|Home | Articles | Forum | Glossary | Books|
In earlier sections, we've covered the philosophy of troubleshooting analog circuits, and the tools and equipment you need to do so. But if you're working on a circuit and are not aware of what can cause component failure, finding the root of your problem could be difficult. Hence, this section covers resistors, fuses, inductors, and transformers; their possible modes of failure; and the unsuspected problems that may occur if you use the wrong type of component. (Capacitors are waiting in the next section. Kind of a shame to segregate them from the resistors . . . .) Troubleshooting circuits often boils down to finding problems in passive components. These problems can range from improper component selection in the design phase to damaged components that hurt the circuit's performance. Resistors, inductors, and transformers can each be a source of trouble.
Resistors are certainly the most basic passive component, and, barring any extreme or obscure situations, you won't usually run into problems caused by the resistors themselves. I don't mean to say that you'll never see any resistor problems, but most of them will be due to the way you use and abuse and mis-specify resistors. In other cases, some other part of the circuit may be causing damage to a resistor, and the failure of the resistor is just a symptom of a larger problem.
You may eventually have to track down a wide variety of problems involving resistors to achieve a working design. Some will seem obvious. For example, your circuit needs a 10 k-ohm resistor. The technician reaches into the drawer for one and instead gets a 1k-ohm resistor, which then mistakenly gets inserted into your board.
This example illustrates the most common source of resistor trouble in our lab.
Consequently, I ask my technicians and assemblers to install resistors so that their values are easy to read. And any time I find a 1k resistor where a 10 k is supposed to be, I check to see how many more 1 k-ohm are in the 10-k drawer. Often there are quite a few!
Sometimes a resistor gets mismarked; sometimes a resistor's value shifts due to aging, overheating, or temperature cycling. Recently, we found a batch of "1%" metal-film resistors whose values had increased by 20 to 900% after just a few dozen cycles of -55 to +125 "C. As it turned out, our QC department had okayed only certain resistors to be used in bum-in boards, and these particular resistors had not been approved. The QC people, too, had spotted this failure mode.
Resistor Characteristics Can Vary Widely
You should be familiar with the different resistor types in order to select the most appropriate type for your application; the most common types and some of their characteristics are summarized in Table A component type that's good for one application can be disastrous in another. For example, I often see an engineer specify a carbon-composition resistor in a case where stability and low TC (Temperature Coefficient) are required. Sometimes it was just a bad choice, and a conversion to a stable metal-film resistor (such as an RN55D or RN60C) with a TC of 50 or 100 ppm/ C max considerably improves accuracy and stability. In other cases, the engineer says. "No, I tried a metal-film resistor there, but, when I put in the carbon resistor, the overall TC was improved." In this case, the engineer was relying on the carbon-composition resistor to have a consistent TC which must compensate for some other TC problem. I have found that you can't rely on consistent TC with the carbon-composition type, and I do not recommend them in applications where precision and stability are required--even if you do see some TC improvement in your circuit.
However. carbon-composition resistors do have their place. I was recently reviewing a military specification that spelled out the necessary equipment for the ESD (electrostatic discharge) testing of circuits. An accurate 1500 ohm resistor was required for use as the series resistor during discharge of the high-voltage capacitor. In this case, you would assume that a metal-film resistor would be suitable; however, a metal-film resistor is made by cutting a spiral into the film on the resistor's ceramic core ( FIG. la). Under severe overvoltage conditions, the spiral gaps can break down and cause the resistor to pass a lot more current than Ohm's Law predicts-the resistor will start to destroy itself. Therefore, the spec should have called for the use of a carbon-composition resistor, whose resistive element is a large chunk of resistive material (FIG. lb). This resistor can handle large overloads for a short time without any such flash-over. Even when you are applying a 200% to 400% overload for just a short time, the nonuniform heating of the spiraled section of a metal-film resistor can cause the resistor to become unreliable. You can also get around this problem by using a series connection of metal-film resistors. If you put fifteen 100 ohm 1/4-W metal-film resistors in series, each individual resistor would not see overvoltage or excessive power.
