Fullscreen HTML5Med Res (??MB)HTML4

Plastic Power PA Amplifier Open-air sporting events like this recent Australia Day surf carnival at Freshwater Beach require plenty of PA muscle. Photo by Andrew McEwen. This article adapts the Plastic Power amplifier module described in the April 1996 issue to public address use. The circuit now includes a 100V line transformer, output transistor protection, a thermal cutout and DC offset adjustment. By ROSS TESTER The “Plastic Power” high-performance amplifier module described in the April 1996 issue has already proved to be a trouble-free design. We foresaw that it would be popular for new amplifier builders and equally sought after as a high-power, high-performance replacement module for many ageing ampli­fiers out there –both commercial and home-built. 24  Silicon Chip And so it has been. But there was one use which we hadn’t really considered – public address or PA. Since the article appeared, we have had a number of enquiries: “can I use this amplifier for PA?” The immediate reaction was “why not?” After all, with power output approaching 200 watts into 4Ω loads, on first glance it would make an excel- lent PA amplifier. But on reflection, it wasn’t quite as simple as that. PA requirements For PA use, there are important requirements which don’t occur in domestic (ie, hifi) applications. Most important of these is the ability to drive a 100V line transformer. A PA amplifier that cannot work into a 100V (or even 70V) line is not considered a PA amplifier – it’s just a toy. But didn’t the specifications box in the April 1996 issue claim “unconditional stability”? Wouldn’t this mean that you could simply bung on a 100V line transformer and the amplifier would be happy? It would be if operating into complex loads was the only problem. But it is not. In fact, it is only a minor considera­tion. By far the most difficult problem to overcome when operat­ing into a transformer of any description is the DC offset at the amplifier’s output. DC offset, as the term implies, is an amount of DC voltage across the speaker output terminals. In a perfect world, or in a perfect amplifier, there would be no DC offset. But in any direct-coupled amplifier there is always some small DC offset voltage at the output and this is mostly due to the mismatch of the differential input transistors. Typically, the DC offset is around 20-50 millivolts and it can be positive or negative, with respect to the “cold” side of the speaker terminals. While this is tolerable in an amplifier intended for hifi or general audio applications where loudspeakers are being dri­ven, it causes a big problem when the load is a 100V line trans­ former. A few quick calculations will show why. For example, if the amplifier is driving a loudspeaker with a voice coil resist­ance of 6Ω (a fairly Performance Output power ........................ 175 watts into 4Ω or 100V line Frequency response ............. -3dB at 30Hz and 17kHz Input sensitivity ..................... 1.15V RMS (for full power into 4Ω) Harmonic distortion .............. <.03% from 20Hz to 20kHz, typically <.01% Signal-to-noise ratio ������������ 101dB unweighted (22Hz to 22kHz); 116dB A-weighted Stability ................................. unconditional typical value), a DC output offset of 50mV will cause 8.3 milliamps DC to flow through the speaker. This will cause a very small mechanical offset of the speaker’s voice coil from its rest position but otherwise no harm will be done. On the other hand, consider that same 50mV DC offset ap­plied to the primary winding of a 100V line transformer. In this case, the DC resistance of the winding is likely to be 100 mil­ liohms (0.1Ω) or less. Now, the DC current which will flow through the primary winding is 500 milliamps or more and this causes really serious problems. Any DC in a transformer winding is bad news. First of all, the transformer can be saturated, which causes awful distortion (hardly what you want when Mr or Mrs High and Mighty steps up to the podium to speak!). Worse, a current of 500mA is much higher than the normal quiescent current in the output stage and it will lead to extra heating, by 20 or 30 watts, depending on the ampli­fier’s supply voltages. This amplifier is capable of delivering 175 watts into 4Ω or a 100V line transformer for PA work. The heatsink shown here is adequate for general use but if the amplifier is to be operated in high ambient temperatures and expected to deliver high power continuously, a larger fancooled heatsink will be required. March 1997  25 PARTS LIST 1 PC board, code 01103971, 99 x 166mm 2 panel mount M205 fuseholders (or 4 20mm fuse clips – see text) 2 5A M205 fuses 1 coil former, 24mm OD x 13.7mm ID x 12.8mm deep, Phillips CP-P26/19-1S or 4322 