. That comparison alone is enough to move inspection away from replacing the battery.
The detection layer also carries a common misunderstanding: a BIOS display of an unknown adapter does not mean the adapter is faulty. Dell public support documentation states that a BIOS adapter wattage of unknown can accompany processor throttling and failure to charge, and that the root causes can sit in three layers: a damaged identification pin, a damaged DC-in socket, or a mainboard charging-circuit fault. A BIOS display of unknown is a clue, not a conviction. The correct comparison is to fit an original adapter of the same wattage and read the BIOS value again. If it still reads unknown, suspicion toward the identification layer rises. That logic is exactly the same as the ACDET layer: identification failure and detection failure can look alike, but they sit in different places, and the inspection report must keep them apart.
4. The Path-Selection Layer: How Adapter and Battery Hand Over Without Fighting
The path-selection layer carries out a task that looks simple and is in fact strict: two power sources, the adapter and the battery, must never both dominate system supply at the same instant. The changeover must not allow both supplies to connect at once and cause a conflict, and it must not allow a supply gap that drops the system. The BQ24702 and BQ24703 use selector feedback to ensure break-before-make operation, disconnecting one side before connecting the other. The newer BQ24780S uses two charge pumps to drive N-channel ACFET, RBFET, and BATFET devices separately, selecting adapter or battery as the system source automatically, and can connect the ACFET within 100 microseconds. An older TI architecture uses ACDRV to drive an external P-channel MOSFET for adapter and battery switching. The drive method differs, but the handover principle is identical: wrong order means trouble.
The ACFET is the adapter input switch, the RBFET is the reverse-blocking switch that stops battery current flowing back into the adapter path, and the BATFET is the battery switch that connects the battery to the system. If any one switch fails or its drive goes wrong, the machine may end up able to use only one side. The standard comparison is the three-state test: adapter only, battery only, and adapter plus battery together. If the machine boots on battery alone but not on adapter alone, suspicion rises toward the selection layer or the input layer. If it boots on adapter alone but dies the instant the adapter is removed, suspicion moves toward the battery or the battery path. If it fails only when both are present, the handover moment itself, with a possible conflict or voltage dip, becomes the focus. Those three states are not casual trials. Each must change one variable at a time and record boot behaviour, charging-light behaviour, and the system battery icon, or the comparison itself is invalid.
The selection layer must also handle battery supplementation. When the adapter is fully loaded, the battery must step in immediately, and that action depends on the BATFET ideal-diode operation and the supplement path described in the next section on hybrid power. At this point, one distinction is enough: a selection-layer fault does not have to present as total loss of power. It can present as unstable system voltage, cut-out only under heavy load, or cut-out only at the instant of plug-in. Those presentations are easily mistaken for a faulty battery or an underpowered adapter, because the reader sees only the result and never the handover. The value of layer thinking is exactly here: the three-state comparison separates a handover problem from a battery problem instead of relying on a feeling about which part to replace.
One more distinction matters for the boundary with the previous guide. USB-C machines and barrel-adapter machines enter the selection layer differently, but the selection logic is shared. A USB-C machine must complete PD negotiation before an input voltage exists, and only then does it enter the selection layer. A barrel machine enters the selection layer directly from the DC-in socket. A successful PD negotiation followed by failure to charge does not mean the negotiation was wasted; it means the fault lies after the selection layer. That boundary was already drawn in the previous guide. This guide only explains the path after the boundary in full, without confusing a PD problem with a selection problem.
5. The Conversion Layer: Charger IC, Inductor, and Bridge Switches Turning Voltage into Charging Voltage
The conversion layer is the work of the charger IC together with its surrounding components: turning the input voltage into a suitable charging voltage and charging current, and holding that output stable while the system load changes. A buck architecture steps a higher input voltage down. A buck-boost architecture switches seamlessly between buck, boost, and buck-boost operation when the input voltage sits above or below the battery voltage. The BQ24715 is an NVDC buck controller, while the BQ25713 and BQ25720 are NVDC buck-boost controllers. The architecture differs, but the task is identical: the output must be stable, the current must be controlled, and protection must stop the process promptly when something goes wrong.
The surrounding components matter just as much, and conversion-layer faults must not all be blamed on the charger IC itself. The inductor stores energy and smooths current, the high-side and low-side MOSFETs perform the switching conversion, and the output capacitors stabilise the output voltage. If any surrounding component fails, conversion can stop or become extremely unstable, presenting as no charging, extremely slow charging, cut-out under load, or abnormal heating at the IC or its surroundings. Repair-community checklists mention the DC-in MOSFET, the charger IC input and output, the battery connector main rail and ground, and a burned or open sense resistor. Those items are inspection concepts only, never a reader checklist for opening the machine.
The conversion layer has one ordering rule that must not be reversed: only after battery, PD, and EC communication have been excluded and charging current is still absent should the conversion layer be suspected. The order cannot be reversed because a faulty battery, a failed PD negotiation, or an EC setting that never reached the IC can all make the conversion layer look dead when it is not. Inspection discipline is to change one variable at a time and exclude layers in order. Replacing the battery without success does not prove the charger IC is faulty, and replacing the charger IC without success does not prove the board must be replaced. That sentence is worth remembering, because most misdiagnoses in this area come from skipped steps rather than from difficult measurements.
The NVDC architecture carries one more practical implication. System voltage follows battery voltage but never falls below the minimum system voltage, which lets the system boot immediately even when the battery is fully empty or removed. Readers can use that property as a comparison, but they must not mistake booting for charging. Booting proves that the system supply path works. Charging proves that the charging branch works. They are different paths, and the monitoring-layer section below uses that distinction again.
6. The Monitoring Layer: How a Sense Resistor Lets the Chip See Current, and Why Blindness Stops Charging
The core of the monitoring layer is the sense resistor, a resistor of extremely low value placed in series with the input or battery path. Current flowing through it creates a tiny voltage difference, which is amplified into a current-monitoring signal. On the BQ24725A, IOUT is 20 times the sense-resistor voltage difference, and the host can use it to monitor input current or charging current. On the BQ25720, 5 milliohm or 10 milliohm sense resistors serve as the charging-current and input-current reference. The resistance is small because a series resistor must not waste much power or voltage drop, but a small voltage difference also means the measurement is sensitive, and any sense-resistor fault distorts the whole current picture.
Sense resistors fail in several ways, and the presentations differ while still pointing to the monitoring layer. A burned-open resistor leaves the chip completely blind to current and stops the charging loop. A drifted value makes the reading systematically too high or too low, so current-limit and protection thresholds land in the wrong place. A failed solder joint makes the reading come and go, presenting as charging that stops and starts. Repair discussions place a main-rail sense resistor near the VBAT rail as a common checkpoint, but its position differs by machine, and readers must not guess which component is the sense resistor from appearance, nor judge burning from appearance. A normal battery health report with erratic charging behaviour is precisely the clue that raises monitoring-layer suspicion, because the battery report describes the battery itself while the monitoring layer describes the live current reading. They are not the same measurement.
Protection and monitoring cannot be separated. The BQ24715 lists overcurrent, overvoltage, and MOSFET short-circuit protection. The BQ25713 lists system overvoltage protection, battery overvoltage protection, battery short-circuit protection, system short-circuit hiccup mode, and thermal shutdown. Protection is a last line of defence, not a guarantee against damage. Sustained high input, surges, and unstable output from a poor-quality adapter strike the input MOSFET and the charger IC first and then reach later stages. When protection triggers, the system stops charging or powers down. That action is protection design, not the fault itself. When protection triggers repeatedly for no apparent reason, however, the monitoring signal or the protection circuit itself may be distorted, and inspection must separate a genuine overcurrent event from a false overcurrent report.
The monitoring layer also includes temperature monitoring and current-monitoring signals. The ISL95521, for example, provides a PROCHOT signal indicating system undervoltage, adapter overcurrent, battery overcurrent, or overtemperature. After receiving such a signal, the host can throttle the processor, limit current, or stop charging. From outside, those actions can look like a machine that suddenly slows down or suddenly stops charging, but behind them the monitoring layer is doing its job. Separating genuine protection from a false report depends on comparisons: whether light-load and powered-off charging are normal, whether protection still triggers after an adapter swap, and whether temperature is genuinely high. Only when comparisons still show repeated triggering should the monitoring layer be suspected.
7. Hybrid Power and Pass-Through Power: Why Falling Battery Level on Adapter Power Is Not Always a Fault
This section is the most misdiagnosed part of the whole guide, so three terms must be kept apart. Different manufacturers use different names, but the mechanisms are related and their purposes are different, and they must not be written as two names for one function. The first term is Hybrid Power Boost, the Renesas wording. The ISL95521 can be configured as HPB or NVDC, and both configurations allow the battery and the adapter to power the system load together when load exceeds adapter capability, in what is called turbo mode. The HPB configuration reverse-boosts battery energy onto the system bus to help the adapter supply power. In plain terms, the battery helping out under heavy load is design behaviour, not a fault.
