I ain't gonna lie. You won't be a master troubleshooter after reading a couple pages here. Hopefully, what I can do is give you the ability to form a logical plan of attack. From that comes self confidence, knowing that there is a deterministic way to solve problems, without a lot of trial and error and without resorting to the dreaded scattergun approach of random part swapping.
Troubleshooting is done mostly with the mind, not with the soldering iron. We all have problems that get the better of us, at least for a while, but approaching the situation as an interesting puzzle, rather than an unpleasant task to be suffered through, will make the work go quickly. Success will come from a 3-prong approach, a combination of understanding the circuit, having the means to test it and being in the right frame of mind to correctly interpret the test data. Never underestimate that third component. Almost all my mistakes come from charging down the "obvious" path, positive that my early tests have revealed the one true solution, only to find that there was more than one explanation for the data and in my excitement and haste, I chose the wrong one.
We have to consider everything from safety to test equipment to how to attack various components. I've tried to break the write-up down into very short sections on each topic so you can go directly to what you need, but please give the whole thing a good read at least once.
Troubleshooting is inherently hazardous. The nature of service requires that equipment covers be removed for access to live circuit components. Because the equipment is faulty in some way, dangerous voltages may appear at unexpected points. Being safe means taking every reasonable precaution to avoid the known dangers, but it also requires taking measures to protect yourself from non-obvious dangers. The following list has been compiled from various sources plus personal experience, but it is by no means complete. Please consider each item carefully.
Here's an example of hazardous voltages being present in unexpected places. I powered up a vintage Stromberg Carlson amplifier to test it. It had passed resistance checks from the chassis to both mains terminals with the power switch on. These amps have no capacitors from line to ground, and are fully transformer isolated, so it should have been safe. In spite of that, when I connected the scope ground to the speaker ground, there was a flash, a bang and the end of the connector was partially burned away. Subsequent measurements again showed no shorts to the line. It turned out that the amp had a broken circuit breaker. A metal washer came loose and jammed between the mains connection and the chassis inside the circuit breaker, putting the chassis at 120 VAC. The connection wasn't good enough to catch with the low voltage ohm meter, but it was a dead short at 120 VAC. You can't catch every hidden problem, but by working one handed until everything is safely grounded you can avoid disaster. If I had grabbed the amp with one hand, and the grounded scope with the other, the outcome could have been far worse.
A good lawyer always has his or her paperwork in order and documents everything. In our case that means finding a schematic for the device in question, plus anything else that might be useful. That would include operators manuals, advertising and technical service bulletins. It would also mean scanning the various Internet forums to see if anyone has traveled this road before. There may be times when you have to work without a schematic but it's never easy. If you can't get the exact schematic, possibly you can find a similar model that has circuitry in common. Do everything in your power to get some kind of documentation.
Now that you've got it, read the documentation including the owners manual. If you don't know how it's supposed to work, how do you know it's broken and how will you tell when it's fixed? This seems like something you could skip over, as I usually did, but I recently helped repair an amplifier where the outputs were wired differently than I was used to. Only by reading the owners manual could one avoid inadvertently shorting them out while testing. Unless you already know the equipment inside out, read the documentation!
Document the work as you go. Draw diagrams, take photos and make notes. One good photo will show where wires went and how capacitors and other parts were installed. Even the best of us get confused about polarity every now and then or forget whether the red wire or the green wire went to that lug. Sometimes PCBs are mislabeled, schematics can have errors and even components like diodes and tantalum caps have been known to have the markings reversed. If you take a photo you'll at least know what you started with.
A good lawyer won't ask the witness a question unless he already knows what the answer should be. When you make a measurement, be sure you have some idea what the answer should be. If you don't have a range of values in mind, you need to study the circuit. Measurements steer our actions so it's important to get clear answers that prove parts are either guilty or innocent. Forward progress depends on this! You need to develop the skills and confidence to declare a part good and move on. If you don't, you'll have continuous doubt about what's working and what isn't; you'll go around in circles or end up replacing perfectly good parts. Shotgun troubleshooting is bad troubleshooting.
Reaching that happy state of confidence means fully understanding the various types of parts and the methods used to test them. We'll cover that in detail further on.
There is a certain minimum collection of equipment needed to efficiently test and troubleshoot electronics. It's possible to do the job with less, but you'll spend more time and be less certain of your results. If you intend to do this regularly, work towards having at least the following on your bench:
A hamfest, eBay, Craigslist or regular electronics supplier will get you most of these items for very little money. In a pinch, the signal source could even be a CD player and good work has certainly been done with a non-controlled 25 watt pencil iron. The one item that isn't cheap, but I consider essential for those serious about this sort of work, is the LCR meter. A new handheld unit with both value and loss will be about $300. Though you can get a "C-only" meter for under $40, that's insufficient by itself. You must combine it with an esr meter for useful measurements. Actually, an esr meter alone is more valuable for troubleshooting than a C-meter without loss. (update- There are now full featured LCR meters, albeit not fully packaged, on eBay for about $25.)
