SSD Electrical Failure & PCB Component Recovery

SSD electrical failure is physical damage to the circuit board components that deliver power to the controller & NAND flash memory. Burnt TVS diodes, blown voltage regulators, and shorted capacitors cut the power path. The NAND chips still hold your data, but the controller can't boot without clean power on the correct voltage rails.

This is different from firmware corruption, where the controller powers on but its internal mapping table is scrambled. Electrical failure means the hardware itself is broken. The drive doesn't show up in BIOS, doesn't spin (SSDs don't spin), doesn't enumerate on the bus at all. Recovery software can't see a drive that has no power.

SSD circuit board repair for electrical damage costs $450–$600 for SATA SSDs and $600–$900 for NVMe drives. If the controller itself is destroyed beyond repair and NAND chips must be transplanted to a donor board, the cost is $1,200–$1,500 (SATA) or $1,200–$2,500 (NVMe), plus donor drive cost. A donor drive is a matching SSD used for its circuit board. Typical donor cost: $40–$100 for common models, $150–$300 for discontinued or rare controllers.

Every case starts with a free evaluation & a firm quote before any paid work begins. No data recovered means no charge. +$100 rush fee to move to the front of the queue.

SSD circuit boards pack the controller, NAND flash, DRAM cache, and power management into a compact PCB. The power delivery components are the first to fail during a surge because they sit between the external power source and the protected silicon.

TVS (Transient Voltage Suppressor) Diodes

TVS diodes clamp overvoltage spikes on the SATA power connector or M.2 3.3V rail. During a surge, they absorb the excess energy and burn out to protect downstream components. A shorted TVS diode pulls the entire power rail to ground, preventing the drive from enumerating. Removing the shorted TVS diode and replacing it restores the rail. The drive was never damaged beyond the TVS diode itself; it did its job.

Voltage Regulators & PMICs

DC-DC converters (buck regulators) and LDOs step down the input voltage to 1.8V, 1.2V, and 0.9V core rails for the controller, DRAM, and NAND. On many SSDs, a single multi-output PMIC handles all rails. When it fails, the controller receives no power or incorrect voltage. Replacing a failed PMIC requires hot air removal of the old chip and precise alignment of the replacement under magnification.

Filter Capacitors

Ceramic capacitors on the output of each voltage rail filter high-frequency noise and store charge for transient current demands. A cracked or shorted capacitor on the 1.2V core rail creates a dead short that prevents the regulator from starting. These are 0402 or 0603 packages, measuring 1mm or 1.6mm long.

Controller BGA Solder Joints

The main controller IC (Phison, Silicon Motion, Samsung, Marvell) connects to the PCB through hundreds of solder balls in a Ball Grid Array package. Thermal cycling from repeated heat-cool cycles or a sudden thermal shock can fracture these joints. The controller loses contact with the PCB traces on specific pins, causing intermittent detection or complete failure. BGA rework with the Zhuo Mao rework station reflows or reballs the solder joints.

ESD-Damaged Controller Pins

Electrostatic discharge during handling (installing an M.2 drive without a grounding strap) can damage the controller silicon's input protection diodes on specific data or power pins. The drive may partially enumerate but fail during data transfer, or fail to initialize entirely depending on which pins were affected.

The goal is to identify the failed component before applying system-bus power, so a shorted rail does not take out adjacent parts during diagnosis. Every step below runs on the same bench, on a drive that is disconnected from any host.

1. Cold resistance sweep. Meter each power rail to ground with the drive unpowered. A reading under 1 ohm on the 3.3V input rail flags a shorted TVS or input cap. A short on an internal rail (1.8V, 1.2V, 0.9V) points to a downstream capacitor or the PMIC itself.
2. Current-limited bench PSU ramp with FLIR overlay. Apply the input rail through a bench supply set to 100mA current limit. A healthy idle draw is tens of milliamps; a shorted component pulls the full limit immediately and heats within seconds. The FLIR thermal camera pinpoints the hot component to a specific 0402 or 0603 package without guesswork.
3. Component lift with the Hakko FM-2032. Remove the suspected failed part with a fine-tip microsoldering iron on the Hakko FM-2032. For small TVS diodes and decoupling capacitors the lift is direct. Repeat the cold resistance sweep: if the short clears with the part removed, the fault is isolated.
4. PMIC or BGA rework with hot air and a BGA station. When the failed part is the PMIC itself or the controller BGA has fractured joints, the Atten 862 hot air station removes the package with a controlled thermal profile. For BGA reball or controller reflow the Zhuo Mao BGA rework station holds the profile across preheat, soak, and reflow stages to avoid damaging the adjacent NAND packages.
5. Replacement from a donor board. TVS diodes and decoupling caps come from a parts library. A PMIC or controller pull comes from a model-matched donor PCB so the package, pinout, and firmware compatibility are identical. A donor drive is a matching SSD used for its circuit board. Typical donor cost: $40–$100 for common models, $150–$300 for discontinued or rare controllers.
6. Power restoration and PC-3000 SSD handoff. Once the rails come up clean on the bench, the drive goes onto the PC-3000 SSD complex. The original controller boots on its own silicon, which preserves the hardware AES-256 Media Encryption Key, and the imaging pass reads the NAND through the normal translator rather than through a raw chip-off path.

