Chip-Level Repair Guide

Interactive version: open the Chip-Level Repair Lab for photo diagnosis, animated blueprint boards, multimeter probing, power-up sequences and fault simulations.

HDD · SATA SSD · NVMe · laptop and desktop motherboards · CPU/GPU/BGA. Diagrams, multimeter checks, and component-by-component explanation.

1. Read this first

Every brand and model uses different part numbers and layouts, but they are all built from the same few circuits: input protection → power converters → controller → memory/data. If you learn how to test those circuits with a multimeter, you can repair any model. For a specific model you then add its schematic and boardview (see section 14) to find exact part locations and expected values.

The universal repair loop (use it on every device)

1. Inspect visually under a microscope or magnifier: burnt parts, cracked caps, corrosion, missing parts, lifted pins, previous repair attempts.
2. Check for shorts with the power off: measure resistance and diode value from each main rail to ground (section 3).
3. Power on with current limiting (bench PSU or USB/DC meter) and watch the current draw.
4. Measure voltages in order of the power sequence. The first missing rail or signal is where the fault is.
5. Look at that circuit: its input, its enable, its output, and its feedback/power-good signal.
6. Replace the faulty part, then re-measure before reconnecting expensive parts (CPU, SSD, screen).

2. Tools & bench setup

3. Multimeter: the full method

3.1 The four modes you will use

3.2 Diode-to-ground measurement (the most important technique)

Board repair technicians measure every point with the probes reversed: red probe on ground, black probe on the test point. You are measuring through the protection diodes inside the ICs connected to that point, so every rail and signal gives a repeatable number.

3.3 Resistance to ground

Use it after diode mode shows something low. Let the reading settle (big capacitors charge up slowly).

Tip: swap the probes. If a reading is low both ways, it is a real short (resistive). If it is very different, you are measuring through a diode junction (IC or MOSFET body diode).

3.4 Voltage measurement tips

• Black probe on ground, red on the test point. Use the output side of an inductor (the side with the big caps) to measure a buck converter output.
• Measure the input, the enable (EN), the output, and power-good (PG) of every regulator. That is enough to judge any power IC.
• A rail that reads slightly low and drops further = load short or weak regulator. A rail that reads the input voltage = shorted high-side MOSFET — disconnect immediately, it will kill everything downstream.
• Pulsing voltage (e.g. 19 V on a 3.3 V rail’s input jumping up and down) = regulator in hiccup mode because of a short on its output.

4. How to test every component

4.1 Passive components

4.2 Diodes

4.3 MOSFETs (the most-replaced active part)

• In-circuit quick test: measure D–S and G–S in Ω mode. D–S near 0 Ω both ways (and the rail is not a core rail) = shorted MOSFET. G–S near 0 Ω = damaged gate.
• Powered test: N-channel switch: gate must be higher than source (by 5 V+ for load switches, or gate = source + ~5–10 V on input MOSFETs driven by the charger IC). P-channel switch (often on 3.3 V/5 V load switches): gate LOW = ON, gate = source = OFF.
• Common laptop MOSFETs: AON7xxx, AO4xxx, SiRA/SiS series, FDMC/FDMS, NTMFS, IRF, plus dual N-channel “power stages” in VRMs (e.g. DrMOS / smart power stages).

4.4 ICs (power ICs, controllers)

• You cannot “test” an IC with a meter directly. You test its pins: power in, enable, output, power-good, and the diode value of each pin vs a good board.
• A pin reading 0.000 when it should read 0.4–0.7 = that IC (or something on that line) is damaged.
• An IC that gets hot with nothing on its output = internally shorted.
• Read the marking, search “marking code + package” (e.g. small chips print a 2–4 letter code instead of the part number) to get the datasheet and pinout.

5. Circuits you will meet on every board

5.1 Buck converter (step-down regulator) — on every HDD, SSD and motherboard

5.2 LDO (linear regulator)

Small 3–5 pin chips (e.g. 1.8 V, 2.5 V, 1.2 V from 3.3 V). Pins: IN, OUT, GND, EN, sometimes PG/ADJ. Output missing with IN and EN present → replace LDO (if the output is not shorted). Gets hot = output shorted or IC shorted.

5.3 Load switch

A MOSFET or small IC that connects an always-on rail to a device only when needed (e.g. +3VS → +LCDVDD, +5V → USB port). Check: input present? enable present? output missing → switch bad, or output short has triggered its protection.

5.4 Rail naming you will see on schematics

6. Finding shorts and dead rails

A rail that measures ~0 Ω to ground has hundreds of parts on it. These methods find the one bad part.