Another possibility, though much slower to operate and requiring a bit more knowledge, is a traditional LCR bridge. The General Radio 1650-A or B is a common example. There are also plans on this site for building your own from scratch. It's a very easy project. An Internet search will turn up various plans for a DIY esr meter as well.
If you don't do this sort of work often, PC-based test equipment is another possibility. My personal feeling is that PC-based test equipment isn't usually as efficient as having dedicated single purpose devices with real knobs and buttons, nor is it as robust, but the capabilities can't be denied. Using the sound inputs or an external USB sound card and a free program like Visual Analyser, you can have an oscilloscope, FFT spectrum analyzer, signal generator and LCR meter for almost no investment at all.
The remarkable thing about initial assumptions is how often they're wrong. We always have our suspicions about the cause of a problem, but don't get too attached to an explanation before you do your testing. Try to keep an open mind and remember that sometimes a fault will be with a cable or some other component, and that what's on the bench will end up with a tag saying, "No Fault Found." Other times there will be multiple faults, sometimes related, sometimes not.
If you're repairing something for someone else, pay close attention to their description of the problem, but again, don't get too attached to it. It's a clue, not guaranteed truth. People may know the symptoms exactly, but will use terminology that leads us astray.
A friend of mine often said, "A bad plan is still better than no plan." He was right, but a good plan is better yet. We all want to go charging in, hoping our suspicion of what's wrong will lead us right to the defect. We'll change the bad part and be on our way. Unfortunately it never seems to work out that way. If you follow a systematic plan that starts with a thorough visual inspection, followed by checks of passive and active components, you'll leave problems few places to hide. When you're done you'll have confidence that all the problems have probably been found.
The exact plan will depend on the unit in question, but the gist of it should go something like this:
At this point you'll probably have inconclusive results on one or more components. If you have two channels, compare them. If the readings match, the problems are likely elsewhere. Naturally this doesn't hold for power amps where both channels are blown. If still in doubt, study the schematic to see what might be shunting the parts to give an unexpected reading. Power transistors will often be shunted by low value resistors. Don't panic if the diode test gives a very low value on one leg. A blown output transistor will usually be shorted all three ways. Don't be too fussy about tolerances at this point since moderate variations rarely cause serious problems.
You may have noticed that much of the work so far is inspection and measurement, not power-on tests. That is as it should be. Troubleshooting is a competitive intellectual pursuit, not a contact sport. You should make enough measurements to have high confidence that a part is defective before even thinking about heating up the soldering iron. You lose points for removing good components from a circuit board. You lose even more for leaving bad ones in!
Try to determine why a failure occurred, as this may lead you to other defective parts. Only when you have high confidence that all the defects have been found and fixed should you consider applying power.
The troubleshooting plan should proceed such that failed components get replaced and are not destroyed again when the unit is powered up. That means using a variable transformer where needed and possibly a "dim bulb" tester. Direct coupled power amplifiers are notorious for "chain failures" where one bad component causes the failure of parts both up and down stream.
Remember that problems common to both channels are often power supply related, protection circuit related or control circuit related. If you're working on a stereo unit don't forget that you can compare readings channel to channel. Even if passive component values are a bit off due to the surrounding circuit, if they match channel to channel the problem is very likely elsewhere. There exceptions to every rule and I've worked on several amps where both channels had identical capacitor failures and exhibited the same symptoms.
Before powering up, and especially after major disassembly, do an ohmmeter check on major grounds and other disturbed connections. Sometimes things that can't possibly go wrong, do. I've had circuit board connections that appeared solidly soldered go open due to cracked traces. IDC connectors can lose contact. Wires can even break inside the insulation where you can't see it. That extra check is cheap insurance that the things you've disturbed or replaced are still intact.
Commercial products are usually well engineered and don't have even subtle issues with noise, hum or imperfect operation. Whenever you see something unexpected, even at very low levels, chances are that some fault remains. You must track down the cause of all unexpected readings and observations because they are an almost certain indicator that you haven't finished the job. Sometimes the problem will be your fault- bad connections to test leads, poor cable routing, CFLs over the workbench or nearby light dimmers. Even a cell phone in proximity can cause interference. Other times it will be a bad ground connection inside the unit, transistors starting to go bad, an undiscovered intermittent or bad solder joint. I can't emphasize enough, if you haven't explained all anomalies the job isn't done and it will come back to bite you!
The mark of a craftsman is that he/she leaves no marks. Pad your bench to protect what you're working on. Get some plastic bins or other containers to hold screws and other hardware you remove. Think twice before picking up a soldering iron. Have you proved with some certainty that the part you're about to remove actually has a problem? If you're working on something that might be ESD sensitive, get a conductive mat and resistor isolated wrist strap. Use nut drivers and wrenches, never pliers that damage nuts and screw heads. Make sure that screwdrivers fit the slots, especially Philips and similar cross point types. Not all screws with a cross-type slot are Philips.
