When we say a loudspeaker is dead, or 'feeling slightly under the weather', invariably we really mean that something nasty has befallen the actual drive units. Despite flippant advertising slogans, most readers will have experienced the dramatic silence that signals the demise of one drive unit or more. Musicians in particular are wary of the event, not knowing quite how or why. A direct result of this trepidation is to fail to the discover the full potential of your speakers. All this is unnecessary: with a little skill and knowledge, drivers can be safely pushed to their limits without prejudicing a long and healthy cone-beat. We'll come back to look at power handling in detail later, but for the moment, let's concentrate on accidental causes of death.

The DC Fault

Scenario: when you turn on your monitoring system, there's a sharp crack. Looking at one of the bass drivers, you notice the cone is well forward from its usual position. You might also hear a loud hum. If the hum promptly ceases, your bass drivers are probably dead. Whether this is the case or not, your amplifier is certainly naff, and should be switched off immediately. Better still, there's a chance of saving the more rugged, high-power bass drivers (a la 15") by tugging out the speaker connections: do this quickly, for every fraction of a second counts!

In case you haven't already guessed, this is the infamous DC fault. Virtually all transistor amplifiers (including MOS-FET types) operate from supply voltages symmetrical about the signal ground, which is also the output ground going to one side of the speaker. (Figure 1 makes this clear.) In electronics parlance, the supply voltage wires distributing the power are known as rails, and in this instance, we're discussing +/-50 volt rails - a fairly typical voltage encountered with high-power amplifiers. That's 50 volts above (+50 volts) and 50 volts below (-50V) the grounded centre rail. Now the other side of the speaker is connected to the 'hot' side of the output terminals, which, going inside the amp, is somewhat precipitously balanced between the two opposing supply rails, hence 0 (zero) volts.

When a signal is present, this midpoint is modulated up and down either side of zero to produce a peak signal of any size 'up to rail', ie. nearly up to +/-50V in this instance, according to how hard we drive the amplifier. This is a very elegant way of designing amplifiers, as it avoids placing a large and messy capacitor in series with the output. However, it's this avoidance that causes the problem: under normal conditions, no significant DC appears across the speaker - only the alternating signal voltage, but as soon as anything drastic happens to the amplifier's circuitry (any one of a number of faults), the midpoint can be flung off balance. More often than not, one of the output transistors will fail, becoming a short circuit which connects the midpoint almost directly to one of the supply rails. The speaker is then subjected to, say, +50 volts DC, though it could equally be -50 volts.

This is equivalent to over 300 watts of continuous dissipation into a nominally 8 ohm drive unit. However, for DC voltages, it's typically 5½ ohms for an 8 ohm unit, so the potential power dissipation is higher, at around 400 watts, assuming the supply rail voltage droops slightly, in view of the heavy load placed upon it. An amplifier with this supply rail voltage, will, incidentally, normally provide about 100 watts into 8 ohms. So, as a rule of thumb, the power dissipated by the hapless drive unit with a DC fault at large will be around four times the nominal maximum; small wonder that DC faults are so frequently fatal...

Scope

DC faults generally afflict bass drivers only in speakers employing passive crossovers; the top, midrange (but NOT the mid bass unit if it's a 2 way speaker) drivers are protected by the series capacitor inherent in the crossover network (Figure 2). In active systems, it's wise to assume that all drive units are directly connected to their respective power amplifiers, unless the manufacturer states otherwise; some tri-amped speakers already feature capacitors wired in series with the midrange and/or top-end drivers for this purpose (Figure 3).

Precautions

As with a lot of risk, you either adopt a belt-and-braces approach, or just take basic precautions. Much depends on how many times you've suffered this particular disaster: some people seem to have more than their share of ill-luck with amplifiers going DC! The most elementary precaution is to plug in the speakers after switching on, for experience suggests that amplifiers frequently go DC when they're first powered-up. If, on first plugging in, there's a loud hum (or at least one different from the usual), and/or you can hear the cones 'pop'; you can then speedily unplug and confirm your suspicions with a test meter.

Reading across the speaker terminals on DC volts, anything over 0.5 volts is unusual, and any reading over 5V indicates a definite fault. On the other hand discernable reading signifies everything is OK. You can also avoid using suspect amplifiers. That means one that's recently been involved in some sort of speaker failure (until it's been thoroughly checked out), or older equipment in - er... less than first class condition.

A few amplifiers incorporate protection against DC faults. Optional DC protection cards are, for example, available for the Quad 405-11, and the late HH V800. Protection may involve a crowbar - circuitry which shorts the output to ground if any excessive DC voltage appears. Alternatively, DC sensing circuitry disconnects the output terminals via heavy-duty relays.

