Real open-hardware PCB repair case studies
Reading theory is only half the skill — real boards have 50 to 500+ parts, multiple power domains, connectors that get yanked, and spills that creep under fine-pitch pins.
Below are spoiler-free diagnostic game plans for all seven open-hardware boards shipped in DEADBOARD. Each walkthrough tells you the customer complaint, the power architecture, and the first measurements a technician makes on the bench — without spoiling the exact faulty component or flag.
Want rail-by-rail reference voltages inside the bench? Open Boards → Mods → Community and install Board notes, which adds expected rail voltages and probe notes for every shipped board.
1. Olimex ESP32-DevKit-LiPo · dead on USB, won’t charge #
- Hardware: 50-part ESP32 development board with USB power and an on-board Li-ion battery charger (no battery plugged in on the bench).
- Customer complaint: Stopped doing anything on USB after the cable was yanked; before that, the battery "never seemed to charge".
- Diagnostic plan (no spoilers):
- Start at the mechanical stress point: A yanked USB cable points straight at connector pads, traces leaving
VBUS, and nearby protection components. Inspect both sides (F) and test Press / flex (B) around the USB input. - Follow
VBUSinto the 3.3 V regulator: With power on (P), probeV⎓from the USB input toward the 3.3 V LDOVINandVOUT. - Inspect the charger circuit separately: A board can boot on USB while its Li-ion charger still refuses to charge. Check the charger IC's
VIN,BAT, and especially its current-programming resistor (PROG) with power off inΩmode.
- Start at the mechanical stress point: A yanked USB cable points straight at connector pads, traces leaving
- Launch lab: Open Olimex ESP32-DevKit-LiPo on the bench →
2. Rana Tadpole V1 · dead on USB #
- Hardware: 56-part RP2040 pocket controller with 20 mouse-switch buttons; USB enters from a breakout board over a flat cable (
U2). - Customer complaint: Plugged into USB, the host sees nothing, and nothing gets warm.
- Diagnostic plan (no spoilers):
- Read bench current first: "Nothing gets warm" usually rules out a dead short and points to an open path, missing rail, or unpowered regulator (
0.0 mAor near-zero draw). - Trace power out of
U2: With power on, probeV⎓from the flat-cable input pins to the3.3 Vregulator input (VIN), enable (EN), and output (VOUT). - Check continuity (
Ω, power off) wherever voltage stops: A single open joint, missing component, or broken trace betweenVBUSand3.3 Vsilences the entire RP2040.
- Read bench current first: "Nothing gets warm" usually rules out a dead short and points to an open path, missing rail, or unpowered regulator (
- Launch lab: Open Rana Tadpole V1 on the bench →
3. Olimex ESP32-C3-DevKit-Lipo · hot board, wrong 3.3 V #
- Hardware: 49-part ESP32-C3 board with USB and Li-ion charger (no battery connected).
- Customer complaint: Fresh off a small production run — gets warm on USB and the chip never starts. The assembler suspects two component reels were swapped.
- Diagnostic plan (no spoilers):
- Measure the
3.3 Vrail immediately (V⎓): Don't assume "warm" always means a dead0 Ωshort — a rail running at the wrong voltage (too high or too low) also heats regulators and prevents the MCU from booting. - Use Thermal view (
Thermal) to locate heat: Note what is getting warm (the regulator itself, a resistor, or a cap?). - Compare feedback & support values against the schematic/BOM: Remember that adjustable regulators set output voltage via their feedback divider (
Vout = Vref × (1 + R_top / R_bot)). If two resistor or part reels were swapped in assembly, inspect values carefully!
- Measure the
- Launch lab: Open Olimex ESP32-C3-DevKit-Lipo on the bench →
4. ESP32-C3 Rust board · rails fine, chip asleep #
- Hardware: 55-part ESP32-C3 training board (sensors, RGB LED, Li-ion charger, USB-C; no battery on the bench).