Removing parts from PCBs is an art. It's usually best to clip the part on top, then heat and remove the short leads from the board. Use Solderwick or a large suction tool to clean things up. Since clipping the part is the best way to avoid PCB damage you become motivated to be sure it's defective, or at least that another one is on hand. Use the same technique on ICs. In difficult situations there's no shame in clipping a part close to the body and soldering the new part to the old leads. This technique was specifically recommended by Hewlett-Packard (I think) for servicing some of their very expensive test gear.
Electrical tape has no place in modern electronics. Always use Teflon sleeving and shrink tubing where appropriate, shrunk with a hot air gun, not a match! Clean up your solder connections with alcohol so there's no evidence that service was done. There's no good reason not to work to the highest possible standards (says the person who just suggested clipping an inaccessible part from the top- it isn't the sin some would have you believe, especially compared to damaging traces).
A lot of equipment will come to you with missing screws or completely wrong screws cross-threaded into ruined tapped holes. Most receivers and amps from Japan will use metric hardware (duh!) and you should lay in a supply of commonly used screws. If the tapped holes are destroyed, it may be necessary to tap for a larger size or even use a suitable sheet metal screw. PEM nuts are very useful if you have the ability to install them and the location is favorable. Bottom line- never try to use a screw of the wrong thread or you'll just damage the unit. Also, pay close attention to the length and style of screws so as to get them back where they came from. Sometimes putting a longer screw in certain locations will actually short something out.
The visual inspection is your critical first step. Use a magnifier and do it under a bright light. You are looking for specific problems like darkened or burned resistors, bulged caps, ruptured caps, bad solder joints, lifted or cracked traces, excessive flux and contaminates, dust and dirt accumulation, oxidation and corrosion, solder shorts, bent tuning cap plates, wire insulation issues and anything else out of the ordinary. Work all controls and note how they feel. Watch dial cords, tuning caps and mechanisms as they move. You are also looking for general issues like evidence of flood damage, beer spills, insect cocoons, mouse damage and corrosion from poor storage conditions. Look for evidence of the unit being dropped like cracked circuit boards or bent sheet metal. Note tar deposits if it was owned by a smoker and any bad smells in general. Toasted transformers and ruptured capacitors have their own distinctive odors as well.
Don't panic if you see a thick brown deposit under caps. It's just the brown glue that manufacturers use to retain the parts and prevent damage in shipping. Some of the glues turned out to be corrosive over time, so if there's evidence of the glue eating away at things, or causing corrosion, consider a glue removal project after the original problems are fixed.
Watch out for problems concealed by wires, chassis parts or just the general style of construction. Gently move wires, look from different angles or even use a mirror to see what's going on in hidden areas. It's always a bit annoying to test your way to a problem that, if you had just looked a bit harder, could have been spotted visually.
Don't fix anything yet, just make a list. If you're taking less than 5-10 minutes to do a visual inspection, you're not looking hard enough. Inspect every solder joint and be sure the solder actually wetted and flowed on the wire. Look for the telltale circle around joints that indicates a crack, common on guitar amps that see vibration. Look at both ends of every cap. Wiggle ground lugs, screw mounted cap terminals and anything else that could get loose. Also take note of anything useful that might be printed on circuit boards and elsewhere. Part numbers, revisions, service info like voltage test points, should all be noted for later use.
Check color and texture. Are any components darker than similar components? Are the color bands a different texture or color? Do you see any plastic capacitor jackets that have shrunk down below the top of the can? Are there any small cracks or holes in transistor or other packages? All these things indicate overheating.
I have not yet perfected the visual inspection and may never do so. There are still a significant number of cases where I ultimately find a problem that could have been spotted visually, had I only looked carefully enough in the right place. Still, I consider the visual inspection extremely important because it allows me to spot potential problems that don't yet have any symptoms.
A surprising number of problems get their start right at the factory. A wire clipping lodged in an unfortunate location, a wire that was nicked when the insulation was stripped, a joint that didn't get soldered. More subtle flaws include screw-mount caps held to a chassis by their screws, with the plastic top under tension. Some of the most respected high end makers, those who should know better, have done this. Are the screws bottoming out in those capacitors? Now and then a circuit design error slips by, maybe an overloaded rectifier or a cap operating outside its voltage range under some unexpected condition, maybe a power resistor that burns the circuit board over time.
It's your job to study best practices and recognize why a failure occurred. Read the application notes from the capacitor companies. Especially read the application notes from the semiconductor people on how to correctly mount power devices. Most people use the wrong hardware and way over-torque, risking cracked dies in the package. Study hardware. Where should you use what kind of lock- washers? Do you know how to replace stripped out sheet metal screws with machine screws and maybe PEM nuts? Do you know how to de-rate resistors and other components for the actual operating conditions? Well known and respected companies usually get their circuitry right and meet their performance goals, but they often fall down on the details of implementation. Interestingly, the high volume Japanese receiver and amplifier builders probably do/did the best of anyone in the world at getting the details right.
