Fullscreen HTML5Med Res (??MB)HTML4

15W/Ch Class-A Stereo Am L Last month, we presented the circuit details of a 15W class-A module with extremely low total harmonic distortion. This month we show how to build two modules into a chassis to produce a stereo power amplifier. In order to obtain the extremely low distortion from a stereo pair, it was necessary to use a com­pletely separate power supply. 72  Silicon Chip AST MONTH, we stated in no uncertain terms that building a pair of these 15W class-A modules into a chassis along with a conventional (unregulated) power supply would be a sure path to disappointment. But little did we know, at the time of writing, just how difficult it would be to get the claimed performance in a stereo amplifier – even with a regulated supply. We already knew that we would have to resort to a fully regulated power supply. We had built a suitable power supply into the intended chassis and we used this setup to produce the graphs and figures featured last month. The only problem was that when we hooked up two amplifier modules and started taking meas­ urements in stereo mode, the results were less than stunning. Distortion at 1kHz and 10W was up to around .001% while the signal-to-noise ratio was only around -80dB or so. Now in any conventional amplifier these results might be regarded as satisfactory. But this was no ordinary amplifier and a signal-to-noise ratio of 80dB is a long way from 113dB. The difference is a low background hum compared to just the faintest hiss. Clearly, we still had a problem with hum induced from the transformer. This was being induced into the common earth loop formed by the signal earths back to the common program source. If we broke the loop, the distortion and noise was back down By LEO SIMPSON mplifier where it should be but that is hardly any consolation when it’s sup­posed to be a stereo amplifier. After trying lots of earthing arrangements and playing with the lead dress of the power supply cables, we came to the conclu­sion that the only practical solution was to build the power supply in its own steel box inside the amplifier chassis. So we duly built the box, rebuilt the power supply, reconnected all the leads and Above: this view of the amplifier chassis shows how the various wires and connections have been routed. This layout has been produced after much trial and error to obtain the best distor­tion, separation between channels and signal-to-noise ratio. August 1998  73 AUDIO PRECISION SCCRSTK XTALK(dBr) 0.0 & XTALK(dBr) vs FREQ(Hz) 23 JUN 98 15:15:55 -20.00 -40.00 -60.00 -80.00 -100.0 -120.0 20 100 1k 10k 20k Fig.1: this is the separation between channels across the frequency range from 20Hz to 20kHz. The curves for each channel were measured with both amplifier inputs connected to the measuring source. so on. Result: no improvement. You can imagine the sheer frustration in the SILICON CHIP workshop. Ultimately, we were forced to the conclusion that the power supply would have to be completely separate from the chassis. So that is what we did next. This works but it is an extra expense that we would have preferred to avoid. So be it. If you want this stereo amplifier to have a typical distortion of .0006% or below, it needs a completely separate fully regulated power supply. We also found it necessary to slightly change the earthing of the input circuit on the PC board. Instead of connecting directly to the “star” earth point on the PC boards, the input circuits of each power amplifier are now connected via 10Ω resis­tors. This reduces the incidence of earth currents circulating via the loop formed by the two input cables and the external program source (eg, a CD player or tuner). With the power supply presented here, the signal-to-noise ratio and harmonic distortion, when measured in stereo mode, is as published last month. The separation between channels is quite respectable, measuring around 80dB at mid-frequencies, although this is not as good as we hoped for. Fig.1 shows the separation 74  Silicon Chip across the frequency range from 20Hz to 20kHz. As far as we can tell, the only way to substantially improve upon this would be to have separate power supplies for each channel. Another crucial development was the necessity to specify good quality gold-plated binding post terminals for the speaker outputs. Initially, we used a set of readily available spring-loaded speaker terminals, on the basis that the overall power output was low and therefore heavy duty speaker connections were not really justified. However, in order to consistently obtain the very low distortion figures that we published last month, the spring-loaded terminals had to be replaced. We found that typi­ cally, they caused a