The second term is NVDC supplementation, the Texas Instruments wording. The BQ25713 and BQ25720 state that when load power exceeds the input supply rating, the battery enters supplement mode to prevent system collapse, while the system still operates when the battery is fully empty or removed. The BQ25713 also provides battery MOSFET ideal-diode operation for supplement mode, meaning battery supplementation needs no reader action and happens automatically in the chip. The BQ24780S Hybrid Power Boost mode, formerly called turbo boost, allows the battery to discharge into the system at the same time when instantaneous system demand exceeds the adapter maximum power, so the adapter does not collapse, with a transient response into boost mode of 150 microseconds. Those figures apply to that part only and must be judged against the datasheet for the machine under inspection.
The third term is Pass Through Mode, the Texas Instruments wording, and it must not be confused with HPB. The BQ25713 and BQ25713B provide a patented Pass Through Mode that improves system supply efficiency and battery fast charging. In PTM, the input supply passes directly to the system. TI discussion also mentions using PTM so the adapter powers the system directly. PTM is an efficiency mode that bypasses conversion losses and passes power directly. It is not a bypass of protection. HPB and supplement mode, by contrast, are about the battery topping up the adapter under heavy load. The purposes differ. The misdiagnosis trap sits exactly here: seeing battery level fall while gaming on adapter power and declaring the charger IC faulty is wrong. First compare whether adapter power is itself below peak system demand, whether a high-performance mode is selected, and whether power-hungry peripherals are attached. Only after those are excluded should the supplement path be suspected.
When the supplement path itself fails, the presentation is a different matter. If HPB or supplement operation fails, for example because the battery MOSFET will not turn on or the charger IC judges the situation wrongly, the system bus voltage falls under heavy load and the machine may cut out suddenly, throttle, or show an abnormal charging light. That is the point at which board-layer suspicion becomes justified. The key discriminator is the load relationship. If idle and powered-off charging are normal and only heavy load loses charge, insufficient adapter power and supplement design are the larger suspects. If light-load and powered-off charging also fail, board-layer suspicion rises. Charging normally when powered off but failing when powered on points toward insufficient supply or the supplement path more than toward the battery itself. That comparison must be made before inspection, because it directly decides which layer inspection starts from.
8. EC and Charger IC Communication: Separating a Faulty EC from a Broken Conversation
SMBus or I2C is the channel over which the host talks to the charger IC, and the host is usually the embedded controller. The BQ24715 uses SMBus for programmable voltage and current settings, with voltage in 16 millivolt steps and current in 64 milliamp steps, at accuracies of plus or minus 0.5 percent and plus or minus 3 percent respectively. The BQ24725A is likewise host-controlled over SMBus, and charging requires several conditions at once, including the SMBus charge-enable bit, a valid ACOK, conducting ACFET and RBFET devices, and no overvoltage, overcurrent, short-circuit, or overtemperature condition. Any one failure stops charging. The ISL95521 is likewise SMBus-programmable and provides a PROCHOT signal for system undervoltage, adapter overcurrent, battery overcurrent, or overtemperature. In other words, whether charging stops depends not only on the IC itself but also on the monitoring signals the host receives.
The BQ24780S adds a communication-layer detail worth explaining. The EC writes input current, charging voltage, charging current, and discharge current limits through SMBus from the system power-management microcontroller, with charging voltage accuracy of plus or minus 0.4 percent in 16 millivolt steps. If no ChargeVoltage or ChargeCurrent write arrives within an adjustable 175-second watchdog window, the watchdog terminates charging or hybrid power mode. That mechanism explains a deeply confusing presentation: the machine charges at first, stops after a while, and then charges again. A communication break does not have to be a permanent broken wire. It can be a settings write that times out, a status read that fails, or a monitoring signal in error, causing the charger IC to stop charging by design. All of this is inspection-concept material. Readers must not use an oscilloscope, a logic analyser, or a multimeter to measure SMBus waveforms. The useful reader contribution is history: inspection technicians will check the communication traces and the supply, and a clear history of water exposure, drops, or a battery change before the fault appeared is genuinely helpful.
Layer logic can be separated with two groups of evidence. A faulty EC itself usually comes with several EC functions failing at once, such as the power button, the keyboard, the fan, and the power sequencing. That scope belongs to the EC-failure guide, is assumed excluded here, and is not rewritten in this guide. A working EC that cannot communicate with the charger IC presents differently: power button, keyboard, and similar EC functions work normally, yet charging settings cannot be written, status cannot be read back, and battery information stalls. Possible causes include a corroded or broken SMBus trace, a charger IC with no supply, or an IC that does not respond. A third possibility is a battery SMBus problem: the gas gauge inside the battery cannot talk to the board, and the presentation can resemble a charger IC fault. Only a known-good battery comparison separates that case, and no single figure from a Windows battery report can settle it.
9. Why It Fails: Five Mechanisms, Not Just the Word Ageing
The first mechanism is overvoltage, overcurrent, and surge. The charger IC itself provides input overvoltage and overcurrent protection, but protection is a last line of defence, not a guarantee against damage. Sustained high input, surges, and unstable output from a poor-quality adapter strike the input MOSFET and the charger IC first and then reach later stages. Public discussion of counterfeit or uncertified adapters and cables notes output markings that do not match actual output, inadequate protection, and missing or falsely marked cable E-marker chips, which can let a device draw more current than the cable can safely carry and produce heat and damage. The wording must stay careful: not every third-party product damages machines. The higher-risk group is products without certification, with unclear markings, or with unstable output, judged against the original specification. Useful history for inspection includes use of an adapter or cable of unknown origin before the fault, a lightning strike or unstable mains supply before the fault, and an adapter that ran hot or smelled scorched. That history is not a diagnosis by itself.
The second mechanism is liquid plus humidity, where corrosion does not kill instantly but eats slowly. The long-term liquid-damage mechanism is electrochemical migration. It needs moisture, ionic contamination, and an applied bias all together. Metal ions dissolve at the anode, migrate toward the cathode, and grow dendrites that eventually bridge neighbouring conductors and cause a short. The process can take hours to weeks, so a machine can fail long after its surface looks dry. Sugary, salty, or coffee liquids are worse than plain water, because solutes remain on the board after the water evaporates and keep absorbing moisture and corroding copper traces and solder joints. The Hong Kong context has a sourced figure: the Hong Kong Observatory 1991 to 2020 climate normals give an annual mean relative humidity of 78 percent. That figure proves only that Hong Kong is persistently humid. It cannot directly prove that humidity caused any one machine charger IC failure. The correct wording is that persistent high humidity makes liquid residue harder to dry thoroughly and corrosion easier to sustain, and that moving between air-conditioned rooms and warmer air can cause condensation. Inspection history should therefore cover water exposure, rain, and placement near windows or air-conditioner outlets. The claim that Hong Kong humidity makes charger ICs fail easily, as a causal assertion, must not be written.
Need a laptop repair expert?
Free diagnosis · Transparent pricing · No fix, no fee · 180-day warranty
The third mechanism is drops, squeezing, and long-term flexing, where a solder joint cracks without breaking at once. Solder-joint failure mechanisms fall into thermal, mechanical, and electrochemical classes, and mechanical failure includes shock and vibration fatigue. For leadless packages, especially large BGAs, a drop flexes the board, the stiffer component forces its joints to carry greater stress, and lead-free solder with a higher elastic modulus tolerates mechanical shock relatively poorly. Another source notes that solder fatigue is progressive: resistance rises slowly, signal integrity degrades, intermittent faults appear first, and permanent failure comes last. That point matters greatly. A cracked joint around the charger IC can present as intermittent charging that works until the machine is moved, and that must be layered against the intermittent-charging guide. Intermittent behaviour is not always a loose cable. It can also be intermittent board-level contact.
The fourth mechanism is heat and sustained heavy load, which is not as simple as using the machine a lot. The charger IC provides thermal shutdown protection, but sustained operation at high temperature, with poor cooling or blocked vents, accelerates solder thermal fatigue and component ageing. Wording must avoid equating gaming or rendering directly with burning out a charger IC. The correct statement is that under heavy load, system demand exceeds adapter capability and the normal mechanism is battery supplementation. If the supplement path itself is defective, heavy load exposes the problem. Heavy load does not itself create a short circuit. Another phenomenon must be kept apart: when system load exceeds adapter supply, battery percentage can fall rather than rise, and Dell official support documentation likewise warns that with an underpowered adapter, charging may not keep up while the machine is in use. That is insufficient supply, not charger IC failure.
