Ailing Electrolytics...

I don't think it's an exaggeration to say that in the last few years a kind of electronic epidemic has been taking place: electrolytic capacitors used mainly for high-frequency filtering and bypassing in switch mode power supplies (SMPSs), the deflection circuitry of TVs and monitors, and similar applications have been failing at an increasing rate, and causing all sorts of trouble.

Typical symptoms of electrolytic capacitor failure I've encountered are: the playback picture of VCRs beginning to disappear behind a curtain of tiny dots, or the system microcontroller continuously resetting with accompanying beeps, right up to 'hiccuping' and mysterious self-destruction of SMPSs, involving the destruction of many expensive semiconductors.

For example, National Panasonic NV-G20A and NV-H65A VCRs (Australian models) and their relatives are notorious for problems and failures due to the electrolytics in their hot-running SMPSs giving up the ghost. I've seen the digital display of an electronic cash register flickering and sometimes disappearing, and TV pictures becoming distorted or spoiled by the appearance of flashing Teletext lines (digital signals) at the top, and in each case the problem was tracked down to defective electrolytic capacitors.

Testing Times

As a service technician (reluctantly- I'd much rather be a full-time designer!), I was just-about tearing my hair out because of the difficulty involved in identifying faulty electrolytic capacitors, and proving which ones were still OK, in SMPSs and other equipment.

My primary technique was to gingerly poke around with an oscilloscope's probe in live circuitry, often full of components at mains (it's 240V in Australia!) or other dangerous voltages, looking for high ripple voltages and/or abnormal waveforms. And all the time keeping an eye on the probe's ground lead clip to make sure it didn't get loose and touch something expensive!

When it was physically impossible to make measurements under operating conditions and I suspected defective electrolytics, I didn't have much choice but to replace every electrolytic in the problem area of the circuit. The first method causes high adrenalin levels, while the second wastes time and money!

High ESR is the problem! I figured there HAD to be a better way! Over a period of months I collected a bag full of electrolytics from SMPSs which had returned to normal operation with new capacitors, then I did some careful tests on them.

Most obviously, none of them had significant electrical leakage. Perhaps surprisingly, a digital capacitance meter indicated that although some of them were a bit below their rated value, most were still within the normal value tolerance for electrolytics, of about -20 to +50%.

The real fault with many of these electros was not so much that their capacitance had dropped (as many people assume), but their Equivalent Series Resistance (ESR) had risen to a high value, seriously degrading their effectiveness as filtering, bypassing and coupling components.

ESR...? So what exactly is an electrolytic capacitor's Equivalent Series Resistance, and why is it so important to the correct operation of associated circuitry?

As you probably know, electrolytics are 'wet' devices in the sense that their operation depends on a water-based electrolyte, soaked into a strip of porous material between the alumin(i)um foil plates. This completes the 'outer' electrical connection to the alumin(i)um oxide dielectric, which coats the anode foil.

Unhappily this layer of electrolyte has electrical resistance which, along with the (negligible) resistance of the connecting leads and alumin(i)um foil plates, forms the capacitor's Equivalent Series Resistance. Under normal conditions the ESR has a very low value which stays that way for many years unless the rubber seal is defective, in which case the electrolyte's water component gradually dries out and the ESR creeps up with time.

Deterioration Acceleration

But if an electro is subjected to high temperatures, especially from heat generated internally by large ripple currents, the electrolyte will start to decompose and the dielectric may deteriorate, causing the ESR to increase far more rapidly. To make things worse, as the ESR increases, so does the amount of internal heating caused by ripple current. This can lead to an upward spiral in the capacitor's core temperature, and the electrolyte actually boiling.

In fact the service life of electrolytic capacitors is approximately halved for every 10 degrees C increase in temperature, and I was very surprised to learn that many electrolytics are designed for a reliable operating life of only a few thousand hours at their maximum rated temperature and ripple current. (Remember that a single year is 8,766 hours!)

SMPSs place some of the most severe stresses on filter capacitors. Because of their compact construction, temperatures are high (that's why your PC's power supply is equipped with a fan), and the capacitors are subjected to large amounts of high-frequency ripple current.

Effects of ESR

Regardless of a capacitor's actual capacitance value, ripple current flowing through its equivalent series resistance causes an AC ripple voltage to be superimposed on its DC voltage, and as the ESR increases, so does the amplitude of the ripple voltage. Low capacitor ESR is crucial to the correct operation and regulation of SMPSs in particular, and filter capacitors with high ESR can allow high frequency noise to escape and find its way into all sorts of signal processing circuitry.

ESR meters unavailable?

Knowing that I badly needed an instrument to measure the ESR of electrolytics, preferably while they were still in circuit and with the power safely turned off, I explored the catalogs here in Australia for such a device. Although I succeeded in finding a large variety of transistor testers and even a crystal checker, I found absolutely nothing to measure what I wanted (which seemed strange because electrolytics fail far more frequently than transistors and crystals put together!)

Eventually I accepted that I was going to have to design my own ESR meter from scratch, incorporating all the features I'd like it to have, and this project is the final result of all my subsequent efforts!

