We define troubleshooting as the process of isolating and correcting a fault in equipment that does not function correctly so that we can restore it to its expected performance level.
There are two quite different approaches to troubleshooting: finding out why something you’ve made doesn’t work, and fixing something that did work and is now faulty.
Troubleshooting in general
A wise old technician I once worked with taught me to use my senses before I take out any tools: look, smell, feel.
Look: About 80% of problems can be found by just having a thorough look at the problem. These may include loose or missing or misplaced components, burned components, foreign connections—solder splatter and a loose bit of wire, short circuits, and broken tracks.
Smell: Burned components have a nasty and characteristic smell and will often be discolored or outright carbonized. Also, the PCB under the component may be discolored.
Feel: Using your finger, feel for any excessively hot components but mind you, don’t burn your finger! If anything is so hot it burns you, it is probably a problem.
Tools: Without a doubt, the best tool is a proper continuity tester, and I don’t mean a multimeter. Multimeters can, of course, measure continuity, but they are usually way too forgiving. A resistance of 1Ω may test OK for the multimeter measuring, say a mains cable, but 1Ω with 15A going through it will cause a voltage drop of 15V and gets very hot!
Here is a very useful circuit of a suitable continuity tester that includes a battery saver cut-off circuit, and will give you very good service.
R9 sets the beep pitch, and R7 allows the detection threshold to exclude resistances above 0.3Ω. LS4 is a piezo high impedance type. C3 should be a tantalum type. The idea is you hit the momentary push button, and C3 is charged, holding Q1 on for about five minutes while it slowly bleeds through R11. You can fiddle with these two to get longer hold times. It works so well I have had the same battery in it for years. The IC’s are not critical, and equivalents are fine.
The probes should first be connected to a resistor of value between 0.22 and 1Ω. The control is adjusted until the beeper sounds with the resistance between the probes. The resistor should then be removed and probes short-circuited, and the beeper should stop. As the low resistance detection value is extremely low, the probes must be kept clean and sharp. This design is very old, probably from Wireless World and/or Elektor.
A multimeter is, of course, needed to measure voltages, resistances, and sometimes, current. I like using both digital and old analog multimeters.
A decent oscilloscope is almost indispensable, as is a good soldering station and solder sucker. A scalpel with sharp blades is also very useful. (There is a separate article on this site about setting up a workbench and tools.)
In general, you can find many problems in the power supply section, so check it out first before continuing. If it starts malfunctioning after a period of warm-up, suspect any electrolytic capacitors that may have partially dried out. Intermittent problems are generally caused by poor connections, wiring, or dry joints.
If something you made does not work
Some of the things to check may seem obvious, but in my 50 years of experience, I still make some of these mistakes! Does the PCB or breadboard match the circuit diagram? Does the circuit have power, and is it the correct voltage and polarity? With a meter, test the voltage on all IC power pins and transistor collectors. Are all the ICs and transistors of the correct type and plugged in the right way round?
A common mistake that catches me out is transistors that have fallen into the wrong drawer, and I have soldered them in without checking the type. If you are a beginner solderer, you might be part of the problem! Go over all joints and see that they are properly soldered, and there are no bridges or dry joints. If it is a PCB, hold it up to a strong backlight and look for broken and bridged tracks.
All of the above is purely visual, and you will often find the problem at once. But if the circuit still does not work, it’s time to engage your main tools—logic and reason.
A good starting point in any new circuit is not to connect it to a battery but rather a bench PSU with a current limit control. Set the voltage to the right amount and set the current limit to, say, 100mA and power it up. If the current goes up to 100 and the voltage goes to zero, you have a short circuit. Follow the power tracks or wiring and look for it visually. Check for any electrolytic caps in the wrong way round—these will present as a short. Feel all the semiconductors and ICs for heat.
If the circuit is still not working, it’s time to recheck if you understand how it was supposed to work and have a good idea of what voltages to expect around the circuit. (It’s easy to cut-and-paste a circuit from the internet and not know how it was supposed to work).