The fifth mechanism is the part itself or its surrounding passive components, and the IC is not necessarily the first to die. Repair-community checklists commonly mention the DC-in MOSFET, charger IC input and output, battery connector main rail and ground, a burned or open sense resistor, an ACDET reference value, and a missing REGN indicating no IC input voltage or a dead IC. Those values come from repair-community write-ups, differ greatly by model, and may appear in this guide only as inspection concepts, never as reader measurement instructions. Shorted ceramic capacitors and shorted MOSFETs are common suspects for power-rail shorts, but the short location can only be identified by isolating stages one at a time. Repair discussion also notes that shorts are most often in capacitors, sometimes in MOSFETs, and relatively rarely in the IC itself unless there is physical damage. This guide preserves that uncertainty and never writes that a short circuit means the charger IC has burned out.
10. Brand and Platform Differences: Light Patterns and Identification Must Not Be Transferred Across Brands
The table below organises only differences shown by brand public support documents and verifiable material. Charging-light definitions, BIOS options, and connectors can differ between models and years within the same brand. Wherever this guide describes a light pattern, it must be read with the words "according to that model manual," and that is not a polite formality. Public write-ups already contradict each other, so transferring a definition across brands is certain to mislead somewhere.
| Brand or platform | How charging is identified | Common fault direction | Manufacturer signals a user can check |
|---|---|---|---|
| Dell, older barrel-adapter machines | Centre identification pin plus 1-Wire adapter identification; BIOS shows adapter wattage, and a display of unknown means the system could not identify the adapter | Damaged identification pin, damaged DC-in socket, or mainboard charging-circuit fault; an unidentified adapter can accompany processor throttling and failure to charge | BIOS overview or battery information adapter status; power-on self-test message that adapter wattage and type cannot be determined |
| Dell, USB-C machines | USB-C PD negotiated voltage and wattage, the main subject of the previous guide and only a boundary here | PD controller or charger IC layer fault, where changing adapter and cable has no effect; must be kept apart from the barrel-machine identification-pin problem | BIOS as above; adapter LED off or flickering is an advanced-investigation symptom |
| Dell, charging and diagnostic lights | Battery status light doubles as a diagnostic light; amber flash groups plus white flash groups form a two-digit code, with a 1.5-second pause inside and a 3-second pause between groups; for example, 2,2 indicates a system board, BIOS, or ROM fault, 2,6 indicates a system board or chipset error, and 3,5 indicates a power-rail fault, using Latitude 3510, 9410, and 5540 manuals as examples, with code tables differing by model | A light code by itself does not mean the charger IC is faulty; codes beginning with 2 or 3 point to system board, memory, display, power rail, and other layers and must be decoded against that model manual, never guessed from colour | Light colour while powered off and plugged in, with the caution that different documents define powered-off light colours differently, so no single definition may be chosen as a universal truth |
| HP | HP support-community investigation still stops at checking the charging light, the Windows "plugged in, not charging" message, swapping an adapter of the same specification, clearing port dust or oxidation, HP battery check, BIOS power settings, and a hard reset; beyond that range, the HP official position is repair, without public charger IC layer detail | DC-in socket or mainboard charging-circuit suspicion only becomes valid after adapter, battery, and port have been excluded | Windows battery icon "plugged in, not charging"; HP battery check result |
| Lenovo ThinkPad and IdeaPad | Manufacturer power-tool charging-limit functions manage a charging ceiling; UEFI BIOS and firmware control charging, and the manufacturer also lists BIOS updates among fixes for battery not charging, not fully charging, or not being recognised | Stopping at the same percentage, for example 60 percent or 80 percent, should first raise suspicion of a charging-limit setting rather than a board fault; only after the limit is switched off, the adapter is swapped, and BIOS is updated without effect does board-layer suspicion begin | Charging-limit switch in the manufacturer power tool; battery icon showing that a limit is active |
| ASUS | Light definitions are not unified across models; one write-up states green means adapter connected and 95 to 100 percent, solid amber means adapter connected and below 95 percent, and flashing amber means no adapter and below 10 percent, while another write-up states flashing amber can also mean charging in progress, poor battery connection, poor battery health, adapter trouble, or hardware fault; the two contradict each other | Flashing amber at high battery level, or green and amber alternating without ever reaching full charge, signals a charging-system abnormality sitting somewhere across adapter, battery, and board, and requires adapter-swap and battery-removed boot comparisons before narrowing | Machine battery light plus taskbar battery icon; flashing amber during heavy-load gaming can interact with power, cooling, and performance modes |
| Apple, older MagSafe and MagSafe 3 | System Management Controller communicates with the adapter through the MagSafe centre pin over a one-wire signal; MagSafe light solid amber means charging or charging on hold, green means full, and repeated flashing amber means a connection or charging-system problem; the MagSafe board can fail independently of the logic board | MagSafe light present but battery never rising, or MagSafe light absent while USB-C charging works, point to different paths; MagSafe board fault symptoms include normal magnetic attachment with no light at all, or amber and green alternating without stable charging | MagSafe light colour plus macOS battery status, including charging on hold, slow charger, and service recommended wordings |
| Apple, USB-C and Apple Silicon | USB-C PD is the main charging method; MacBook Pro models from 2021 onward carry both MagSafe 3 and Thunderbolt USB-C, and the two paths differ | USB-C charging works while MagSafe fails points more toward the MagSafe path; both paths failing points more toward power-management firmware, battery, or logic-board layers; Apple Silicon machines have no separate SMC to reset, so older SMC reset key combinations do not apply and an ordinary restart is the corresponding action, while Intel T2 machines do have an SMC reset procedure | Cross-testing the same known-good adapter and cable across every USB-C port and the MagSafe port, recording which ports work and which do not |
11. Symptom Comparison Table: For Narrowing, Not for Diagnosis
The table carries a usage rule that must be preserved. One symptom can map to several layers. The table states only which layer is more likely and which comparison is most useful next. It must never be read as "symptom X means part Y is faulty." Wherever the table says "points toward," the exclusions that remain must be read with it.
| Symptom the user sees | Layer more likely to be involved | What still needs excluding, or the next useful comparison |
|---|---|---|
| Adapter LED goes out or flickers on plug-in; battery never rises | Adapter itself, DC-in, mainboard input layer | Change socket, change adapter, check adapter LED behaviour on another machine or another socket |
| BIOS or boot display shows adapter unknown or wrong wattage; machine slows down and fails to charge | Identification layer: identification pin, DC-in, mainboard identification circuit | Fit an original adapter of the same wattage; if it still reads unknown, identification-layer suspicion rises |
| Powered-off plug-in shows solid white or green light but battery level never rises overnight | Charger IC, BATFET, or battery-path layer; adapter recognised and system powered, but the charging branch has no output | Battery-removed, adapter-only boot comparison; if it will not boot without a battery either, the layer estimate must be revised |
| Powered-off plug-in shows solid amber or orange light but battery level does not rise | Charging in progress with declining health, charging on hold, a charging limit, or current limited by the limiting layer | Check manufacturer charging-limit settings and the battery health report; observe again under light load or powered off |
| Charging light repeatedly flashes amber or orange, or alternates green and amber, while battery level wanders | Unstable connection, charging-system abnormality, MagSafe board or DC-in contact layer | Clean the port exterior, swap cable and adapter, record the flashing rhythm; Dell diagnostic codes require counting amber and white flashes |
| Machine boots on battery alone and runs normally on adapter power, but the battery never charges | Charging-branch layer: charger IC, BATFET, sense resistor, or EC-to-charger-IC communication; normal system operation proves the ACFET and main supply rail are not fully dead | Adapter comparison, charging-limit check, inspection measurement of charging output |
| Machine will not boot on adapter power, but the charging light or adapter LED is on | Main supply rail, power-selection layer, or power-rail short-circuit suspicion; power reaches the board but the system rail never establishes or is shut down by protection | Stop using the machine and send it for inspection; this symptom must not be tested with repeated plug-in attempts |
| Instant cut-out, automatic shutdown, or adapter voltage pulled down on plug-in | Input-layer short-circuit suspicion: main-rail short, shorted capacitor, shorted buck high-side MOSFET | Stop using it immediately and do not plug it in again; before inspection, record appearance and smell only, with no powered tests |
| Machine works on adapter power but battery level falls under heavy load such as gaming or rendering | Insufficient adapter power, hybrid-power supplementation, or PD negotiated wattage shortfall, primarily in the previous guide scope | Compare an original adapter of the rated wattage and charging behaviour under light load |
| Machine dies the instant the adapter is removed, although battery level is displayed | Battery, BATFET, or battery-path layer; "works on adapter, dies on removal" commonly points first to battery or battery path rather than the charger IC itself | Battery comparison is subject to a written quotation after inspection; built-in batteries must not be removed by the reader |
