Silicon ChipAUDIO OUT - November 2023 SILICON CHIP
  1. Outer Front Cover
  2. Contents
  3. Subscriptions: PE Subscription
  4. Subscriptions
  5. Back Issues: Hare & Forbes Machineryhouse
  6. Publisher's Letter: Upping your electronics housing game
  7. Feature: Where’s my pneumatic car? by Max the Magnificent
  8. Feature: Net Work by Alan Winstanley
  9. Feature: The Fox Report by Barry Fox
  10. Project: LC Meter Mk3 by Charles Kosina
  11. Project: DC Supply Filter for vehicles by John Clarke
  12. Project: PM (particulate matter) ‘Dust’ Sensors by Jim Rowe
  13. Project: Anodising Aluminium at home by Phil Prosser
  14. Feature: Arduino Bootcamp – Part 11 by Max’s Cool Beans
  15. Feature: Circuit Surgery by Ian Bell
  16. Feature: AUDIO OUT by Jake Rothman
  17. Project: Electronic Building Blocks by Julian Edgar
  18. PCB Order Form
  19. Advertising Index

This is only a preview of the November 2023 issue of Practical Electronics.

You can view 0 of the 72 pages in the full issue.

Articles in this series:
  • (November 2020)
  • Techno Talk (December 2020)
  • Techno Talk (January 2021)
  • Techno Talk (February 2021)
  • Techno Talk (March 2021)
  • Techno Talk (April 2021)
  • Techno Talk (May 2021)
  • Techno Talk (June 2021)
  • Techno Talk (July 2021)
  • Techno Talk (August 2021)
  • Techno Talk (September 2021)
  • Techno Talk (October 2021)
  • Techno Talk (November 2021)
  • Techno Talk (December 2021)
  • Communing with nature (January 2022)
  • Should we be worried? (February 2022)
  • How resilient is your lifeline? (March 2022)
  • Go eco, get ethical! (April 2022)
  • From nano to bio (May 2022)
  • Positivity follows the gloom (June 2022)
  • Mixed menu (July 2022)
  • Time for a total rethink? (August 2022)
  • What’s in a name? (September 2022)
  • Forget leaves on the line! (October 2022)
  • Giant Boost for Batteries (December 2022)
  • Raudive Voices Revisited (January 2023)
  • A thousand words (February 2023)
  • It’s handover time (March 2023)
  • AI, Robots, Horticulture and Agriculture (April 2023)
  • Prophecy can be perplexing (May 2023)
  • Technology comes in different shapes and sizes (June 2023)
  • AI and robots – what could possibly go wrong? (July 2023)
  • How long until we’re all out of work? (August 2023)
  • We both have truths, are mine the same as yours? (September 2023)
  • Holy Spheres, Batman! (October 2023)
  • Where’s my pneumatic car? (November 2023)
  • Good grief! (December 2023)
  • Cheeky chiplets (January 2024)
  • Cheeky chiplets (February 2024)
  • The Wibbly-Wobbly World of Quantum (March 2024)
  • Techno Talk - Wait! What? Really? (April 2024)
  • Techno Talk - One step closer to a dystopian abyss? (May 2024)
  • Techno Talk - Program that! (June 2024)
  • Techno Talk (July 2024)
  • Techno Talk - That makes so much sense! (August 2024)
  • Techno Talk - I don’t want to be a Norbert... (September 2024)
  • Techno Talk - Sticking the landing (October 2024)
  • Techno Talk (November 2024)
  • Techno Talk (December 2024)
  • Techno Talk (January 2025)
  • Techno Talk (February 2025)
  • Techno Talk (March 2025)
  • Techno Talk (April 2025)
  • Techno Talk (May 2025)
  • Techno Talk (June 2025)
Items relevant to "Stewart of Reading":
  • Bookshelf Speaker Passive Crossover PCB [01101201] (AUD $10.00)
  • Bookshelf Speaker Subwoofer Active Crossover PCB [01101202] (AUD $7.50)
  • Bookshelf Speaker Passive and Active Crossover PCB patterns (PDF download) [01101201-2] (Free)
  • Bookshelf Speaker System timber and metal cutting diagrams (PDF download) (Panel Artwork, Free)
Articles in this series:
  • Easy-to-build Bookshelf Speaker System (January 2020)
  • Building the new “bookshelf” stereo speakers, Pt 2 (February 2020)
  • Building Subwoofers for our new “Bookshelf” Speakers (March 2020)