doubling of the measured distortion! So while the heavy gold-plated terminals might look like an unnecessary expense, they are needed. How does it sound? The writer feels a little uncomfortable in answering this question because it requires a subjective answer. In my listening setup, I am using the SILICON CHIP Stereo Control Unit described in September & October 1993 combined with the 100W per channel amplifier described in February 1988. The loudspeakers are the highly regarded Dynaudio Image 4s while the CD player and tuner are current models by Sony (CDP-XE300 and ST-SE200 respectively). The amplifier/speaker combination has compared very well with any number of other systems over the years but when the class-A 15W/ channel amplifier was substituted for the 100W unit and the levels carefully matched, there was a distinct improvement. Put simply, the 15W class-A amplifier sounded cleaner; quite a lot cleaner in fact. And yet, going back to the 100W amplifier, it still sounded very good. Further listening seemed to indicate that the instruments spread across the “sound stage” between the two loudspeakers were more distinct, and occupying a more precise location. After considerable testing, we believe that the perceived improvement in sound quality may not be solely due to the considerably improved distortion of the new amplifier but to greatly improved separation between channels. We hope to report on this aspect further in a future issue but it appears that audio equipment which has nominally good separation under the conventional IHF-201 test method actual­ly has degraded performance when connected to “real” stereo program sources such as CD players. Amplifier case The new amplifier is mounted in a 2-unit high rack-mounting case with large finned heatsinks on both sides. On the front panel is a headphone socket, volume control and LED power indica­tor. On the rear panel is a pair of RCA sockets for the left and right channel inputs and gold-plated binding post terminals for the power amplifier output connections. The separate power supply is mounted in a standard plastic instrument case measuring 260 x 82 x 190mm. This has a bare front panel apart from the power switch. On the rear panel is a large single-sided heatsink, a fused IEC power socket and the output cable for the DC supply rails. Stereo amplifier circuit Fig.2 shows the circuit of the complete 15W per channel stereo amplifier minus the power supply. Both channels are shown, with the transistor numbering in the second channel running Q101, Q102, etc. Fig.2: this is the complete circuit of the stereo power amplifier except for the separate power supply. Note the 10Ω isolating resistors in the input earth returns for both channels. August 1998  75 Fig.3: the power supply circuit uses a toroidal power transformer with two 21V secondaries to feed a bridge rectifier and two 4700µF 50VW filter capacitors. These then feed identical positive and negative regulator circuits comprising an adjustable 3-terminal regulator and a power transistor. Fig.4: the PC board component overlay and wiring diagram for the power supply. Take care with the polarised components. 76  Silicon Chip There are a number of differences from the circuit published last month. First, there are two errors which have been corrected: (1) Q8 & Q9 are specified as BC547 and not BC546; and (2) trimpot VR1 is 200Ω, not 500Ω. The amended circuit shows the 20kΩ (log) ganged potentiome­ t er which acts as the volume control for the amplifier. We think this feature will appeal to those who want to operate the ampli­fier as a very simple no-frills system with just a CD player. Later on, if there is a demand from readers, we may develop a stereo control unit with matching performance. A stereo headphone socket is included, fed by a 330Ω 1W resistor from each channel output. The head­ phone socket incorpo­ rates speaker switching, so that if the headphones are plugged in, the speakers are switch­ed off. Interestingly, while investigating an increase in distor­tion which was eventually blamed on the spring-loaded speaker terminals, as noted above, The power transformer and bridge rectifier are mounted on a metal baseplate inside the case. The rear panel is also metal and has a large heatsink for the regulators and power transistors. we also checked whether the headphone/speaker switching caused any distortion. It didn’t. The amended circuit also includes the change to the input circuitry whereby a 10Ω resistor is connected in series with the input and feedback earthing for the differential pair, Q1 & Q2. Finally, the LED power indicator and its 2.2kΩ resistor is shown connected to the -20V supply rail. involves an LM317 and Q1, a TIP42 PNP power