Carbon-film resistors are now quite inexpensive and have become the most common type of resistor around most labs. Their main drawback is that they are very similar in appearance to metal-film resistors and have some similar characteristics: Carbon-film resistors have 1 % tolerances, are normally manufactured with spiral cuts, and have the same kind of voltage-overload limitations as metal-film types. But, carbon-film resistors have much higher TCs-500 to 800 ppm/ C. It's easy to erroneously insert a drifty carbon-film resistor for the intended metal-film type. Don't confuse the two.
Table 2: Typical Resistor Characteristics
Precision-film resistors, on the other hand, are available with greatly improved accuracy and TC. Compared to ordinary RN55D and RN55C resistors with TCs of 100 or 50 ppm/ degr. C, these resistors have TCs as good as 20, 10, 5 or 2 ppm/C and accuracies as good as 0.01%. These resistors are comparable to small precision wirewound resistors but are generally smaller and (slightly) less expensive. They also have much less inductance than the wirewound types and, thus, are suitable for higher-speed operation. A few spirals on a film substrate add negligible inductance compared to the hundreds or thousands of turns on a wirewound resistor's bobbin.
Precision-film resistors are also available in matched sets of discrete resistors whose relative accuracy and TC tracking are better than those of individual resistors.
You can also buy custom thin-film resistor networks on a single substrate if your requirements are critical and your bankroll large. A more economical route is to use four, seven, or eight matched precision thin-film resistors in a DIP. I have found the TC tracking of these devices from several manufacturers to be better than 1 ppm/ C. These sets are ideal for precision amplifier stages and D/A converters. (For a good example of where they are useful, refer to the matched resistors in the thermocouple amplifier in FIG. 10.) Also, when you buy film resistors, leave them on the tape.
When you need matched sets, you can pull off adjacent resistors and be reasonably confident that they will match and track well.
Thick-film resistors are usually found in hybrid circuits, but are also available as small networks. They are made of Cermet or other proprietary materials, and are baked and fired after being screened onto a ceramic substrate. Their TC is not quite as good as thin-film resistors, but they are popular because they have good TC tracking, and they are inexpensive and easy to trim to 1 or 1/2%.
Traditionally, the best, most stable resistors have been wirewound resistors. These days, precision film resistors can match wirewound resistors quite well for almost any set of specifications. However, for a resistor whose value is between 200 k-ohm and 1 M-ohm, wirewound resistors are more expensive and come only in larger packages. Wirewound resistors also have one major disadvantage: the inductance of an ordinary wirewound resistor makes achieving fast (sub-microsecond) settling almost impossible. However, you can specify a special winding pattern that can greatly cut down the inductance of the windings. This type is listed in several manufacturers' catalogs as "Type HS." But I've found that there are two different types of HS: One type has almost zero inductance and greatly increased interwinding capacitance; the other type has low inductance and low capacitance and is well suited for fast-settling amplifiers. Be cautious of manufacturers' oversimplified statements.
A tricky problem popped up a couple of years ago when we assembled a precision amplifier with wirewound resistors. The output was drifting all over the place, but the amplifier, zeners, and transistors were stable. What was drifting? It turned out that a wirewound resistor was "drifting" because we had mistakenly used a special temperature-compensating resistor with a TC of +3300 ppm/ C. This type of temperature-compensating resistor is often used for correcting the TC of transistor-logging circuits, but it wasn't labeled in an obvious way. When we put this resistor in a circuit where a low-TC resistor was required, it took us a couple of hours of troubleshooting to pinpoint the problem.
Match the TC to the Application
Diffused resistors, commonly used in ICs, have some strange characteristics. Their TC is high-around +1600 ppm/ C -- and includes a nonlinear, or quadratic, term.
Thus, the resistance goes up faster at high temperatures than it falls at cold temperatures. These resistors would be useless except for one minor detail: They track at the rate of approximately +1 ppm/ C. Since it is very inexpensive to make matched pairs or sets of these resistors in a monolithic integrated circuit, their use is popular among IC designers. If you're not designing an IC, though, you probably won't meet up with diffused resistors very often.