The short answer is because most receivers and such are designed with 10 lbs of you-know-what in a 5 lb sack. If the whole circuit were laid out in front of you, flat on proto boards, it would be a simple matter to measure voltages, currents, active and passive parts, and quickly zero in on the problem. Instead, one is often looking at a microscopic schematic, with no clue which wires are what and where critical voltages can be measured. Or, one knows where, but the leads are inaccessible or the risk of shorting something out is too great.
The only advice I can offer here is to be methodical and make yourself some aids. Zoom in on schematics and print out the details you need. If necessary, tape several pages together. Next, trace out critical wiring and mark test points in red on the board layout printouts (if you have them) so you know where to probe. One trick I do is photograph the component side of a board, then flip or mirror the image. Print it out and outline caps and such in red. Now you can look at the back of the board and quickly probe or desolder components without constant flipping of the board or whole unit. This not only saves wear and tear on wires, but will help you maintain your focus.
Modern carbon film and metal resistors are very good. Unlike carbon composition resistors of old, they rarely drift far with age. If you're working on modern equipment you'll rarely find a bad resistor that isn't obviously burned or damaged. If you're working on an old guitar amp, check every carbon composition resistor in the circuit no matter how good it looks. They typically drift high, but fortunately the circuitry is usually tolerant of them. When they exceed tolerance, replace 'em. Watch out for plate resistors and screen resistors in tube equipment. They seem to have an unusually high failure rate, often without any visual evidence. Remember, if a resistor burns up, it's never the fault of the resistor. Something else failed, put too much voltage across the resistor, and then it burned up.
As said above, you want to know the correct answer before you measure, so learn the 3-bar and 4-bar color codes so you don't have to look everything up. Short out all electrolytic capacitors for a few seconds with a clip lead having 20-100 ohms built into it because small residual voltages will confuse your DVM. I move across the board measuring each resistor with an autoranging DVM. Modern DVMs use a very low voltage to measure resistance so they don't turn on associated transistors and other semiconductors. Only rarely will you need to lift one leg to confirm a resistance value, but it does happen. Most resistors can be measured in-circuit, but there are usually a few that will read low due to other parts of the circuit being in parallel. Look at the schematic to see why. Resistors can't be made to look high by other circuit components (unless you have a still-charged capacitor somewhere) so any resistor that reads higher than its labeled value is suspect, as is any resistor where you can't explain the observed value based on the schematic. Compare channels if possible. Identical readings, even if unexpected due to shunting by other parts, almost always indicate that all is well.
A useful mnemonic is, "Better Be Right Or Your Great Big Venture Goes West." This lets you remember the standard color code of black(0), brown(1), red(2), orange(3), yellow(4), green(5), blue(6), violet(7), gray(8) and white(9). Thus, a 3-band resistor with yellow, violet and red bands will be 47 plus 2 zeros, or 4700 ohms. The part may also have a gold, silver or other band to indicate tolerance, gold being 5% and silver being 10%. Rarely it could also have yet another band indicating reliability. 1% or better tolerance resistors will use a 4-band scheme. A resistor having yellow, violet, green and brown bands will be 4750 ohms. Since the 4-band scheme has an extra digit, the last band will be one less than you're used to with the 3-band scheme. Much depends on the vintage of the equipment and the type of resistors used.
In my opinion, if you do a lot of service, there's no substitute for a good auto-ranging capacitance meter that displays both value and loss. The least expensive hand-held unit is several hundred dollars, but worth every penny. It's easy to move across a board, checking all the caps and noting anything suspicious. Almost all capacitors can be checked in-circuit with sufficient accuracy to determine if they're the cause of a problem.
Capacitors can also develop low DC resistance (leakage). In most modern circuits the leakage would have to be quite large to cause a problem, and might show up as excessive loss on the C-meter. In other circuits, say coupling to the grid of a tube, even a slightly leaky cap will change the bias and increase the distortion. You'll have to remove the cap to do a proper leakage test at the actual operating voltage. Your fancy bridge or LCR meter won't help you here unless you own one of the more specialized units that can do biased tests at full operating voltage, but you still have to remove the part to avoid damage to attached circuitry. Once you've removed it, it's usually more efficient to just replace it.
Capacitors don't cause as many problems as people tend to think. You'll certainly see some random failures, but until equipment is 20-30 years old, capacitor failures are infrequent. At 30-40 years old you're on the rising side of the well known "bathtub" shaped failure curve and should be measuring value and loss carefully. When losses rise or values are lower or higher than expected, it's time to change 'em.