DIY Protection

If your amplifier lacks these facilities, external protection can be readily added. For tweeters and midrange drivers in active speaker systems, you simply wire capacitors in series with the signal path (Figure 4). The capacitor not only blocks the passage of DC, there's also a highpass filtering action which limits the bass levels that can be fed into the drivers (especially the tweeter!) should you mix up the output leads from the amplifiers.

At the same time, the breakpoint - the frequency where the filtering action begins to take effect - should obviously lie well below the crossover frequency, to avoid upsetting the slopes. As a rule of thumb, if the tweeter comes in at 3kHz, the breakpoint set by the DC protection capacitor should be around 1kHz for a -24dB/octave crossover slope. Overall, this means using larger than anticipated values, for example, 22uF for an 8 ohm tweeter, or 220uF for an 8 ohm midrange unit with a 300Hz lower crossover point. For 4 ohm impedances, you should double these values; for 15 or 16 ohms, halve them.

For best results, plastic film (non-polarised) capacitors should be used - polyester, polycarbonate or polypropylene types in order of ascending preference. On account of their relative expense, these types aren't feasible for overall values much above 30uF, but they are a sensible buy for top-end protection, and this, fortunately is the place where they're needed most.

The largest commonly available value in plastic film is 4.7uF, so to make up 22uF approximately, you'll need to wire five in parallel. Exact values aren't important, and for lowest cost, try to make most use of the larger values, as these are proportionally cheaper. The overall cost should not exceed £15, which is a good investment if it saves the price of just one replacement tweeter, not to mention shipping costs, no sound for a week, and the general hassle involved.

From a practical viewpoint, the capacitor array must be securely mounted. One way is to glue the caps upside down onto a small piece of plywood. The leads can then be bent over in two rows, and soldered (Figure 5); this arrangement also provides a secure termination for the leadout wires.

Alternatively, you could make a simple PCB, using small tie-wraps and/or adhesives to make the assembly vibration proof. In either case, you simply screw the baseboard to the inside of the cabinet, and wire one side of the tweeter (usually the positive terminal) in series.

Midrange

For midrange drivers, values around 220uF or larger make electrolytic capacitors the only sensible proposition. Electrolytics are undoubtedly capable of afflicting the sound quality if used or chosen carelessly, so follow these hints:

1) Avoid 'Non-polarised' electrolytics; some are good, but many are of execrable quality.

2) Instead, use conventional electrolytics of top-quality manufacture. Many of the modern Japanese types are extremely good. These should be wired back-to-back (Figure 6) to enable them to handle the alternating polarity of signals. If you do not do this, they are likely to explode because of the series connection, you will need to double the value of each capacitor to maintain the original overall value. So, for 220uF (overall), you would use two 470uF capacitors.

3) Electrolytic capacitors are not unduly expensive, so use the highest rating you can find, eg. 160 volt. This promotes longevity; there's little use in DC protection if it is likely to fail. And along the same lines...

4) Don't be afraid to use larger values than you calculated if this is convenient; it will do no harm to even double the overall value(s).

5) If you envisage doing any low level monitoring, it is a good idea to bias the capacitors. You can do this by wiring their centre-point via a network to the power amplifier's +ve supply rail (Figure 5). In this instance, it'll be most convenient to mount the protection capacitors inside the dedicated midrange amplifier. If you're not sure how to do this, ask a competent engineer to help you. And don't forget the need to double the voltage rating you first thought of.

Guarding Woofers

Bass-end protection is needed for all speaker systems (passive or active) and can take one of several forms. To begin, a pair of 4700uF or better, 10,000uF capacitors wired back-to-back will provide protection of the kind we've just discussed. However, many speakers, especially Thiele and horn-loaded types will exhibit a change in the bass sound under these circumstances, so other means must be sought.

Fusing is one obvious possibility, but for discrimination. That means NEVER blowing a fuse in response to a genuine music signal however loud or bass-ridden, whilst ALWAYS blowing in the event of a real DC fault. Considering the wide tolerance of most fuses, this isn't so simple as it sounds, and the only satisfactory way to discover the right fuse value for a particular amp/speaker combination is to carry out a series of experiments, selecting the lowest value that never blows in response to music with lots of low bass, and testing several samples of this fuse value to confirm that no nuisance blowing will occur. There's also a problem if you discover that an 8.5A fuse is what's needed - you won't find a value like this at all easily.

If this sounds like a lot of trouble, a third possibility is to install a crowbar. This circuit connects across the driver, so unlike series capacitors, it has no effect on bass performance. Essentially, the crowbar involves thyristors which clamp the voltage very quickly, when excess DC is detected causing a massive current surge which guarantees rupturing of a conservatively rated fuse of nominal value. We hope to present a DIY crowbar circuit in a future edition of HSR.