Many ICs, such as D/A converters and voltage references, are made with thin-film (sichrome or nichrome) resistors on the chip. Compared with most other resistor types, these resistors have the somewhat lower TCs of 50 to 350 ppm / C, closer ratios, better long-term stability, better TC tracking, and less nonlinearity of the "voltage coefficient." This last term refers to the nonlinearities in Ohm's Law that occur when there is a large voltage drop across a resistor; the effect is most common in resistors with large values and small values--ones built with high densities.
Therefore, when you drive the reference input to a D/A converter, you should be aware that the Rin will only shift 1-3% over the entire temperature range. However, there may still be a broad tolerance, as it is not easy to keep tight tolerances on the "sheet rho," or resistivity, during the IC's production. For example, a typical D/A converter's R_in specification is 15 k-ohm +33%. These film resistors have even better tracking TC than diffused resistors, often better than 1 ppm/ C. In addition to the TC, you might also be concerned with the shunt capacitance of a resistor. Recently (back in Section 2), I was trying to build a high-impedance probe with low shunt capacitance. I wanted to put a number of 2.5 M-ohm resistors in series to make 10 M-ohm. I measured the shunt capacitance of several resistors with our lab's impedance bridge. A single Allen-Bradley carbon-composition resistor had a 0.3 pF capacitance, so the effective capacitance of four in series would be down near 0.08 uF--not bad ( FIG. 3). Then I measured a Beyschlag carbon-film resistor. Its capacitance was slightly lower, 0.26 uF. The capacitance of a Dale RN60D was 0.08 uF; the capacitance of four in series would be almost unmeasurable.
It would be an improper generalization to state that certain resistor types (films) always have less shunt capacitance than others. However, the main point is that if you need a resistance with low shunt capacitance, you can connect lower-value resistors in series; and if you evaluate several different manufacturers' resistors, you may find a pleasant surprise.
Variable Resistors and Pots
As with the fixed resistors discussed so far, there are many kinds and types of variable resistors, such as trimming potentiometers, potentiometers, and rheostats.
These resistors are made with many different resistive elements, such as carbon, cermet, conductive plastic, and wire. As with fixed resistors, be careful of inexpensive carbon resistors, which may have such poor TC that the manufacturer avoids any mention of it on the data sheet. These carbon resistors would have a poor TC when used as a rheostat but might have a good TC when used as a variable voltage divider or a potentiometer. Recently I ran an old operational amplifier where the offset trim pot had a range of 100 mV. Yet for 4 hours in a row, the amplifier's offset held better than +/- 10 uV. That's an amazing +/- 0.1% stability for a carbon composition pot! On the other hand, some of the cermet resistors have many excellent characteristics but are not recommended for applications that involve many hundreds of wiper cycles.
For example, a cermet resistor would be inappropriate for a volume control on a radio.
The major problem area for variable resistors is their resolution, or "settability." Some variable resistors claim to have "infinite" resolution; but, if you apply 2 V across a variable resistor's ends and try to trim the wiper voltage to any or every millivolt in between, you may find that there are some voltage levels you can't achieve. So much for "infinite resolution." As a rule of thumb, a good pot can usually be set to a resolution of 0.1 %, or every 2 mV in the previous case. Thus, counting on a settability of 0.2% is conservative.
Good settability includes not only being able to set the wiper to any desired position but also having it stay there. But, I still see people advertising multi-turn pots with the claimed advantage of superior settability. The next time you need a pot with superior settability, evaluate a multi-turn pot and a single-turn pot. Set each one to the desired value, tap the pots with a pencil, and tell me which one stays put. I normally expect a multi-turn pot, whether it has a linear or circular layout, to be a factor of 2 to 4 worse than a single-turn pot because the mechanical layout of a single-turn pot is more stable and balanced. Does anyone know of an example in which the multi-turn pot is better? A full year after this statement was originally published, nobody has tried to contradict me, although people who sell multi-turn pots still brag in the vaguest possible terms about "infinite resolution"--bleah!