  • Stewart of Reading (October 2023)
  • Stewart of Reading (November 2023)
  • ETI BUNDLE (December 2023)
  • Active Subwoofer For Hi-Fi at Home (January 2024)
  • Active Subwoofer For Hi-Fi at Home (February 2024)
AUDIO OUT AUDIO OUT L R By Jake Rothman Discrete audio op amp – Part 3 PR1, 2 5kΩ miniature 5mm round preset Bourns 3321H Capacitors All 20% tolerance unless otherwise stated. R8 R9 R11,12 110Ω 5% 5.6kΩ 5% 1Ω 5% Power supply GND + – C12 C13 cbe A* C11 ebc C7 + OA+ C8 TR2 + C4 TR5 C2 TR1 TR4 Notes Op amp components have black labels External components have red labels ‘OA’ denotes op amp connections ‘–IN’, ‘+IN’ and ‘OUT’ are audio connections TR3 cbe B* ebc PR1 R23 TR9 PR2 + TR13 Thermal pad C9 TR11 C10 C5 + TR6 R20 R21 R5 R6 R8 R7 –IN OA– TR7 C14 R18 R19 R1 R2 R3 R4 R16 R26 + + R13 R14 + Op amp input GND +IN R15 R27 C1 R17 TR12 LED C6 + 56 180Ω 1% 330Ω 1% 470Ω 5% 2.2kΩ 5% 10kΩ 5% 220Ω 5% 6.8kΩ 5% 3.9kΩ 5% 12Ω 1% (2.2Ω 5%*) 2.7kΩ 5% 1.5kΩ 5% 43kΩ 5% 5.6kΩ 5% 47kΩ 5% 22kΩ 1% 1kΩ 1% 4.3kΩ 1% 22kΩ 1% 47Ω 5% Low impedance, low-voltage option Changes to above are as follows: Mute Resistors All 1% metal film 0.25W, otherwise 5%, for non-critical devices. R1,2 R3,4 R5 R6 R7 R8 R9 R10 R11,12 R13 R14 R15 R16 R17 R18 R19 R20 R21 R22 Miscellaneous L1 3.9-15µH axial inductor 20% 20ºC/W heatsinks plus M3 nuts, bolts and washers x2* 0.1-inch single-sided terminal pins x7 0.35-inch wire links (low power only) x2 C1 220nF, ceramic 2.5mm X7R C2 1µF, tantalum bead 10V C3 39pF, 5mm NP0 5% / 47pF* C4 10µF, tantalum bead 10V C5 100nF, ceramic 5mm X7R C6 1µF tantalum 35V C7 4.7µF electrolytic 35V C8 150pF ceramic 5mm NP0 5% C9 100µF electrolytic 10V C10 100µF electrolytic 25V C11 10pF ceramic 5mm NP0 5% C12 47pF 5mm NP0 5% / 22pF* C13 680pF polypropylene 10% R10 R9 Added components marked * are for the high-power ±25V, 180Ω version. Values for the low-impedance design are given separately later. Semiconductors TR1,2,6,7,8,9,12 BC546B NPN small-signal medium-voltage TR3,4,5,10,11 BC556B PNP small-signal medium-voltage TR13 BD140, NPN high-voltage power* TR14 BD139, PNP high-voltage power* LED1 red, low-current 5mm C3 Components list C14 10µF electrolytic 25V C15 100pF, ceramic 2.5mm NP0 5%* R23 100k Ω 5% R24,25 link (560Ω 5%*) R26 3.3kΩ 1% R27 330Ω 5% (link for max headroom, insert for best PSRR) External power transistor b c e TR8 Op amp output OUT TR10 C15 R11 R12 R24 R25 L1 R22 O ver the last two issues I’ve covered the theory and ideas behind my Discrete Audio Op Amp, so now it’s time to start the ‘soldering bit’ for a standard non-inverting x6-gain amplifier. The good news is that this is pretty easy. The ‘science bit’ will be testing all the possible variations and values to tweak it for your application. If you can wait, I intend to use this discrete op amp as an optimised module in many specialised audio designs. In this article I have provided construction details for three variations. First, the standard low power ±25V, 600Ω version, and then two high-power types using extra output transistors. One is a high-voltage ±25V, 180Ω version and the other is a low-voltage ±12V, 30-150Ω design, mainly for headphones. Last, but not least, I’ve included plenty of R&D data for experimenters and tinkerers. OA Out TR14 b c e External power transistor SMD dual transistors If you are using dual bipolar transistors instead of through-hole devices then: *A is for TR1 and TR2 – see Fig.23 in Part 2. *B is for TR4 and TR5 – see Fig.23 in Part 2. Fig.35. PCB overlay for the discrete op amp – insert all components for the non-inverting amplifier. Feed the input into +in and use the OUT pin. (Note there are a few minor changes from the version 1 PCB shown in Fig.16 in Part 2) C12 and C13 were reverse numbered, R27 and C15 added). Practical