Fullscreen HTML5Med Res (??MB)HTML4

Part 2: by Nicholas Vinen Complete Kit (SC6885; $70): includes the case but not a power supply Compact HiFi headphone Amplifier Introduced last month, our new Stereo Headphone Amplifier fits in a neat package and has two sets of inputs with individual volume controls. Having described its performance and how it works, we’ll go over some notes on the PCB design before getting into construction and testing. T he Headphone Amplifier circuit is fairly basic and uses all low-cost and common parts, but it delivers great performance in a small package. It’s suitable for relative beginners, with nothing being terribly tricky during the assembly process. Despite that, it still gives a very professional result. PCBs for hifi circuits are always a bit challenging to design due to the tiny levels of distortion and interference that are required to achieve good performance. So let’s take a brief look at what was involved in designing this one. PCB design It was a little tricky to fit everything into a relatively small (148 × 80mm) PCB using through-hole components, but we managed that, and the result is shown in Fig.8 and the photos. The power supply section has been kept on the left side, with the input section in the middle and the amplifier section on the right. The incoming signals arrive at the RCA connectors at the top of the board, flow down through the filtering and coupling components to the buffer op amps at lower middle, then to the volume control pots. They go to the mixer op amp to the right, and up to the transistor buffer section above, then right to the output filter and down to the output sockets. This arrangement keeps all the signal tracks relatively short, to minimise siliconchip.com.au the chance of picking up EMI or magnetic/electric fields from other parts of the PCB. It also keeps the component arrangement neat. As power needs to flow from the supply on the left side to the transistors at upper right, the positive and negative supply tracks are kept fairly wide and close together so that the magnetic loop is small. That reduces the amount of supply-ripple-induced distortion entering the sensitive signal tracks in the middle of the PCB. The output transistors have local 100μF bypass capacitors (shared between the channels) to help reduce the effect of the resistance and inductance of those supply tracks. All major ground returns are kept separate back to the power supply common point (similar to star Earthing) so that half-wave rectified currents don’t get into the signal grounds and increase distortion. If you’re wondering why only the NPN output transistors have small heatsinks attached, it definitely isn’t because we didn’t check whether there would be enough room for all four heatsinks to fit side-by-side on the PCB! Actually, during testing we found that even with reasonably high quiescent currents, the output transistors didn’t get terribly warm. Four resistors were added between VR1 & VR2 in the final version. Australia's electronics magazine January 2025  33 Still, as there was room to fit small heatsinks to the NPN output transistors (Q3 & Q5), we did so. That’s because these transistors have the Vbe multipliers (Q7 & Q8) mounted on top, so they won’t be able to dissipate heat as effectively as the PNP output transistors (Q4 & Q6) will. Also, the PCB is designed to draw heat away from all the transistors that are mounted on it (including those in the power supply). However, as the PCB’s ability to absorb, distribute and radiate heat is limited, we figured that by keeping Q3 & Q5 cooler with small heatsinks, that will reduce the total heat load on the board and thus effectively improve cooling for Q4 & Q6 as well. The heatsinks are actually sandwiched between each NPN output transistor and its associated Vbe multiplier transistor, with thermal paste in between. As the thermal resistance of the heatsink is low, that shouldn’t have any significant impact on thermal tracking for the Vbe multipliers. While we’re on the topic of output transistor ratings, we also need to keep in mind their continuous current limits of 1.5A each, especially during plugging and unplugging headphones. The output transistors have an hFE (current gain) of around 50 times at their limit of 1.5A, regardless of the junction temperature. That means, to exceed their 1.5A current limit would require a base drive of over 30mA (1.5A ÷ 50). While the NE5532 data sheet says it can typically source or sink 38mA, that’s with a ±15V supply and under short-circuit conditions. In practice, due to supply droop and other factors, with our recommended 