Don't Exceed Your Pot's I and V Ratings
How do variable resistors fail'? If you put a constant voltage between the wiper and one end and turn the resistance way down. you will exceed the maximum wiper current rating and soon damage or destroy the wiper contact. Note that the power rating of most variable resistors is based on the assumption that the power dissipation is uniformly distributed over the entire element. If half of the element is required to dissipate the device's rated power, the pot may last for a short while. However, if a quarter of the element is required to dissipate this same amount of power, the pot will fail quickly. For example, many years ago, the only ohmmeters available might put as much as 50 mA into a 1 R resistor. When a 50 k-ohm, 10-turn precision potentiometer (think of an item costing $20) was tested at incoming inspection using such an ohmmeter, the test technician would turn the pot down to the end where the 50 mA was sufficient to bum out the delicate wirewound element. Then he would write in his report that the potentiometer had failed. What a dumb way to do incoming inspection! Some trimming potentiometers are not rated to carry any significant DC current through the wiper. This DC current-even a microampere-could cause electro-migration, leading to an open circuit or noisy, unreliable wiper action. Other trim-pots are alleged to be more reliable if a small amount of current--at least a microampere DC-is drawn through the wiper, to prevent "dry failures." Carbon pots are not likely to be degraded by either of these failure modes. If you have any questions about the suitability of your favorite trimming potentiometers for rheostat service, you or your components engineer should ask the pot's manufacturer.
How do you spot resistor problems? The most obvious way is to follow your nose.
When a resistor is dying it usually gets quite hot, and sometimes the strong smell of phenolic leads right to the abused component. Just be careful not to bum your fingers. You may also encounter situations in which a resistor hasn't truly failed but doesn't seem to be doing its job, either. Something seems to be wrong with the circuit, and a resistor of the wrong value is the easiest explanation. So, you measure the resistor in question, and 90% of the time the resistor is just fine-usually the trouble is elsewhere. A resistor doesn't usually fail all by itself. Its failure is often a symptom that a transistor or circuit has failed; if you just replace the resistor, the new one will also bum out or exhibit the same strange characteristics.
Around our lab, if anybody smells an "overheated resistor," he makes sure that we understand what it is. Usually when I holler, "Who has a resistor overheated??," an engineer or technician will sheepishly say, "I just cooked my circuit . . . ." But sometimes it is a failure in a piece of un-attended equipment, and the sooner we can turn off its power or fetch a fire extinguisher, the better.
How do you check for resistor errors? If you're desperate, you can disconnect one end of the resistor and actually measure its value. It's often easier to just measure the 1 X R drops in the network and deduce which resistor, if any, seems to be of the wrong value. If one resistor is suspected of being temperature-sensitive, you can heat it with a soldering iron or cool it with freeze mist as you monitor its effect. In some solid-state circuits, the signals are currents, so it's not easy to probe the circuit with a voltmeter. In this case, you may have to make implicit measurements to decide if a resistor is the problem. Also, remember that a sneak path of current can often cause the same effect as a bad resistor.
When you are trying to make precision measurements of resistors, you should be aware that even the best ohmmeters--even the ones with 4-wire connections and lots of digits on the DVM--do not have as good accuracy or resolution as you can get by forcing a current through a stable reference resistor R_REF and then through the Rx and comparing the voltages. This is especially true for low resistor values. See FIG. 4. You also have complete control over the amount of current flowing through the Rx.
FIG. 4. If you use a good voltmeter to measure Vref and V, and take the ratio, you can resolve the Rx a lot better than in the OHMS mode.
Watch Out for Damaged Components
Damaged resistors can also be the source of trouble. A resistor that's cracked can be noisy or intermittent. When resistors are overheated with excess power, such as 2 or 3 W in a 1/4-W resistor, they tend to fail "open"--they may crack apart, but they don't go to low ohms or to a short circuit. The accuracy or stability of a high-value resistor ( 10^8 to 10^12 ohm) can be badly degraded if dirt or fingerprints touch its body.
Careful handling and cleaning are important for these high-value resistors and high-impedance circuits.