021 30362 - see text 1 4Ω/100W toroidal output transformer (Altronics M1124 or equiv­alent) 2 metres 0.8mm enamelled copper wire 1 thermal circuit breaker 80°C, 10A (Altronics S5610 or equivalent) 1 large single-sided finned heatsink, at least 300mm long, 0.7°C/W 2 TO-126 heatsinks (Altronics H-0504 or equivalent) 4 TO-3P transistor insulating washers 3 TO-126 transistor insulating washers 1 200Ω 10-turn vertical trimpot (Bourns 3296W series or equival­ent) 1 100Ω 5mm horizontal mounting trimpot 13 PC board pins 4 3mm x 20mm screws 5 3mm x 15mm screws 9 3mm nuts Worse still, such a high current can easily lead to thermal runaway in the output devices, and their eventual destruction. The DC offset problem has been known for a long time, ever since direct coupled amplifiers were first produced. In fact, some years ago, National Semiconductor brought out the LMC669 as the ideal answer to this problem and SILICON CHIP featured a circuit using it in the September 1989 issue. Alas, the IC now appears to be unobtainable, so other means need to be found to cure the DC offset problem. Fig.1 shows the modified circuit of the Plastic Power amplifier. It is capable of delivering around 175 watts into a 100V line. Now let’s consider the problem of DC offset and how it is corrected. First, we include provision for adjusting the DC offset to zero (or as close as we can achieve) with a trimpot connected between the emitters of the differential pair, Q1 and Q2. This will allow any minor differences between the two “sides” of the circuit to be nulled out. The emitter resistors of Q1 and Q2 were reduced from their original value of 150Ω to 100Ω and a 100Ω trimpot placed between them. Adjustment is simple: when the amplifier is completed set the trimpot to its centre position, then adjust it so that the DC voltage across the speaker output terminals (as measured on a digital multimeter set to its lowest voltage range) is zero or as close as possible. The board pattern, incidentally, allows for either a verti­cal or horizontal mounting 5mm trimpot. A horizontal mounting pot is preferred, for ease of adjustment. Second, we have modified the PC board slightly to allow Q1 & Q2 to be thermally bonded together. Thus any tendency for one transistor to get hot, which may cause increased DC imbalance, will be reflected in the other transistor. We also did the same with Q4 and Q5, the current mirror stage. 26  Silicon Chip Semiconductors 2 MJL21194 NPN power transistors (Q12, Q13) 2 MJL21193 PNP power transistors (Q14, Q15) 2 MJE340 NPN driver transistors (Q9, Q10) 1 MJE350 PNP driver transistor (Q11) 1 BF469 NPN transistor (Q8) 1 BF470 PNP transistor (Q6) 4 BC546 NPN transistors (Q4, Q5, Q7, Q16) Reduced bandwidth Sometimes a high performance amplifier is simply “too good” for PA. If you think about it, PA is one of the worst-case audio applications: (a) Long speaker leads can act as magnificent RF antennas for any local radio or TV station or even close-by two-way 4 BC556 PNP transistors (Q1, Q2, Q3, Q17) 2 1N5404 power diodes (D5, D6) 4 1N914 diodes (D1, D2, D3, D4) 1 3.3V 0.5W zener diode (ZD1) Capacitors 4 100µF 63VW electrolytic 1 22µF 16VW electrolytic 1 0.33µF 250VAC MKP 1 0.33µF 50VW MKT 5 0.1µF 63V MKT 1 .0012µF MKT or ceramic 1 100pF 100V ceramic Resistors (0.25W, 1%) 2 18kΩ 1 180Ω 1 15kΩ 1W 2 160Ω 1 6.8kΩ 3 100Ω 1 5.6kΩ 1W 1 68Ω 1 1.5kΩ 1 47Ω 1 820Ω 3 12Ω 1W 1 470Ω 4 0.47Ω 5W 2 390Ω 2 560Ω 5W 3 220Ω radios (and many sports, coaches, etc, use two-way). (b) They’re often used in portable situations, and every location has its own share of problem electrical noises which may or may not be treatable. (c) If it is a portable setup, speaker lines may be temporary and therefore not too secure against either shorts or cuts. Speaker cabling is often exposed to the elements, with joins, plugs & sockets, etc which may be corroded, even with the best “weather­proofing”. With these problems in mind, it is wise to limit the over­all bandwidth of a PA amplifier. This can assist in reducing interference, especially electrical noise picked up by the speak­er leads. Therefore, the input RC filter and the output RLC filter have been modified. The result is that both the bass response and the high frequency response have been deliberately curtailed: -3dB at 30Hz and 17kHz, as depicted in Fig.2. This shows the frequency response of the complete amplifier, including the 100V line transformer. Protection circuitry This is something of a thorny Fig.1: the circuit is essentially the same as that published in the April 1996 issue except that it has been adapted for PA use. The main changes include the addition of a 100V line transformer, DC offset adjustment (using VR1) and current limiting. The latter is provided by transistors Q16 & Q17, which monitor the emitter currents of Q12 & Q14 respectively. Note that the frequency response has been deliberately limited to ensure reliability under PA conditions. March 1997  27 AUDIO PRECISION SCFREQRE AMPL(dBr) vs FREQ(Hz) 15.000 21 JAN 97 11:00:02 10.000 5.0000 0.0 -5.000 -10.00 -15.00 20 100 1k 10k 20k Fig.2: this graph shows the overall frequency response of the power amplifi­er, including the 100V line transformer, at a power level of 10 watts. The bass and high frequency response has been deliberately curtailed. subject, so let’s get straight into the blackberry bushes! Some designers of hifi amplifiers will have nothing to do with protection circuitry in the output stages, claiming that it causes distortion even before it becomes active and then causes severe distortion as it acts to limit current. Indeed, where foldback current limiting is used in amplifi­er output stages, it can cause squealing from tweeters, and in severe over-drive condition, can cause tweeter burnout. PA amplifiers, on the other hand, are a different kettle of fish. First, ultimate low distortion figures are of minor importance (although this amplifier is pretty good in that respect, even with protection). Second, PA amplifiers are often subjected to serious abuse. Years of experience has taught us that people can be abso­lutely ruthless when it comes to their personal enjoyment: they sit in front of a PA speaker, then complain that the PA is too loud! We have had many occasions at sporting functions where the speaker lines have been deliberately cut or shorted. Bring on the protection! Transistors Q16 & Q17, in conjunction with diodes D3 & D4, provide the protection feature. Q16 monitors the current flow through the 0.47Ω emitter resistor of output transistor Q13, via a voltage divider consisting of 390Ω and 160Ω resistors. What happens is that normally Q16 (and Q17) are off and play no part in the circuit operation. However, if the current through the 0.47Ω emitter resistor of Q13 exceeds about 4.4 amps, Q13 begins to turn on and it shunts the base current from Q10, the associated driver transistor. In turn, the drive to Q12 & Q13 is limited so that the output current does not exceed about 4.5 amps peak. The same process happens with Fig.3: suggested power supply for the amplifier. The power trans­former should be rated at 300VA or more. 28  Silicon Chip Q17 which monitors the cur­rent flow through the 0.47Ω emitter resistor of output transistor Q14. Diodes D3 & D4 are included to prevent Q16 & Q17 from shunting the signal when they are reverse-biased; this happens for every half-cycle of the signal to the driver transistors. Diodes D5 & D6 are included as part of the protection cir­cuitry although their function is ancillary. They prevent large voltage spikes from the transformer, generated when the current limiting circuitry acts to turn off the output transistors, from actually damaging the transistors. D5 does this, for example, by clamping any spike voltage to 0.6V above the positive supply rail. Similarly, D6 clamps any spike voltage to 0.6V below the negative supply rail. Normally, both diodes are reverse biased and play no part in the amplifier operation. Note that this protection circuitry provides simple current limiting, not foldback protection, where the current drops back to a low value to limit power dissipation in the output stages (and with attendant serious distortion, as outlined previously). With this simple current limiting, the transistors are protected from sudden death in the case of serious over-drive or short-circuits, although the fuses may blow before this happens. While the output transistors are protected against imme­diate destruction, their dissipation is greatly increased over what it would be if the amplifier was simply delivering full power. In fact, the output transistors can dissipate four or five times as much power as in normal operation. Hence, they get very hot very quickly and eventually, if the over-drive or short-circuit condition is not corrected, they will fail; probably sooner than later. To prevent this eventual failure, we have included a ther­mal cutout which is mounted on the heatsink. When the heatsink temperature exceeds 80°C, the thermal cutout opens and is not restored until the heatsink cools down again. Heatsink selection Note that the thermal cutout is there for a secondary reason and that is to prevent over-dissipation in the output transistors under continuously high power conditions. To elabo­rate, the maximum dissipation in a class-B amplifier occurs when it is deliver- ing about 35 to 40% of the maximum output power. Under this condition, the power dissipated in the output transistors can be expected to be about 30% more than the maximum