I believe in testing because I find capacitors that need to be replaced before they cause any symptoms. I save money by not replacing expensive screw mount types. That said, many people find it easier to just replace smaller caps on general principle. I have no problem with that in terms of a rebuild, but not as a troubleshooting procedure. The odds of introducing a new problem are too high, as evidenced by the large number of people on Internet forums who have "recapped" a working piece of equipment, only to find it now has some mysterious problem or doesn't work at all. Fix it first, and only then modify or upgrade it!
You'll find quite a bit more detail on testing capacitors, plus plans for an easy and inexpensive bridge that can measure losses, elsewhere on this site.
In this category we have the film caps like Mylar, polystyrene and polypropylene. We also have ceramic, mica and the non-aluminum types like tantalum. They are all very reliable and don't age to any significant degree. Polystyrene can be damaged by solvents used to clean circuit boards and by heat. The other films are quite robust. Tantalum caps are very reliable when used correctly, but will fail due to fast current spikes or any reverse voltage. The problem is they fail short circuited, often destroying the circuit board underneath. If a tantalum is warm or changes color slightly during operation (light yellow changing to darker yellow) it's probably damaged or shorted. In most audio equipment they are best changed to high quality (long life) aluminum electrolytics, or films if smaller in value.
Modern tantalum capacitors are very reliable if used properly. That includes having a series resistance of at least 0.1 to 3 ohms in the circuit, derating the voltage to about 60% maximum of the rated voltage and keeping the temperature to a reasonable value. They must never, even briefly, be exposed to any reverse voltage. I suspect older tantalums were basically less reliable and the failure modes less well understood. Some manufacturers used tantalum caps where very low DC leakage was needed, usually as coupling caps between amplifier stages. These can get leaky with time and should be replaced with film caps. If the value is too large for a film, or it won't fit, special low leakage aluminum electrolytics can be purchased. Don't use standard electrolytics because the leakage may be too high, or they may develop excessive leakage as they age. Look for leakage specs of less then 0.01CV. I don't recommend using tantalum capacitors at all, unless space prevents replacement with anything else. The audio qualities of tantalums suspect at best, and downright poor if they're not applied correctly.
Connector contacts are usually reliable, but their fastening to wires and PC boards can be less so. It's not unusual to find poorly crimped contacts and header pins that have flexed PC board traces to the point of failure. The fixes are usually obvious. Watch out for intermittent IDC (insulation displacement connector) wires. These types of connectors are quite fussy about wire gage and insulation thickness and type. If everything wasn't perfect during assembly, over time they can oxidize and lose contact. They can be difficult to repair by soldering (the metal may not wet and the housing may melt) and sometimes replacement is the only quality option. You may have to make or buy a suitable tool to properly force the wire into the terminal slot.
Switches seem to account for more than their fair share of trouble. At best, they get dirty or oxidized and a shot of a good contact cleaner like DeOxit will make them reliable again. At worst, the internal contacts eventually wear out and replacement is the only option. There are also self-latching switches with small wire loop mechanisms and springs that cause them to alternately lock and unlock. These are best replaced, but sometimes a bit of epoxy can renew broken parts.
Guitar amps and PA equipment often use 1/4" phone jacks with automatic shorting to reduce noise on unused inputs. The shorting contacts commonly lose tension or get dirty. Clean these by slipping a piece of stiff card saturated with DeOxit between the contacts and sliding it in and out. Retension contacts if needed. The newer plastic jacks can often be fixed by unsoldering the contact from the PC board, pulling it up and out of the switch body, retensioning it and then reinstalling it.
Solid wire is prone to breaking so inspect equipment built with solid wire carefully. Breaks inside insulation can be the most difficult problem you'll run across. Sometimes when contacts are crimped on, the wire is weakened or cut where you can't see it. The ohmmeter and lots of patience are the only answer.
Look for burned insulation and try not to create any yourself! Look for abrasion and cuts. Look for insulation that moved back on wires when they were soldered, and can now short on adjacent parts. Wires also have a tendency to get caught when PC boards are screwed down or pinched between brackets and chassis parts. It can take a while for the insulation to flow and fail, so a unit might get through final test just fine, then fail some months down the road. For the same reason, never pull wire ties too tight.
Inductors rarely go bad. If they do, the smaller ones usually go open. The only other possibilities are shorted turns and shorts to the frame. A resistance check will usually do the trick, otherwise you'll have to resort to the LCR meter or bridge. If the inductor exhibits low Q, it may be shorted.
Transformers are tough to test because you rarely have any factory specifications to go by. Quality transformers will be quiet and run cool. Cheap transformers with low quality iron may buzz and run hot, even with minimum load. If you can isolate the secondaries of a power transformer you can bring it up with the Variac and check the AC output and input power consumption. Buzzing or excess power consumption usually means a shorted winding, primary or secondary. No output means an open winding, easily determined by a resistance check.
Because transformers are expensive and often unavailable, you want to be very sure they're bad before replacing them. When all else has been eliminated, to be absolutely sure, I like to check the primary and secondary inductance in various combinations to determine the leakage inductance and coupling. The Handy Formulas sheet on this site will give you the necessary instructions.
