Fullscreen HTML5Med Res (??MB)HTML4

Don’t let this happen to you! Build a . . . Loudspeaker Protector T HE SPEAKER in the above photo is well on fire and this could be the result of a relatively minor fault in an audio amplifier. As anyone who has been following our series on the Ultra-LD Mk.3 Amplifier would be well aware, big power amplifiers have big power supplies and so a relatively 34  Silicon Chip minor fault in one channel of a stereo amplifier can result in a large DC voltage being applied to one of your precious (and expensive) loudspeakers. Once that happens, your speaker is kaput! If you’re lucky, the woofer’s voice coil will quickly burn out and that will be the end of it – it’s unlikely that the amplifier’s fuses will blow in time to save your loudspeaker. You will then probably have to scrap the loudspeaker although if you are lucky and if it is economic, you might be able to have it repaired. But you might not be that fortunate and the consequences could be a siliconchip.com.au D1 1N4004 A K CON1 *20V DC + INPUT – 33k 100k *SEE TEXT FOR HIGHER 33k 100k VOLTAGE OPTIONS CON2 AC SENSE INPUT (50V AC MAX.) D2 1N4004 A K 100 10k B 0.5W A K C E Q1 BC546 100 C B E 470nF MKT 12k E Q2 BC546 Q3 BC546 D4 1N4148 A K D5 1N4004 K 47 F 63V D3 1N4004 10 A ZD1 12V 1W POWER-UP DELAY LOSS-OF-AC DETECTION Q4 BC556 C R2 0 (LINK) C B E B R1 2.7k K RLY1 24V 650 A RELAY DRIVER CON3 OVER-TEMP SWITCH INPUT FROM RIGHT AMP SPEAKER OUTPUTS TO RIGHT SPEAKER RSPKIN+ RSPKIN– Q5 BC546 C E 47k 1W RLY1a 47 F 50V NP RSPKOUT+ LSPKIN+ TO LEFT SPEAKER LSPKOUT– C C E E C C E Q7 BC546 B 22k LSPKIN– DC DETECTION Q8 BC546 C E 47k 1W RLY1b 47 F 50V NP Q9 BC556 B Q10 BC546 B 22k LSPKOUT+ BC546 BC556 DIODES (D1-D5 & ZD1) A 2011 B E RSPKOUT– FROM LEFT AMP SPEAKER OUTPUTS SC  Q6 BC556 K B E C SPEAKER PROTECTION & MUTING SUPPLY RAIL R1 20V 2.7k 0.25W R2 0  (LINK) 40V 4.7k 0.25W 470  5W 52.5V 8.2k 0.5W 820  5W 57V 8.2k 0.5W 1k 5W 70V 12k 1W 1.2k 10W Fig.1: each channel of the amplifier is monitored for DC faults by three transistors – Q5-Q7 for the right channel & Q8-Q10 for the left. If DC is detected, these pull Q3’s base low which then turns off Q4 and the relay to disconnect the speakers. Q1 & Q2 provide the switch-off muting feature, while CON3 accepts a normally-open temperature switch. whole lot worse. Now, instead of suddenly burning out, the voice coil stays intact and gets red-hot, as you would expect it to – it is dissipating many hundreds of watts. After all, voice coils are quite small and they normally operate in the very confined space of the speaker magnet’s voice coil gap. With a large DC fault voltage across the voice coil, the speaker will either jump forward out of the gap or jump back as far as it can go. The latter is probably the worse scenario since the voice coil can then get even hotter and soon sets the speaker cone on fire. siliconchip.com.au The sequence of photos shown elsewhere in this article show how the whole speaker cone can catch fire within just a few seconds. A few seconds later and those flames were producing copious amounts of smoke. If we hadn’t been on the spot to put the fire out by laying the speaker face down on the concrete, the fire could have spread to who knows where. If that happened in your home and you were not present to take very quick action, you could lose your home and everything in it. This sort of thing really can and does happen! Don’t let it happen to your stereo system. This loudspeaker protector and muting circuit can prevent such disasters. Main features Originally designed for the Class-A Stereo Amplifier describ­ed in the JuneSeptember 2007 issues of SILICON CHIP, the 2-channel loudspeaker protector described here (in slightly modified form) is also eminently suitable for use with the new Ultra-LD Mk.3 Amplifier module. In fact, it can be used with just about any audio amplifier, October 2011  35 Q5 BC546 C E B Q6 BC556 C B 7002 Q4 BC556 E R2* 5W/10W 33k ZD1 100k D4 63V 12V/1W D1 1N4004 100 B