A dead circuit has no LEDs, no serial output, and no sign of life. The cause is almost always one of a few things: no power, a bad connection, a blown protection part, or a stuck reset line. Guessing wastes hours. A fixed order of checks, from the power source to the signal path, isolates the fault in minutes.
Quick Takeaways
- Work in order: safety, visual inspection, power input, voltage rails, clock/reset, signal path, component test.
- A multimeter is enough for about 90% of dead-board faults. An oscilloscope covers the rest.
- Measure first, replace second. Never swap parts until a measurement points to them.
- Change one thing at a time and write down every reading.
The Core Method at a Glance
| Stage | What You Check | Tool | Typical Finding |
|---|---|---|---|
| 1. Safety | Isolation, polarity, current limit | Bench supply | Reversed supply, short to ground |
| 2. Visual | Burns, cracks, cold joints | Eyes, magnifier | Scorched resistor, lifted pad |
| 3. Power input | Voltage at the connector | DMM | 0V: dead adapter, open fuse |
| 4. Rails | 3.3V, 5V, core voltages | DMM | Failed regulator, shorted rail |
| 5. Clock and reset | Crystal, RESET/EN pin | Scope | No oscillation, reset held LOW |
| 6. Signal path | I2C, SPI, UART, GPIO | Scope, logic analyzer | Missing pull-ups, wrong logic level |
| 7. Component test | Diodes, caps, MOSFETs | DMM | Shorted capacitor, open diode |
Step 0: Safety Before Measurement
Disconnect mains-powered circuits from the wall before you probe anything. Use an isolation transformer when you must measure live mains-side parts. For low-voltage work, use a current-limited bench power supply.
Set the limit before you connect the board:
- 5V logic boards: start at 100mA, raise it only if needed.
- 3.3V microcontroller boards: start at 50mA.
- Motor or LED driver boards: start at 250mA and watch the current meter.
If the supply hits its limit and the voltage collapses, you have a short. That is your first clue.
Discharge large electrolytic capacitors before you touch them. A 470µF cap charged to 24V can pop a meter fuse or burn a finger.
Step 1: Visual Inspection
Spend two minutes with a magnifier and good light. Look for:
- Burn marks or brown discoloration around resistors, regulators, and MOSFETs.
- Bulging or leaking electrolytic capacitors. A domed top means failure.
- Cracked ceramic capacitors. These often short to ground.
- Cold solder joints: dull, grainy, or ring-shaped cracks around pins.
- Lifted pads and snapped traces near connectors and mounting holes.
- Solder bridges between fine-pitch IC pins.
- Reversed polarity parts: diodes, electrolytics, and ICs installed backwards.
Smell the board too. A sharp, acrid scent marks the spot where a part overheated.
Step 2: Verify the Power Source
Most dead circuits are dead because the power never arrives. Test the supply before you blame the board.
- Set the DMM to DC volts.
- Measure the adapter or battery with no load. A “5V” wall adapter should read 4.9V to 5.3V.
- Measure again at the board’s input connector with the board attached.
- Compare the two readings.
| Reading at Connector | Likely Cause |
|---|---|
| 0V | Dead adapter, broken cable, open fuse |
| Correct unloaded, drops under load | Short circuit or weak supply |
| Negative voltage | Reversed polarity |
| Correct voltage | Power input is fine, move to the rails |
Also check the input protection chain: fuse, reverse-polarity diode, TVS diode, and ferrite bead. Measure voltage on both sides of each. A part that shows full voltage on one side and 0V on the other is open.
Check the Fuse the Right Way
Never judge a fuse by sight alone. Use the continuity mode. A good fuse reads under 1Ω. An open fuse reads OL. If a fuse blew, find out why before you replace it. A new fuse in front of a shorted rail will blow again.
Step 3: Trace the Voltage Rails
Follow the power from input to every regulator output. Put the black probe on GND and move the red probe along the path.
Common Rail Test Points
| Rail | Expected Value | Acceptable Range | Where to Probe |
|---|---|---|---|
| 5V USB/logic | 5.0V | 4.75V to 5.25V | Regulator output pin, VCC pins |
| 3.3V logic | 3.3V | 3.15V to 3.45V | LDO output, MCU VDD pins |
| 1.8V core | 1.8V | 1.71V to 1.89V | Core regulator output |
| VBAT | Per cell | 3.0V to 4.2V (Li-ion) | Battery connector |
If the input is good and the output is 0V, the regulator is suspect. Check three things before you replace it:
- Enable pin: many switching regulators need EN held HIGH.