9V AC plugpack, we were unable to get our prototype to get anywhere near the limit. Having said that, we didn’t deliberately short-circuit the output, so we can’t promise it’s short-circuit proof. But we think, if you are careful not to abuse it, it should be OK. Construction The Headphone Amplifier is built on a double-sided PCB coded 01103241 that measures 148 × 82mm. The same PCB is used regardless of which version you are building. Fig.8 is the component overlay diagram that includes all components for building the full version of the Amplifier, with two sets of stereo inputs. Fig.9 shows the same arrangement as Fig.8 but without the two buffer op amps. If you’re building it from a kit, you might as well build the full version as they are included, but it is possible to leave those two op amps out and save a few dollars. There will be more interaction between the volume controls, though. Fig.10 shows the PCB with just the components needed for one stereo input. We’ve chosen to retain CON2, but you could keep CON3 instead and fit the resistors, capacitors and potentiometers in the positions to the right instead. Regardless of which version you’re building, start by fitting all the smaller (¼W and ½W) resistors. They have colour-coded stripes that you can decode with the aid of the table in the parts list. Still it’s safer to check each set’s value with a DMM set to measure ohms before installing them. All the smaller resistors are laid flat on the PCB, so bend their leads, insert them, solder them and trim the excess. For the four 100W resistors, slip a ferrite bead over one of the leads before inserting it into the board. Solder the shorter end, then pull the other lead with a pair of pliers so it’s tight before soldering it. That should stop the ferrite bead from rattling if you move the board. Next, solder the two diodes, which are the same type. Make sure that both have their cathode stripes facing up, towards Q2. If using IC sockets, solder them in place now, ensuring the notches all face up as shown on the overlay diagrams. Otherwise, solder the ICs directly to the PCB, again ensuring that their notch or pin 1 dot faces up. Fig.8: use this overlay diagram as a guide to where to mount each component. This shows the full version with two buffered stereo inputs. Don’t forget to add the ferrite beads to the 100W resistors before soldering them and watch the orientation of the diodes, ICs and electrolytic capacitors. 34 Silicon Chip Australia's electronics magazine siliconchip.com.au This is important as they won’t work if reversed! Now is a good time to fit CON4 if you are using it. Once its pins are lined up with the pads, it should slot right into place. Solder it flat on the PCB. Next mount transistors Q2, Q4 & Q6. These are all the TTA004B PNP type. Make sure the writing is on the top side, then bend the leads down a few millimetres from their bodies so that they fit through the PCB pads while the mounting hole on the tab lines up with the one on the board. Add a small amount of thermal paste to the underside of each transistor, then feed a 10mm M3 machine screw up from underneath and push the transistor body over its shaft. Add a flat washer and hex nut on top and tighten while stopping the transistor body from rotating. Check the body is aligned properly, then solder and trim the leads. Use the same procedure to fit Q1, which is a TTC004B. Leave the other transistors off for now. Next, mount the two trimpots. They are the same type and only fit one way. Then move on to the capacitors, starting with the ceramics, which are not polarised, so they can go in either way around. Two of the 100nF capacitors are recommended to be MKT types; fit them next. They are also unpolarised. The The output filter inductors are wound on the bodies of the 1W resistors they’re paralleled with. You could add heatshrink tubing on top if you want. other 100nF capacitors can be MKT, ceramic or multi-layer ceramic, none of which are polarised. Then move on to the electrolytic capacitors, all of which are polarised. In each case, the longer lead goes into the pad next to the + symbol, with the stripe on the can facing the opposite way. The only thing to watch out here, apart from the polarity and the values being correct, is that there are three different types of 100μF capacitors specified. The four or eight capacitors marked 50V (in the middle of the board) should ideally be 50V types, to make the inputs as robust as possible. They could be lower-rated (eg, 35V) if absolutely necessary. The two low-ESR 100μF capacitors in the power supply section and two more at upper-right