Troubleshooting Flowcharts
Decision flowcharts for a dead circuit, a microcontroller that will not run, a motor or load that does not work, and noisy or unreliable behavior.
A method that works
Fault finding is a search problem: every measurement should halve the number of places the fault can be. These flowcharts encode the usual order of checks, from the cheapest and most likely to the most specific. They assume the basic tools in Test Tools and Fault Finding: a multimeter, and an oscilloscope for the later steps.
- Look first. Burnt parts, cracked solder, reversed polarity, a wrong part, a bent pin. Smell for hot or burnt components. Many faults are visible.
- Check power before anything else. A circuit with the wrong supply tells you nothing else. Measure at the chip pins, not at the connector.
- Divide the circuit in half. Find a point in the middle where you can tell whether the signal is right, then work toward the bad half.
- Compare with a good board or the schematic’s expected voltages; a difference points to the fault.
- Change one thing at a time and write down what you did. Undo changes that did not help.
- Believe the measurement, not the assumption. The part you are sure is fine is often the fault.
Flowchart 1: the board is dead
Use this first for any “nothing happens” fault. Most dead boards fail at the first or second question: no power at the board, or a supply that collapses because of a short.
- Supply short: measure resistance between the supply and ground with the power off (a few ohms or less is a short). Find it by feeling for the warm part with a current-limited supply set to a low current, or by cutting the supply into sections.
- Regulator output low: remove the load (lift the regulator output) and see whether the voltage recovers. If it does, the load is shorted; if not, the regulator is the fault or its input is too low (see dropout).
- Thermal clues: a part too hot to touch is the problem or is fed by it. Power on for a second only, or use a thermal camera or isopropyl alcohol that evaporates faster over the hot part.
Flowchart 2: the microcontroller will not run
- All the supply pins (including analog and reference pins) must be present, with decoupling capacitors at each. A missing one can cause strange behavior.
- The reset pin must idle at the correct level (a pull-up for most) and the reset line must not be held by a programmer or a floating signal.
- Clock: check the crystal with a high-impedance scope probe (it may disturb it: a 10× probe at the output pin shows the oscillation). A wrong fuse or clock setting (the internal oscillator chosen when no crystal is fitted, or the reverse) is a classic cause.
- Programming: no response from the programmer usually means a wiring fault, a supply problem, a locked device or the wrong target selected. Try a slower programming speed.
- Boot pins (BOOT0 on STM32, GPIO0 on ESP32) set the start mode; a floating one starts the chip in the wrong mode.
- Code: blink an LED in a minimal program. If that works the hardware is fine and the fault is in the software.
Flowchart 3: a load does not switch
- Control signal: measure the output pin; if it is wrong the problem is in the software or the pin setup. If it is right, go to the driver.
- Drive stage: a BJT needs base current (about one tenth of the collector current in saturation); a MOSFET needs enough gate voltage for its RDS(on) and a pull-down so it stays off. A wrong pinout (a transistor fitted backwards) is common.
- Supply to the load: measure across the load while it should be on. Full supply across the load and no action means a faulty load. A low voltage means a high drop somewhere upstream.
- Common ground: the load supply and the control circuit must share ground, otherwise the transistor sees no signal.
- Flyback diode: a missing or backwards diode across an inductive load can destroy the driver.
Flowchart 4: a noisy or unstable reading
- Look at the signal on the scope with a short ground spring. Noise at a fixed frequency (the switching converter, mains hum at 50 or 60 Hz) points to its source.
- Supply and reference: ripple on the ADC reference appears directly in the reading. Decouple it and keep digital switching away from it.
- Source impedance: an ADC input needs a source impedance low enough to charge its sampling capacitor; add a buffer or a capacitor near the pin.
- Averaging takes out random noise: averaging N readings reduces it by √N, but not offsets or interference locked to the sampling.
- Grounds and cables: see EMC and Noise at Board Level. Try moving the cable, a battery supply or a ferrite to see whether the noise changes.
Worked cases
Case 1: a 12 V board is dead and the supply shows its current limit
- The 12 V bench supply is set to a 500 mA limit and the display shows 500 mA at nearly 0 V: a short. The equivalent resistance is at most 12 V ÷ 0.5 A = 24.0 Ω for the unlimited case; with the supply folded back it is lower.
- Resistance between supply and ground with the power off reads 0.4 Ω: confirmed. Lowering the limit to 100 mA and touching components with a finger finds one capacitor too warm: a tantalum capacitor fitted backwards. Removing it clears the short.
- Lesson: always set the supply’s current limit for the first power-up of a new board, so a fault does not destroy the part that failed.
Case 2: a 5 V regulator is hot and the output reads 3.9 V
- A linear 12 V to 5 V regulator feeding 400 mA dissipates (12 − 5) × 0.4 = 2.8 W. With no heat sink that is a rise of around 100 °C in a small package, and it goes into thermal shutdown: the output sags.
- The fix is not a replacement part but a lower input voltage, a heat sink, or a switching regulator (see Power Electronics: Buck, Boost and Thermal Design).