B C Q8 BC546 E C C E B BC546 1N4148 100 E 470nF + B Q9 BC556 B E Q2 C BC546 C B E Q1 22k BC546 C 1747 07 02 F 10 100k Q10 50V NP E 10 47k 1W 47 F C 1N4004 D3 OPERATION LINK FOR 20-24V D5 1N4004 33k Q3 BC546 12k E C 10k C D2 1N4004 22k BC546 50V NP 47 F 47k 1W +TUOKPSR LEFT SPEAKER OUT+ TO LEFT SPEAKER B B 1N 4148 LEFT SPEAKER IN+ E Q7 2.7k R1* 24VDC 10A +NIKPSR FROM LEFT AMP SPEAKER OUTPUTS -NIKPSR LEFT SPEAKER IN/OUT– RLY1 -NIKPSL RIGHT SPEAKER IN/OUT– +NIKPSL RIGHT SPEAKER IN+ FROM RIGHT AMP SPEAKER OUTPUTS +TUOKPSL RIGHT SPEAKER OUT+ TO RIGHT SPEAKER CON3 * SEE TABLE ON FIG.1 FOR RESISTOR VALUES OVER-TEMP SWITCH INPUT CON2 AC SENSE INPUT CON1 + – 20V DC POWER INPUT Fig.2: follow this layout diagram to install the parts on the board. Note that the final version of the PCB supports both double-ended spade connectors (attached using M4 screws & nuts – see Fig.3) and the solderable PC-mount vertical spade connectors as shown in the photo. Refer to the table on Fig.1 for the values of resistors R1 & R2. either mono or stereo. It provides the following functions: (1) it protects the loudspeakers against catastrophic failure in the amplifier, eg, if an output transistor goes short circuit or one supply fuse blows; (2) it provides muting at switch-on and switch-off, to prevent thumps from the loudspeakers; and (3) it provides an input for an overtemperature switch to disconnect the loudspeakers if the output stage heatsink rises above a certain temperature. In the latter case, disconnecting the loudspeaker from a class-B amp­ lifier immediately reduces the current through the output stage to the quiescent current setting. This is typically around 50-200mA, assuming that there’s no fault in the amplifier. So for a class-B amplifier, it makes sense to use over-temperature sensing. If the heatsink to which the output transistors are attached gets too hot, disconnecting the loudspeaker immediately reduces the dissipation to just a few watts, which allows the heatsink to cool. Protecting against fire As mentioned at the start of this article, by far the biggest reason for incorporating speaker protection into an amplifier is to prevent further damage and possible fire in the case of a serious amplifier fault. In the Ultra-LD Mk.3 Amplifier, the main supply rails are ±57V DC. As a result, if an output transistor fails (or if one side of the output stage turns hard-on due to a fault elsewhere in the amplifier) and there’s no loudspeaker protection, this could apply one of the full 57V DC rails to the loudspeak­er. If the on-board supply fuse didn’t blow fairly quickly (a strong possibility), the voice coil would quickly become red-hot. This risk applies to any audio power amplifier of more than about 40W per channel. Muting the thumps Muting switch-on and switch-off thumps is another important function of this unit. Switch-on thumps are eliminated by using a simple circuit to delay the Resistor Colour Codes o o o o o o o o o o No. 2 2 2 2 1 1 1 2 1 36  Silicon Chip Value 100kΩ 47kΩ 1W, 5% 33kΩ 22kΩ 12kΩ 10kΩ 2.7kΩ 100Ω 10Ω 4-Band Code (1%) brown black yellow brown yellow violet orange gold orange orange orange brown red red orange brown brown red orange brown brown black orange brown red violet red brown brown black brown brown brown black black brown 5-Band Code (1%) brown black black orange brown not applicable orange orange black red brown red red black red brown brown red black red brown brown black black red brown red violet black brown brown brown black black black brown brown black black gold brown siliconchip.com.au relay from turning on when power is first applied. This allows the amplifier modules to power up and settle down before the relay switches on (after about five seconds) to connect the speakers. By contrast, switch-off thumps are eliminated by using an “AC Sense” input to monitor the secondary AC voltage from the transformer (30VAC in the case of the Ultra-LD Mk.3 and up to 50VAC maximum). When this AC voltage disappears