Fullscreen HTML5Med Res (??MB)HTML4

20W Class-A Amplifier; Pt.2 This fully assembled chassis shows what the Stereo Class-A Amplifier will look like. The preamplifier/remote volume control module and the loudspeaker protection module will be described in following issues. In Pt.2 this month, we present the construction details for matching left and right channel mirror-image modules, together with the circuit and construction details of the power supply. Pt.2: By Leo Simpson 20  Silicon Chip siliconchip.com.au Fig.6: follow this parts layout diagram to build the left channel power amplifier module. Be sure to use the correct part at each location and make sure that all polarised parts (transistors, diodes and electrolytic capacitors) are correctly installed. T HE NEW PC boards are longer than the original SILICON CHIP July 1988 modules. This is mainly to spread the two power output transistors further apart on the large single-ended heatsinks. This has the effect of spreading the central hot spot produced by the original module and allows us to use the heatsinks more efficiently. This time around we have also designed mirror image PC boards, for the left and right channels. This has been done to achieve a better wiring layout within the amplifier chassis and also to optimise the residual noise performance in both channels. Both PC boards measure 146 x 80mm and are coded 01105071 (left) and 0110572 (right). To ensure reliable connections to the PC boards, we have specified chassis-mount QuickConnect single-ended male spade terminals which have a mounting hole for an M4 screw. These connectors are normally used for high current connections but we are using them here because we want to ensure very low resistance connections. They have the advantage over normal soldered connections to the PC board in that they can be repeatedly connected and disconnected without siliconchip.com.au problems, when an amplifier is being assembled and checked. By the way, we do not recommend staked Quick-Connect spade lugs for this application as they are not as reliable, particularly after they have been reconnected a few times. With the same thought in mind about reliable terminations, the audio signal connection to each module is made via an on-board RCA socket. This is much better than using soldered connections for shielded cable, as they are bound to look messy after being disconnected and reconnected just once. One module or two? Before we start on the assembly details, there are a few other points to note. The first pertains to whether you are building a single PC board module to be used as a mono amplifier (unlikely, but we have to consider it). If so, note that R1 is a 0W link, as shown on the circuit of Fig.5 in last month’s issue. Alternatively, if you are building left and right modules for a stereo amplifier, R1 must be changed to a 10Ω resistor (in each channel). This is done to reduce the possibility of circulating currents in the completed stereo amplifier which could compromise the performance, particularly separation between channels. Transistor quality To ensure published performance, the MJL21193 & MJL21194 power transistors must be On Semiconductor branded parts, while the 2SA970 lownoise devices must be from Toshiba. Be particularly wary of counterfeit parts, as reported by us in the past. We recommend that all other transistors used in this project be from reputable manufacturers, such as Philips (NXP Semiconductors), On Semiconductor and ST Microelectronics. This applies particularly to the BD139 & BD140 output drivers. The component layouts for the mirror reverse boards are shown in Fig.6 (left) and Fig.7 (right). Begin each board assembly by installing the wire links, the two 1N4148 diodes (D1 & D2), and the resistors and capacitors. The resistor colour codes are shown in Table 1 but we strongly advise that you also check each value using a multimeter before it is installed. Make sure that the diodes and elecJune 2007  21 Fig.7: this is the layout for the matching right-channel power amplifier module. It’s almost a mirror image of the left-channel board. trolytic capacitors are installed with the correct polarity. That done, you can then install the fuse clips. Note the each fuse clip has a little lug on one end which stops the fuse from moving lengthways. If you install the clips the wrong way around, those lugs will stop you from fitting the fuses. Next, install the two 0.1W 5W resistors followed by trimpot VR1 (this must go in with its adjustment screw oriented as shown). The small-signal (TO-92) transistors (2SA970s, BC546s & BC556s) can then be installed. As supplied, these transistors usually have their leads in a straight line, although the centre lead may sometimes be cranked out. They have to be splayed outwards and cranked to fit nicely into their allocated positions. The way to do this is as follows. First, grip the three leads adjacent to the transistor body using a pair of needle-nose pliers and bend the centre lead back and up by about 70°. That done, grip each of the two outer leads in turn and bend it outwards and up by about 70°. Finally, grip each lead in turn at the end of the pliers and bend it downwards again – see photos. Install each transistor on the PC Fig.8: follow this diagram to attach the BD139 and BD140 transistors to their respective heatsinks. Note that each transistor is electrically isolated from its heatsink using a silicone insulating washer. 