must be rated at least 25V, although higher-voltage types are suitable if they will fit. The two or four 100μF capacitors near VR1/VR2 can be 16V types, Fig.9: here are the difference if you’re building the two-input version without the buffer op amps. Fit the four links instead of the ICs and leave off the four 100kW resistors. siliconchip.com.au Australia's electronics magazine January 2025  35 although a higher rating certainly won’t hurt, as long as they will fit. Now is a good time to solder the two-pin header for JP1 in place. After that, fit VR1 and/or VR2, making sure they are pushed fully down and their shafts are perpendicular to the edge of the PCB. Also fit the barrel socket, again making sure it is straight and flat before using generous amounts of solder to attach it. The RCA sockets need the projection on the top cut off. It’s easiest to do it before mounting them on the board. Use a hacksaw or rotary tool to cut them off in line with the top edge of the socket face, then file or sand off any burrs or projections. Snap them into the PCB and make sure they’re flat before soldering the pins. Similarly, mount the on/off switch next. The LED goes next to the switch, with its lens at the same height as the switch shaft. Bend its leads by 90° about 3mm from the band of the lens, ensuring that when it’s inserted into the PCB, its longer (anode) lead will be to the right, as shown in the overlay diagrams. Insert and solder it so that its lens is at the same height as the switch and pot shafts. M3 machine screws. This bit can get a little fiddly and messy, so keep a damp cloth on hand, along with needle-nose pliers and angled tweezers. The mounting arrangement is depicted in Fig.11. First, bend the leads of all four transistors down so that they will fit into the PCB pads with the tab mounting hole in the correct position and the writing on the top. Make sure they can be inserted easily and that the tab hole is properly aligned, as that will make the rest of the job much easier. Insert a machine screw up through the bottom side of the PCB, then add a thin layer of thermal paste on both sides of one of the transistors. This will be Q3. Insert its leads and push it most of the way down to the PCB, then add a heatsink over the top, with the longer section projecting to the right (over Q3’s leads). Next, add thermal paste to the bottom side (only) of another transistor and add it on top of the heatsink (Q7). Place a flat washer over the screw shaft, then do up a nut on top. Hold the transistor bodies steady as you tighten the nut, then solder and trim all six leads. Repeat for the other transistor pair. Heatsinks Winding the inductors All four remaining transistors are TTC004Bs, and they are held to the board using 15mm or 16mm long We used 0.4mm diameter enamelled copper wire (ECW) to wind the inductors, although you could use a smaller diameter (down to about 0.25mm) if you happen to already have it. Cut it into two 1m lengths, then use a sharp hobby knife or emery paper to strip the insulation off the ends by 2-3mm. The inductors are wound using the bodies of the 10W 1W resistors as formers. Clamp a resistor in some sort of holder (we used the type that has mini grabbers), then add some solder to the leads on either end of the body. Solder one end of the ECW to that point, with the rest going past the body, then start winding it around the body. Try to keep it neat and closely spaced at first, although it’s basically impossible to keep it neat after the first layer. The good news is that there aren’t a huge number of turns required, so it hopefully won’t end up a jumbled mess by the time you have finished. Keep it wound tightly around the body, then solder the remaining stub close to the other end of the resistor body. Use a DMM to measure the resistance across the resistor. It should have dropped to around 0.2W (depending on your DMM lead resistance). If it’s close to 10W, that suggests the solder joint at one end (or both) is bad, so fix it. Repeat for the other resistor, then bend the leads, insert them into the PCB and solder them at similar Fig.10: if building the single-channel version, you can leave off either channel; here we’re showing CON2 fitted and CON3 not. Only one IC needs to be linked out in this case. In place of the two 1MW resistors, use 100kW instead. 