at switch-off, the circuit switches the relay off in less than 100ms. This is much faster than simply relying on the collapsing DC supply rail to turn to the relay off. In practice, this could take 0.5 seconds or more as the main filter capacitors discharge – more than long enough for any switch-off thumps to be audible. Circuit details Refer now to Fig.1 for the circuit details. It’s virtually the same as the circuit published in July 2007 with just a couple of minor modifications. We’ll come to those shortly. As shown, each channel of the amplifier is con­nected to the NC & NO (normally closed & normally open) con­tacts of a relay. The relay wipers and NC contacts then each respectively connect to the positive and negative loudspeaker terminals. Each channel of the amplifier is monitored for DC faults by a triplet of transistors – Q5, Q6 & Q7 for the right channel and Q8, Q9 & Q10 for the left channel. We’ll look at the operation of the right channel only, as the circuit for the left channel is identi­cal. As shown, the active signal from the amplifier’s right channel is fed to a low-pass filter consisting of a 47kΩ resistor and a 47μF 50V bipolar (BP or NP) electrolytic capaci­tor. The original circuit used a 2-pole filter consisting of two 22kΩ resistors and two 47μF bipolar capacitors in this position but we’ve modified it to a single-pole filter to achieve a faster response – see panel. The low-pass filter network removes any audio frequencies so that transistors Q5-Q7 simply monitor the output of the amplifier for DC voltages (if present under fault conditions). This is done because we don’t want audio signals to trip the protection cir­cuit. The low-pass filter output is connected to the emitter of Q5 and to the base of Q7. Q5 monitors the amplifi­er’s siliconchip.com.au This assembled PCB differs slightly from the version shown in Fig.2, as it suits the Altronics PC-mount vertical spade connectors. Modifying The Circuit For A Faster Response As stated in the main article, we modified the original Loudspeaker Protector circuit published in July 2007 to reduce the switch-off delay in the event of an amplifier fault. This was done by removing one of the two cascaded RC low-pass filters at each input, which means that the circuit now uses singlepole filters instead. In practice, this simply involved re­ moving a 22kΩ resistor and a 47μF bipolar capacitor from both the left and right inputs and replacing the resistors with wire links. The filter corner frequency is essentially unchanged as the resistor value in the remaining filter in each channel is doubled from 22kΩ to 47kΩ. The logic behind the original 2-pole filter was that the -12dB/octave slope allowed a higher -3dB point than is possible with a single-pole filter (-6dB/ output for negative DC signals while Q7 monitors for positive DC signals. In operation, Q7 turns on if a DC signal of more than +0.6V is present on its base. Similarly, Q5 turns on if a DC signal of more than -0.6V is present on its emitter. This in turn pulls Q6’s base low and so Q6 also turns on. Normally, in the absence of any am- octave). This should provide a faster response to DC faults while preserving the necessary large amplitude bass signal rejection. However, this ignores the 2-pole filter’s larger phase shift of 180° rather than 90°. The delay created by this phase shift more than offsets the benefit from the higher corner frequency! According to Douglas Self, the singlepole filter has a 78ms response time, compared to 114ms for the 2-pole filter (see Audio Power Amplifier Design Handbook, Fifth Edition, Chapter 17: “Amplifier and Loudspeaker Protection”). That does not include the relay switching time, which will be around 10ms. But it is clearly a worthwhile improvement and also reduces the parts count and the cost. So making the change is a “nobrainer”. plifier faults, transistors Q5-Q7 are all off and Q3 is biased on via the 100kΩ resistor connected between its base and the