transistor. The LM317 is set to deliver 20V by virtue of the 120Ω and 1.8kΩ resistors connected to its ADJ (adjust) terminal. Because of the way it is connected across the 3-terminal regulator, the TIP42 transistor is forced to follow the LM317. This happens in the following way. All the current passing through the LM317 must first pass through the associated 0.22Ω resistor and diode D1. The total voltage drop across these two components becomes the Fig.5: actual size artwork for the power supply PC board. Power supply circuit Fig.3 shows the details of the power supply circuit. It uses a toroidal power transformer with two 21V secondaries to feed a bridge rectifier and two 4700µF 50VW filter capacitors. This develops unregulated supply rails of about ±29V and these are fed to identical positive and negative regulator circuits comprising an adjustable 3-terminal regulator and a power tran­sistor. To see how these work, let us consider just the positive regulator which August 1998  77 Parts List Amplifier chassis 1 2-unit high rack-mounting case 2 single-sided heatsinks, 300 (W) x 75 (H) x 49mm (D) (Altronics H-0545, DSE H-3406 or equivalent) Note: these heatsinks form the sides of the rack mounting case. 2 PC boards, SC01207981, 118 x 81mm 8 20mm fuse clips 4 M205 2.5A fuses 2 coil formers, 24mm OD x 13.7mm ID x 12.8mm long (Philips 4322 021 30362) 4 metres, 1mm dia. enamelled copper wire 1 0.5-metre length of 0.7mm dia. tinned copper wire for board links 6 2-metre lengths, medium duty hookup wire, (6 different colours) 1 2-metre length of figure-8 twin shielded audio cable 1 stereo headphone socket, insulated, DPDT switched (Altronics P-0074 or similar) 1 dual 20kΩ log, 26mm dia. potentiometer (VR2) 2 200Ω trimpots VR1,VR101; Bourns 3296W or similar 23 PC stakes 4 adhesive rubber feet 2 3-way insulated terminal blocks 4 TO-3P insulating washers 4 TO-18 heatsinks (Farnell 170-061 or equivalent) 4 100mm standoffs tapped for 3M screws 8 3M x 20mm screws 2 3M x 10mm screws 10 3M nuts 4 3mm flat washers 1 cord-grip grommet 10 BC547 NPN transistors (Philips or Motorola) (Q5, Q6, Q8, Q9, Q10, Q105, Q106, Q108, Q109, Q110) 2 BC337-25 NPN transistors (Philips) (Q11, Q111) 2 BC327-25 PNP transistors (Philips) (Q13, Q113) 2 MJL21193 PNP power transistors (Motorola) (Q12, Q112) 2 MJL21194 NPN power transistors (Motorola) (Q14, Q114) 2 BZX55C3V3 3.3V 0.5W zener diodes (ZD1, ZD101) 1 3mm green LED and LED bezel holder Capacitors 8 100µF 25VW electrolytic 2 47µF 16VW electrolytic 2 2.2µF 16VW electrolytic 2 0.15µF 100V MKT polyester or Philips MKC 2222 344 21154 10 0.1µF 100V MKT polyester 2 .0012µF MKT polyester or ceramic 2 100pF NP0 ceramic Resistors (0.25W, 1%) 4 18kΩ   4 180Ω 2 8.2kΩ   4 150Ω 2 3.3kΩ   4 120Ω 3 2.2kΩ 12 100Ω 2 1.8kΩ   2 10Ω 2 390Ω 16 1Ω 0.5W 2 330Ω 1W 2 1.8Ω 5W (for setting bias) Power Supply Semiconductors 10 BC557 PNP transistors (Philips or Motorola) (Q1, Q2, Q3, Q4, Q7, Q101, Q102, Q103, Q104, Q107) 1 plastic instrument case 260 x 82 x 190mm (W x H x D, with metal rear panel) (Jaycar HB-5910 or equivalent) 1 metal baseplate, 167 x 225mm (1.6mm aluminium in prototype) 1 power transformer, toroidal, 160VA, 2 x 21V secondaries (see text) 1 SPST mains power switch (Jaycar SK-0984 or similar) base bias voltage of the TIP42. In effect, the voltage drop across D1 is matched by the base-emitter voltage of Q1 which is then forced to repro- duce the voltage across the 0.22Ω resistor across its own 0.1Ω emitter resistor. So if the current flowing through 78  Silicon Chip 1 IEC fused power socket (Altron­ics P-8324, Jaycar PP-4004) 1 IEC mains power cord 1 M205 3A fuse 1 single-sided heatsink, 110mm x 75mm x 48mm (W x H x D) 4 adhesive rubber feet 1 PC board, 04208981, 94 x 76mm 1 3-way insulated terminal block 1 3 or 4-pole matched automotive connector set 1 4M x 20mm screw 1 4M nut 1 4mm flat washer 10 3M x 20mm screws 4 3M x 10mm screws 14 3M nuts 4 3mm flat washers 4 TO-220 mounting kits (mica insulators, insulating bushes) 1 cordgrip grommet 5 PC stakes Semiconductors 1 KBPC1004 400V 10A bridge rectifier (BR1) 1 LM317-T variable positive regulator (REG1) 1 LM337-T variable negative regulator (REG2) 1 TIP42 PNP power transistor (Q1) 1 TIP41 NPN power transistor (Q2) 2 1N5404 power diodes (D1,D3) 2 1N4004 power diodes (D2,D4) Capacitors 2 4700µF 50VW electrolytics 2 100µF 25VW electrolytics 2 10µF 35VW electrolytics 2 0.1µF 100V MKT polyester Resistors (0.25W 1%) 2 1.8kΩ 2 10Ω 2 120Ω 2 0.22Ω 5W wirewound 2 0.1Ω 5W wirewound Miscellaneous Heatshrink tubing, tinned copper wire for board links. the LM317 causes a vol­ tage drop of 0.15V across the 0.22Ω resistor, the same voltage will be produced across the 0.1Ω resistor and so Q1 will deliver 1.5A to the output. So Q1 is