36 Silicon Chip Australia's electronics magazine siliconchip.com.au heights. There’s no significant dissipation in these devices, but it’s easier to solder them spaced off the PCB, so you might as well. Finally, if you’re using the 6.35mm jack socket, CON5, solder it now. It will need to be a low-profile version to fit in the case. Jaycar’s PS0190 is unfortunately too tall, but many others like it sit lower. There are several suitable parts available from Altronics. Push it down fully and solder it in place using generous amounts of solder for good mechanical retention. Testing Adjust VR3 & VR4 fully anti-clockwise and ensure switch S1 is in the up (off) position. Plug in the plugpack and switch it on at the mains. Nothing should happen since the switch is off. Set your DMM to alternating current (AC) measurement mode (not DC!) in the amps range and connect the probes appropriately. Hold one against switch S1’s pad that’s closest to the large capacitor (ie, the one at the back & top). While watching the multimeter, touch the other to the middle pad for S1 for a second or two. If you’ve used IC sockets and the chips are not inserted, you should see a current draw of only a few tens of milliamps at most, and LED1 should light up. If all three op amps are soldered to the board, the current draw will be closer to 150mA. If you have fewer op amps installed, it will be in between (~50mA for one and ~100mA for two). If the current draw is a lot higher than that, or LED1 doesn’t light up, you have a problem. Disconnect the power supply and check the board for faults like pads bridged with solder, incorrectly orientated components, components in the wrong location etc. If it seems OK, set your DMM to measure DC volts and hold the black probe to a convenient ground point, such as the left-most pin of JP1 or the bottom-most end (closest to the PCB edge) of one of the row of four 100kW resistors between VR1 & VR2. Hold the red probe on pin 8 of one of the ICs and switch the power back on. You should get a steady DC voltage reading of around 13V DC for a 9V plugpack or 17V for a 12V plugpack. Then touch the red probe to pin 4 of the same IC, and you should get a negative voltage of a similar magnitude. Next, check the AC voltages at those siliconchip.com.au two pins. The reading should be no more than about 10mV AC (our prototype measured almost exactly 10mV with the ICs in-circuit). If you are using IC sockets and haven’t inserted them yet, switch off the power and wait for LED1 to extinguish. Install all the ICs you require, ensuring that pin 1 goes towards the upper-left corner, near the notches on the sockets. Now measure the DC output voltages relative to ground. They are available at the bottom ends of the two 10W 1W resistors that have the ECW wrapped around them. Measure those points relative to ground with the power on (see earlier for convenient ground points) and confirm that the readings are under 50mV (with either polarity). Our prototype measured around -25mV on both channels. If they are much higher than that, something is wrong, so switch off and search for faults. Rectify any problems you find and re-check the output voltages to verify they are under ±50mV before proceeding. Adjustment Connect a DMM set to measure millivolts between TP1 and TP2. The reading should be close to zero initially. Slowly rotate VR3 clockwise and by the time it reaches its midpoint, the voltage reading should start to rise. Adjust it for a reading close to 25mV (meaning 25mA quiescent current). Move the probes to TP1 & TP3 and the reading should be similar. Now connect the probes between TP4 & TP5 and perform the same adjustment using VR4. You can then check that the reading is similar between TP4 & TP6. At this point, you are ready for a listening testing. Switch off the power, rotate VR1 and VR2 fully Fig.11: the mounting arrangements for the power transistors and heatsinks. anti-clockwise and plug headphones or earphones into one of the sockets. Don’t put them over or onto your ears yet. Connect a low-level stereo audio signal source to one of the inputs, cue it up and switch the amplifier back on. Slowly wind up the volume pot associated with the channel you’re using (VR1 for CON2 or VR2 for CON3) and check that you can hear audio by moving the headphones/earphones closer to your ears. If it sounds normal, try putting them over/into your ears and adjust the volume to a comfortable level. Verify that the audio sounds normal and undistorted, with similar levels for both channels. If it sounds strange, switch off and look for faults on the PCB. Jumper option Before assembling the case, decide if you want to put a jumper shunt on JP1. With it out, if you plug headphones into both sockets, audio will only come from CON5 (CON4 will be disconnected). With it in place, the headphones will be connected in parallel and both will get audio (but possibly not at the same volume!). A