Electronics | November | 2023 220Ω 5% 6.2kΩ 5% 6.8kΩ 5% omit 100Ω 5% L1 –IN C12 TR2 C4 + TR5 TR3 C3 + C2 TR1 TR4 PR1 R23 TR8 Op amp inverting amplifier output OUT TR10 C15 TR14 R11 R12 R24 R25 L1 R22 C11 TR9 PR2 Thermal pad TR6 C5 TR7 C13 + TR11 C10 + C14 R10 R9 + R16 R26 Op amp inverting amplifier input R15 R27 C1 R13 R14 R17 TR12 LED C6 R1 R2 R3 R4 TR1,2,6,7,12 BC549C NPN small-signal low-noise TR3 MPSA63/4 PNP small-signal Darlington TR4,5,11 BC559C small-signal low-noise TR8,9 BC337 NPN medium-power TR10 BC327 PNP medium-power TR13 BD436 PNP power TR14 BD435 NPN power GND + – + C3 100pF 5% NP0 ceramic C9 100µF 16V bi-polar electrolytic, Suntan CD71 or equivalent. C10 220µF 16V bi-polar electrolytic, Suntan CD71 or equivalent. Omit for DC coupling. C12,13,15 omit C14 100µF 16V electrolytic Power supply R20 R21 R5 R6 R8 R7 R24,25 R10 R15 R26 R27 Notes Op amp components have black labels External components have red labels ‘OA’ denotes op amp connections ‘–IN’, ‘+IN’ and ‘OUT’ are audio connections Inverting amplifier Remove C7,C9, R18, R19 Replace C8 with link R21 14µH, low-Z choke V+ R20 PCBs and kits containing the harder-to-find parts for the basic Discrete Op Amp are available from the PE PCB Service – www.electronpublishing.com Remember, that all parts, including any unusual ones mentioned in these articles are available separately from the AO Shop – see advert on p.64. OA– Input – OA+ Output + OAO Gain = –R21/R20 V– Discrete op amp 0V PCB stuffing Fig.36.(above) Discrete op amp inverting amplifier connections. Fig.37. (below) Discrete op amp differential amplifier connections. Power supply Mute TR2 C13 + TR5 C11 RGND C7 TR4 R19 R1 R2 R3 R4 TR1 C5 TR3 C2 PR1 R23 R10 R9 C4 + Thermal pad C12 TR9 PR2 C3 3 +IN C14 TR6 TR11 C10 R20 R21 R5 R6 R8 R7 –IN + + R13 R14 TR7 + Practical Electronics | November | 2023 + R16 R26 2 R15 R27 C1 TR8 OUT OP amp differential amplifier output TR10 C15 R11 R12 R24 R25 L1 R22 OP amp differential amplifier input Power op amp The output current and hence power of the op amp can be increased by adding an extra pair of output transistors, TR13 and TR14. These are coupled to the original output transistors by adding 560Ω collector load resistors, R24 and R25. This forms a complementary follower pair (CFP) output. This means we don’t need thermal sensing on the output transistors, as we would if a second push-pull emitter-follower stage was added on. Make sure the metal part of the transistor case faces inwards to the board, as shown in Fig.38. Small heatsinks are a good idea, such as those shown in Fig.39. The emitter resistors R11 and R12 are reduced to 2.2Ω for the high-power op amp, so more current can be delivered. The quiescent current has to be set to a minimum of 13mA for low distortion, giving 30mV across the resistors. The distortion curve is shown in Fig.40. R17 TR12 LED C6 XLR 1 GND + – + If you are using the SMT transistors on the input stage, solder these first; you don’t want other components getting in the way. If you’re making the low-power version, fit wire links into positions R24 and R25. Although it’s normal practice to solder semiconductors last, I think it’s best to fit the output transistor TR9 and TR10 along with the associated temperature-tracking bias transistor TR8 on the thermally conductive pad while the space around is clear. A bit of thermal paste under the transistors is a good idea. Bend, position and hold them down flat before soldering. Fig.35 shows the overlay for the x6-gain non-inverting amplifier. Fig.36 and Fig.37 show how to configure the PCB for the standard inverting and differential configurations respectively, as described in Part 2. To make a buffer, just link R21 and leave out the other feedback parts, C9 and R20. Notes Op amp components have black labels External components