Systematic Fault Finding: the short answer
Random poking wastes hours. A method finds faults faster and teaches you something each time.
- Level
- Advanced
- Time
- About 6 min
- Topic
- Tools and measurement
Find faults in complex boards quickly with a repeatable method: define the symptom, check power first, split the problem and prove the cause.
Before this: Systematic Troubleshooting, Using an Oscilloscope
Random poking wastes hours. A method finds faults faster and teaches you something each time. Start by writing down the exact symptom and the conditions: what should happen, what actually happens, and when it started. Was it ever working? What changed: new code, a new part, a different supply, a dropped board? A fault that appears only when warm, only on battery or only with Wi-Fi on points straight at a cause.
Always check power before anything else. Measure every supply rail at the chip pins, not at the connector, with the circuit running. Check the current drawn against what you expect; a board drawing far too much points to a short or a reversed part, and touching parts carefully (or using a thermal camera) finds the hot one. Look at the rail on an oscilloscope too, because a multimeter averages away dips and ripple. Then check clocks and resets: a microcontroller with a stopped crystal or a reset pin held low will do nothing at all.
Then split the problem in half, again and again. In a signal chain from sensor to amplifier to ADC to code to display, measure the middle: if the signal is correct at the ADC input, the fault is after it. Substitute known-good parts and known-good code (a minimal test sketch) to separate hardware from software. Change one thing at a time and write down each result. When you find a cause, prove it: put the fault back and see the symptom return, then fix it properly and test under the worst conditions. Finally, record the fault and fix in your project log, because faults repeat.
Work from what you know
Begin by describing the fault exactly: what happens, when, and what it should do. Ask what changed since it last worked. Check the simple things first: power, connections, the right file uploaded. A large share of faults are a loose wire or a supply that sags.
Divide the system and test the halves. If the input is right and the output is wrong, the fault is between them, so test in the middle, and keep halving.
Measure instead of guessing
Check the supply voltage at the faulty part while it is working, not only at the source. Compare with a known good board if you have one. Change one thing at a time and write down what you tried, since otherwise you may repeat a test or forget what already ruled out a cause.
Use the right tool: a multimeter for voltages and continuity, an oscilloscope for fast signals and supply noise, a logic analyser for bus data.
Intermittent faults
A fault that comes and goes is often a loose connection, heat, a supply dip or a timing problem. Wiggle connectors, warm or cool the part, and watch the supply on an oscilloscope while it fails. Log events with timestamps, and look for what changes at the moment of failure.
- Changing several things at once and not knowing what fixed it.
- Checking the supply at the source and not at the faulty part.
- Not writing down what has already been tested.
Practise this lesson with Pro
The explanation above is free for everyone. The practice for Advanced lessons in every domain are part of the Pro plan at ₹999 a month.
With Pro you also get: a worked example, a hands-on task, a 3-question check.