- Customer complaint: Came back from a workshop after a drink spill; another participant noted it "only really worked on the battery" before that. On USB it now stays silent, even though rails look fine.
- Diagnostic plan (no spoilers):
- Verify the power path on USB (
VBUS→3.3 V): Confirm why USB power differed from battery power (check input diodes, jumpers, and joints on the USB side). - Think like a microcontroller: When
3.3 Vis present yet the chip stays asleep, check the reset / enable (EN/CHIP_PU) pin and pull-up network! - Treat the liquid spill (
0Isopropyl): Dried drink residue creates leaky parasitic resistance between adjacent pads or pins — pulling anENor strap line low without looking like a0 Ωdead short. Inspect closely and clean residue with IPA (0).
- Verify the power path on USB (
- Launch lab: Open ESP32-C3 Rust board on the bench →
5. Rana Tadpole V2 · dead controller, warm chips, one dead button #
- Hardware: 147-part RP2040 pocket controller with USB-C and 20 mouse-switch buttons.
- Customer complaint: Host sees nothing on USB, the middle of the board gets warm (is the RP2040 itself dead?), and before it died, button
SW2had stopped registering. - Diagnostic plan (no spoilers):
- Don't condemn the main IC prematurely: A warm area near the RP2040 does not prove the RP2040 is dead — a shorted decoupling MLCC or struggling LDO right beside it produces the same symptom.
- Check
ΩtoGNDon the3.3 Vrail (power off), then pinpoint the hot spot: Use Thermal view, Freeze spray (9), IPA (0), or Rosin (R) under current limit /1 Vinjection to see which exact part heats up first. - Track down
SW2separately: Once the controller powers up, selectSW2(1) and trace its two pads (Ωpower off, and pressSW2while probing) to see why that single switch never reaches the MCU/GND.
- Launch lab: Open Rana Tadpole V2 on the bench →
6. Antmicro Jetson Nano baseboard · no 3.3 V, flaky 5 V #
- Hardware: 369-part compute-module carrier board powered from
12 V. - Customer complaint: The compute module never starts:
3.3 Vis missing, and the5 Vrail "sometimes" drops out when the board is handled. - Diagnostic plan (no spoilers):
- Map the power tree (
12 V→5 V→3.3 V): Use the Schematic and click pads to trace how the5 Vand3.3 Vconverters receiveVINandEN. - Reproduce the "flaky when handled" fault (
BPress / flex +OScope Roll): Clip the oscilloscope (O) in Roll mode onto the5 Vrail and probe/flex around the5 Vconverter and passive path (B) to catch the cracked joint or intermittent drop. - Diagnose missing
3.3 Vmethodically: Power off → check3.3 VrailΩtoGNDfor a short. Power on → check3.3 VregulatorVIN,EN, and switch node (SW) on the scope.
- Map the power tree (
- Launch lab: Open Antmicro baseboard on the bench →
7. MNT Reform 2 mainboard · dead 3.3 V rail #
- Hardware: 522-part open-hardware laptop mainboard; bench supply clipped to
30V_GATE(30 V). - Customer complaint: The board powers up on the bench supply, yet the system never starts because the
3.3 Vrail is missing. - Diagnostic plan (no spoilers):
- Check the rail hierarchy: Compare
+5V,+3V3, and+1V2voltages with power on (V⎓), and watch total current on the30 Vbench supply. - Check
+3V3resistance toGND(power off): Is the3.3 Vbuck converter disabled/dead, or is a shorted component on+3V3dragging the rail down and forcing protection? - Use the oscilloscope (
O) on the3.3 Vswitch node + thermal localization: If the rail is shorted, inject1 V(via Bench → Supply) onto+3V3and scan with Thermal view or Rosin (R) to pinpoint the exact part out of 522 without guessing.
- Check the rail hierarchy: Compare
- Launch lab: Open MNT Reform 2 mainboard on the bench →