The "diode check" setting of your DVM will display the voltage on the probes at a very low current. Placed across a diode you should see the voltage drop of the junction at that very low current. For standard silicon diodes that will be about 0.4-0.7 volts. For Schottky diodes expect about 0.1-0.3 volts. For germanium diodes like 1N34 and 1N270 expect 0.2-0.3 volts. Remember that these numbers are not the expected voltage drop under full load operating conditions, just what you can expect to see when measuring with the typical DVM at room temperature.
Test these with the diode check function of your DVM. For NPN transistors place the positive lead of the meter (the red one, we hope) on the base of the transistor (or someplace connected to the base). You should get one silicon diode drop when you touch the other lead to the collector, and the same to the emitter. Next, measure from the collector to the emitter in both directions. You should see a high impedance or open circuit. Any sort of low reading means a defective device. Note that other circuit components, especially in power amp circuits, can give you false readings. You may have to remove the device to be absolutely sure.
The classic case is the audio power amp. You measure collector to emitter, hear the continuous beep of the meter and see a zero volt reading. You almost certainly have shorted output transistors. Note that amps with paralleled transistors may have only one bad device and it can be difficult to determine which one it is without removing them. Sometimes the offending device can be located by bringing the amp up on a Variac to a very low voltage and seeing which device heats up more quickly. I've also done this with a current limited bench supply when the amp can't be powered at all. Note that some circuits will have the transistors shunted by low value resistors. Those will fail on diode check, so switch to ohms. If you see 50-500 ohms, the transistors are likely good.
Use the DVM to test small signal BJTs as described above. Small signal devices usually have enough resistance in the associated bias and other circuitry that you can make the measurements in-circuit. Power devices are usually obvious if they're bad, but can give confusing reading if they're good because of associated low impedance circuitry. Darlington devices seem to be falling out of favor, but note that darlington devices will have two diode drops from the base to the emitter.
There should be no ohmic connection between the gate of a MOSFET and the source or drain. Shorts between the source and drain will be obvious, just as for BJTs. A failed power MOSFET will often short through the gate as well, and damage the driver circuitry.
Don't try to test a MOSFET on the bench with the gate floating. It could be in any state of conduction and there's always the risk of static damage. Tie the gate to the source when looking for shorts.
The resistance of negative temperature coefficient (NTC) thermistors goes down when they heat up. Those are used for surge protection in the AC line of many amplifiers. The resistance of positive temperature coefficient thermistors (PTC) goes up when they heat up. You'll find those in lieu of fuses in many locations, often the high voltage and heater supplies in tube guitar amps.
Both types will occasionally fail. The usual symptoms are amplifiers shutting down or going into protection sooner than they should, noise due to erratic resistance or any other expected problem due to extra resistance in a supply line. If the device is just marginal or intermittent it may be difficult to determine its condition with absolute certainty. Try heating or cooling it with hot air or freeze mist. Replacements are inexpensive.
You'll also find small bead thermistors sometimes used for bias control or other compensation in power amps. They can be tricky to replace, as exact specifications are usually unavailable. They are often mistaken for multiple diodes, which are also used and can look the same. Sometimes they can be replaced with multiple diodes, but check forums to see what people have done with success in the past.
The perfect solder joint is shiny and wets the wire, forming a nice fillet. A dull joint or a joint where the solder seems to bead up away from the wire, rather than wetting it, is cause for suspicion. Resolder the suspicious ones, adding a bit of rosin flux if necessary, but be aware that bad joints are usually obvious and bulk resoldering efforts are usually a mistake, as is reflowing every exposed trace. The pitfall is that the joints and traces aren't usually the problem and you may end up masking a thermal issue, think the unit is repaired, and discover the truth some time down the road.
Newer equipment will have been built with lead free solder. This tends to have a duller appearance and may not wet the leads as well. Equipment built during the transition period may have defective joints as the materials and process weren't as well established then as they are today.
Fuse issues will be obvious and the only caution here is to be sure you've got the right fuse installed. It needs to be the right value and the right speed- fast blow or slow blow, of which there are various subtypes. It is always a mistake to try a larger fuse to see if that might "fix" the problem or because you don't have the right one. If you have any doubt about the meaning of the word "always", this repair game might not be for you.
Potentiometers, "pots", get dirty, get noisy and eventually wear out. If you can get access, a shot of Caig DeOxit will often bring them back to health, but the big picture is more complicated. A new pot will have an unworn wiper and a smooth resistive track. It will probably have a bit of special lubricant on the track. It may have a completely different lubricant on the shaft to give it a smooth damped feel. When you use a contact cleaner the lubricants may be washed out, increasing the wear and changing the feel of the pot. You may want to buy suitable lubricants for both the track and the shaft if you want to preserve the original feel. Caig Faderlube is one such contact lubricant. Various motion control greases are available from Nye Corp. for shaft lubrication, but getting the grease into the shaft area can be difficult. A bit of solvent may help carry it in.