| Swollen battery, deformed chassis, keyboard or base cover pushed up | Battery safety layer, taking priority over all charging diagnosis | Stop using the machine immediately, do not charge again, do not press on it, and follow manufacturer or recognised recycling routes |
12. Common Misdiagnosis Traps
| Misdiagnosis | Why it seems plausible | Why it may be wrong | Correct comparison |
|---|---|---|---|
| Mistaking it for battery failure | Failure to charge feels like a dead battery | Battery health report normal, a replaced battery behaves the same way, and adapter-only boot also fails, all point to the board layer rather than the battery | Known-good battery comparison; check whether BIOS detects the battery; record the adapter-only state |
| Mistaking it for a PD negotiation problem | Any USB-C charging failure invites a PD explanation | PD negotiation can succeed, with the adapter recognised and voltage correct, while the path after the charger IC is broken, so charging still fails; PD sits at L0 to L1 while the charger IC sits at L2 to L4 | Original adapter and cable comparison, direct machine-port connection; only when PD succeeds yet no charging current appears should the board layer be suspected |
| Mistaking it for EC failure | The EC manages power, so power problems get blamed on the EC | A working EC with a broken charger IC link, or a charger IC with no supply, can look like EC failure; normal EC functions elsewhere are important counter-evidence | Separate two evidence groups: whether other EC functions are normal, and whether charging settings and status can be read |
| Mistaking it for insufficient adapter power | Falling battery level on adapter power feels like an underpowered adapter | Adapter power can be sufficient while a failed supplement path leaves the battery unable to help under heavy load, producing the same fall and cut-out | Compare idle against heavy load and powered-off against powered-on charging; compare adapter power with the manufacturer specification rather than guessing |
| Mistaking it for a charging limit or software problem | Stopping at one percentage looks like a setting | A charging-limit function is design behaviour; a board-layer fault usually means no charging at all, repeated stopping and starting, or instant cut-out on plug-in, not a stable stop at a limit value | Check manufacturer battery-tool limit settings and BIOS charging modes; enter the board layer only after exclusion |
| Mistaking board replacement for a certain cure | A faulty board feels simplest to replace entirely | The short-circuit source may be a single capacitor, one MOSFET, or one corroded trace; full board replacement can cost more than chip-level repair and can hide the true cause | Ask the inspection report to state which layer failed and the advantages, disadvantages, and written quotation for board replacement and chip-level repair separately |
| Mistaking a dry surface for no water damage | A machine that works after drying feels undamaged | Electrochemical migration corrosion can appear days or even weeks later; a dry surface does not mean the board underside, connector underside, or shield underside is dry | Any water history must be reported, even when the machine worked afterwards, with liquid type and timing |
13. Layered Exclusion in Three Stages, Without Teaching Disassembly
The first stage is checks without disassembly, which readers can safely perform. First, record the three states: adapter only, battery only, and adapter plus battery together, noting boot behaviour, charging-light behaviour, and the system battery icon for each. Second, record the load relationship: whether charging works at idle, whether faults appear only under heavy load, and whether powered-off charging is normal. Charging normally when powered off but failing when powered on points toward insufficient supply or the supplement path more than toward the battery itself. Third, compare adapter and cable by using an original or known-good adapter and cable of the same specification, plugged directly into the machine charging port, without a dock, a hub, or an extension strip, changing one item at a time. Dell, HP, and Lenovo official guidance all begin by checking socket, adapter, cable, and charging port for damage and debris before looking at battery health. Fourth, read BIOS or UEFI and manufacturer diagnostics: whether the adapter is recognised, whether displayed power reads unknown or below its rating, and whether the battery is detected. Lenovo guidance states that entering BIOS to check battery detection separates hardware from software problems, and Dell likewise states that adapter status should be read in BIOS. A BIOS display of unknown is a clue, not a conviction. Fifth, compare with the charging-limit function switched off. A manufacturer battery-tool charging limit can make a machine look as if it cannot reach full charge, and Lenovo official explanation states that failure to reach 100 percent can be a design function rather than a fault. This guide assumes the earlier guide already excluded that case, but the reminder belongs here so readers do not mistake a limit for board failure. Sixth, stop immediately on safety signals: a swollen battery, a scorched smell, smoke, an adapter light that goes out on plug-in, or abnormal machine heating. Stop using the machine, disconnect power, and send it for inspection. Do not test with repeated plug-in attempts. HP official guidance likewise warns that a swollen battery is a safety risk requiring the machine to be stopped and taken to the manufacturer or an authorised repairer.
The second stage is comparison before inspection, where the reader supplies evidence rather than performing a repair. Prepare a comparison table for inspection: which adapter and cable were tried, which ports were tried, which states produced the fault, whether there was water exposure, a drop, an adapter change, lightning, or unstable mains before the fault, whether the fault appeared suddenly or gradually, and whether it is intermittent or permanent. Back up data before inspection. Board-level inspection may involve disassembly and board replacement, so data risk must be handled first. That sentence is a reminder only and is not expanded here into a data-recovery guide. Ask about the quotation structure: inspection, chip-level repair, board replacement, and machine replacement are four different options, and a written quotation should be requested before any decision. When sending the machine for inspection, bring the machine itself, the original adapter and cable, and the failing adapter and cable set. If the problem relates to a dock, bring the dock as well. If it relates to a multi-port combination, bring the cable set that produced the fault.
The third stage is board-level inspection concepts, described as concepts only, never as operations. The technician way of thinking follows the supply path and narrows it stage by stage: whether input reaches the board, whether ACDET or ACOK recognises the adapter, whether the path MOSFETs turn on, whether the charger IC has supply and output, whether the sense resistor and monitoring signals are reasonable, whether a main system rail is shorted and dragging the whole board down, and whether SMBus communication responds. The identification principle is that replacing the battery without success does not prove the charger IC, and replacing the charger IC without success does not prove board replacement. The discipline is to exclude layers in order and change only one variable at a time. Every step of that process can reveal a second fault point, so this guide never writes that replacing one IC fixes the machine immediately, and it gives no fixed repair time.
14. First Distinctions for Power-Rail Shorts: Concepts Only, Not Measurement Instructions
Repair-community general logic states that a main rail, such as VBAT, 3.3 V, or 5 V, with resistance to ground near zero indicates a hard short, a definite but abnormally low resistance indicates a soft short, and a fault appearing only under heating or load indicates an intermittent short. Common short locations concentrate near the DC-in socket, coils, MOSFETs, and ceramic capacitors near the charger IC. A short is not necessarily in the IC itself. Capacitors are suspected first in most cases, MOSFETs second, and the IC itself relatively rarely unless there is physical damage. This guide must state at the same time that in-circuit MOSFET measurement is easily disturbed by parallel paths and misjudged, and that a multimeter in-circuit test alone can hardly confirm a diagnosis. Isolation against a circuit diagram or boardview is required, and that requirement is precisely why readers should not open the machine to measure. It is a reason, not a lesson.
Hard, soft, and intermittent shorts call for different handling directions, but readers neither need nor should attempt to distinguish them personally. A hard short shows most clearly and triggers adapter protection fastest, which makes it easiest to notice but also most dangerous, and repeated plug-in attempts must stop absolutely. A soft short may present as extremely slow charging, cut-out under load, or localised heating, and readers easily mistake it for insufficient adapter power or battery ageing, delaying inspection until the short spreads. An intermittent short is hardest to catch, because it may appear only under heat or mechanical stress, and reader history such as faults appearing only when hot, only when moved, or only during gaming is extremely useful for narrowing inspection. Whatever the short type, the correct reader action is identical: disconnect power, stop using the machine, record observations, and send it for inspection.
Bench short-circuit investigation concepts include visual inspection for burning and corrosion first, then main-rail resistance to ground, then a controlled method for finding hot spots. This guide must carry a warning: those are concept descriptions only. Readers must not open the board, measure the board, inject power, or solder the board. Wrong handling can spread a short, burn more components, and create a safety risk. Power injection to find hot spots is a bench operation requiring a circuit diagram, a current-limited supply, and experience, with a high in-circuit misjudgement rate. A reader attempting it can expand the damage. That warning is not written to frighten readers. It reflects the fact that short-circuit detection is itself a staged isolation procedure, and one wrong isolation step expands the fault area across the whole board.