22  Silicon Chip board after dressing its leads. Note that transistor pairs Q1 & Q2 and Q3 & Q4 are installed with their flats facing each other. Make sure that you don’t install the TO-92 transistors in the wrong positions. Inadvertently swapping 2SA970s for BC556s will not have any dire consequences, except that the amplifier will not be as quiet as it would have been. But swapping BC546 NPN transistors for BC556 or 2SA970 PNP transistors will cause serious damage when the amplifier is first powered up. You have been warned! The idea is to work carefully and This close-up view shows one of the BD139 transistors (left-channel amplifier board). A second BD139 mounts on the other side of this heatsink. siliconchip.com.au Follow this photo in conjunction with Fig.7 when building the right-channel amplifier module. Note that the output transistors (Q12 & Q14) must be electrically isolated from the heatsink using thermal washers (see Fig.9). patiently through the assembly process. Check each step against the diagrams and photos as you go. Care and patience now will be rewarded later when you turn the amplifier on. The TO-126 transistors Q10, Q11 & Q13 are fitted to U-shaped flag heatsinks before they are soldered to the PC board. More specifically, Q10 and Q11, both BD139s, are mounted on opposite sides of the same flag heatsink (see Fig.8) while Q13, a BD140, is mounted on a separate flag heatsink. Note that each transistor must have a silicone rubber pad to isolate it from the heatsink – see Fig.8 and the photos. Note also that the 100pF ceramic capacitor at the collector of Q9 should be an NPO type (ie, with zero temperature coefficient). NPO capacitors have a black spot or strip across the top. If your 100pF capacitor does not have this black labelling, it is not NPO. siliconchip.com.au Other types may change their capacitance markedly with temperature, which is undesirable. Winding jig The next step is to wind the 6.8mH inductor. To do this, you need about 1.5m of 1mm enamelled copper wire which is close-wound onto a plastic bobbin. This bobbin may have an in- ternal diameter of either 11.8mm or 13.8mm, depending on the supplier. As shown in the photos, we made up a small winding jig for the bobbin, as this enables a really neat job. It consists of an M5 x 70mm bolt, two M5 nuts, an M5 flat washer, a piece of scrap PC board material (40 x 50mm approx.) and a scrap piece of timber (140 x 45 x 20mm approx.) for the handle. The leads of the TO-92 transistors are cranked to fit the PC board using a pair of needle-nose pliers. These photos show how it’s done. June 2007  23 ➊ ➋ ➌ ➍ Above: these photos show how to make a simple jig from scrap material to wind the 6.8mH inductors (see text). First, the bobbin is slipped over the collar on the bolt (1), then the end cheek is attached and the wire threaded through the exit slot (2). The handle is then attached and the coil wound using 25.5 turns of 1mm enamelled copper wire (3). The finished coil (4) is secured using a couple of layers of insulation tape and a band of heatshrink tubing. In use, the flat washer goes against the head of the bolt, after which a collar is fitted over the bolt to take the bobbin. This collar should be slightly less than the width of the bobbin and can be wound on using insulation tape. Wind on sufficient tape so that the bobbin fits snugly without being tight. Next, drill a 5mm hole through the centre of the scrap PC board material, followed by a 1.5mm exit hole about 8mm away that will align with one of the slots in the bobbin. That done, the bobbin can be slipped over the collar and sandwiched into position between the washer and the PC board (which acts as an end cheek). Align the bobbin so that one of its slots lines up with the exit hole in the end cheek, then install the first nut 24  Silicon Chip and secure it tightly. The handle can then be fitted by drilling a 5mm hole through one end, then slipping it over the bolt and installing the second nut. Winding the choke Begin by feeding about 40mm of the wire through one of the bobbin slots and the exit hole in the jig (loosen the handle if necessary to do this). Bend this end back through 180° to secure it, then tighten the handle and wind on 25.5 turns as evenly and tightly as possible. Finish by bending the remaining wire length through 90° so that it aligns with the opposite slot. The windings can now be secured using a couple of layers of insulation tape, after which the bobbin can be removed from the jig. Cut off the excess Here’s another view of the fullyassembled right-channel power amplifier module, attached to its heatsink. After mounting the output transistors, it’s a good idea to use a multimeter (set to a high ohms range) to confirm that they are correctly isolated from the heatsink. You should get an open-circuit reading between the heatsink and each