have red labels ‘OA’ denotes op amp connections ‘–IN’, ‘+IN’ and ‘OUT’ are audio connections Differential amplifier Remove: C9, R18 Replace C8 with a resistor (RGND) R21 All resistors same value for unity gain, eg 1kΩ R20 Input– V+ OA– OA+ Input+ – Output + OAO R19 This resistor (RGND) in C8 position V– Discrete op amp 0V 57 Fig.38. TO126 transistor orientation – note PNP is on the left, NPN on the right. Fig.40. Distortion curve for high-power version of the discrete op amp using MJE243/53 output transistors (TR13 and TR14). Ratings are: ±25V, 180Ω load, 8Vpk-pk. most, including my lovely 50Ω Sennheiser HD515s, open circuit in microseconds. When increasing the current output it is usually necessary to increase the VAS current to avoid premature clipping on the negative cycle. This can easily be done by reducing R8 to 110Ω, increasing the current to 9mA. It’s also necessary to increase the quiescent current to 50mA, by adjusting PR1. Cheap low-voltage output transistors can be used, such as the BD135/6. Many headphone amplifiers use an output series resistor of 100Ω to equalise the power between different impedances and give extra short-circuit protection. This is important because shorts often occur in the jack plug. This resistor is also a good idea with the high-power op amp, since these headphones only need about 3Vrms for full volume. Warning: using a high-voltage amp with sensitive headphones, such as the famous 0.004%, compared to 0.002% for the lower-power version The board is not designed for standard second-order compensation, as shown in Fig.27 last month. However, if you want to experiment, it can be connected as shown in Fig.41. GND + – R16 R26 TR7 + C12 TR2 OA+ TR5 TR3 C2 C11 GND +IN + C4 + TR1 C7 TR4 Notes Op amp components have black labels External components have red labels ‘OA’ denotes op amp connections ‘–IN’, ‘+IN’ and ‘OUT’ are audio connections C5 PR2 C9 C8 TR6 PR1 TR13 TR9 R13 R14 –IN OA– TR11 C10 External power transistor b c e TR8 OP amp output OUT TR10 C15 R11 R12 R24 R25 L1 R22 OP amp input + + C14 R23 + Thermal pad C6 R15 R27 C1 R10 R9 R17 TR12 LED + 58 Power supply Mute + Adding an extra set of output transistors increases the open-loop high frequency loss and phase shift. This can cause oscillation when negative feedback is applied unless the compensation is optimised. The main problem with the CFP configuration is that each driver and output transistor pair have their own individual feedback loop. This loop itself can sometimes become unstable. The low-cost, ever popular BD139/140 pair commonly used in discrete op amp outputs can be unstable, which was the case here. One problem is that the PNP BD140 (TR13) is slower, having a transition frequency (Ft – where the gain falls to one) of 70MHz, compared to the BD139 (TR14), which is 250MHz. I suspect this is the cause of the negative side of the cycle bursting into oscillation, a common problem with CFP stages. It is usually fixed by adding a 100pF capacitor (C15) across the base-collector junction of the driver transistor (TR10), slowing down the negative section of the output stage to match the positive side. Further tweaking was also done to the compensation, with C3 increased to 47pF and C12 decreased to 22pF; basically, the inclusive Baxandall second-order compensation had to be reduced. The distortion was the same in the higher-power option as the lower-power version with most transistors, except the BD139/40 which gave around This is possibly the most important application for the discrete audio op amp. The standard output transistors are fine for traditional high-impedance headphones, such as the 600Ω Sennheiser HD480s. The common Beyerdynamic DT150 studio headphones are normally 250Ω, so these will need the extra output transistors. Some more