15. Dangerous Operations That Must Never Be Attempted
First, never keep using an adapter or a machine that has shown a scorched smell, burn marks, smoke, abnormal heating, or an adapter voltage pulled down, and never plug and unplug repeatedly to see whether it recovers. Second, never puncture, flatten, bend, prise up, or dismantle a swollen or damaged lithium battery. Dell manuals state explicitly that a swollen battery stuck in place must not be prised out with tools, that a damaged or swollen battery must not be reinstalled, and that a swollen battery should be stopped and handled through manufacturer or recognised recycling routes. Third, never insert a metal tool into a live charging port to clear dust or straighten pins. A metal tool plus a live port carries a short-circuit risk. Cleaning is limited to powered-off exterior inspection, and internal port cleaning belongs to inspection. Fourth, never open the base cover, disconnect a battery cable, measure the mainboard, or inject voltage into a main rail to find hot spots. Power injection for short finding in repair literature is a bench operation requiring a circuit diagram, a current-limited supply, and experience, with a high in-circuit misjudgement rate, and reader attempts can expand damage.
Fifth, never replace a charger IC, a MOSFET, or a sense resistor personally. In-circuit MOSFET testing is difficult to confirm, and replacing the wrong part, damaging the board during soldering, or choosing the wrong compatible part can turn a repairable board into an unrepairable one. Original and compatible part selection must be checked against part number and circuit diagram, never substituted on similar appearance or similar numbering. Sixth, never use an adapter of mismatched wattage or voltage as a long-term stand-in, especially never run heavy loads on a low-wattage adapter while expecting charging. PD negotiation will limit current, but a sustained deficit only makes the problem harder to separate. Seventh, never keep charging in wet conditions, after water exposure, or with corrosion visible in a port. Disconnect power first, stop using the machine, and send it for inspection. Do not blow it dry with a hot hair dryer and plug it in immediately. Eighth, never update BIOS or EC firmware as a desperate measure. A BIOS update itself needs stable power, and a power cut during an update on a machine whose charging circuit is already abnormal can worsen the problem. Firmware updates should be considered only under manufacturer guidance and with stable power.
16. Hong Kong Context: Humidity, Commuting, Third-Party Adapters, and Local Parts Reality
For humidity, the Hong Kong Observatory 1991 to 2020 climate normals give an annual mean relative humidity of 78 percent. That figure proves only that Hong Kong is persistently humid and cannot directly prove that humidity caused any one machine charger IC failure. It can be used to explain why liquid residue is hard to dry thoroughly and why corrosion sustains more easily, but it cannot be used to assert a failure rate. A further dimension is salt spray and corrosion. An EDN printed-circuit-board corrosion case notes that in severe corrosion samples from South China, Hong Kong, and Japan, the main corrosion elements were sulphur and chlorine, that chloride near the sea may come from sea spray, and that a server-room cycle of daytime air conditioning at 20 degrees Celsius and 50 percent relative humidity rising overnight without air conditioning to about 30 degrees Celsius and 80 to 85 percent relative humidity accelerated corrosion. The general PCB corrosion mechanism is that atmospheric salt and moisture form compounds on copper traces and component pins, eroding metal and causing open circuits, while ionic contamination plus moisture can also grow dendrites on copper traces and short neighbouring traces. Hong Kong repair discussion also notes that silver-plated parts stored in a non-climate-controlled Hong Kong warehouse can turn black within weeks, and that although ENIG boards are relatively tolerant, non-wiping contacts lose contact quality after long idle periods. The writing boundary is that all of this proves only that a humid Hong Kong environment with salt spray raises the prior likelihood of corrosion and poor contact. It does not mean every Hong Kong machine that fails to charge is corroded. This guide must state that confirmation requires inspection appearance or microscopic examination, never writing an environmental factor as a case diagnosis. When green or white corrosion is visible at a charging port, the only supportable wording is suspected corrosion with inspection recommended, never confirmed corrosion causing charger IC damage.
For commuting, daily backpack carrying, crowded-train squeezing, and cable pulling during charging keep the charging port, solder joints, and cable under long-term stress. Those are reasonable usage-context descriptions, not fault statistics for any brand. Carrying a machine by one corner, squeezing it against hard objects in a backpack, and letting the machine hang from its cable all transfer force to the charging port and the board, and accumulated force is one source of solder-joint fatigue. The reader prevention available at the usage level is limited: support the machine body while plugging in, never carry it by its cable, separate hard objects inside a backpack, and send a loosening charging port for inspection early rather than waiting for total failure. Prevention cannot reverse a joint that has already cracked, but it can reduce new stress.
For third-party adapters and cables, a replacement adapter must be compared against original voltage, current, power, connector, and PD specification, never judged only by whether the plug fits. Cables without certification, with unclear markings, or with missing or falsely marked E-markers can let a device draw more current than the cable can safely carry, and that risk has public discussion behind it. When an office shares one adapter, cable and adapter must be compared together, not adapter alone, because the cable changes the tier list the adapter offers. That point was already made in the previous guide. Here it matters only as a reminder that board-layer inspection must start from a known-good combination, or the inspection starting point itself is wrong.
For local parts and pricing reality, only principles may be written, with no invented figures. Charger IC part numbers differ by machine, and some older parts may be discontinued or need ordering. Before inspection, readers should ask clearly about parts supply, repair labour, and how inspection charges are calculated. Apart from the four confirmed prices, every other item is subject to a written quotation after inspection. The chip-level repair background flow is inspection, appearance and power-rail first distinction, circuit-diagram or boardview comparison, isolated measurement of suspect components, replacement followed by re-measurement of charging voltage and current, and light-load and powered-off charging verification. Every step of that flow can reveal a second fault point, so this guide never writes that replacing one IC fixes the machine immediately, and it gives no fixed repair time.
17. Decision Framework: Chip-Level Repair, Board Replacement, or Machine Replacement
The decision framework is best written as four questions, all answered before deciding, never on a single price alone. The first question is age and residual value: is the machine still worth saving with chip-level repair, and if parts beyond the board, such as screen, keyboard, battery, and storage, are all near end of life, do all pending repairs added together exceed what the machine is worth? The total must be counted, not the charger IC item alone. The second question is fault scope: does the inspection report limit the fault to the charger IC or its immediate surroundings, or has it spread across multiple power rails, processor or memory supplies, and input and output circuits? Wider scope means greater uncertainty for chip-level repair success and durability, and higher reasonableness for board replacement. The third question is data and downtime: is there unbacked-up data, how long will inspection take, and is there a spare machine? Data recovery and downtime costs belong in the decision, not only the repair charge. The fourth question is recurrence risk: has the root cause been removed, for example an uncertified adapter not yet replaced, water corrosion not yet cleaned, or a cooling problem not yet addressed, since an unremoved cause can break the repair again? At inspection, readers should ask what the cause is, how recurrence can be avoided after repair, and whether the adapter, cable, or usage environment needs to change at the same time.
The three options are compared without specific board-replacement or machine-replacement prices, which are always subject to a written quotation after inspection. Chip-level repair targets localised faults in the charger IC, MOSFETs, sense resistors, and corroded traces. Its advantage is keeping the original board and data path, with a cost usually below full board replacement. Its disadvantage is dependence on parts supply, board-layer damage extent, and corrosion range, so not every board is suitable. Board replacement suits multi-layer damage, damage to traces under the IC, or instability after chip-level repair. Its advantage is clearing the fault scope in one step. Its disadvantage is higher cost and the need to confirm that the new board is compatible with original accessories, serial number, and system licensing, with price subject to a written quotation after inspection. Machine replacement suits high machine age, several large repairs stacking together, discontinued parts, or a total repair cost already approaching a reasonable replacement range. Before deciding, back up data and check the new machine charging specification, including barrel or USB-C PD type and power requirement, so a wrong adapter is not bought and the same mistake repeated. No absolute conclusion such as repair always cheaper than replacement or board replacement always better than repair may be written. The decision must be made from the inspection report plus a written quotation plus machine condition together.
The pricing principle is repeated once more because it is the easiest place in this guide to break the rules. The only confirmed prices that may be used are battery from HK$580, screen from HK$1,280, board chip-level repair from HK$880, and keyboard from HK$480. Every other repair, inspection, board-replacement, and machine-replacement price is subject to a written quotation after inspection. No estimated price, no price range, and no price carried over from an earlier guide may be written. Asking about the quotation structure before inspection is itself part of the decision framework, not an extra step. For the price range, see pricing; for how an inspection is arranged, see diagnostic service.
18. Inspection Information Checklist and Closing
The inspection information checklist belongs at the end of the guide, so readers can follow it item by item and inspection does not have to repeat the same investigation. The checklist covers model, adapter specification, comparisons tried, water or drop history, whether the fault appeared suddenly or gradually, whether it is intermittent or permanent, charging-light behaviour, and BIOS adapter or battery display. The three power-state comparison, the idle against heavy-load against powered-off comparison, the adapter, cable, socket, and direct-connection comparison, and the charging-light video should all be handed over together, so inspection can start from the most likely layer rather than from the beginning. Complete information does not guarantee a successful repair. Its function is to reduce repeated investigation and let the inspection report name the failed layer sooner.