- Output capacitor: a shorted cap pulls the rail to ground and makes a healthy regulator look dead.
- Regulator temperature: a part that gets hot within seconds points to a short on its output.
Finding a Short on a Rail
Switch the DMM to its lowest resistance range and measure from the rail to GND with power off. Typical results:
- Above 100Ω: normal for most logic rails.
- Under 10Ω: likely a fault.
- Near 0Ω: a hard short.
To locate the shorted part, inject a low-voltage, high-current source (for example 0.5V at 1A) into the rail. Then touch the board with a fingertip or use a thermal camera. The shorted part warms first. Keep the injected voltage below the part’s operating rating.
Step 4: Check Clock and Reset
A powered board can still look dead if the processor never starts. Two culprits dominate: the clock and the reset line.
Reset and Enable Pins
Measure the RESET, NRST, or EN pin with the board powered.
- On an Arduino Uno (ATmega328P), the RESET pin should sit near 5V (HIGH via the 10kΩ pull-up).
- On an ESP32, the EN pin should sit near 3.3V.
- A pin stuck at 0V keeps the chip in permanent reset. Look for a shorted reset button, a bad capacitor, or a failed supervisor IC.
Crystal Oscillator
Put an oscilloscope probe on the crystal pin. A working 16MHz crystal shows a sine-like wave at roughly 16MHz. No signal means one of these:
- Dead or cracked crystal.
- Wrong or missing load capacitors (typically 12pF to 22pF).
- Wrong fuse bits that select an external clock when none is present.
A DMM cannot see a crystal. Use a scope or a frequency counter.
Step 5: Test the Signal Path
Once the chip is powered and running, confirm it can communicate. This step needs code to prove life. Upload a minimal sketch that blinks the LED and prints to serial. It separates hardware faults from firmware faults.
Firmware Heartbeat Test (Arduino C++)
// Minimal "is the board alive?" test for Arduino-compatible boards.
// Blinks the onboard LED and prints an ADC reading of the supply rail.
const int LED_PIN = LED_BUILTIN; // Onboard LED, usually GPIO13 on Uno
const int ADC_PIN = A0; // Wire A0 to a known 3.3V or 5V rail via divider
void setup() {
pinMode(LED_PIN, OUTPUT); // Configure LED pin as output
Serial.begin(115200); // UART at 115200 baud for fast logging
Serial.println(F("Board alive. Starting heartbeat."));
}
void loop() {
digitalWrite(LED_PIN, HIGH); // LED on
delay(250);
digitalWrite(LED_PIN, LOW); // LED off
delay(250);
int raw = analogRead(ADC_PIN); // 10-bit ADC: 0 to 1023
float volts = raw * (5.0 / 1023.0); // Convert to volts (5V reference)
Serial.print(F("A0 raw: "));
Serial.print(raw);
Serial.print(F(" Volts: "));
Serial.println(volts, 2); // Print with 2 decimal places
}
Read the result:
- LED blinks, serial prints: the MCU, clock, and USB-serial path work. The fault is in your external circuit.
- LED blinks, no serial output: check the TX/RX wiring, baud rate, and USB-serial chip.
- Nothing happens: return to power, clock, and reset. Check the bootloader if the chip was recently replaced.
Common Signal-Level Faults
| Symptom | Protocol | Likely Cause | Fix |
|---|---|---|---|
| Sensor not found | I2C | Missing pull-ups, swapped SDA/SCL, wrong address | Add 4.7kΩ pull-ups, run an I2C scanner |
| Garbage characters | UART | Baud mismatch, 3.3V/5V mismatch | Match baud, add a level shifter |
| No data returned | SPI | Wrong mode, floating CS line | Set the right CPOL/CPHA, tie CS correctly |
| Pin stuck HIGH or LOW | GPIO | Shorted pin, wrong pinMode | Check continuity to GND/VCC, review code |
Logic level mismatch kills more boards than most people expect. A 5V signal into a 3.3V pin can stress or destroy the input. Use a level shifter or a resistor divider.