Pots controlling AC signals shouldn't have DC voltages present. If you have trouble with noisy pots and offsets that change with pot settings, look for leaky coupling capacitors feeding or isolating the pot. Another subtle flaw is a broken resistive element. If the pot adjustment isn't acting as expected, check across the pot for the correct resistance.
Hardware isn't usually the direct cause of a problem, but you'll be dealing with a lot of it just the same. Always have something available to put hardware in as it's removed. Note differences in screw styles and lengths- the manufacturer used them for a reason. Make notes if necessary as to what goes where. When replacing screws start them carefully and don't force them if they don't spin in easily. Figure out what the problem is before stripping the threads out of a hole or damaging a screw. Wood and sheet metal screws will often start at two positions, only one of which is correct from the original self-tapped installation. If the screw doesn't go in easily, back it out half a turn and see if it will start in an alternate position. Always re-start wood and sheet metal screws by going backwards until you feel the slight drop into alignment with the thread, then go forward to tighten. You do not want to cut a new thread because that will cause the screw thread to strip out. Be sure your screwdrivers fit the screws properly so they don't slip and damage the heads or the unit being serviced.
I seem to encounter a lot of missing hardware. It pays to keep a supply of some common items around. The screws holding many receiver and amp covers are M4-5 or M4-10. A few dollars will buy you 100 from Digikey or other suppliers. An assortment of sheet metal screws is useful, and you may have to go up one size if the original hole is stripped out. I haven't found a good source of black hardware, but you can stick them in a piece of cardboard and spray them black with a good epoxy paint.
The most frustrating and time consuming problems are often intermittents. You can only fix things that occur on the test bench and intermittents are famous for staying hidden until the customer reinstalls the unit. The problem then recurs and the customer calls to inform you of what an incompetent idiot you must be. There are several things you can do to encourage intermittents to reveal themselves:
This is where you need to explain all anomalies. If you thought you saw some glitch or noise spike, you probably did, and it will come back to haunt you unless you track it down.
Noise problems are common with older equipment. Though we tend to think of transistors are being completely stable and lasting forever, that's not always the case. Some types, like the infamous 2SA458 "low noise" part develop leakage and high noise levels later in life. Resistors, particularly carbon composition types, can become noisy. Capacitors rarely become noisy, being self filtering, but it can still happen. I've seen old silver-mica and film caps get noisy, but only a couple times in several decades of troubleshooting.
The problem with troubleshooting noise is locating the exact component or components causing it. All solid state circuits use feedback, so noise at any point in the circuit tends to be fed back to the beginning and appear everywhere. In simple low level two transistor amplifiers don't be surprised if the noisy part is the one following the one you think it is, since there is often feedback to the emitter resistor of the first transistor. Sometimes you can heat and cool various parts with freeze mist and a soldering iron, to see if the noise level changes. That's the most straightforward method, but sometimes it just doesn't work. Sometimes you can temporarily disconnect the feedback (if the circuit is still stable), or you may have to just take a guess and replace some parts. Shotgun troubleshooting stinks, but at the same time you can't make a lifetime project out of what should be a simple repair.
Watch out for leaky coupling capacitors as they can allow DC to appear on volume pots (or other places) and cause noise when the control is adjusted. The rule is no DC current should pass through the wiper connection of a pot used for signals. In some circuits the amount of leakage that can cause a problem is quite small. Use your DVM on a sensitive range to check for DC where it shouldn't be.
The final test needs to be the final test. That means covers screwed on and everything buttoned up just as the unit will be used. If you have to open the unit or disturb any hardware, the final test must be repeated. This is most important if you do work for others because you can't honestly say everything was in perfect working order when the unit was finished unless you tested it that way. Even manufacturers mess up on this one. It's just not that uncommon for something unexpected to happen if the final assembly happens after final test.
A quick comment on the "diode check" mode of DVMs, because it's so important in troubleshooting. Any DVM worthy of use will have either a special diode check setting, or a resistance range (or two) marked with a small diode symbol. When you measure resistance with a DVM, it applies a low voltage across the unknown and measures the current. The internal processor does the simple Ohm's law division of voltage divided by current to give you the resistance in ohms. The applied voltage is cleverly kept below about 0.6 VDC, so any transistors and diodes are not turned on. Thus, you can measure many things in-circuit, without semiconductor junctions contaminating the result. In diode check mode, the meter applies a small current, but allows the voltage to rise above the 0.6 VDC semiconductor turn-on voltage. No calculation is done; the meter simply reads out the voltage.