One problem that occurs with all resistors is related to the Seebeck effect: the production of an EMF in a circuit composed of two dissimilar metals when their two junctions are at different temperatures. In precision circuits, you should avoid thermal gradients that could cause a large temperature difference across a critical resistor. For example, don't stand a precision resistor on end, as in an old transistor radio-if it has any dissipation, it might get a lot hotter on one end than the other.
Many precision wirewound and film resistors have low Seebeck coefficients in the range 0.3 to 1.5 kV/ C. But avoid tin oxide resistors, which can have a thermocouple effect as large as 100 kV/ C. If you are going to specify a resistor for a critical application where thermocouple errors could degrade circuit performance, check with the manufacturer.
So, you ought to know that resistors can present challenging troubleshooting problems. Rather than re-inventing the wheel every time, try to learn from people with experience.
When Is a Resistor Not Just a Resistor?
When it's a fuse. Obviously, when a low-value resistor is fed too much current and fails "open," that is sometimes a useful function, and the multi-million-dollar fuse industry thus serves to protect us from trouble. But the fuses themselves can cause a little trouble. They don't always blow exactly when we wish they would. As Ian Sinclair put it in his book Passive Component User's Guide. "If you thought that a 1A fuse would blow when the current exceeded 1 A, then you have not been heavily involved in choosing fuses." (Ref. 1) Fuses are generally guaranteed to carry 100% of their rated current indefinitely, and most will carry 120% for several hours.
Even the fast-blow ones cannot open up much faster than 10 ms if overloaded by 1OX their rated load, or 100 ms if overloaded by 2X. You may be able to get faster response than that if you shop for the new semiconductor-rated fuses with very fast blow time. If somebody in your organization-a components engineer or an old-timer-can help you find the right information in a catalog of fuses, he can save you a lot of time. Without that kind of help, you will probably not be able to find a catalog from a fuse maker, or to figure it out when you get it. The curves of various ratings are a little obscure until you get used to them.
You may not use fuses much-modem solid-state circuits have such good current-limiters and thermal limiters that you may not see fuses every day. So when you do see fuses, you may be surprised. The low-current ones act pretty soft-resistive.
Some fuses just happen to fail unprovoked. The one in my clothes-dryer fails every 3 or 4 years, leaving my wife perplexed. Finally I wrote down the list of symptoms, so any time the fuse goes out and there is no heat, we at least save time by recognizing the symptoms. When my microwave oven quit working recently, I was a little concerned because the label on its back said, "No user-serviceable components inside." When I opened it up, there was a fuse clip with a blown fuse. After shopping unsuccessfully at several electrical supply houses, I finally went in to a Radio Shack. They had them and, I realized, that was the first place I should have gone. I replaced the fuse and turned on the power-would the fuse blow for a good reason, or had the old fuse just fatigued out? The new fuse has held for several months, so it was just a fatigue failure.
Most fuses are fully rated for 115 or 230 VAC, but not more than 32 V of DC. That's because the alternating current flow gives time for an arc to be extinguished, which would not happen with DC. So for high-voltage DC, the answers aren't so simple. Some circuit breakers are rated for as much as 65 VDC, but often that's not enough. There is a CD Series that is good up to 125 VDC, and a larger GJ Series that is rated up to 150 V, available from Heinemanm. Another approach in circuits with rectified power is to put the sensing coil in the DC circuit, but connect the breaker into the AC circuit. That's no help if you just have a 120-V battery supply.
These days, high-powered MOSFETs can be used to make such a good high-voltage high-current switch that you can build your own fast-turn-off switch, activated by over-current-an electronic equivalent of a fuse. I built one of these-and it didn't work very well, the first time. It blew out the FET. Twice. I haven't really given up on it, and when I get some time after I put this Guide to bed, I'll go back and get it running. When I get it running, I'll publish it somewhere where you can all see it: The equivalent of a solid-state fuse that can handle as much as 200 V of DC. Meanwhile, when you need some fuse protection on a DC power supply, just put a fuse in the secondary of the transformer, so it sees AC current flow and AC voltage, rather than DC.