output power. This amplifier will actually deliver about 175 watts before clipping and the maximum dissipation in the output transistors can be expected to be about 230 watts, depending on the supply regulation and the actual value of the load. 230 watts equates to almost 58 watts per transistor which means that the largest possible heatsink should be used. Ideally, if you anticipate rigorous operating conditions, the heatsink should be fan-cooled. We have specified a fairly large heatsink with a rating of 0.7°C/W but to cope fully with a total dissipation of 230 watts, the heatsink needs to be much larger, at 0.3°C/W. Hence, with the specified heatsink, the thermal cutout is a worth­while safety feature in case the amplifier’s operating conditions become a little torrid. The remainder of the circuit description is as featured in the April 1996 issue of SILICON CHIP. A suggested power supply is shown in Fig.3. The transformer should be rated at 300VA or more. AUDIO PRECISION SCTHD-W THD+N(%) vs measured 10 LEVEL(W) 15 JAN 97 11:18:24 1 0.1 0.010 0.001 0.5 1 10 100 300 Fig.4: THD versus power at 1kHz into a 4Ω load. AUDIO PRECISION SCTHD-W THD+N(%) vs measured 10 LEVEL(W) 15 JAN 97 11:10:15 1 Performance The amplifier’s performance is summarised in a separate panel and as you can see, it is very respectable for PA use. Fig.4 shows the harmonic distortion versus power output into a 4Ω load while Fig.5 shows the distortion versus power with the 100V line transformer connected. There is very little difference between these curves, indicating that the transformer is a high quality unit which degrades the signal very little. Construction The procedure for assembling the PC board is quite similar to that the for the original amplifier described in the April 1996 issue but there are enough differences to justify giving the complete assembly and setting-up procedure. The component overlay for the PC board is shown in Fig.6. Before starting the PC board assembly, it is wise to check the board carefully for open or shorted tracks or undrilled lead holes. Fix any defects before fitting the components. 0.1 0.010 0.001 0.5 1 10 100 300 Fig.5: THD versus power at 1kHz with a 100V line transformer. The load resistance was 57Ω (two jug elements wired in series and immersed in water)! This done, you can start the assembly by inserting the PC pins and the resistors, followed by the diodes. When installing the diodes, make sure that they are inserted with correct polarity and don’t confuse D1-D4 (1N914 or 1N4148) with the 3.3V zener diode (BZX79-C3V3 or equivalent). You should also take care to ensure that the electrolytic capacitors are all installed the right way around on the PC board. Note that the 100pF compensation capacitor from the collec­tor of Q8 to the base of Q7 should have a voltage rating of at least 100V while the 0.33µF capacitor in the output filter should have a rating of 250VAC. The 4Ω resistor in the output filter is comprised of three 12Ω 1W resistors March 1997  29 Fig.6: install the parts on the PC board as shown in this diagram. Note that while provision for on-board fuses has been made (as in the hifi version of the amplifier) external chassis-mounted fuses are more practical for PA use. in parallel. Choke L1 is wound with 19.5 turns of 0.8mm enamelled copper wire on a 13mm plastic former. Some readers who built their own version of the original amplifier (ie, not from a kit) experienced difficulty in obtaining the correct former. The one used in our prototype is a Philips CP-P26/19-1S (previously known as a 4322 021 30362). If your supplier cannot obtain this part, a possible replacement is the plastic bobbin some parts suppliers still have to suit FX-2240 pot cores. This is marginally different in size but the inductance of the coil wound (with RESISTOR COLOUR CODES                   No. 2 1 1 1 1 1 1 2 3 1 2 3 1 1 3 4 2 30  Silicon Chip Value 18kΩ 15kΩ 1W 6.8kΩ 5.6kΩ 1W 1.5kΩ 820Ω 470Ω 390Ω 220Ω 180Ω 160Ω 100Ω 68Ω 47Ω 12Ω 1W 0.47Ω 5W 560Ω 5W 4-Band Code (1%) brown grey orange brown brown green orange brown blue grey red brown green blue red brown brown green red brown grey red brown brown yellow violet brown brown orange white brown brown red red brown brown brown grey brown brown brown blue brown brown brown black brown brown blue grey black brown yellow violet black brown brown red black brown not applicable not applicable 5-Band Code (1%) brown grey black red brown brown green black red brown blue grey black brown brown green blue black brown brown brown green black brown brown grey red black black brown yellow violet black black brown orange white black black brown red red black black brown brown grey black black brown brown blue black black brown brown black black black brown blue