Step 6: Test Individual Components
When measurements point to a specific part, test it out of circuit if you can. In-circuit readings can mislead because parallel paths change the result.
| Component | DMM Mode | Good Reading | Bad Reading |
|---|---|---|---|
| Resistor | Ohms | Within 5% of marking | OL (open), drifted value |
| Diode | Diode test | 0.5V to 0.7V forward (silicon), OL reverse | Short both ways, OL both ways |
| Electrolytic capacitor | Capacitance | Within 20% of rating | Low value, high ESR, shorted |
| Ceramic capacitor | Resistance | Charges then reads OL | Near 0Ω (shorted) |
| BJT | Diode test | Two junctions, one forward each | Shorted junction |
| MOSFET | Diode test | Body diode only | D-S short, gate leakage |
| Inductor | Continuity/Ohms | Low resistance, a few ohms | OL (open winding) |
Test a MOSFET in this order: gate to source, then drain to source. A drain-to-source reading near 0Ω in both directions means the part is shorted. This is a very common failure on motor drivers and power switches.
Real-World Workflow: Reviving a Dead ESP32 Sensor Node
Here is the method applied to a board that shows no activity after a battery swap.
- Safety: connect a bench supply at 3.7V, limit 100mA.
- Visual: find a darkened diode near the battery input, a reverse-polarity protection part.
- Input: measure 3.7V at the battery connector but 0V after the diode. The diode is open.
- Fix: replace the diode with a Schottky type rated for at least 1A, which also loses less voltage than a standard silicon diode.
- Rails: measure 3.3V at the LDO output. The rail is now alive.
- Reset: the EN pin sits at 3.3V. Good.
- Firmware: flash the heartbeat sketch. The LED blinks and the serial port prints.
- Signal: run an I2C scanner. The temperature sensor answers at address 0x76.
The whole diagnosis used a multimeter and one sketch. The root cause was one failed part in the input protection chain.
Protecting the Rebuilt Board
Prevent a repeat failure:
- Add a polyfuse (resettable fuse) on the main supply.
- Add a TVS diode across the input for voltage spikes.
- Place 0.1µF decoupling capacitors close to every IC VCC pin.
- Use an optocoupler or level shifter between mismatched voltage domains.
Tools Worth Having on the Bench
| Tool | Budget Option | Best Use | Priority |
|---|---|---|---|
| Digital multimeter | Basic auto-ranging DMM | Voltage, continuity, diode test | Must have |
| Bench power supply | Adjustable 0–30V, 0–3A | Current-limited power-up | Must have |
| Oscilloscope | 2-channel, 50MHz+ | Clock, reset, signal edges | High |
| Logic analyzer | 8-channel USB | I2C, SPI, UART decoding | High |
| Hot air station | Entry-level rework | SMD part replacement | Medium |
| Thermal camera | Clip-on phone model | Locating shorted parts | Medium |
Common Mistakes to Avoid
- Probing with a wide tip on fine-pitch pins. You can short two pins together. Use a sharp probe or a test hook.
- Measuring resistance on a powered board. You get wrong readings and can damage the meter.
- Replacing multiple parts at once. You will never learn which one was the fault.
- Trusting a firmware result before checking hardware. Confirm power, clock, and reset first.
- Skipping the ground connection on scope measurements. A missing ground clip gives noisy, meaningless traces.
Troubleshooting Checklist
- [ ] Power disconnected, bench supply current-limited
- [ ] Visual inspection done under magnification
- [ ] Input voltage confirmed at the connector
- [ ] Fuse, diode, and protection parts verified
- [ ] Every voltage rail measured against its target
- [ ] Rail-to-GND resistance checked for shorts
- [ ] Clock and RESET confirmed
- [ ] Heartbeat sketch uploaded and running
- [ ] Communication buses scanned and confirmed
- [ ] Failed parts replaced one at a time, readings logged
FAQ
Why does my circuit have power but still not work?
Power at the connector does not prove power at the chip. Measure the voltage on the IC’s VCC/VDD pins directly. Then check the reset line and the clock. A powered processor held in reset or missing its crystal looks identical to a dead board.
How do I find a short circuit on a PCB?
Power the board off and measure resistance between the suspect rail and GND. A reading under 10Ω signals a short. Then inject a low voltage at high current into the rail and look for the part that heats up first. A thermal camera or a drop of isopropyl alcohol, which evaporates first over the hot part, speeds this up.
Can I test components while they are still on the board?
You can test some, but results may be inaccurate. Parallel paths through other parts skew resistor, capacitor, and inductor readings. Diode and continuity checks work well in-circuit. For a reliable result, lift one lead or remove the part.
What is the first thing to check on a dead Arduino or ESP32 board?
Check the USB cable and the supply voltage. Many cables carry power only or are damaged. Then measure 5V or 3.3V at the board and confirm the onboard LED lights. If it does not, inspect the regulator and the RESET/EN pin.