recent studio headphones, such as the Beyerdynamic DT770 can be as low as 32Ω or 80Ω, which will need a high-current output. Low-voltage rails are also a good idea here, using ±25V rails would render R20 R21 R5 R6 R8 R7 Instability and compensation Headphone amplifiers R18 R19 R1 R2 R3 R4 Fig.39. Heatsinks are needed for highpower low-impedance operation. OA Out TR14 b c e External power transistor Second-order compensation 3.3kΩ R26 82pF C12 omit C3, C13 Added capacitor 82pF (shown in green above) Fig.41. Second-order compensation can just about be added, but it’s a ‘bit messy’. It is an interesting experiment for tinkerers though. Practical Electronics | November | 2023 +12V R27 100Ω TR3 MPSA63/4 R3 330Ω + R4 330Ω DC Offset Input + C14 C7 4.7µF 100µF 16V R19 1kΩ R18 22kΩ C8 150pF OA+ Noninverting input 2.7mA PR1 5kΩ TR4 BC559 C1 220nF R24 220Ω 1mA R1 180Ω R2 180Ω TR6 BC549C + C4 10µF R9 6.8kΩ 1mA TR1 BC549C TR2 BC549C R5 470Ω TR9 BC337 C3 100pF TR5 BC559 R10 6.2kΩ OA– Inverting input PR2 5kΩ R7 10kΩ R6 2.2kΩ C2 1µF 10V TR13 BD436 with heatsink TR8 BC549 TR12 BC549 R11 1Ω TR11 BC556 R14 1.5kΩ Iq set C5 100nF 54mA TR10 BC337 R12 1Ω R22 47Ω OAO 4.5mA R8 110Ω R23 100kΩ R15 6.8kΩ R16 5.6kΩ R17 C6 47kΩ 1µF + 35V C10 220µF 16V Bipolar Output Thermal link TR7 BC549 LED red low I + 10V L1 3.9 to 15µH 54mV Low resistance R13 2.7kΩ Mute (Pull to 0V or V–) R25 220Ω 2.7mA TR14 BD435 with heatsink –12V R20 4.3kΩ C9 100µF 16V Bipolar R21 22kΩ C11 22pF 0V Fig.42. Low-impedance 32Ω headphone op amp driver circuit. Note the use of a Darlington transistor for TR3 and value changes white 400Ω DT100 headphones used by musicians in studios for decades, can cause noise-induced hearing damage. I know, I now have severe high frequency loss from recording vocals with them. Low-impedance version Op amps have a similar circuit topology to power amps, which are basically op amps which can drive low impedances. In other words, it’s easy to use this PCB to build mini power amps. The low-cost BD135/6 output transistors are useful for synthesisers, spring-line drivers, low-impedance headphone amps and other audio systems where a small low-voltage power amp is required. Small high-impedance monitor speakers can also be driven. If using a power amp chip, such as an LM380 with a standard 8Ω speaker, an additional high-current ‘odd-voltage’ power rail is needed. By using an 80Ω speaker, the 50mA current consumption from standard op amp ±12V power rails can usually be accommodated, avoiding this power supply issue. The output power is 600mW. For lower impedances, such as 8-32Ω, use lower voltage rails from ±9V to ±12V and use the higher current transistors BD435/6. With these low-voltage rails, high-gain, low-Vce transistors can be also Practical Electronics | November | 2023 be employed in the earlier stages, such as BC549C and BC559C. TR3 was upgraded to an MPSA63 Darlington giving the least distortion. (Note this device has to be fitted to the PCB rotated by 180°). TR9 and T10 were upgraded to higher-current (BC327/37) devices to increase the drive to the output transistors. The Iq is increased to 54mA and the VAS current to 9mA. Using ±12V you get 1.4Wrms into 32Ω, or 2.13W into 15Ω. A lot of component value changes were needed (see Fig.42. and parts list), giving an indication of the work required to optimise the circuit for a particular application. With these ‘slow’ power transistors the distortion is a little higher at 10kHz (see Fig.43), but much less than standard power-amp chips. The completed board is shown in Fig.44. Preliminary testing A dual power supply with a current limit is needed. Preferably, if there is Fig.43. Distortion curve for Discrete Audio Op amp low-impedance headphone driver – load 33Ω, 2W resistor, ±12V supply. 