Inductors and Transformers Aren't So Simple
Inductors and transformers are more complicated than resistors-nonlinearity is rife.
Their cores come in many different shapes and sizes, from toroids to pot cores and from rods to stacks of laminations. Core materials range from air to iron to any of the ferrites. I am not going to presume to tell you how to design an inductor or transformer or how to design circuits that use them, but I will discuss the kinds of trouble you can have with these components. For example, you can have a good core material; but, if there is an air gap in the core and you don't carefully control the gap's width, the energy storage and the inductance of the component can vary wildly. If someone has substituted a core of the wrong material, you may have trouble spotting the change; an inductance meter or an impedance bridge can help. But even with one of those tools, you're not home free.
For most inductors and transformers with cores composed of ferromagnetic materials, you had better make sure that the test conditions--the AC voltage and the frequency that the measuring instrument applies to the device under test--closely approximate those the component will see in your real-life application. If you fail to take such precautions, your inductance measurements stand a good chance of seriously misleading you and making your troubleshooting task much more frustrating.
The phenomena you are likely to run into as a result of incorrect test conditions include saturation, which can make the inductance look too low, and core loss, which can lower the Q of an inductor. For transformers, make sure you understand which of the inductances in the device's equivalent circuit you are measuring.
Equivalent Circuits Demystify Transformers
You can represent a transformer with a turns ratio of N as a "T" network. N equals N1/N2, where N, is the number of secondary turns and N, is the number of primary turns. However. if you plan to make measurements on transformers, it's helpful to keep the equivalent circuit shown in FIG. 5b in mind. For example, the inductance you measure between terminals A and B is quite large if you leave terminals C and D open, but the measured inductance will be quite small if you short terminals C and D together. In the first case, you are measuring the mutual inductance plus the leakage inductance of the primary. But because the leakage inductance is normally much, much smaller than the mutual inductance, you are measuring the leakage inductance of the primary plus the reflected secondary leakage in the second case.
When you work with inductors or transformers, you have to think in terms of current: In any transformer or inductor, flux is directly proportional to the current, and resistive losses are directly proportional to the current squared. Therefore, be sure to have several current probes, so you can observe what the current waveforms are doing. After all, some of the weirdest, ugliest, and most nonideal waveforms you'll see are the waveforms associated with inductors. (Especially in a switch-mode regulator. . . )
In the absence of an instrument designed to measure inductance, parallel the inductor with a known capacitance to create a parallel resonant circuit. If you use a high-impedance source to apply a current pulse to this circuit, you can determine the inductor's value from the resonant frequency and the capacitance: f = 1/ (2 pi _/LC). If you look at the inductor's waveform on a scope, you can compare it to the waveform you get with a known good inductor. This technique is also good for spotting a shorted turn, which reduces inductance nearly to zero. The L meter and the similar Q meter can help you ensure that good inductors haven't been damaged by saturation.
Incredible as it may sound, you can permanently damage an inductor by saturating it. Some ferrite toroids achieve their particular magnetic properties by means of operation at a particular point on the material's magnetization curve. Saturating the core can move the operating point and drastically change the core's magnetic properties.
The likelihood of your being able to return the material to its original operating point is small to nonexistent. In other cases, as a result of applying excessive current, the core temperature increases to a point where the core's magnetic properties change irreversibly. Regardless of the mechanism that caused the damage, you may have to do as I once did-package the inductors with a strongly worded tag to demand that nobody test them at Incoming Inspection.
Bob Widlar had a good solution to that. He would instruct the Incoming Inspection Technician to count the number of leads. Don't measure anything, just count the number of leads. If they follow that instruction, they probably won't wreck the transformer.
If you choose too small a wire size for your windings, the wire losses will be excessive. You can measure the winding resistance with an ohmmeter, or you can measure the wire's thickness. But if the number of turns is wrong, you can best spot the error with an L meter-remember that L prop. to N2. Be careful when using an ohmmeter to make measurements on transformers and inductors--some ohmmeters put out so many milliamps that they are likely to saturate the component you are trying to measure and at least temporarily alter its characteristics. Select an ohmmeter which puts out only a small amount of current.