positive supply rail (ignore Q1 & Q2 for the time being). As a result, Q3 pulls Q4’s base down (via resistor R1) to just over 12.6V, as set by diode D4 and zener diode ZD1, and so Q4 and relay RLY1 are also on. Now let’s consider what happens October 2011  37 Temperature Sensors Temperature sensors are variously called “thermostat switches”, “thermal cutouts” and “thermal circuit breakers” and are available in a range of trip temperatures from 50-100°C. Both NO and NC (normally-closed) temperature sensors are available but the temperature sensor used with this unit must be a normally open (NO) type. A thermal cutout rated at 70°C (eg, Jaycar ST3831) is suitable for the Ultra-LD Mk.3 amplifier (and for earlier versions of this module). Alternatively, use the Altronics S5591 which is rated at 60°C. if an amplifier fault condition results in DC being present at its output. In this case, either Q6 or Q7 turns on and pulls Q3’s base low via a 10Ω resistor. And when that happens, Q3, Q4 and the relay all immediately turn off, disconnecting the speakers. Diode D5 protects Q4 by quenching any back-EMF spikes that are generated when the relay switches off. Transistors Q8, Q9 and Q10 monitor the left channel of the amplifier and they switch Q3, Q4 and the relay in exactly the same manner. Relay specifications The relay selected for the job is a 24V DC DPDT type with contacts rated at 10A. There are two reasons for this high contact rating. First, we want the contact resistance in the relay to be as low as possible so that it has negligible effect on the amplifier’s per­formance, as regards to distortion, damping factor and so on. Second, the relay contacts have to pass and break the heavy DC current which would otherwise flow through the loudspeaker if a fault occurs in the amplifier. However, we don’t merely use the relay to disconnect the amplifier’s output from the speakers. If we simply did this, it’s possible that the contacts would just arc across and so the heavy DC current would continue 38  Silicon Chip to flow through the loudspeaker. That might seem unlikely but when you have a heavy DC cur­rent, an inductive load and a high DC voltage pushing it along, it can be quite hard to break the circuit. This problem is solved by shorting the moving relay contacts to the loudspeaker ground lines (via the otherwise unused NC contacts) when the relays turn off. This diverts the arc current to chassis and ensures that the fuses blow on the amplifier. By the way, the specified relay (Altronics S-4313) has an in-built green LED that lights when the relay turns on. This lets you quickly check the status of the relay during testing but is not really necessary. Muting delay at switch-on Switch-on muting is achieved using a delay circuit. This consists of a 100kΩ resistor and a 47μF capacitor connected to Q3’s base, along with diode D4 and zener diode ZD1. When power is first applied, the 47μF capacitor is discharged and Q3’s base is held low. As a result Q3, Q4 and the relay all remain off. The 47μF capacitor then charges via the 100kΩ resistor until, after about five seconds, it reaches 13.2V. This forward biases Q3 which then turns on Q4 and the relay to connect the loudspeakers. This 5s delay is more than sufficient for the amplifier modules to achieve stable operating conditions. Switch-off muting Transistors Q1 & Q2 and diodes D2 & D3 provide the switch-off muting function. D2 & D3 rectify the AC voltage that’s fed to the “AC Sense” input (at connector CON2) from a transformer secondary winding. Provided this AC voltage is present, the rectified output forward biases Q1 and keeps it turned on. This holds Q2’s base low and so Q2 is off and Q3 functions normally. The 100kΩ resistor and the 470nF capacitor on Q2’s base form a time constant. This is long enough to ensure that Q2 remains off when Q1 very briefly turns off around the AC zero crossing points. However, if the AC signal ceases (ie, at switch off), Q1 immediately turns off and Q2 