The closing returns to layer logic. Failure to charge is not a diagnosis. It is a symptom, and behind that symptom sit seven possible layers, from external power to the power rails, each with its own signals and comparisons. After basic checks are done, the most useful reader work is not guessing which part failed, but completing the three-state comparison, the load comparison, the parts comparison, the light recording, and the history recording, and then stopping and sending the machine for inspection. The value of board-level inspection is to follow the supply path stage by stage, name the failed layer, and then decide between repair, board replacement, and machine replacement from the inspection report plus a written quotation plus machine condition together. The shop is Room 1209, Dali Plaza, Shek Mun, Sha Tin, telephone +852 2186 8010, WhatsApp +852 6558 6806, opening hours 12:00-19:00. For the price range, see pricing; for how an inspection is arranged, see diagnostic service.
19. Reading an Inspection Report: Four Questions to Ask and Three Common Traps
When an inspection report comes back, readers do not need to read a circuit diagram, but they do need to ask four questions. First, which layer failed? Does the report name the detection, selection, conversion, monitoring, communication, or power-rail layer, or does it only say the mainboard is faulty? A report that only says the mainboard is faulty carries too little information for a decision, and the reader should ask which layer, which location, and which signal was abnormal. Second, how wide is the fault scope? Is only one layer involved, or has the fault spread across multiple power rails? Scope decides the reasonableness of chip-level repair against board replacement, and the clearer the scope is stated, the more stable the decision. Third, what is the root cause? Is it water corrosion, a drop-cracked solder joint, an overvoltage surge, an uncertified adapter with unstable output, or a part that failed by itself? Without a stated cause, recurrence risk cannot be assessed, and the repair can easily fail again. Fourth, how much verification was done? After replacement, were charging voltage and current re-measured, were light-load and powered-off charging verified, and was the machine observed over a period of time? Verification is not a single sentence saying the machine is repaired. It must state which states were tested, because an idle test does not prove heavy-load behaviour and a powered-off test does not prove powered-on behaviour.
Three common traps deserve care. The first trap is replacing a whole group of parts at once while the report never states which part was the cause. Staged exclusion, changing one variable at a time, is inspection discipline rather than slowness. It is accurate. Replacing everything at once looks faster, but the true cause is then never identified, and the next failure has no starting point. The second trap is substituting a similar-looking part. Charger ICs and MOSFETs must be matched by part number, circuit diagram, package, and parameters. Similar numbering, similar pin layout, and similar appearance never prove substitutability. A wrong substitute can leave charging behaviour abnormal or burn more components and turn a repairable board into an unrepairable one. The third trap is ignoring a second fault point. Short-circuit and charging faults often have multiple fault points, and a second point may only appear after the first is repaired. That is not necessarily a repair error, but the report must state whether other layers were checked and whether a second suspicious location was found. Asking about those three traps is not a challenge to the inspector. It is how a decision gains enough information.
Board replacement carries compatibility questions that must be asked clearly. Has the new board been checked for compatibility with original accessories, serial number, and system licensing? Can the original memory, storage, wireless card, keyboard, and screen all be reused as they are? Will system licensing need to be handled again because of the board change? All of those questions belong before a written quotation, not after replacement reveals a mismatch. Machine replacement carries a parallel check. The new machine charging specification must be compared, including barrel or USB-C PD type, power requirement, port position, and cable requirement, so the wrong adapter is not bought and the same mistake repeated. An old adapter may not suit a new machine, and a new cable may not suit an old adapter, so the whole combination must be compared together rather than one item alone.
For data handling, backing up before inspection is a necessary reminder, but its limits must be stated honestly. If the machine will no longer boot at all, reader backup may already be impossible, and the data-handling option and its charge are likewise subject to a written quotation after inspection. Readers must not remove a storage device personally or force power to read it. At inspection, readers should state which data matters most, which data is already backed up, and which data is not backed up, so the inspection procedure can take data risk into account. Data risk is not part of the charging circuit, but it is part of the inspection decision, and it is what the third decision question addresses.
For daily usage habits, several actions can reduce stress on the charging circuit. First, use an original adapter and cable combination or a known-good combination of the same specification, and do not rely long-term on a replacement with unclear markings. Second, support the machine body while plugging in and never carry the machine by its cable, reducing force on the charging port and solder joints. Third, do not block cooling vents long-term. High-temperature charging accelerates solder thermal fatigue, and although heavy load alone does not burn out a charger IC, sustained high temperature with poor cooling does age components faster. Fourth, after water exposure, do not power on immediately to test. Disconnect power, stop using the machine, and send it for inspection. A dry surface does not mean the board underside is dry, and the cost of delayed corrosion far exceeds the cost of sending the machine a day earlier. Fifth, a manufacturer charging-limit function is design, intended to slow battery ageing. A machine stopping at one percentage should first be checked against the limit setting rather than treated as board failure at once. Those five habits cannot guarantee a machine never fails, but they reduce misdiagnosis and recurrence risk and make the next inspection history clearer.
20. How to Describe the Fault So Inspection Starts in the Right Place
The quality of a fault description at inspection can directly decide which layer inspection starts from. The most useful descriptions share three qualities: state facts, state comparisons, and do not state conclusions. Stating facts means stating what was seen: when the fault began, whether it appeared suddenly or gradually, whether it is intermittent or permanent, how the charging light flashes, what the system battery icon says, and how the adapter light reacts. Stating comparisons means stating the combinations tried: which adapter, which cable, which port, adapter only or battery only, idle or heavy load or powered off, and whether the result changed. Not stating conclusions means not opening with a claim that the charger IC has burned out, that the mainboard is completely dead, or that the battery is finished, because a conclusion pulls the inspection direction off course. The inspector then spends time overturning the conclusion before returning to staged exclusion. The reader job is to supply evidence. The inspector job is to draw conclusions from evidence. When that division is clear, inspection is faster and more accurate.
A side-by-side example makes the difference obvious. A vague description reads: the machine will not charge, it has been broken for a long time, and many things have been tried without success. That sentence carries almost no information. How long is unknown, whether days or months. Which things were tried is unknown. Whether failure to charge means no charging at all or intermittent charging is unknown. A useful description reads: the fault began last Wednesday, the machine was caught in rain the day before, it worked after drying, and from this Monday it stopped charging entirely. It boots on battery alone but not on adapter alone. An original adapter and a different cable were both tried without change. Powered-off charging overnight produced no rise in battery level, and the charging light stays solid amber without flashing. That paragraph carries a timeline, a water history, a three-state comparison, a parts comparison, a load comparison, and a light description. An inspector reading it knows to start from the input layer and the detection layer rather than from the battery. Both descriptions may describe the same machine, but their information content differs enormously, and so does the inspection starting point.
Reading an original adapter label is also worth doing before inspection. A label usually carries input voltage and current, output voltage and current, power, and connector type, and a USB-C adapter also carries PD tier information. Read the label line by line rather than copying a single total wattage figure, because wattage is voltage multiplied by current, and different tiers can share a wattage figure while differing in voltage. The machine may need one particular tier rather than the total. For comparison, place three documents together: the original adapter label, the replacement adapter label, and the machine specification page charging-input requirement, and compare them line by line. A difference in any single line can be the problem. Photograph the label so every output line is visible, not only the large print on the front, and hand the photographs to inspection so the inspector does not have to ask for them again.
Communication at inspection deserves a few proactive sentences of its own. First, state whether important data inside the machine is unbacked up, so the inspector can take data risk into account when planning the procedure. Second, state whether a spare machine is available and whether the machine is urgently needed, so the inspector can explain the procedure order clearly rather than assuming urgency is known. Third, state how decisions should be handled: whether work must wait for a written quotation, and whether chip-level repair is acceptable or only board replacement is acceptable. The earlier those positions are stated, the fewer misunderstandings follow. Fourth, when the report arrives and anything is unclear, ask. Ask which layer, ask the scope, ask the cause, and ask the verification. Those are the four questions from the previous section. Ask them clearly before deciding, rather than approving on feeling at once. Good communication produces a clearer inspection report, and a clearer report produces a steadier decision.
There is one more communication habit worth building: keep the inspection evidence together as a single package. The three-state comparison, the load comparison, the parts comparison, the BIOS or manufacturer-tool readings, the charging-light video, the adapter-label photographs, and the history should travel together, in the same handover, rather than arriving in separate messages over several days. Inspection is a staged process, and staged processes depend on complete starting information. A missing piece discovered halfway through forces the process back to the beginning, or forces a second round of questions that could have been answered on day one. A single complete package does not guarantee a repair, and this guide never claims that it does. Its aim is only that inspection starts from the best available evidence, and that is the most useful contribution a reader can make before any board is opened.