59 Fig.44. (above) Low-impedance version of the discrete op amp – this uses the MPSA63 Darlington for TR3; notice reverse orientation of the device. Fig.45. (right) A good power supply set up using cheap radio rally equipment. A Weir 762.1 dual-rail power supply is wellworth seeking out for audio work. Old fashioned ex-college 150mA moving-coil ammeters complete the set-up. You can hear them hit their end-stops if the current jumps! overcurrent on one rail it should shut off the other. It is also desirable to have current meters on each output. My power supply test set-up is shown in Fig.45. To start with, set the PSU to a low voltage, say ±9V. The current limit should also be set at around 200mA. Before turning on, put the offset preset PR1 to its central position and the quiescent current preset PR2 to minimum current (maximum resistance, fully clockwise) and check if the power supply is connected the right way. Now for the DC conditions check – do not connect a load at this stage, since if there’s a fault it can provide a low-resistance current path to ground, aiding possible destruction. In directly coupled circuits such as this, just one transistor wired up wrong Zero-point crossing – output transistors off Subtle testing V +3V Distortion t –3V Distortion residual V 100mV a) t Distortion residual V Spikes removed Fig.46. Setting Iq: a) Bad crossover spikes, b) Clean residual. 60 For distortion testing, a load resistor is required. Low levels of distortion are very difficult to measure without specialist instruments. I used to use a John Linsley-Hood notch filter and thermistor-controlled oscillator, but this could only measure down to 0.02% THD. I later upgraded to a second-hand Audio Precision SY-2712 from Stuarts of Reading which produces the curves shown here: excellent hardware, awful software. High distortion is usually caused by instability, excessive loading or insufficient quiescent current. Looking at the distortion residual (that is what’s left after filtering out the fundamental sinewave) can show the optimum quiescent current setting. The crossover distortion is revealed by characteristic spikes, as shown in Fig.46. Adjust the preset until they just disappear. Noise –100mV b) or a PNP inserted instead of an NPN, can cause it to latch to a power rail and destructive currents to flow. Switch on at a low voltage. The LED should light up and the current drawn should be between 10 and 18mA, and equal on both rails. It’s then safe to slowly ramp up the supply voltage. If the current suddenly jumps that is usually an indication of oscillation. The LED will extinguish if this or overloading occurs. The next stage is to check the output offset voltage is below ±0.2V. Now it’s time for AC testing. Use a 1kHz sinewave and oscilloscope. Check it clips symmetrically and there is no oscillation. Look for ringing and overshoot with a square wave. Then try it with a 620Ω load resistor. This all sounds like a lot, but it’s just standard industry testing for any audio amplifier development. Next, set the offset preset (PR1, if fitted) and make sure the DC output is tweaked down to within a few mV either way. Finally, using PR2 set the quiescent current to 70mV across the emitter resistors, R11 and R12. If all is well, increase drive up to clipping at full voltage with a suitable load resistor. t The way I test noise level is to set the gain of the op amp to 1000x and put it in a metal box powered by two PP3 batteries. An input load resistor reflecting the source is a good idea, enabling the LTP transistors (TR1 and TR2) to be selected and the current adjusted using R5. The set-up can then be tweaked for minimum noise. I have a 470mH Toko 10RBcoil and 300Ω resistor in series to represent a moving-magnet phono pick-up. The output of the op amp is connected to a scope via a