6.1 Voltage injection (best method)

1. Remove battery, adapter, CPU (if socketed), RAM, SSD. Nothing expensive should be connected.
2. Set PSU voltage to no more than the rail’s normal voltage. For CPU/GPU core rails and anything ≤1.8 V, start at 1.0 V. Never inject 19 V into a 1 V rail.
3. Set current limit to 1 A; raise slowly to 3–5 A if nothing heats up (core rails may need 5–10 A).
4. Connect red to the rail (a cap or inductor on it), black to ground.
5. Watch the thermal camera / freeze spray. Remove the hot part and re-measure the rail.

6.2 Other methods

• Isolation: lift inductors, 0 Ω resistors, or MOSFETs that split the rail into sections. The section that stays shorted contains the fault.
• Milliohm / voltage-drop method: with current injected, measure millivolts along the rail with the meter; the voltage drops toward the short.
• Remove caps one by one (last resort): replace with new ones after.

7. HDD repair (3.5″ and 2.5″, all brands)

7.1 HDD PCB block diagram

7.2 Identify the fault by symptom

7.3 Multimeter checks on an HDD PCB (power OFF)

1. 5 V input to ground: measure on the SATA power connector 5 V pins (or on the TVS). Normal: tens/hundreds of Ω rising, diode ~0.4–0.6. 0–2 Ω = short.
2. 12 V input to ground (3.5″ drives). Same check.
3. Find the TVS diodes near the power connector (SMA/SMB package, 2 pins, marked with a code). Desolder the TVS on the shorted rail. If the short disappears → the TVS was the fault (it sacrificed itself). Replace it with the same type (5 V: SMAJ5.0A-class, 12 V: SMAJ12A-class). Running without TVS works but leaves the drive unprotected.
4. Short still present after TVS removed? Inject 1–3 V (≤ 5 V) with current limit and find the hot part — usually the motor driver IC.
5. Fuse / 0 Ω resistors in the 5 V and 12 V paths: must beep.
6. Spindle motor windings (PCB removed, measure on the motor contact pads on the HDA): 3 or 4 pads. Between each pair of the 3 phases you should read equal low values (typically about 1–3 Ω on 3.5″, a few Ω on 2.5″). Open or very different = bad motor (clean-room job). Equal values = motor fine.
7. Motor driver outputs on the PCB (to the motor pads) in diode mode: all three phases should read the same. One different = damaged driver.

7.4 Powered checks

• Power from a bench supply (5 V and 12 V, current limit ~1.5 A each). 3.5″ drive spin-up draws ~1.5–2 A on 12 V for a few seconds, then ~0.3–0.6 A.
• Measure the regulator outputs: typically ~3.3 V, ~2.5 V/1.8 V and ~1.0–1.2 V core around the MCU (generated by the motor driver or separate buck/LDO). Missing core rail = regulator or its load is bad.
• Check crystal oscillation with a scope if the board has power but does nothing.

7.5 PCB swap and ROM transfer (very important)

1. Match the donor PCB exactly: same board number (printed on the PCB, e.g. WD 2060-771xxx-00x; Seagate 100xxxxxx REV x), same MCU, and for Seagate the same firmware family.
2. If your original board has an 8-pin ROM chip (25-series SPI flash, common on many WD and Seagate boards): desolder it and solder it onto the donor in the same location, or read it with a programmer and write it to the donor’s ROM.
3. If there is no 8-pin chip (ROM is inside the MCU): either move the MCU itself (BGA/QFP rework), or read the ROM via the drive’s terminal/vendor commands with PC-3000/MRT and write it to the donor. On some boards the empty 8-pin footprint can be populated so the MCU loads ROM from it.
4. On Seagate, read ROM via the serial terminal (TX/RX/GND jumper pins) with a USB-TTL adapter at the correct level — only if you know the commands; mistakes can make data unrecoverable.
5. Toshiba/HGST/Samsung: same principle — ROM is unique; match board revision and transfer ROM.

7.6 Brand quick notes

8. SATA SSD repair (2.5″ and M.2 SATA)

8.1 SSD block diagram

8.2 Symptoms

8.3 Step-by-step multimeter checks

1. Input to ground (5 V for SATA): normal = no short, diode ~0.4–0.7. Short → check TVS and input caps, then inject ≤5 V / 1 A.
2. Fuse after the connector: must beep. Open fuse = something downstream shorted — check rails before replacing.
3. Each output rail to ground (measure on the inductors near the PMIC). Typical diode values: 3.3 V NAND rail ~0.3–0.6, 1.8 V ~0.3–0.5, core ~0.1–0.3. 0.000 on one rail = shorted cap or the chip on that rail.
4. Power on (bench 5 V, 1 A limit). Idle draw: SATA SSD ~0.05–0.3 A. Measure each rail: 3.3 V, 1.8 V (or 1.2 V), DRAM 1.35/1.2 V, core ~0.9–1.1 V.
5. One rail missing → its regulator (check EN and input). All rails missing → PMIC input/enable or the PMIC itself.
6. All rails present but not detected → crystal, controller (bad solder / dead), or firmware. Try ROM / safe mode: many controllers have two pads/pins marked ROM, SAFE, JP1 or FW. Shorting them at power-on makes the controller appear in vendor tools (data is generally not touched by entering this mode, but flashing firmware usually wipes the translator).