turns on and quickly discharges (within a millisecond or so) the 47μF timing capacitor via a 100Ω resistor. As a result, Q3, Q4 and the relay all turn off and the loudspeakers are disconnected, thereby eliminating any switch-off thumps. Over-temperature input Connector CON3 is the temperature sensor input. It relies on the use of a normally-open (NO) thermal switch that’s bolted to the heatsink used for the amplifier’s output transistors. Basically, this input is wired in parallel with transistors Q6 & Q7 (and Q9 & Q10) and it controls transistor Q3 in exactly the same manner. When the temperature reaches a preset level (set by the switch itself), the contacts inside the thermal switch close and pull Q3’s base low via a 10W resistor. As a result, Q3 turns off and this switches off Q4 and the relay. When the heatsink subsequently cools down, the thermal cutout opens again and Q3, Q4 and the relay turn on to reconnect the loudspeakers. In a stereo amplifier, two identical temperature sensors are used – one for each heatsink (assuming the power amplifiers use separate heatsinks). These are simply wired in parallel and connected to CON3. Power supply Power for the Loudspeaker Protection circuit is derived from a suitable DC rail within the amplifier. This can range anywhere from about 20V DC up to 70V DC. In the case of the Ultra-LD Mk.3 Amplifier, the +20V and 0V rails from the power supply board are used. The “AC Sense” signal is picked up directly from the AC terminals on the bridge rectifier on the same board. Note that the values shown for R1 & R2 on Figs.1 & 2 assume a 20-24V supply rail. If the DC supply rail is higher than this, then resistors R1 and R2 must be changed to ensure a base current of about 3-5mA for Q4 (as set by R1) and to ensure that no more than about 24V DC is applied to the relay (set by R2). In the latter case, it’s just a matter of selecting R2 so that the relay current is about 37mA (the relay has a coil resistance of about 650Ω). The table included with Fig.1 shows the resistor values to use with a number of supply rails. Building it The parts for the Speaker Protection & Muting Module are all mounted on a PCB coded 01207071. This board is the siliconchip.com.au Attaching The Spade Lug Connectors Fig.3 (right) shows how the doubleended spade lugs are mount­ed. Each lug is secured using an M4 x 10mm screw, a flat washer (which goes against the PC board pad), an M4 star lockwasher and an M4 nut. The trick to installing them is to first do the nut up finger-tight, then rotate the spade lug assembly so that it is at a right-angle to the PC board. A screwdriver is then used to hold the M4 screw and the spade lug stationary while the nut is tightened from below using an M4 socket and ratchet. same as the one used in July 2007 – it’s simply a matter of leaving out the two unwanted capacitors and installing a couple of wire links (or 0Ω resistors) in place of the deleted 22kΩ resistors. The other two 22kΩ pull-down resistors (to the left of Q5 and to the left of CON3) remain in circuit. Fig.2 shows the revised PCB layout. Mount the resistors and diodes first, taking care to ensure that the diodes are all oriented correctly. Install a 2.7kΩ 0.25W resistor for R1 and a link for R2 if you are building the unit for the Ultra-LD Mk.3 Amplifier. Alternatively, select these resistors from the table shown in Fig.1 if you intend using a supply rail greater than 24V. If the supply rail is between the values shown in the table, simply scale the resistor values accordingly and use the nearest preferred value. The six double-ended spade lugs for the speaker input and output terminals are attached using M4 x 10mm screws, flat washers, star washers and nuts. Fig.3 shows the details. Note that, ideally, the double-ended spade lugs supplied should be 90° types. If you are supplied with 45° types, just bend the lugs to 90° before installing them on the board. Alternatively, the final version of the PCB will also have provision