of the transistor leads. siliconchip.com.au Fig.9: this diagram shows the mounting details for the output transistors (left), along with the heatsink drilling diagram (above). Note that the transistors are mounted with a lead length of 9mm using the method detailed in the text. Be sure to deburr the mounting holes using an oversize drill, to prevent punch-though of the insulating washers. siliconchip.com.au June 2007  25 Fig.10: the power supply circuit uses a centre-tapped transformer with 16V windings to drive a bridge rectifier and six 10,000mF filter capacitors. wire at each end, leaving about 10mm protruding. Finally, complete the choke by fitting some 20mm-diameter (9mm wide) heatshrink tubing over the windings. Be careful when shrinking it down with a hot-air gun though – too much heat will damage the plastic bobbin. You can now test fit the finished inductor to its PC board, bending its leads as necessary to get the bobbin to sit down flush on the board. It’s then just a matter of stripping the enamel from the wire ends and tinning them before soldering the choke in place. Power transistors The two output transistors must be installed with their plastic bodies exactly 9mm above the surface of the PC board. In practice, you have to first mount the two transistors on the heatsink. Fig.9 shows the mounting details for each device. Note that it is necessary to use a thermal insulating washer to electrically isolate each device from the heatsink. First, check that the mounting areas are smooth and free of metal swarf (deburr the holes if necessary using an oversize drill), then loosely secure each device to the heatsink using an M3 x 20mm machine screw, flat washer and nut. That done, cut a couple of 9mm wide cardboard spacers about 40mm long – these will be used to space the transistor bodies off the PC board. Next, turn the heatsink assembly upside down and slip the PC board (upside down) over the transistor leads. Push the board down so that the cardboard spacers are sandwiched between the board and the transistor bodies, then line everything up square and lightly tack solder the centre lead of each device. It’s important to now check that everything lines up correctly. The PC board should sit exactly 10mm below the edge of the heatsink, while each end of the board should be 77mm from its adjacent heatsink end (it helps to mark these points beforehand). Make any adjustments as necessary, then complete the soldering and trim the device leads. That done, you can tighten the mounting screws that secure the transistors to the heatsinks, making sure that the insulating washers are correctly aligned. These screws should be tight to ensure good thermal Table 1: Resistor Colour Codes o o o o o o o o o o o o o o No. 1 4 3 1 1 1 1 1 8 3 1 1 1 26  Silicon Chip Value 1MW 10kW 2.2kW 2.2kW 1W 5% 1kW 680W 510W 270W 100W 68W 16W 10W 1W 5% 6.8W 1W 5% 4-Band Code (1%) brown black green brown brown black orange brown red red red brown red red red gold brown black red brown blue grey brown brown green brown brown brown red violet brown brown brown black brown brown blue grey black brown brown blue black brown brown black black gold blue grey gold gold 5-Band Code (1%) brown black black yellow brown brown black black red brown red red black brown brown NA brown black black brown brown blue grey black black brown green brown black black brown red violet black black brown brown black black black brown blue grey black gold brown brown blue black gold brown NA NA siliconchip.com.au The power supply module carries the six 10,000mF 35V filter capacitors plus two LED circuits to indicate that the supply is working correctly. Fig.11: here’s how the build the power supply board. Install the Quick-Connect terminals first so that there’s no risk of damaging the expensive 10,000mF capacitors if a tool slips while tightening the screws. The capacitors can then go in, followed by the resistors and the LEDs. coupling between each device and the heatsink. Finally, check that each device is electrically isolated from the heatsink using a multimeter. You should get an open-circuit reading between each device lead and the heatsink metal. By the way, we recommend highefficiency thermal insulating washers for the MJL21193 & MJL21194 output devices (see parts list last month). Typical low-cost silicone rubber wash­ers performed poorly in our lab tests, resulting in at least 5°C higher transistor running temperatures. On a similar theme, adequate airflow siliconchip.com.au up through the heatsink fins is vital to amplifier survival and long-term reliability. This means that the amplifier must be operated in a well-ventilated area – those heatsinks do get hot (typically 30°C above ambient). That completes the assembly details of the power amplifier modules. Next, we need to discuss the power supply circuit and construction of the power supply module. Shielded power transformer As noted last month, this new design dispenses with the regulated power supply and uses a bridge recti- fier and a bank of filter capacitors. Fig.10 shows the circuit. As can be seen, it employs a centre-tapped transformer with 16V windings to drive