8.4 TRIM warning

Once the OS deletes files on a working SSD, TRIM erases the blocks. Deleted-file recovery from SSDs is usually impossible — power off the drive as soon as a customer reports accidental deletion.

9. NVMe M.2 SSD repair

Same blocks as the SATA SSD above, but input is 3.3 V from the M.2 slot, the interface is PCIe, and the controllers run hotter (common failure). The most frequent board-level faults: shorted 3.3 V input capacitor, shorted PMIC, and shorted core-rail capacitor.

9.1 M.2 M-key pinout (what to measure)

9.2 NVMe step-by-step

1. 3.3 V input to ground: resistance and diode mode. A good NVMe typically shows no short (diode ~0.3–0.6). 0–1 Ω = short.
2. Inject 3.3 V (current limit 1–2 A, start at 1 V for safety) into the 3.3 V input. Hot part = usually an input MLCC (remove, test again) or the PMIC.
3. PMIC replaced or cap removed, short gone? Power the drive on a bench adapter or in a PC: measure each buck output (core 0.8–1.0 V, DRAM 1.2/1.1 V, VCCQ 1.2/1.8 V, NAND 2.5/3.3 V).
4. Core rail shorted (very low diode, gets hot): first check the caps around the controller; if the controller itself heats up, it is dead — data unrecoverable for encrypted controllers.
5. All rails good, not detected: check crystal, PERST#/CLKREQ# continuity from fingers to controller, and the controller’s BGA (cracked solder from heat cycling). Try ROM/safe-mode jumper with vendor tools if only the drive (not data) matters.
6. Detected but 0 GB / wrong ID → firmware fault, not hardware (same as SATA).

10. Laptop motherboard (all brands)

HP, Dell, Lenovo, Asus, Acer, MSI, Toshiba and others are mostly designed by a few ODMs (Compal, Quanta, Wistron, Inventec, Pegatron, Foxconn, LCFC). The board number (e.g. LA-xxxxP, DAxxxxMB, NM-Axxxx, 448.xxxxx) is printed on the board and is what you search for the schematic and boardview. The logic below works on all of them.

10.1 Power sequence (Intel / AMD laptops)

10.2 Measurement table for each stage

10.3 Current-meter (ammeter) diagnosis — the fastest first test

Power the laptop through a DC bench supply (19 V, 4 A limit) or a USB-C meter, battery disconnected, and read the current.

10.4 Dead laptop — complete flow

Checking for shorts before powering a dead board

Before applying power to a liquid-damaged or dead board, measure resistance / diode to ground on: B+ (19 V), 3VALW, 5VALW, VDDQ, VCCIN/VCORE, GPU core (if dedicated GPU), +3VS, +5VS. Any short here explains the dead board — fix it first.

10.5 Charging problems

10.6 USB-C powered laptops

• USB-C first provides only 5 V. The PD controller (e.g. TI TPS65987/65988/65994, Cypress CCG, Realtek RTS54xx, Apple CD3215/CD3217) negotiates 9–20 V over the CC lines.
• Laptop stuck at 5 V (0.0x A) on a USB-C meter → no negotiation: PD controller has no power, its firmware SPI chip is bad, CC line damaged, or shorted VBUS behind the PD controller’s MOSFETs.
• Measure: CC1/CC2 (one shows ~1.6–2.6 V with a charger attached), PD controller 3.3 V / LDO outputs, VBUS before and after the power path MOSFETs.
• Liquid in the USB-C port corrodes the CC and VBUS pins — the most common USB-C charging fault. Check the port pins in diode mode vs the other port.

10.7 Powers on (fan spins), no display

1. Connect an external monitor (HDMI). Picture on external = internal screen/cable/backlight problem (go to 10.9). No picture anywhere = board doesn’t boot.
2. Read the current (10.3). Stuck low current = not running code.
3. Remove RAM, try one stick in each slot, clean contacts. Listen for beep/LED codes (Dell/HP/Lenovo have diagnostic codes).
4. Reset CMOS: unplug RTC battery 5+ minutes (or short CMOS jumper).
5. Check all S0 rails, VCCIN/VCORE, VDDQ and the power-good chain → PLTRST#.
6. If everything is present: reflash BIOS (10.8). Corrupt BIOS is very common after failed updates or power loss.
7. Still nothing: onboard RAM, CPU/PCH BGA, GPU (if dedicated GPU drives the internal display).