Start with the transistors. Measure the voltage between emitter and base. They should be close to 0.6V, and there should be something reasonable on the collectors, like between half and full voltage if the transistor is off and close to zero if the transistor is on. If not, there may be a problem here—wrong type or reversed or even dead. Measure the supply voltages using BOTH probes on the IC pins themselves. You might have left off the ground, and if you measure with the -ve probe to common ground, you will not spot it. So put your scope on AC and 5mV/div and look at the power rails—are there hum spikes or glitches? If all the static DC voltages are correct, it’s time to feed in a signal at the input and look for it with the scope along the path (assuming an audio type circuit). Make sure there is gain where there should be. If the circuit is a digital one, check that the logic levels are correct for the logic family in use and if it’s CMOS, make sure there are no unconnected inputs. For both analog and digital circuits, see that there is adequate decoupling for both high and low frequencies (0.1 and 100uf caps).
If you have done all the above diligently and still won’t work, pack it away and come back tomorrow. Remarkably, this often works—taking a step back for a while gives your brain a chance to mull over the problem and think outside the box for a solution.
Something that was working but is now faulty
A lot of the above will, of course, apply here too, but there are some other considerations—what happened? What was observed? Was there smoke, sparks, etc.?
All electrical and electronic components are manufactured with a little bit of ‘magic smoke’ inside them, and if that smoke is allowed to escape, the device will never work again! Lighten up—it’s a joke. (Read more about it here.)
Suspect anything that moves—where a cord enters a housing or inside the mains plug, knobs, switches, rotary selectors, pots, pushbuttons, belts, or pulleys and where connectors are soldered to the PCB. All these can become fatigued. Motors have brushes and commutators that wear out. Look for and test any fuses or thermal cutouts—some fuses may look okay, but never trust that. Please remove them and test for continuity. Look for burned tracks from lightning or static. If the device is older than, say, 30 years and has electrolytic capacitors, it is often the case that they have dried out and become partially shorted. Replace them with equivalent values. Semiconductors, ICs, etc., that have survived a few years are unlikely to be faulty unless stressed by lightning or overheating.
Finding a short
The general technique is messy: cut and repair PCB tracks along the path until it clears, then the short will be just before your latest cut. But there is a more advanced method if the path does not have too many loops. Assuming the short is on the +12V line. Set the PSU to the current limit at about 1A. Connect one side of the multimeter to the +12V at the PSU source, set the meter to the smallest DC voltage, and with the other probe, start moving away along the +12V line and carefully watch the voltage (it will give you mV or uV if you have a good meter) as you travel towards the short. The voltage will rise due to the PCB track resistance and the fact that a large current is flowing. If it stops rising as you move, you have just passed the short or the fork where the +12 feeds into another loop.
Finding an open circuit
The general technique in finding an open circuit is to start applying shorts along the path until you “see” it from the source, then the open circuit will be between your current short and the previous short. With the power off, connect one side of the continuity tester to the source, and with the other probe, start moving away along the line until you no longer get a beep.
Fixing pads and tracks
Everybody has their own ideas here, but after some experience, I just use a very fine enameled copper wire of the solder-able type (usually pink or green, not the varnished type used in coil winding) and bypass the damaged or lifted track. (See the bottom right corner of the image below ). Any fine wire will do. If the pad is lifted, you can try making a small loop with the end of the wire, or if you are putting in a new component, keep the lead long, bend it over and make a track out of it (see top left corner below).
Some articles suggest gluing copper to the board, but I have found this overkill. If you want to read more about it, check here.
Often getting a new project to work will need patience and perseverance. Almost always, the problem will be something silly you have done, and you will want to kick yourself afterwards. Repairing previously working devices is more difficult, especially if the device has complex electronics and cannot get a circuit diagram. But remember these two rules: Suspect the power source first, and look, smell, and feel.