Audio demands on scopes are minimal. Even a low bandwidth service scope can do most of what's required. There are so many scope models that it's hard to make specific recommendations. Some of the classics are getting pretty old and will need service. The Tektronix 465 is a favorite but falls in that category. The big old tube boat anchors even more so. Some slightly newer scopes have great features but also contain unobtainable parts. Do your homework before buying and don't pay too much for a scope that could easily become unfixable. I don't consider early digital scopes desirable for audio work. The resolution is too poor to see traces of noise. There are great digital scopes, and the prices are coming down. Hobbyists won't be able to justify the latest and greatest from Tek or HP/Agilent/Keysight, but 2nd tier products like Rigol are becoming very capable and are affordable.
I confess to having a powerful secret weapon. Most scopes have a maximum sensitivity of 5 mV/division. Even if they could go lower, high frequency noise would hide anything you were trying to see. Various Tektronix scopes that used plug-in modules could accept high gain differential amplifiers. For the old tube units it would be the 1A7A. For the 7000 series it was 7A22. No doubt they had various others. These plug-ins combined a maximum sensitivity of 10 uV/division with adjustable hi-pass and lo-pass filters to kill noise. Using such a plug-in, you can follow the signal from an amplifier all the way back to the input, even all the way back to a phono cartridge, observing noise and waveform quality.
You can use a differential preamp, or even just a single input preamp, with any scope. There may be some commercial models, or you can build one with a few opamps and other parts. Just be sure to include adjustable filtering so you can see the desired signals without noise contamination.
There are lots of inexpensive audio signal generators. I recommend a function generator that can produce sine, triangle and square waves. The typical function generator won't have very low distortion; it usually has a small peak on the top and bottom of the sine wave, but that's of little consequence unless you intend to use it with a THD analyzer. It's usually better to have a dedicated low distortion sine generator for that.
Personally, I like the older Wavetek function generators, but they often have minor service issues. By now they'll need new capacitors and switch cleaning, but once you get them into shape they provide decent wide bandwidth waveforms. Whatever you get, look for something that can get down to 1 Hz or less, and up to 2 MHz or more. Decent power amps will have their -3dB point at just a few Hz and it's nice if the same generator can be used to align AM tuners in a pinch, though a true RF generator is a better choice.
This section is included for those with a Harbor Freight DVM and not much else. I started out with just a Heathkit VTVM and my brain, and still managed to fix a lot of stuff, so you can too. On the downside there was also a lot of stuff that either took forever or I couldn't fix at all. There was even a list of things I flat out wrecked. You have to be realistic about what you can do with limited resources.
With just a simple meter you can still check resistances, diodes and semiconductors as described above. You can also do a rough check of capacitors by knowing how fast your meter charges up a given value on the ohms setting. Compare with some known good parts on the bench. Know your meter. It might even pay to "troubleshoot" a working unit to get a feel for how the meter reacts to typical diodes, transistors and shunted components.
Beyond that, you'll probably have to do more powered testing. You can measure the AC ripple on power supplies. If voltage and ripple are good, move on to the rest of the circuit. Understand the basics of ohms law so you can see if circuit voltages make sense. Think in terms of current; that's sometimes more useful than voltage. Measure voltage drops across resistors to determine current and see if the numbers make sense for the circuit. There should always be about 0.7 VDC between the base and emitter of transistors and on forward biased diodes. Zener diodes should have the zener voltage present across them unless used for signal limiting.
If you have some parts and a DVM, you may have more test equipment than you think. Some handy tools can be made from a few caps, resistors and semiconductors using on-line schematics. You can build a simple oscillator to inject a signal. You can build a capacitance bridge. You can build an inverse RIAA network. You can build an attenuator that will let you measure input impedance. None of those things have more than about ten parts and can be made for a few dollars.
Sometimes even a rudimentary signal is sufficient to determine if a circuit is working. An old trick with audio RCA inputs is to hold an input cable by the shield with your thumb and third finger, then lightly touch the center pin with your index finger. Because your hand is grounded, the signal should be just a moderate amount of hum, but enough to tell if an amplifier circuit is working or dead.
An FM tuner that doesn't work isn't very useful, so it makes sense to try and repair it. Special equipment is almost always required to align the RF sections and stereo demultiplexer, so take care not to disturb coils, transformers or any frequency determining component. The adjustment are probably correct where they were, unless someone has already used the "fiddle" method to try and fix it. In general you can change resistors, larger value capacitors and even semiconductors, without disturbing the alignment significantly. Do not use anything on the tuning capacitor plates that leaves a residue, as that will change the capacitance.
If you want to do tuner alignment you'll need a sweepable signal generator that can also be frequency modulated. You'll also need a multiplex signal generator. All these things are contained in the old Sencore SG-165, which is good enough for most purposes, though not for the highest performance tuners. What's more difficult than collecting the necessary equipment is understanding the theory and various methods used for coupling signals and the alignment itself. The manufacturers instructions invariably assume some prior knowledge of the matter, are often cryptic or incomplete, and sometimes misidentify test points or other details. It doesn't make sense for most people to try and do this, but if you have a good grounding in the fundamentals you may find it an enjoyable and useful pursuit.
last edit Dec. 26, 2015