a bridge rectifier and six 10,000mF 35V electrolytic capacitors (30,000mF on each side) to provide balanced ±22V DC supply rails. Also included in the power supply circuit are two LEDs and two 2.2kW resistors to provide a visible indication that power is present on the supply rails. This is very handy when you are working on the amplifier. Finally, there are two 100nF MKT polyester continued on page 30 June 2007  27 Measuring Ultra-Low Harmonic Distortion How good are our new Class-A audio amplifier modules? Well, they are too good to measure on our Audio Precision test gear, as we shall see. Back in 1998 in the class-A amplifier article, we noted the great difficulty in measuring the very low distortion of the circuit. The main problem is that, at lower power levels, circuit noise tends to completely obliterate the measurement. Even at full power (20W), the noise in the signal is quite significant. To put that into perspective, the signal to noise ratio of the new amplifier with respect to full power is -115dB unweighted (ie, with a noise bandwidth from 22Hz to 22kHz) which is very, very low. How low? Think of a noise signal which is only 22 microvolts! Compare that with the total harmonic distortion which is typically .0006% (-104dB or 76mV) and you can see that noise is a significant part of the measurement. In the July 1998 article we demonstrated a method to remove the noise component of a THD (total harmonic distortion) signal using the averaging feature of a Tektronix TDS360 digital scope. The noted audio designer, Douglas Self, devised this method. This technique can filter out virtually all the random noise signal to leave the harmonic content displayed. Fast-forward nine years to June 2007 and we can do the same procedures using our vastly more capable LeCroy WaveJet 2Gs/s 200MHz digital oscilloscope. We often feature screen grabs from this scope to demonstrate circuit performance. However, the LeCroy WaveJet does not allow us to perform normal sampling and averaging on the same signal simultaneously and we wanted to do this in order to more clearly demonstrate the dramatic effect of noise averaging using a digital scope. What to do? It turns out that LeCroy have a much higher performance scope which would let us do this procedure. So, thanks to Charles Holtom of Trio Smartcal (phone 1300 853 407), we managed to gain access to a LeCroy WaveRunner 10Gs/s 600MHz scope. We performed three tests to dem­ onstrate the extremely high performance of our new amplifier. The accompanying three scope screen grabs each show three signal traces. In each case, the top trace is the fundamental – ie, a 1kHz sinewave. The trace below that is the residual THD signal after the fundamental Scope1: the THD measurement of the amplifier at 1kHz and 20W. Note the much cleaner averaged bottom trace (green). 28  Silicon Chip siliconchip.com.au 1kHz sinewave has been nulled out by our Audio Precision automatic distortion test set. Both these traces are displayed using normal scope sampling so all the noise in the signal is clearly shown as a large random component. The bottom trace is displayed using the averaging technique and is in fact the average of 128 sweeps of the trace. Furthermore, we have applied a degree of digital filtering to limit the noise in the displayed signal. Scope1 shows the measurement of the new amplifier at 20W. The total harmonic distortion was .00056%. To explain this, the middle trace represents an RMS voltage which is .00056% of 12.69V, the signal level needed for 20W into an 8-ohm load. As presented on the scope, the middle trace has a mean (ie, average) value of 4.54mV RMS. Now look at the averaged trace (bottom). Not only is it almost completely devoid of random noise (revealing the true harmonic content) but its RMS value is only 1.96mV RMS. This enables us to recalculate the true harmonic dis- This photo shows the prototype modules under test using the LeCroy WaveRunner scope and our Audio Precision test set. These tests also allowed us to optimise the wiring layout in the test chassis. tortion to be around .00024%! Wow. By the way, the scope displays a full set of measurements for channel 3 (blue) and channel 4 (green), including instantaneous value, mean, min, max and standard deviation. Scope2 is even more dramatic as it demonstrates the THD measure- ment at a power level of 1W. Here, the measurement is .001%, much worse than for full power but in this case the fixed residual noise level of around 22mV is much more significant compared to the THD residual which is 56mV. In this case, the THD . . . continued next page Scope2: the THD measurement of the amplifier at 1kHz and 1W. Here the residual noise (trace 2 – blue) is much greater and the averaged trace (green) is much cleaner. siliconchip.com.au June 2007  29 Scope3: the THD measurement of the Audio Precision test set at 1kHz and 600mV. Measuring Ultra-Low Distortion: continued from previous page trace is 2.2mV RMS compared to the averaged trace (bottom) of 531mV. Recalculating the harmonic distortion in the same way again gives a result of .00024%. This clearly shows that the harmonic distortion does not increase