close-up photo of the way the heatsinks are fitted. This is from the opposite side to that shown in Fig.11. Australia's electronics magazine January 2025  37 If you’ve only fitted CON5, it doesn’t matter if you put a jumper on JP1. If you’ve only fitted CON4, you must add the jumper, or it won’t work. While CON4’s ground is disconnected without JP1 if a plug is inserted in CON5, due to the way headphones are wired, you might still get some sound out of headphones still plugged into CON4. It’s unlikely to be anywhere near full volume, though. If it bothers you, simply unplug the unused pair. Case preparation & installation Preparing the case is relatively straightforward: all the holes to be made are in the front and rear panels, and they are all round, so you can use a drill (a stepped drill bit makes it easier). The locations are shown in Fig.12. There are six holes to make in the front panel and five at the rear, from 3mm to 10mm in diameter. You can download a PDF of Fig.12 from siliconchip.au/Shop/19/7406, print it out at actual size, cut it out and stick it to the panels using weak glue or scotch tape. Drill small pilot holes as accurately as you can in the centre of each location, then remove the templates and drill them out to the sizes shown. Deburr the holes and check that the panels fit over the assembled PCB in the case. You may need to slightly enlarge some holes if their locations are not perfect. The bottom of the case can be identified as it has four small circular recesses for feet. Stick small rubber feet in or near those locations, then secure the PCB to the base using four small self-tapping screws. Remove the nuts from the jack sockets, slot the lid on top, then push the front and rear panels in place. After that, you can attach the knobs. Our initial prototype was designed with the potentiometer and socket shafts essentially being flush with the front panel, so we couldn’t reattach their nuts. We didn’t think that was a problem as it seemed robust enough without them. Still, we made some adjustments to the final PCB so that the on/off switch, volume control pots and 3.5mm jack socket are closer to the front. That means you should be able to get the nuts back on the pots, which will provide a bit of extra rigidity, and it will make plugging into the 3.5mm socket easier, although you probably won’t be able to get its nut on. We have kept the front of the 6.35mm socket close to being flush with the front panel as we think it’s neater, and it’s mechanically secure enough without it. Using it It’s generally a good idea to wind VR1 & VR2 back to zero (or close to it) before playing audio if you don’t know if the levels set previously are appropriate. Then slowly advance the volume control associated with the input (VR1 for CON2 & VR2 for CON3) until you reach a comfortable volume level. It’s best to avoid ‘live plugging’ headphones as they can short the outputs when doing so. It will probably be OK, but it’s safer to switch the device off before plugging or unplugging. We also suggest you remove the headphones/earphones when switching the amp on or off to avoid any painful clicks or pops that may occur. This will also protect you in case you switch it on and the volume level is set too high. The amp draws no power when switched off, although AC plugpack will draw some power from the mains even when it has no load. So if you want to minimise power consumption when the amp is off, switch off the plugpack at the wall or unplug it when not in use. If you ever have to get the case apart again, it’s a bit tricky but it can be done. Remove the knobs and nuts, then detach the front panel on the switch side. The rear panel is almost impossible to remove once assembled as the RCA sockets prevent you from flexing it in such a way to release the tabs, so don’t try. Once you have the front panel off on one side, pull at the bottom on the jack socket side and squeeze the main part of the case in, and it should pop off. You can then gently lever the top off and pull it forwards to release the SC rear panel. Fig.12: the front (top) and rear (bottom) panel drilling details. Depending on how accurately you drill the holes, you may need to enlarge some slightly before the panels will snap into place. It’s best to start them all small and then increase them by a couple of millimetres at a time until they’re at full size. If building a single-channel version, only drill the two 9mm holes corresponding to the RCA sockets you have fitted (and the same for the 7.5mm potentiometer holes at the front). 38 Silicon Chip Australia's electronics magazine siliconchip.com.au