to accept 6.3mm PC-mount vertical spade connectors (Altronics H2094), as shown in the photo. The transistors, the electrolytic capacitor and the bipolar capacitors can now be installed. The two 47μF bipolar capacitors can go in either way around but watch the orientation of the polarised 47μF 63V electrolytic capacitor. siliconchip.com.au Fig.3: attach the six double-ended spade lugs to the PCB as shown in this diagram. Do the nut up nice and tight to ensure a good connection and to ensure that the assembly does not rotate. Finally, you can complete the board assembly by fitting the three 2-way terminal blocks and the DPDT relay. Testing If you have a suitable DC supply, you can test the unit prior to installing it. To do this, connect the supply to CON1 and install a wire link between one of the CON2 “AC Sense” input terminals and the positive supply rail (this is done to ensure transistor Q1 turns on). Do not connect anything to the temperature switch input or to the speaker terminals at this stage. Next, apply power and check that the relay turns on after about 5s. If it does, temporarily short the temperature switch input – the relay should immediately switch off. Similarly, the relay should immediately switch off if you disconnect the link to the “AC Sense” input. The next step is to check that the relay switches off if a DC voltage is applied to the loudspeaker terminals (this simulates an amplifier fault condition). To do this, apply power, wait until the relay switches on, then connect a 3V (2 x 1.5V cells in series) or 9V battery (either way around) between the ground (-) terminal of CON1 and the LSPKIN+ terminal. The relay should immediately switch off. Repeat this test for the RSPKIN+ terminal, then reverse the battery polarity and perform the above two tests again. The relay should switch off each time the battery is connected. Note that we don’t connect to the LSPKIN- or RSPKIN- terminals for this test because these two inputs are fully floating at this stage. That changes when the Speaker Protector module Parts List 1 PCB, code 01207071, 112.5mm x 80mm 1 10A 24V DPDT PC-mount relay (Altronics S-4313) 3 2-way 5mm or 5.08mm pitch terminal blocks (CON1-CON3) 4 M3 x 10mm tapped spacers 4 M3 x 6mm pan head screws 6 M4 x 10mm pan head screws 6 M4 flat washers 6 M4 shakeproof washers 6 M4 nuts 6 6.3mm double-ended 45° or 90° chassis-mount spade lugs (Altronics H-2261, Jaycar PT-4905) OR 12 x 6.3mm PC-mount vertical spade connectors (Altronics H2094) 0.7mm diameter tinned copper wire for links Semiconductors 7 BC546 NPN transistors (Q1-Q3, Q5, Q7, Q8 & Q10) 3 BC556 PNP transistors (Q4, Q6 & Q9) 4 1N4004 diodes (D1-D3, D5) 1 1N4148 diode (D4) 1 12V 1W zener diode (ZD1) Capacitors 1 47μF 63V PC electrolytic 2 47μF 50V non-polarised (bipolar) electrolytic (Altronics R-6580, Jaycar RY-6820) 1 470nF 50V metallised polyester (MKT) Resistors (0.25W, 1%) 2 100kΩ 1 10kΩ 0.5W 2 33kΩ 1 2.7kΩ 2 22kΩ 2 100Ω 1 12kΩ 1 10Ω 2 47kΩ 1W 5% is installed in a chassis and the loudspeaker leads are connected, because the negative loudspeaker terminals on the amplifier are connected to chassis (via the power supply). Troubleshooting If the relay doesn’t activate when power is applied, switch off immediately and check for wiring errors, eg, incorrect supply polarity, a transistor in the wrong location etc. If this doesn’t locate the fault, switch on and check the supply voltage, then check the voltages around the transistors. October 2011  39 Setting The Loudspeaker On Fire These three photos, taken at 3-second intervals, illustrate just how quickly the fire takes hold once the cone ignites. To obtain the sequence of photos for this article, we engaged in some deliberate vandalism! To simulate what can happen when a DC fault occurs in a big stereo power amplifier, we actually connected the nominal 57V DC rail of our prototype power supply for the Ultra-LD Mk.3 power amplifier to a loudspeaker. This power supply employs a 300VA transformer