Considering the effect of each component will help you understand the results of your measurements.
Protect Transistors from Voltage Kick
There is one trouble you can have with an inductor or relay coil that will not do any harm to the magnetic device, but will leave a trail of death and destruction among its associated components: When you use a transistor to draw a lot of current through an inductor and then turn the transistor off, the "kick" from the inductor can generate a voltage high enough to damage or destroy almost any transistor. You can avoid this problem by connecting a suitable snubber, such as a diode, an RC network, a zener, or a combination of these components, across the inductor to soak up the energy. The use of a snubber is an obvious precaution, yet every year I see a relay driver with no clamp to protect the transistor. The transistor may survive for a while, but not for long.
The tiniest inductors are called beads. They are about the same size and shape as beads worn as jewelry, they are available in various types of ferrite material, and they have room for only one or two or four turns of wire. Beads are commonly used in the base or emitter of a fast transistor to help keep it from oscillating. A bead not only acts inductive but also acts lossy at high frequencies, thus damping out ringing. In general, the choice of a bead is an empirical, seat-of-the-pants decision, but designers who have a lot of experience in this area make good guesses. This topic is one that I have not seen treated (except perhaps one sentence at a time) in any book or magazine. You'll just have to get a box of ferrite beads and experiment and fool around.
Transformers usually are susceptible to the same problems as inductors. In addition, the turns ratio may be wrong, or the winding polarity might be incorrect. And, if your wire-handling skills are sloppy, you might have poor isolation from one winding to another. Most ferrite materials are insulators, but some are conductive.
So, if you've designed a toroidal transformer whose primary and secondary windings are on opposite sides of the toroid and you scrape off the core's insulating coating, you could lose your primary-secondary insulation. If the insulating coating isn't good enough, you might need to wrap tape over the core.
Fortunately, it's easy to establish comparisons between a known-good transformer and a questionable one. If you apply the same input to the primaries of both transformers, you can easily tell if the secondaries are matched, wound incorrectly, or connected backwards. If you're nervous about applying full line voltage to measure the voltages on a transformer, don't worry--you can drive the primary with a few volts of signal from a function generator (preferably in series with a resistor and/or a capacitor, to prevent saturation and overload) and still see what the various windings are doing.
Two general problems can afflict power transformers. The first occurs when you have large filter capacitors and a big high-efficiency power transformer. When you turn the line power switch on, the in-rush current occasionally blows the fuse. You might install a larger value of fuse, but then you must check to make sure that the fuse is not too high to offer protection. As an alternative, you could specify the transformer to have a little more impedance in the secondary: Use smaller wire for the windings or put a small resistor in series with the secondary.
Another approach, often used in TV sets, is to install a small negative-TC thermistor in the line power's path. The thermistor starts out with a nominal impedance.
so the surge currents are finite. But then the thermistor quickly heats up, and its resistance drops to a negligible value. Thus, the efficiency of the circuit is quite good after a brief interval. If the circuit is a switch-mode power supply, the control IC should start up in a "soft-start" mode. In this mode, the IC makes sure the switcher won't draw any extreme currents in an attempt to charge up the output capacitors too quickly. However, you must use caution when you apply thermistors for in-rush current limiting: Beware of removing the input power and then re-applying it before the thermistors have had a chance to cool. A hot thermistor has low resistance and will fail to limit the current; thus, you are again likely to blow a fuse-or a rectifier.
The second general problem with a line transformer occurs when you have a small output filter capacitor. In our old LM317 and LM350 data sheets, we used to show typical applications for battery chargers with just a 10 uF filter. Our premise was that when the transformer's secondary voltage dropped every 8 ms, there was no harm in letting the regulator saturate. That premise was correct, but we began to see occasional failed regulators that blew up when we turned the power on.