grey black gold brown yellow violet black gold brown brown red black gold brown not applicable not applicable Fig.7: this diagram shows the heatsink mounting details for the driver and output transistors. After mounting, switch your multi­meter to a high Ohms range and check that each device has been correctly isolated from the heatsink (there should be an open circuit between the heatsink and the transistor collectors. the same number of turns) will be close enough. If installing the on-board fuse clips (see text about external fuses below), note that they each have little lugs on one end which stop the fuse from moving. If you install the clips the wrong way, you will not be able to fit the fuses. The 560Ω 5W wirewound resistors can also be installed at this stage; they are wired to PC stakes next to each fuseholder and are used when setting the quiescent current. Next, mount the smaller transistors such as BC546 & 556, BF469 & 470. Note that the transistor pairs Q1/Q2 and Q4/Q5 are thermally bonded; the pairs are mounted on the board so that their flat surfaces are touching, with heat transfer between them assisted by a smear of heatsink compound. Solder in one of the pair so that it is angled very slightly towards where its mate will go and then spread a thin film of heatsink compound over the flat surface. This done, solder in the collector and emitter of its mate and push the flat surfaces together before soldering the base, to lock the transistor in place. Repeat this process for the other pair of transistors. Both Q6 & Q8 need to be fitted with U-shaped heatsinks. The four output transistors, the driver transistors (Q10 & Q11) and the Vbe multiplier Q9 are mounted vertically on one side of the board and are secured to the heatsink with 3mm machine screws. Perhaps the best way of lining up the transistors before they are soldered to the board is to temporarily attach all of them to the heatsink; don’t bother with heatsink compound or thermal washers at this stage. This done, poke all the transistor leads through their appropriate holes in the PC board and line it up board so that its bottom edge is 10mm above the bottom edge of the heatsink. This is so that the board will be horizontal when fitted with 10mm spacers at its front corners. Note that you will have to bend out all the transistor leads by about 30°, in order to poke them through the PC board. The heatsink will need to be drilled and tapped to suit 3mm machine screws. The relevant drilling details were included in the April 1996 article (Fig.12). You can now solder all the power transistor leads to the PC board. Having done that, undo the screws attaching the transis­tors to the heatsink and then fit mica washers and apply heatsink compound to the transistor mounting surfaces and the heatsink areas covered by the mica washers. The mounting details for these transistors is shown in Fig.7. Alternatively, you can dispense with mica washers March 1997  31 Note the thermal cutout fitted to the heatsink. This interrupts the speaker line if the heatsink temperature rises above 80°C. Q6 & Q8, which are BF470 and BF469 respectively, are fitted with U-shaped flag heatsinks, as shown here. and heatsink compound and use silicone impregnated thermal washers instead, as can be seen in the photos. Whichever method you use, do not overtighten the mounting screws. With your multimeter switched to a high Ohms range, check that there are no shorts between the heatsink and any of the transistor collector leads. If you find a short, undo each transistor mounting screw until the short disappears. You can then re­mount the offending transistor, having fixed the cause of the short. The thermal cutout is mounted on the heatsink close to one of the output transistors. The leads connecting the thermal cutout switch to its appropriate PC pins should be rated at 10A. Double-check all your soldering and assembly work against the circuit of Fig.1 and the component layout diagram of Fig.6. Finally, connect the primaries of the output transformer to the output terminals, exactly as shown on the circuit diagram of Fig.1. Note that the 32  Silicon Chip primaries are connected in parallel while the secondary windings are connected in series – watch out for the colour-coding. Adjustments With no fuses in position, set trimpot VR2 fully anticlock­wise so that it is set for minimum resistance and set trimpot VR1 to its centre position. A 560Ω 5W resistor should have been soldered across each on-board fuseholder (or more correctly, the PC pins alongside). Assuming that the amplifier passes the “smoke test” when you apply power, set your multimeter to about 20-50V DC and connect it across a 560Ω resistor. Slowly adjust trimpot VR2 so that the multimeter reads 14V (equivalent to a quiescent current of 25mA or 12.5mA through each output transistor). The voltage across the other 560Ω resistor should be virtu­ally identical. Now connect the multimeter, on its lowest DC voltage range, across the output terminals on the PC board –that is, in paral­lel with the output transformer primary. Carefully adjust trimpot VR2 for minimum voltage (a digital multimeter is best for this purpose). You should be able to set VR2 so that the DC offset voltage is less than ±2mV DC. Once this has been done, leave the amplifier running for 10 minutes or so and check both voltages again. Adjust VR1 if necessary – changing this should not have any effect on the output DC offset voltage but if your DC offset has risen (in either direction) adjust VR2 once again to achieve the minimum possible. Finally, install the 5A fuses. External fuses As you may have noticed, the original module used on-board fuses for the supply rails. While not suggesting for a moment that the fuses be left out, fuses inside a public address ampli­fier are a pain in the proverbial! When the inevitable happens, it is invariably only a few minutes before the keynote speaker is due to make his/ her ad­dress, or the competitors turn Fig.8: this is the full size artwork for the PC board. Check your board care­fully for any defects before installing the parts. for their last lap in the final! Searching around for a screwdriver to open up a case can be a tad embarrassing in these circumstances. We suggest that external (ie, rear of case) fuseholders be provided and cable of the same diameter/rating as the power supply cabling used to connect these to the board. This way, the on-board fuseholders could be eliminated, with the 560Ω resistors still used to set up the module in the suggested way. Why 100V lines? In this article, we have talked about 100 volt lines as if they were “de rigueur” in PA applications. But what is a 100V line and why is it used so extensively for public address? Is a 100V line essential? Let’s answer the last question first. No, but . . . Of course “ordinary” 4Ω or 8Ω speakers could be and often are used in PA applications. In a small hall, for example, a few low impedance speakers connected appropriately will often be satisfactory. The key word here is “appropriately”. First of all, you need to worry about the overall impedance. You have to work out the various series and parallel combinations which will bring you back to 4Ω or 8Ω to suit the amplifier. Then there’s the problem of power – can the individual speakers handle the amount of power being fed to them? And are the power ratings correct for the way you want to connect them? It’s not hard to get into a mess! All of these problems are solved by the use of a 100V (or less commonly, 70V) line. Each speaker, together with its own stepdown transformer, is merely connected across the 100V line (ie, in parallel). As far as power ratings are concerned, you simply add up the wattage of the individual speakers and ensure that the total does not exceed the power rating of your amplifi­er. Even if it does, most speakers for 100V line use have multiple taps – if you want more speakers in the system (for example, to fill a sound “hole”) then select a lower wattage tap on some of your speakers to allow the extras. It really is that simple. But there is a more important reason to use 100V lines for PA use: less power loss (commonly known as I2R loss). It’s exact­ly the same reason that power authorities use high voltage for long distance transmission of elec- tricity; higher voltage means lower current and lower current means lower loss. In a typical PA installation for a sporting field or large hall there could easily be 1000 metres of speaker cable; often much, much more. Assuming that the speaker cable used was of reasonable quality, you could expect a resistance of about 2.5Ω per 100 metres. That means 1000m of cable would have an overall resistance of about 25Ω. This would be totally impractical for a 4Ω or 8Ω system but is not a serious problem for a 100V line system. Are 100V lines dangerous? Finally, let’s dispel one furphy: that 100V speaker lines are dangerous. Yes, they will give you a bit of a bite if you get across them while the announcer is waxing eloquent or the music is reaching a crescendo. But – and the but is important – the 100 volts is not constant like the 240VAC mains supply which often does kill. The full 100VAC is only present when the ampli­fier is delivering its full power. Most of the time, the voltage is only a few volts. Of course, it’s better if you don’t get yourself across a 100V speaker line, especially if a hyperventilating sports commenta­tor is getting excited at the SC other end of the signal chain! March 1997  33