The data recovery industry grew up around firmware tools. PC-3000, MRT, and similar platforms communicate with functioning controllers to extract data from corrupted NAND. When the controller is electrically dead, these tools have nothing to talk to.

Fixing the power delivery path requires a different skill set: board-level component identification, surface-mount soldering on 0402-package components, BGA rework for controller reflow, and thermal profiling to avoid damaging adjacent NAND chips during hot air work. Most data recovery labs don't employ technicians with microsoldering training and don't stock Hakko FM-2032 irons, hot air rework stations, or BGA rework equipment.

Rossmann Repair Group started as a board repair operation in 2008. MacBook logic board repair, component-level microsoldering, and BGA rework were the business before data recovery was added.

The soldering infrastructure and trained technicians were already in place. Electrical SSD failure is where those two capabilities intersect: the firmware tools to read the NAND and the soldering skills to make the controller boot.

Modern SSDs use AES-256 hardware encryption where the Media Encryption Key is bound to the original controller silicon. If someone replaces the entire controller, the new chip has a different key and the NAND data is unreadable ciphertext.

Board-level repair preserves the original controller and its encryption key. We replace the support components around the controller (TVS diodes, regulators, capacitors), not the controller itself. When the original controller boots with clean power, it decrypts the NAND using the key that was baked into its silicon at the factory.

This is the only viable path for encrypted SSDs with electrical damage. If the controller silicon is cracked or destroyed, the encryption key is lost and the data is unrecoverable. We'll tell you that during the free evaluation.

For cases where the controller is destroyed and the drive doesn't use hardware encryption, chip-off NAND extraction is the last-resort option: desolder the NAND chips, read them raw, and reconstruct the file system from flash page data.

Power-section diagnostics on a dead SSD work from the input rail inward: confirm the host rail, then probe each downstream rail at its inductor output or dedicated test point. The goal is to find which rail is missing, sagging, or short to ground before any rework, so the lift target is one specific 0402 cap, one LDO, or one PMIC package rather than guesswork.

Voltage rail measurement procedure

The drive sits on an ESD-safe bench, disconnected from any host. A current-limited bench PSU supplies the input rail (3.3V on M.2, 5V on a 2.5" SATA drive). A FLIR thermal sweep runs first so a dead short is localized before any voltage probe touches the board. Once the cold short is either ruled out or lifted, a DMM in DC mode probes the inductor outputs on each buck converter and the dedicated test pads where the PMIC vendor exposes them. Reading the inductor output rather than the bulk cap pad avoids loading the rail through the probe tip.

A rail that sits at the wrong voltage with no thermal hotspot points to a buck regulator out of regulation or an LDO that has failed open. A rail at zero with a thermal hotspot points to a downstream short. A rail at the host input voltage instead of the regulated voltage points to a buck FET failing closed and passing the input directly through; that condition destroys whatever silicon sits on the downstream rail within seconds, so the bench PSU stays current-limited until the rail is confirmed clean.

TVS diode and PMIC failure modes

TVS diodes sit on the host input rail in 0402, 0603, or SOD-323 packages. M.2 2280 and 2230 PCBs use 0402 and 0603 footprints because of the spatial constraint; 2.5" SATA and U.2 enterprise boards often use SOD-323 because the 5V or 12V input rails carry more energy and need a part with higher peak power dissipation. Vishay and ProTek Devices engineering documentation classifies TVS failure into three modes.

TVS: closed circuit (short)

Most common mode. Thermal runaway during a surge fuses the junction into a dead short to ground. The drive will not power on until the shorted TVS is lifted. The Atten 862 hot air station or Hakko FM-2032 removes the part; the cold resistance sweep then confirms the input rail is no longer short.

TVS: open circuit

Catastrophic surge vaporizes the metallization. The drive may still operate because downstream silicon survived, but the input rail no longer has surge protection. The TVS is replaced before the drive goes back into service.