- A reading of 3.9 V with a hot part also fits an overloaded output: remove the load first to tell the two apart.
Case 3: a motor does not turn from the driver board
- The gate of the low-side MOSFET measures 0 V when the microcontroller should be driving it. The microcontroller pin measures 3.3 V, so the signal is lost between: a 10 kΩ pull-down is fitted where a series resistor should be, forming a divider that holds the gate low.
- A motor of 6.0 Ω winding resistance at 12 V would draw 2 A: with the MOSFET not on, the drop across the load is zero and the whole 12 V appears across the transistor, matching the measurement.
- Moving the resistor to the right position restores the drive. The measurement at the transistor pin found the fault in one step.
Case 4: an LED is dimmer than expected
- A 5 V supply with a 150 Ω resistor and a red LED (about 2.0 V) should give (5 − 2.0) ÷ 150 = 20.0 mA. The LED is faint and the resistor voltage measures only 1.2 V (8 mA).
- The supply is actually 3.9 V under load (a sagging regulator): the LED current is (3.9 − 2.0) ÷ 150 = 12.7 mA. The LED was fine; the fault was the supply, found by measuring it at the LED.
- Meter drops: with 0.35 A in a lead of 0.8 Ω the lead drops 0.28 V, which can be enough to matter on a 5 V rail: measure at the load.
Case 5: an ADC reading jumps by several counts
- A 12-bit ADC on a 5 V reference has 1 LSB = 5 V ÷ 4096 = 1.22 mV. Ripple of 300.0 mV on the reference ( 6.0 percent) is about 246 counts of change at full scale; the scope shows a 500 kHz ripple from the nearby converter.
- A 10 µF and 100 nF decoupling pair at the reference pin, and moving the converter’s inductor away from the analog input, reduces it to under 2 counts.
Quick symptom table
| Symptom | First checks | Often |
|---|---|---|
| No power LED | Supply at the connector, fuse, polarity protection | Blown fuse, reversed supply, a broken connector |
| Supply goes into current limit | Resistance supply to ground; warm parts | Reversed capacitor or diode, solder bridge, a failed part |
| Part gets very hot | Its supply voltage and current; orientation | Fitted backwards, overloaded output, dead short |
| Works on the bench, fails in the box | Supply drop on the cable, temperature, vibration | Voltage drop, thermal limit, a cracked joint |
| Works for a few minutes, then stops | Temperature of regulators and drivers | Thermal shutdown, an electrolytic capacitor drying out |
| Intermittent | Tap and flex the board; reflow suspect joints | Cold solder joints, connectors, cracked traces |
| Resets randomly | Supply dips on the scope; the reset pin | Brown-out, noise, inrush from a load |
| Wrong readings on one channel | Input wiring, reference, a leaking capacitor | Contamination (flux), a damaged input, a floating pin |
Safety while probing
- Never probe mains-powered circuits without an isolation transformer and proper training. A scope ground clip connected to a live point shorts it through the mains earth.
- Discharge capacitors in power supplies before touching them; large ones keep dangerous charge for minutes.
- Use a current-limited supply and the lowest current that lets the circuit start.
- One hand in a pocket when probing live circuits, and keep metal jewelry off.
- Use short ground connections and the right probe attenuation, and never exceed the probe or meter rating.
- Batteries can deliver very large currents: lithium cells can burn if shorted, so fuse the pack.
Related
- Design, Testing and Troubleshooting: the overview of this guide.
- Test Tools and Fault Finding: multimeter and oscilloscope use, and the fault flow chart for first checks.
- EMC and Noise at Board Level: noise and grounding problems.
- Reading Datasheets: checking the values you expect to see.
- Power Electronics: Buck, Boost and Thermal Design: converters, regulators and thermal faults.
- Ohm’s Law and Power Calculator for quick checks.
FAQ
Where should I start with a board that does nothing?
Measure the supply at the chip’s power pins. A large fraction of “dead” boards have a supply problem: no input, a short, a tripped regulator or a polarity error.
How do I find a short circuit?
Power off, measure resistance between supply and ground. Then feed a few hundred milliamps from a current-limited bench supply and look for the warm part, or use a milliohm meter and follow the lowest reading to the fault.
Why does the fault disappear when I probe it?
The probe’s capacitance or the pressure on the board changes the circuit: look for a marginal oscillation or a bad joint. Reflow suspect joints and try again.
Should I replace parts on suspicion?
Only after measurement. Replacing parts without evidence adds new faults (heat damage, wrong parts) and teaches you nothing about the cause.
References
- Horowitz, P. and Hill, W., The Art of Electronics, 3rd ed., Cambridge University Press: debugging and test.
- Pease, R. A., Troubleshooting Analog Circuits, Butterworth-Heinemann.
- Manufacturer application notes on probing and measurement (Keysight, Tektronix, Texas Instruments).
Last reviewed: October 2026. Values are typical: check the datasheets of the parts you use. These flowcharts are general guides; follow the manufacturer’s service instructions for specific equipment, and the safety rules for your supply voltage.