and while its continuous rating is 300VA it can deliver a lot more than that in the short term. If you get a DC fault in a power amplifier, the normal result is that it applies the full DC supply (positive or negative) to the loudspeaker. You cannot rely on the amplifier fuses to blow quickly; to blow quickly, they need to carry a current which is two times their rating or more. As preparation for this staged disaster we first connected one of the 57V rails across an 8-ohm resistive load. Under these conditions, the supply dropped to about 47V. This means that 276W was dissipated in the load; quite enough to cause a fire in the wrong circumstances. And note that the fault current of 5.9A would definitely not blow the on-board 6.5A fuses in the Ultra-LD Mk.3; they will happily run all day with that current. Q3’s emitter should be at about 12.6V and its collector at 12.8V, while both Q3 and Q4 should have base-emitter voltages of 0.6V. Similarly, Q1 should have a baseemitter voltage of 0.6V (provided the link between the “AC Sense” Input and the positive supply terminal is in place) but transistors Q2 & Q5-Q10 40  Silicon Chip We then measured the DC resistance of the loudspeaker victim (actually the Minstrel 2-way 8-inch loudspeaker we described in the February 1989 issue). It was about 7.6Ω. Again, doing the calculations, a DC fault in the Ultra-LD could be expected to deliver almost 300W into the 25mm voice coil of the poor unsuspecting loudspeaker. We duly set up the test with the loudspeaker on a stand and with cameras at the ready, one of them being set to video the event. After a quick measurement, we switched on the fault. This produced a loud click from the loudspeaker and the cone jumped out about 2cm; probably to the limit of suspension travel. There was a fairly pronounced hum for a few seconds and then silence. Bugger! The voice coil had obviously burnt out! We quickly felt around the front of the voice coil dust cap and noted that it was quite hot and also had a distinct burning smell. Well that was that but we still needed some photos to demonstrate what can really happen if the voice coil stays intact for just a bit longer. Being ever resourceful (and using generous journalistic licence), should all be off – ie, they should have base-emitter voltages of 0.2V or less. If Q3’s base voltage is low (around 0.2V), it could mean that Q2 is on and Q1 is off, possibly due to no voltage being applied to Q1’s base. Alternatively, one of the transistors in the speaker input monitoring circuits (ie, Q5-Q10) could be faulty (short circuit). we just happened to have a propane gas lighter handy. We lightly touched its very small flame to the speaker’s cone, just below the dust cap. It was already so hot that it immediately caught fire and within seconds the whole cone was well alight with lots of flame and smoke, as can be seen in the sequence of photos. Within just a few seconds more, this would be enough to set a whole room alight with really dire consequences for anyone in the house. Need we say more. Do not imagine for a moment that this sort of thing cannot happen to you. In fact, our calculations show that the bigger the power amplifier, the more risk of a catastrophic fire if the amplifier does not have an in-built loudspeaker protector to disconnect the speakers in the case of a DC fault. After we had extinguished the fire by putting the loudspeaker face down onto the concrete in our company parking area, we took some additional photos which showed that the bonded acetate fibre filling behind the speaker had been already alight. When this stuff burns it produces copious quantities of thick black acrid smoke. Enough said. You can quickly isolate which circuit section is at fault by disconnecting the 10W and 100W resistors to Q3’s base. Just remember that all transistors that are turned on will have a baseemitter voltage of about 0.6V. This should enable you to quickly locate SC where the trouble lies. siliconchip.com.au