10.8 BIOS / SPI flash chip

10.9 Screen and backlight

10.10 USB, keyboard, touchpad, audio, Wi-Fi

• USB no power: 5 V on the port VBUS → USB power switch (load switch, e.g. SY6288/AP2xxx/RT9715/G547) EN pin and input 5 V. Port data: diode-mode D+ / D− vs a good port.
• Keyboard / touchpad dead: connector pins (diode mode vs known good), 3.3 V supply on connector, liquid corrosion on the EC (keyboard matrix goes straight to EC).
• No audio: codec power (3.3 V, 5 V, 1.8 V), codec crystal/clock from PCH (HDA/I2S), speaker amp power and SD/EN pin, headphone jack detect switch.
• No Wi-Fi: 3.3 V at the M.2 E-key slot, W_DISABLE# line, try another card.
• Liquid damage: remove all shields, ultrasonic clean with IPA, dry, then do the short checks for every rail. Look under the EC, PCH, and around connectors for green/white corrosion.

10.11 MacBook notes (Intel and Apple Silicon)

11. Desktop motherboard

Same methods, different input: the ATX power supply gives 5VSB (always on), 12 V, 5 V, 3.3 V (after PS_ON#).

Super I/O (ITE IT86xx, Nuvoton NCT67xx) plays the role of the EC. Bulging capacitors around the VRM on old boards = replace with same-or-better polymer caps.

12. Processor, GPU and BGA chips

12.1 Is it the CPU or the board?

1. Measure VCCIN/VCORE to ground. Note the Ω and diode value; compare to a good board.
2. Socketed CPU (desktop / old laptops): remove the CPU and measure again. Short gone → CPU is shorted. Short still there → board (MOSFET/cap).
3. BGA CPU (soldered): inject 1 V with current limit. If the CPU package heats evenly and no MOSFET/cap heats first → CPU short. Check the small caps on the CPU package itself (top or bottom): a shorted package cap is common on many Intel and AMD mobile chips — remove it and re-measure. If the short disappears, the CPU often works without that single cap.
4. Run a known-good CPU in the socket if you have one — the definitive test.

12.2 Typical processor-related faults

12.3 BGA reballing process

1. Preheat board from below to ~150–180 °C. Protect nearby plastic parts with Kapton/aluminium tape.
2. Remove the chip with hot air or a BGA station (lead-free solder melts ~217–220 °C; peak ~235–245 °C).
3. Clean the chip and board pads with flux and wick — flat, shiny, no bridges, no ripped pads.
4. Apply flux on the chip, place the correct stencil and ball size (0.3/0.4/0.45/0.5/0.6 mm), fill balls, heat until they form round balls.
5. Clean, flux the board, align chip by silkscreen corners, reflow (leaded balls ~200–215 °C peak). The chip self-aligns when solder melts.
6. Let cool slowly; never move the board while hot. Check rails for shorts before powering.

13. Soldering & rework settings

• Always mix leaded solder into lead-free joints — they melt lower and flow better.
• Too much heat lifts pads. If a pad lifts, scrape the trace and bridge with a thin wire (jumper) secured with UV mask.
• Clean flux residue with IPA — some fluxes are conductive or corrosive when left.
• Replace parts with the same value and voltage rating (caps), same R_DS(on)/voltage/package (MOSFETs). Cap voltage rating must be equal or higher.

14. How to work on any specific model

1. Find the board number: printed on the PCB (laptop: LA-, DA0…MB, NM-, 6050A…, 820-xxxxx for Apple; HDD: PCB number; SSD: controller part).
2. Get the schematic (PDF): shows every rail, IC, signal name and the power sequence page. Search the board number + “schematic”.
3. Get the boardview (.brd, .bdv, .fz, .cad): shows physical location of every part. Open with OpenBoardView (free) or FlexBV. Some boardviews include diode values and voltages.
4. Open the schematic power sequence / power block diagram page — it maps exactly onto the table in 10.2.
5. Look up datasheets for every IC you suspect (charger, PWM, VRM, PD controller) to learn pin functions.
6. Keep your own log: model, symptom, faulty part. The same models fail the same way again and again.

15. Quick reference sheet

Values in this guide are typical ranges. Always confirm exact voltages, pinouts and part numbers from the schematic and datasheet for the model on your bench.