After extensive investigations, we found the problem in the transformer: If the line power switch was turned off at exactly the wrong time of the cycle, the flux in the transformer's steel core could be stored at a high level. Then, if the line power switch was reconnected at exactly the wrong time in the cycle, the flux in the transformer would continue to build up until the transformer saturated and produced a voltage spike of 70-90 V on its secondary. This spike was enough to damage and destroy the regulator. The solution was to install a filter capacitor of at least 1000 pF, instead of just 10 pF. This change cut the failure rate from about 0.25% to near zero.
Another problem occurred when the LM317 was used as a battery charger. When the charger output was shorted to ground, the LM317 started drawing a lot of current.
But, the transformer's inductance kept supplying more and more current until the LM317 went into current limit and could not draw any more current. At this point, the transformer's secondary voltage popped up to a very high voltage and destroyed the LM317. The addition of the 1000 pF snubber also solved this problem.
Inductors, Like Resistors, Can Overheat
How do you spot a bad inductor or transformer? I have already discussed several mechanisms that can cause the inductance or Q of an inductor to be inferior to that of a normal part. And, as with a resistor, you can smell an inductor that is severely overheating. Overheating can be caused by a faulty core, a shorted turn, incorrect wire gauge, or anything else that causes losses to increase. An open winding is easy to spot with an ohmmeter, as is a short from a primary to a secondary. If the pattern of winding has been changed from one transformer to another, you may not see it unless you test the components in a circuit that approximates the actual application.
However, you may also be able to see such a discrepancy if you apply a fast pulse to the two transformers. Changes in winding pattern--even clockwise vs. counterclockwise--have been observed to cause significant changes in transformer performance and reliability.
Tightly-coupled windings, both bifilar and twisted pairs, have much better magnetic coupling and less leakage inductance than do well-separated primary and secondary windings. As the magnetic coupling improves, the capacitance between windings increases--but high capacitance between windings is often an undesirable effect in a transformer. An experienced transformer designer weighs all the tradeoffs and knows many design tricks--for example, the use of special pi windings and Litz wire. Mostly, you should know that these special techniques are powerful; if you ask the transformer designers the right questions, they can do amazing tricks.
I recently read about an engineer who designed an elegant shield made of mu-metal. However, the shield was difficult to install, so the technician had to tap on it with a hammer. When the engineer operated the circuit, the shielding seemed nonexistent-as if the shield were made of cardboard. After a lot of studying, the engineer realized that the mu-metal--which costs about $2 per 15 square inches, the same as a $2 bill--had been turned into perfectly worthless material by the pounding and hammering. In retrospect, the engineer had to admit that the mu-metal, when purchased, was prominently labeled with a caution against folding, bending, or hammering. So remember, in any area of electronics, there are problems with inductors and magnetic materials that can give you gray hair.
Consider the Effects of Magnetic Fields
One problem recently illustrated the foibles of inductor design: Our applications engineers had designed several DC/DC converters to run off 5 V and to put out various voltages, such as +15V and -15 V DC. One engineer built his converter using the least expensive components, including a 16-cent, 300 FH inductor wound on a ferrite rod. Another engineer built the same basic circuit but used a toroidal inductor that cost almost a dollar. Each engineer did a full evaluation of his converter: both designs worked well. Then the engineers swapped breadboards with each other. The data on the toroid-equipped converter was quite repeatable. But, they couldn't obtain repeatable measurements on the cheaper version. After several hours of poking and fiddling, the engineers realized that the rod-shaped inductor radiated so much flux into the adjacent area that all measurements of AC voltage and current were affected.
With the toroid. the flux was nicely contained inside the core. and there were no problems making measurements. The engineers concluded that they could tell you how to build the cheapest possible converter, but any nearby circuit would be subject to such large magnetic fields that the converter might be useless.
When I am building a complicated precision test box, I don't even try to build the power supply in the main box because I know that the magnetic fields from even the best power transformer will preclude low-noise measurements, and the heat from the transformer and regulators will degrade the instrument's accuracy. Instead. I build a separate power-supply box on the end of a 3-ft cable: the heat and magnetic flux are properly banished far away from my precision circuits.
|Top of Page||PREV.||NEXT||Related Articles||HOME|