It is also worth stating plainly what inspection cannot do from a description alone. No description, however detailed, can confirm which layer failed. A description can raise suspicion toward a layer, and it can rule layers out, but confirmation requires staged bench work: appearance and power-rail first distinction, circuit-diagram or boardview comparison, isolated measurement of suspect components, replacement followed by re-measurement of charging voltage and current, and light-load and powered-off charging verification. Readers who understand that boundary will not be disappointed when inspection reports a different layer from the one they suspected, and will not treat a revised layer as a contradiction. A revised layer is what staged exclusion is for. The description points inspection in a direction. The bench work decides the layer. Keeping those two roles apart is the difference between a useful handover and a misleading one, and it protects both the reader and the machine from decisions made on a guess.
Finally, the same discipline applies after repair as before it. When the machine comes back, the reader should repeat the comparisons that originally showed the fault, in the same states and under the same loads, before treating the repair as complete. Idle charging, powered-off charging, and heavy-load behaviour should each be checked, because a repair verified only at idle has not been verified under the conditions that exposed the fault. If any original symptom returns, the reader should record it in the same factual style: which state, which load, which adapter and cable, what the light did, and what the system icon showed. That record is far more useful for any follow-up than a general statement that the problem came back. It lets inspection pick up exactly where the previous staged exclusion left off, instead of starting again from the beginning. Layer thinking does not end when the machine is returned. It is the method for describing, deciding, and verifying every charging fault from the first symptom to the final check.
21. Glossary: What Inspection-Report Terms Mean in Plain Language
| Term | Plain-language meaning | The one point readers should remember |
|---|---|---|
| ACDET | Adapter-detect input pin, with a threshold set by a resistor divider, used to judge whether an adapter is valid | Adapter output does not mean ACDET recognises the adapter; a drifted divider resistor or a broken trace can make the judgement wrong |
| ACOK | Adapter-valid signal, driven high only when all conditions are met, telling the host that charging may begin | ACOK staying low means charging never starts; that is a detection-layer clue, not proof that the adapter is faulty |
| REGN | Internal low-dropout regulator output in the BQ family, supplying charger IC internal circuits and MOSFET drive | A missing REGN is an inspection concept indicating no IC input voltage or no IC response; it must not be measured by readers |
| ACFET, RBFET, BATFET | Adapter input switch, reverse-blocking switch, and battery switch, responsible for the adapter-to-battery handover | If any one fails to turn on, the machine may use only one side; the three-state comparison exists to separate this layer |
| NVDC | Architecture in which system voltage follows battery voltage but never falls below the minimum system voltage | With the battery empty or removed, the system can still boot immediately; booting does not prove charging works |
| HPB, supplement mode | Design behaviour in which the battery helps the adapter supply power under heavy load | Falling battery level on adapter power can be normal design; compare idle and powered-off charging before suspecting the board |
| PTM | Efficiency mode in which input power passes directly to the system | Its purpose is to reduce conversion losses, not to bypass protection, and it must not be confused with HPB |
| SMBus, I2C | Channel over which the EC talks to the charger IC, carrying charging voltage, current limits, and status | Settings that cannot be written and status that cannot be read, while other EC functions work normally, point to the communication layer rather than a faulty EC |
| Sense resistor, IOUT | Low-value resistor in series with a power path for measuring current, whose voltage difference is amplified into a monitoring signal | An open or drifted sense resistor leaves the chip blind to real current, so protection can trigger for no apparent reason |
| Hiccup mode | One presentation of system short-circuit protection, repeatedly attempting and then stopping | Repeated attempts do not mean the machine is about to recover; while a short is unexcluded, repeated plug-in attempts must stop |
| Break-before-make | Handover principle of disconnecting one side before connecting the other | Wrong handover order causes a supply conflict, and that is why the path-selection layer exists |
| Watchdog timeout | Charger IC behaviour of stopping charging or hybrid power when expected host writes do not arrive within a set window | Charging that starts, stops after a while, and starts again can reflect a communication timeout rather than a permanently broken wire |
| Ideal-diode operation | Battery MOSFET behaviour that lets the battery supplement the system bus without a manual switch | Supplementation is automatic in the chip; readers do not enable it, and its failure under heavy load is a board-layer matter for inspection |
That same caution applies to the whole guide. Layer names, part families, and protection terms are tools for narrowing a fault, not shortcuts to naming a failed part. The guide has deliberately kept every figure tied to its part number, every light pattern tied to its model manual, and every short-circuit statement tied to staged isolation at inspection. Readers who carry that discipline into inspection will find that reports become easier to read, questions become easier to ask, and decisions become easier to defend. The alternative, jumping from a symptom straight to a part name, feels faster and costs more in wrong parts, repeated visits, and expanded board damage. Layer by layer is slower in words and faster in outcomes, and that is the trade this guide has tried to make throughout.
One further principle belongs with any report reading: a term by itself is not a conclusion, and only a measurement result is a conclusion. When a report states that a signal is abnormal, it should also state the measurement context and the comparison result, such as the power state under which the measurement was taken, whether the result changed after any item was swapped, and whether other layers showed the same abnormality. If a report offers terminology without measurement context, readers should ask for that context rather than guessing from the term alone. Asking is not distrust of the inspector. It is how a decision gains enough information, and it is the final step of the layer logic carried through this whole guide: enter by symptom, narrow by comparison, identify by report, and decide from a written quotation together with machine condition.
Frequently Asked Questions
Q1: A replacement battery still does not charge. Does that mean the charger IC is dead?
No. A replacement battery that still does not charge only excludes the battery itself; it does not point to any one board layer. Six layers remain: the detection layer, the power-path layer, the conversion layer, the sensing layer, the communication layer, and the power-rail layer. A detection circuit that fails to recognise the adapter, a path MOSFET that fails to turn on, a charger IC with no output, a sense resistor that has gone open-circuit and distorted the current reading, a broken EC-to-charger-IC link, or a downstream rail short dragging the whole board down can all present as failure to charge. Declaring the charger IC dead before step-by-step inspection risks missing the real cause and replacing the wrong part. The useful action is to record the three power states and the charging-light behaviour, attach the adapter specification, and let the inspection report state which layer failed.
Q2: The adapter clearly outputs power, so why does the machine act as if there is no adapter and the charging light stays off?
Because an adapter producing output and a board recognising an adapter are two different things. The board uses the ACDET pin to judge whether the adapter is valid. On the BQ24725A, for example, ACDET must sit between 2.4 V and 3.15 V, the supply must clear its undervoltage threshold, and the supply must exceed the sense reference by more than 275 mV before ACOK goes high and charging is permitted. A drifted ACDET divider resistor, a corroded or broken detection trace, or damage to the detection pin itself can make the system believe there is no adapter, so it does not charge and the light behaves abnormally. Those figures apply to that part only; other parts must be judged against the datasheet for the machine under inspection. Readers should not open the machine to measure the detection pin; recording that output is normal but the system does not recognise the adapter is already valuable inspection evidence.
Q3: The machine boots on battery alone but not on adapter alone. Which layer does that point to?
That comparison points toward the input and power-path selection layers, but it is not a diagnosis. Booting on battery alone shows that the battery path, the downstream rails, and the power sequencing broadly work. Failing to boot on adapter alone means something in the adapter input path, the path MOSFETs, adapter detection, or charger-IC supply has not come up. Detection failure can make the system refuse adapter power; an input MOSFET that will not turn on has the same outward effect; an input-rail short triggering protection is a third possibility. The next useful comparison is a known-good adapter of the same specification, plugged directly into the machine port, without an extension strip or a dock, changing one variable at a time. If the result does not change, stop repeated plug-in attempts, stop using the machine, and send it for inspection so the input path can be narrowed stage by stage.
Q4: With the battery removed, the machine boots on adapter power but will not charge. What does that mean?
This is a valuable comparison: the system power path broadly works, so the fault is more likely concentrated in the charging branch. Under an NVDC architecture the system can boot immediately even with the battery fully empty or removed, so booting proves nothing about the charging path. The charging branch can fail through a charger IC with no output, a battery MOSFET that will not turn on, an open sense resistor, a charging setting from the EC that never took effect, or a battery communication fault. Conversely, if the machine will not boot on adapter power even without a battery, the layer estimate must be revised: the fault may sit on the main power rail or the input layer rather than the charging branch alone. Machines with built-in batteries must not be opened for this comparison; use only an operation the manufacturer manual permits, or send the machine for inspection instead.
Q5: Battery level still falls while gaming or rendering on adapter power. Is the charger IC faulty?