when the power level of the amplifier is reduced. Well, that’s great but it is not the whole story because when we measure the Audio Precision distortion test set itself, its THD is .0004% at 1kHz at a level of 600mV. Scope3 shows the equivalent process and after averaging the harmonic distortion, the reading is capacitors to provide a high frequency bypass filter on each supply. However, the real feature of the power supply is the magneticallyshielded toroidal power transformer. Most people would be aware that standard toroidal power transformers have quite a low leakage inductance and therefore little hum radiation 30  Silicon Chip .00024%. But isn’t that the same as the above readings for the amplifier? Yep. So in fact, we don’t know how good the amplifier really is. Based on these figures, it might be less than .0001% but we have no way of knowing. As a further exercise, we were able to do spectrum analysis using the LeCroy WaveRunner’s FFT facility. However, while that showed the first harmonic content at down below .0001% for the Audio Precision’s generator and similar low figures for the amplifier, the tests simply did not let us make any further estimates. By the way, measuring a level of .0001% with respect to a 600mV signal actually refers to a signal comwhen compared to conventional EI laminated transformers. That is correct but the hum radiation from a standard toroidal power transformer is still not low enough when used in conjunction with these high performance class-A amplifier modules, as we found with our 1998 design. Because of the constant power ponent of just 6mV. The FFT analysis was able to measure harmonics out to the 19th, at much lower levels, so we were looking at harmonic components as little as -130dB with respect to the fundamental signal level. This is far below the amplifier’s residual noise level; such is the capability of the LeCroy WaveRunner oscilloscope. It has 11-bit precision, enabling accurate measurements even at just a few microvolts. So when you look at the overall harmonic distortion figures published in Pt.1 (and to be published in future months for the completed stereo amplifier) remember that they don’t tell the true story. This amplifier is actually too good for us to measure properly. demand of about 100W drawn by the two modules, the transformer still has quite a significant hum field and this is a real problem when it is operated in close proximity to the amplifier modules. Our solution in the 1998 design was to use a separate power box, to keep the transformer well away from the modules. siliconchip.com.au This life-size view shows the fully assembled left-channel amplifier module. Note that some minor changes were made to the PC board (just to the right of the RCA audio input socket) after this module was assembled This time around, we are specifying a shielded toroidal transformer, to keep the leakage inductance much lower. This employs a number of long strips of grain-oriented steel wound around the outside of the finished transformer and then covered in several layers of insulation. The unit looks just like any other toroidal transformer but the hum field is much lower. In addition, the transformer is oriented to give the best performance when it is finally installed in the chassis. As shown in the specifications panel last month, the end result is excellent, with extremely impressive signal-to-noise ratios and harmonic distortion figures. We will discuss this further in a future article. Power supply assembly The PC board for the power supply accommodates the capacitors, the siliconchip.com.au two LEDs, their resistors and that’s it. The 35A bridge rectifier mounts on the chassis which is necessary to remove the significant amount of heat produced by it. The power supply PC board is coded 01105073 and measures 135 x 63mm. As with the amplifier modules, all the connections to it are made via chassismount Quick-Connect male spade terminals which have a mounting hole for an M4 screw. Fig.11 shows the parts layout on the PC board. Install the Quick-Connect terminals first. As shown, three doubled-ended terminals are installed at the DC end of the board (ie, the same end as the LEDs), while three singleended termainals are installed at the bridge rectifier end. Once all the Quick-Connect terminals have been tightly secured to the PC board, you can then install the six PC-mount electrolytic capacitors. Make sure that you mount them with the correct orientation otherwise there will be an almighty bang when you first turn on the power! Finally, mount the MKT capacitors, the resistors and the two red LEDs. That’s it – the power supply board is complete. Next month Next month, we will describe the Loudspeaker Protector module, with the Preamplifier & Remote Volume Control Module to follow. And just in case you are wondering, the remote volume control will be achieved using a motorised pot and will work with a standard universal remote transmitter. In the meantime, don’t be tempted to power up the amplifier modules – there’s a set procedure to follow with regards to setting the quiescent current SC through each output stage. June 2007  31