effectively a “current follower” and the ratio of the current delivered by the LM317 to the current from Q1 is set by the ratio of the two resistor values, 0.22Ω and 0.1Ω. This ratio is 2.2:1 and so Q1 always delivers 2.2 times the current of REG1 while always re­maining under its control. The negative regulator circuit, involving REG2 and Q2, is identical in operation. Building the power supply Since the power supply has to be up and running before you can run the amplifier, we will describe its construction first. The power transformer and bridge rectifier are mounted on an aluminium baseplate which is secured into the integral pillars in the base of the case. Our prototype’s power transformer was supplied with 18V secondary windings so we added 15 turns of 1.25mm enamelled copper wire for each secondary. These turns were wound bifilar (ie, two wires at a time) using a shuttle made from a piece of PC board copper laminate. We wound a layer of clear insulation over the extra winding to protect it. The dual regulator circuit fits onto a PC board measuring 94 x 76mm and coded 04208981. The 3-terminal regulators and two power transistors are along one edge so that they can be easily mounted on the metal rear panel. Fig.4 shows the PC board compon­ ent overlay for the power supply. Mount the resistors and diodes first, followed by the elec­trolytic capacitors, the regulators and the power transistors. Note that the electrolytics and diodes must go in the right way around otherwise the circuit is likely to be damaged at switch-on. For the same reason, do not get the regulators and transis­tors mixed up. Note that the spacing between the power transistors and regulators on the PC board matches the fin spacing on the speci­ fied single-sided heatsink. This is necessary to allow the tran­sistor mounting screws to pass right through the heatsink and the metal rear panel. Fig.6 shows the detail of the heatsink mount­ing. Also on the rear panel is the fused IEC power socket, an earth solder lug and the exit hole for the three-core DC output cable. These holes will need to The power supply case should be positioned at least 600mm away from the amplifier chassis in order to keep the induced hum to an absolute minimum. The power supply is connected to the amplifier using a 3-pole or 4-pole automotive matched connector set. The large finned heatsink is necessary to cool the power supply regulators. Fig.6: this diagram shows the detail of the heatsink mounting for the TO-220 devices in the power supply. After mounting the devices, use your multi­meter to check that there is an open circuit between the heatsink and the device collectors. August 1998  79 Fig.7: chassis wiring diagram for the power supply. 80  Silicon Chip Fig.8: this is the amended PC component overlay for the amplifier module. Take care to ensure that all transistors are correctly oriented and note that transistors Q11 and Q13 should be fitted with finned heatsinks to keep them cool. be drilled and cut as necessary. Fig.7 shows the wiring of the power supply. All the mains supply wiring must be run in 250VAC-rated hookup wire and all wiring terminals should be sleeved with heatshrink sleeving to prevent accidental contact. The three-way DC output cable was run in a short length of 250VAC three-core cable, terminated directly to the PC board at the power supply end. The other end of the cable was fitted with a 4-way plug which mates to a socket on a cable from the power amplifier. Once all your assembly work is finished, check it carefully against the diagrams of Fig.3, Fig.4 and Fig.7. Then apply power and check that the outputs are +20V and -20V DC. Then you can turn your attention to the amplifier chassis. Amplifier assembly Last month we discussed the assembly of the amplifier PC boards. In Fig.8 we show the amended PC board layout which in­cludes the 10Ω input earthing resistors referred to above. Finned heatsinks must be fitted to the TO-92 driver transistors, Q11 & Q13. Fitting these heatsinks is not easy. They are made of springy beryllium-copper to fit TO-18 metal can transistors but they will fit TO-92 transistors provided they are openedup a little as they are fitted over the plastic encapsulation. We were able to Table 2: Capacitor Codes ❑ Value IEC Code EIA Code ❑ 0.15µF   150nF   154 ❑ 0.1µF   100nF   104 ❑ .0012µF   1.2nF   122 ❑ 100pF   100p   101 do this with the aid of a pair of longnosed pliers. The devices we used are supplied by Farnell Electronic Components Pty Ltd (Cat No. 170-061). Fig.10 shows the chassis wiring diagram for the amplifier. It must be followed exactly, in order to obtain the claimed performance. You should Table 1: Resistor Colour Codes ❑ No. ❑   4 ❑   2 ❑   2 ❑   3 ❑   2 ❑   2 ❑   2 ❑   4 ❑   4 ❑   4 ❑ 12 ❑   2 ❑ 16 Value 18kΩ 8.2kΩ 3.3kΩ 2.2kΩ 1.8kΩ 390Ω 330Ω 180Ω 150Ω 120Ω 100Ω 10Ω 1Ω 4-Band Code (1%) brown grey orange brown grey red red brown orange orange red brown red red red brown brown grey red brown orange white brown brown orange orange brown brown brown grey brown brown brown green brown brown brown red brown brown brown black brown brown brown black black brown brown black gold gold 5-Band Code (1%) brown grey black red brown grey red black brown brown orange orange black brown brown red red black brown brown brown grey black brown brown orange white black black brown orange orange black black brown brown grey black black brown brown green black black brown brown red black black brown brown black black black brown brown black black gold brown brown black black silver brown August 1998  81 Repeated from last month, this photo shows one of the assembled power amplifier modules. Note that the module has been amended slightly since the photo was taken, with the addition of two extra resistors (2.2kΩ and 10Ω) and finned heatsinks to Q11 and Q13. look closely at the photograph of the amplifier chassis to see how the various wires and connections have been routed. These are not arbitrary; the layout has been produced after much trial and error to obtain the best distor­tion, separation between channels and signal-to-noise ratio, so be sure to follow the diagram exactly. There are a number of features of the wiring which require particular comment. First, the input wiring from the RCA sockets to the volume control must not be earthed to the Fig.9: this is the full-size etching pattern for the amplifier PC board. 82  Silicon Chip Fig.10: this is the chassis wiring diagram for the amplifier. Note that it must be followed exactly, in order to obtain the claimed performance. August 1998  83 The amplifier employs a volume control so that a CD player can be connected without the need for a stereo control unit. fully against the diagrams of Fig.2 and Fig.10 and the chassis photos. chassis. It must be run exactly as shown in Fig.10. Second, the DC input cable from the power supply is clamped after it enters the chassis and then terminated in a 3-way insu­lated terminal block. The 0V line is connected to chassis via an adjacent solder lug. The three supply wires to each amplifier module are tightly twisted and laid flat against the chassis. This is to minimise any harmonic radiation from the supply leads into the input circuitry of the modules. Third, the loudspeaker wires to and from the headphone socket are tightly twisted and laid flat against the chassis. Again, this is to minimise any Setting up radiation into the input circui­try. The speaker earth wires are terminated to an insulated terminal block adjacent to the headphone socket but there is no connection to the chassis at this point. Note that the headphone socket itself is insulated from the chassis. Both the active and earth speaker terminal posts are insu­lated from the chassis but short wires run from both speaker earth terminals to an adjacent solder lug on the rear panel. Again, this might seem like an arbitrary wiring arrangement but it must be followed if the best performance is to be obtained. When you have completed all the chassis wiring, check your work care- Heavy duty gold-plated loudspeaker terminals were specified in order to obtain the lowest distortion. If the loudspeaker connec­tions are poor, the distortion performance can be degraded. 84  Silicon Chip Before the amplifier can be run with signal, the quiescent (no signal) current must be adjusted on each module. To do this, remove the fuses on both modules and wire 1.8Ω 5W wirewound resistors across the adjacent PC stakes. This done, apply power and use your multimeter to check that ±20V is present on the supply rails of both amplifier modules. Next, adjust trimpot VR1 (VR101) to obtain a voltage of 1.8V DC across one the 1.8Ω 5W resistors, on both modules. This sets the quiescent current at 1A. Leave the amplifier to run for five minutes or so and then check the voltage again. It should not drift by much but if it does, readjust VR1 to obtain 1.8V again. Then leave the amplifier to run for half an hour or so and then re-check the readings. During this time the amplifier heatsinks will become quite warm and the heatsink on the power supply case will become warmer still but that is normal. Finally, check the DC voltage at the output of each ampli­fier. It should measure less than ±50mV. You can then remove the 1.8Ω 5W resistors from both amplifier modules, reinstall the fuses and place the cover on the amplifier. You can now hook up your CD play­ er and loudspeakers and sit down to enjoy some very pleasant music. SC