Not necessarily. Many architectures are designed so the battery supplements the adapter: when system load exceeds what the adapter can supply, the battery automatically helps power the system. Texas Instruments calls this supplement mode and Renesas calls it Hybrid Power Boost, and its purpose is to stop the system bus voltage collapsing. Falling battery level on adapter power under heavy load can therefore be normal design behaviour, especially when the adapter rating is below the peak system demand, a high-performance mode is selected, or power-hungry peripherals are attached. Compare idle and powered-off charging first, then compare the adapter rating with the manufacturer specification. If idle and powered-off charging are normal and only heavy load loses charge, insufficient adapter power and the supplement design are more likely suspects than a board fault; if light load and powered-off charging also fail, board-layer suspicion rises.
Q6: The adapter light goes out the moment it is plugged in, or the machine cuts out instantly. What should I do?
Stop using it immediately and do not keep plugging it in to test. This presentation points more toward an input-layer or power-rail short: a hard short with near-zero resistance to ground on a main rail, a shorted ceramic capacitor, or a shorted buck high-side MOSFET can pull the adapter output down and trigger adapter protection. Repair discussions mention cases where an adapter output was pulled from 19 V down to 3 V or near zero, but case figures must not be transferred to your own machine. The short location can only be identified by isolating stages one at a time; shorts are most often suspected in capacitors first, then MOSFETs, and in the IC itself relatively rarely unless there is physical damage. The only reader actions are to disconnect power, note the light behaviour and any smell at the moment of plug-in, and send the machine for inspection. Injecting power to find hot spots and measuring rails on an opened board are bench operations, not reader steps.
Q7: The charging light flashes amber, or alternates green and amber. Can the colour alone identify the failed part?
No. Charging-light definitions vary by brand, model, and year, and the same colour can mean entirely different things on different machines. On Dell machines the battery status light can double as a diagnostic light, combining amber flash groups and white flash groups into a two-digit code that must be decoded against that model manual rather than guessed from colour. For ASUS, different write-ups contradict each other on flashing amber: charging in progress, poor connection, poor battery health, adapter trouble, or hardware fault have all been offered, which itself proves that flashing amber alone cannot diagnose. On MagSafe, solid amber can mean charging or charging on hold, while repeated flashing amber points more toward a connection or charging-system problem. Record a 10 to 15 second video of the light, count the flashes and pauses, and hand the model number and the video to inspection so the code can be judged against the correct manual.
Q8: Every other EC function works, so why can the machine still fail to charge?
Because a working EC and an EC that can talk to the charger IC are two different things. The charger IC receives its charging voltage, charging current limit, and minimum system voltage from the EC over an SMBus or I2C channel, and reports status back. On the BQ24715, for example, voltage is programmed in 16 mV steps and current in 64 mA steps, and charging requires several conditions at once, including the charge-enable bit, a valid ACOK, conducting path MOSFETs, and no overvoltage, overcurrent, short-circuit, or overtemperature condition; any one failure stops charging. A corroded or broken SMBus trace, a charger IC with no supply, or an IC that does not respond can all block settings from being written and status from being read while EC functions such as power button, keyboard, and fan control remain normal. That distinction matters: normal EC functions elsewhere are important counter-evidence against blaming the EC itself.
Q9: The battery health report looks normal, so why does charging behaviour keep going wrong?
Because a battery health report describes the battery itself, its capacity and cycle information, not the live current reading of the charging circuit. The charger IC measures input and charging current from the tiny voltage difference across a sense resistor. On the BQ24725A, for example, the IOUT signal is 20 times the sense-resistor voltage difference for host monitoring; on the BQ25720, the design uses 5 m ohm or 10 m ohm sense resistors as its reference. A sense resistor that has burned open, drifted in value, or lost a solder joint leaves the IC blind to real current, so current limiting, load shedding, and protection can trigger for no apparent reason, presenting as a frozen battery percentage or charging that repeatedly stops and starts. A normal battery report with erratic charging behaviour is therefore a clue pointing toward the sensing layer, but final identification still requires inspection; the sense resistor must not be judged by appearance alone.
Q10: The machine got wet, dried out, and worked, so why does it suddenly fail to charge weeks later?
Because liquid damage is often a slow process rather than instant death. Electrochemical migration needs three ingredients together: moisture, ionic contamination, and an applied bias. Metal ions dissolve at the anode, migrate toward the cathode, and grow dendrites that eventually bridge neighbouring conductors and cause a short, a process that can take hours to weeks, so a machine can fail long after its surface looks dry. Sugary, salty, or coffee liquids are worse than plain water because solutes remain on the board after the water evaporates and keep absorbing moisture and corroding copper traces and solder joints. Hong Kong annual mean relative humidity is 78 percent, which makes liquid residue harder to dry thoroughly and corrosion easier to sustain; that figure only proves Hong Kong is persistently humid and cannot by itself prove the cause of any one machine failure. Any water history should be reported, even when the machine worked afterwards, with the liquid type and timing stated as facts.
Q11: Charging works intermittently and stops when the machine is moved. Must that be a loose cable or port?
Not necessarily. Intermittent charging can be a loose port or cable contact, but it can also be an intermittent solder joint at board level. Solder fatigue is progressive: resistance rises slowly, signal integrity degrades, intermittent faults appear first, and permanent failure comes last. For leadless packages, especially large BGAs, a drop flexes the board, the stiffer component forces its joints to carry more stress, and lead-free solder with a higher elastic modulus tolerates mechanical shock relatively poorly. A cracked joint around the charger IC can therefore present exactly as charging that works until the machine is moved. The distinguishing test is to hold the plug still and change cable, adapter, and direct-to-machine connection one at a time: if it still drops with the plug held still, board-joint suspicion rises; if it only drops when the plug is moved, port and cable suspicion is higher. Either way, do not wiggle the plug hard to test, because force on the port transfers to the board and can worsen the joint.
Q12: How should I choose between chip-level repair, board replacement, and replacing the machine?
Decide with four questions together, never on a single price. First, age and residual value: is the machine still worth saving with chip-level repair, and if screen, keyboard, battery, and storage are all near end of life, do all pending repairs added together exceed what the machine is worth. Second, fault scope: does the inspection report limit the fault to the charger IC or its immediate surroundings, or has it spread across multiple power rails or processor and memory supplies, where wider scope means greater uncertainty for chip-level repair. Third, data and downtime: is there unbacked-up data, how long will inspection take, and is there a spare machine, because data and downtime costs belong in the decision. Fourth, recurrence risk: has the root cause been removed, for example an uncertified adapter not yet replaced, corrosion not yet cleaned, or a cooling problem not yet addressed, since an unremoved cause can break the repair again. The full trade-off table and pricing principle are in the decision-framework section of this guide, and every final charge is subject to a written quotation after inspection.
Q13: Can the charger IC for an old machine be out of stock, and does that mean it cannot be repaired?
That supply variable is real, but it cannot be generalised. Charger IC part numbers differ by machine, and some older parts may be discontinued or need ordering. The public example is the TI BQ24703 product page showing that part as out of stock, and the BQ24702 and BQ24703 platform dates from 2003 with a 2007 revision, which proves only that supply for an old platform can vary, not that all parts are unavailable. At inspection, ask three things: whether the part is available, whether it is an original or a compatible part, and how the final charge is calculated. Charger ICs and MOSFETs must be matched by part number, circuit diagram, package, and parameters, never substituted on similar appearance. Actual availability and charges are subject to a written quotation after inspection; neither availability nor unavailability should be assumed before inspection.
Q14: What should I prepare before inspection, and what are the shop details and starting prices?
Prepare seven groups of information: the model and adapter specification; the three power-state comparison, meaning adapter only, battery only, and adapter plus battery, each with boot behaviour, charging-light behaviour, and system battery icon; the three load comparisons, idle, heavy load, and powered-off charging; the one-variable-at-a-time results for changing adapter, cable, socket, and direct machine-port connection; the adapter recognition and battery detection status copied from BIOS or the manufacturer tool; a video of the charging-light pattern; and the history, water exposure, drops, adapter or battery changes, lightning, or unstable mains, stated as facts only. Bring the machine itself, the original adapter and cable, and the failing adapter and cable set. For prices, battery from HK$580, screen from HK$1,280, board chip-level repair from HK$880, keyboard from HK$480, and every other item is subject to a written quotation after inspection. The shop is Room 1209, Dali Plaza, Shek Mun, Sha Tin, telephone +852 2186 8010, WhatsApp +852 6558 6806, opening hours 12:00-19:00. For the price range, see pricing; for how an inspection is arranged, see diagnostic service.
Need Professional Laptop Repair?
Laptop King has 15+ years of repair experience in Shek Mun, Sha Tin. Free diagnosis, transparent pricing, no fix no fee. 180-day warranty for your peace of mind.