Microcontroller-based

After a few fairly unsuccessful attempts at designing an ESR meter around analog circuitry, I realized that this was an ideal application to take advantage of the speed and versatility of a Zilog Z86E0408 microcontroller, using a simple pulsed measurement technique. The Z86E0408 already has two comparators and two flexible counter/timers built in, so all I additionally needed to complete a full-featured ESR meter with digital readout was one cheap CMOS IC, a 5V regulator, a couple of dollars worth of transistors, and a handful of passive components. But of course the complex part of the design is hidden away in the microcontroller's software!

Features

All the circuitry is contained on a single PCB, bolted to the lid of a compact 'Jiffy box', which it shares with a 9V alkaline battery. This instrument has three ESR ranges: 0 to 0.99 ohm (0.01 ohm resolution), 0 to 9.9 ohms (0.1 ohm resolution), and 0 to 99 ohms (1 ohm resolution).

The micro always selects the most appropriate range automatically, so your hands are free to hold the test leads on the electro under suspicion. A single '-' on the left-hand display indicates a reading in excess of 99 ohms, and the accuracy of the prototypes was better than 5% of displayed reading, +/- one digit.

Readout is on two big 0.5 inch (12.7mm) seven-segment LED displays plus two 3mm decimal point LEDs, needed because the decimal points of the displays are on the wrong side for this application. Besides, 3mm decimal points are very hard to overlook, even after a long day's servicing!

Auto switch off

If you forget to turn the power off, the micro itself will do it for you when the displayed reading has remained the same for two minutes. But since this circuit can be powered by a plugpack (DC power supply), and automatic switch off then would be more of a nuisance than a help, this feature can be disabled.

Low battery warning

When the battery voltage is nearly too low for the 5V regulator to function correctly, the micro 'latches' into a 'low battery' mode until the power is switched off. During this time the power to the LED displays is reduced by 50% to minimize the battery load and give you a bit more operating time, while a 'b' flashes on the right-hand LED display once per second in the 'off-scale' condition, to warn you to look for a new battery.

Operation

Using the ESR and Low Ohms Meter couldn't be much simpler: The single push button has three functions: one press turns the power on, and another press will switch it off again if the measured resistance is one ohm or more. A push of the button with the leads shorted together will cause their resistance value to be subtracted from all subsequent readings, as long it's less than 1 ohm.

Then all you need to do is make sure the electrolytic under test is discharged (I find a 100 ohm 5W resistor does this well), and connect the test leads to it either way around, as the meter doesn't output a DC voltage. If your capacitor is still in circuit (with the power OFF!), you're likely to get quite an accurate reading because it should be the component with the far lowest high-frequency impedance, which is what this meter is looking for.

Due to the timing of the measurement pulses, any small non-electrolytic bypass capacitors in parallel with the one you're measuring will have little effect on the accuracy of the reading. The test signal coming from the circuit has a peak open-circuit voltage of 600mV (maximum 100mV probe-tip peak at full-scale reading), so it won't make diodes or transistors conduct and cause erroneous readings.

The front panel table

Once you have a reading, refer to the front-panel table to get an idea of whether your electrolytic's ESR is about normal or above it. The approximate worst ESR figures on the table were taken from the Nippon Chemi-Con Aluminum Electrolytic Capacitors Catalog Number 4, for their 'SXE', 'SXG' and 'LXA' capacitor series. They're actually 100KHz impedance figures, but I've measured the ESRs of lots of new electros of assorted brands, styles, sizes and ages, and the readings I obtained agree pretty closely with them.

Due to limited space, the table only lists a broad 'sprinkling' of capacitances and voltages, but it will enable you to estimate whether the ESR of an unlisted capacitor is acceptable or excessive.... From my experience when doing actual fault-finding with the prototypes, an electro's ESR needs to be many times the table value before it's likely to cause problems.

Low ohms uses

During development of this little instrument, it quickly became obvious that it's very handy for measuring low values of resistance, too. The only catch is that because this meter uses a pulsed measurement technique, it can't give you a sensible indication of the DC resistance of inductive components such as transformer windings or chokes. I've already used it to 'roll my own' low value resistors by simply measuring off the required number of (milli) ohms of resistance wire, which I then formed into a spiral by wrapping it around the shank of a drill. Very convenient if you need to replace a burnt-up 0.33 ohm 2W power amplifier emitter resistor on a Sunday evening!

You can also use it to locate short circuits on printed circuit boards by measuring the actual track resistance. If the reading increases as you probe further along the track, you know you're going in the wrong direction! This technique is a bit more civilized than locating shorted components by connecting a high-current power supply to a PCB's power tracks to see what catches fire and/or explodes (or simply goes open-circuit so you can't find it)! You could also use it to confirm solid continuity of the power lead earth conductor on mains-operated equipment, etc etc....

Don'ts

There are a few minor points to keep in mind when using this meter.... First, it's quite useless for identifying leaky and short-circuited capacitors; that's what the resistance ranges of normal multimeters are for! Also, avoid using those self-retracting spiral-wound test leads, as their inductance can cause small measurement errors. Lastly, don't use it right next to an operating TV set or computer monitor because the high amplitude pulses radiated by the horizontal output stage can be picked up by the test leads (and the person holding them!) and cause unstable readings.