TVS: degraded (leaky)

The most subtle mode. Sub-lethal surges or sustained heat degrade the junction over time, leaving the diode with elevated reverse leakage at the nominal rail voltage. The drive runs at idle but drops out under heavy workload as the leakage current pulls the rail down. On the bench the TVS shows abnormal forward and reverse readings under diode-test mode; under load the FLIR shows the package warming when it should be cold.

PMIC: catastrophic short to ground

An internal switching MOSFET or LDO pass element fails closed, pulling the host input rail directly to ground. The host motherboard sees the short and refuses to start, or the bench PSU pegs at the current limit immediately. FLIR shows the PMIC package heating uniformly across its body rather than at a single pin.

PMIC: loss of regulation (overvoltage)

The PMIC fails to regulate a lower-voltage rail and passes the full input through to the downstream block. A 3.3V or 5V input arriving on what should be a 1.2V core rail will destroy the controller silicon. This mode is why bench bring-up runs current-limited; without the limit, a single PMIC fault takes out the controller during the diagnostic step.

PMIC: LDO failure (missing rail)

A single internal LDO dies open. The host input rail is fine, the drive does not short, but one downstream rail (commonly VCCQ on the NAND bus) is missing. The controller starts to boot, fails the NAND handshake, and does not enumerate on the bus. Diagnosis is a DMM sweep across the regulated outputs, looking for the one rail that reads zero with no short to ground.

PLP capacitor discharge and replacement

Enterprise and prosumer SSDs (for example the Samsung PM9A3 and Micron 7450 class of drive) carry Power Loss Protection capacitor banks: tantalum polymer arrays that hold the rails up for roughly 20 to 50 milliseconds after an unclean shutdown, long enough for the controller to flush the in-flight DRAM FTL state to NAND. The dominant failure mode is dielectric breakdown into a short. A shorted tantalum on the PLP rail trips the PMIC overcurrent protection and the drive refuses to initialize.

Before any rework on a PLP-equipped drive, the capacitor bank has to be safely discharged. Healthy PMICs include an active discharge path that bleeds the tantalum bank when the input rail drops below the undervoltage lockout. When the PMIC itself is the failed part, that path cannot be trusted and the discharge has to happen manually.

1. Measure first. Probe each tantalum across its terminals with a DMM in DC volts. A bank that reads near zero is already discharged; a bank that reads at or near the rail voltage is live and needs to be brought down before any iron touches the board.
2. Bleed through a resistor. Place a 100 ohm, 5W bleed resistor across the capacitor terminals for several seconds. Never short a charged tantalum with metal tweezers; the discharge current can weld the tips, lift adjacent pads, and pit the PCB copper.
3. Confirm zero before rework. Re-probe with the DMM and confirm the bank is at zero volts before the soldering iron or hot air goes near the area.
4. Two-sided heat for the lift. PLP capacitor pads sit on large copper thermal planes that wick heat away faster than a single-sided tool can deliver it. A Zhuo Mao precision bottom-side heater holds the board at a controlled preheat while the Atten 862 hot air station works the top side. Top-only heat warps the PCB and lifts adjacent pads on a PLP array.
5. Replace from a parts library. Tantalum polymer replacements are sourced by capacitance, voltage rating, and package. The new part goes on under the same two-sided thermal profile to avoid the same warp on installation.

Board-level repair versus chip-off NAND

The choice is dictated by whether the controller silicon is still alive. Board repair works when the controller is intact but a support component on the power tree has failed: TVS, decoupling cap, LDO, PMIC, or BGA solder joint. Each of those is a discrete part on the PCB that can be lifted and replaced while the controller die stays in place. Once the rails come up clean, the controller boots on its own silicon, the Hardware Unique Key is still on the die, and the Media Encryption Key unwraps normally for imaging through PC-3000 SSD.

Chip-off NAND is the path when the controller die itself is destroyed: cracked silicon from a heavy impact, ESD-killed core, or a die burned through from a loss-of-regulation event that put the host input voltage directly on the core rail. With the controller gone, there is nothing to revive. On modern hardware-encrypted controllers (Phison E18/E26, SMI SM2264, Samsung Elpis and Pascal, Marvell with AES-256 XTS), raw NAND reads come back as ciphertext and no donor controller can decrypt them, because the HUK was fused into the original die. The hardware-encryption page covers the key-binding mechanics in full; chip-off NAND covers the path for older non-encrypted parts where raw reads do yield usable data.

The decision happens at the free evaluation, not at the quote stage. If the controller is alive, the work goes onto the board repair path at $600–$900 for NVMe or $450–$600 for SATA. If the controller die is gone and the drive is hardware-encrypted, we tell you the data is unrecoverable rather than bill for a chip-off run that cannot succeed.