BNC connector. Practical Electronics | November | 2023 Fig.47. Distortion curve of 2N5564 JFET low-power version: load 600Ω, 6Vpk-pk output. More transistor options month). Remember to link out the emitter resistors R1 and R2. I tried increasing the current to 2mA by reducing R5 to 220Ω and correspondingly reducing the current mirror resistors R3 and R4 by half. This made no difference, but I was surprised to see that with the current mirror resistors linked out, distortion was much worse. Interesting – I think the resistors linearise the mirror. Using a 2N5564 dual JFET at ±15V, the distortion is a bit higher than the bipolar version, but at the full rail voltage of ±25V it was better, as shown in Fig.47. The offset was very low without trimming, around 3.4mV. A photo of the board is shown in Fig.48. With random BF244A devices the offset was high in the order of 0.5V, although it was easy to select matched devices using sockets. Toshiba do a dual SMT JFET, the 2SK2145-Y, which is cheap (50p from Mouser). It’s got a strange pin-out where the sources are joined together. The 2N5564 dual JFETs (pin outs shown in Fig.49) are also available from Digikey and Keytronics (in the UK). The SOI-23-6 adapter board was designed by Rex Harper at www.QRPme. com – a US site. It’s also available from Telford and District Amateur Radio Society (TADAR) and the AO Shop. This JFET op amp I only have two leaded dual JFETs in my AO Shop stock, the 2N5564 and the E402. So I thought it was worth trying those first in the long-tailed pair. The E402 had low transconductance so I thought it best to keep selling those for Moog synthesiser oscillators. The cost of discrete dual JFETs from distributors, such as the Hi-Fi 2SK389, is shocking. Possibly the best input devices in the world are the InterFET devices IF3601 (£22) and the dual IF3602 (£60) from Mouser. If I was designing a hydrophone pre-amp for the military they might be a good choice. Increasing the drain current in JFETs increases the transconductance, but unlike bipolar transistors, there’s no increase in bias current to worry about. Also, the noise voltage in proportion to current decreases up until Idss (the saturation current) is reached. Thus, the operating current (Id) through JFETs may be quite high: 1 to 5mA per device. If you want a cheap JFET for experimenting, the BF244A is a good choice because it has a centre gate pin and is inserted in the PCB with the same orientation as a BC546 (see Fig.25 last 2N5564 TO-71 is shown in Fig.50. Stop press. I’ve just designed a drop-in mini PCB created by Mike Grindle which I’ll put at the end of the article. I’ve only just received them, see Fig.56 for a photo. John Linsley-Hood The late John Linsley-Hood, a long time writer for PE, had some interesting ideas, which I just had to try. He used a VN1210M MOSFET for the VAS stage (TR3) in his popular power amps (Electronics Today International, July 1984). This provides a minimal load on the current mirror output (TR4’s collector), maximising the open-loop gain. It should give the performance of using a VAS stage with an input buffer. The transconductance of MOSFETs is also adequate. The high noise level of MOSFETs is not a problem, since it is in the second stage. When I substituted a ZVP2106A P-channel MOSFET for TR3 the distortion at 10kHz was doubled compared to the BC556B, A ferrite bead on the gate connection was also needed for stability. Sadly, at this stage I wouldn’t recommend MOSFETs based on this experiment, but I will try some other types. Linsley-Hood also used small-signal monolithic Darlington transistors (MPSA13/63) in his Liniac, an early dis- 2SK2145-Y 5 4 Q1 Q2 G2 S1 6 1 1 D1 Fig.48. Dual JFET op amp using the 2N5564 in an unusual 6-pin TO18-sized package, called a TO71. The gate pins have to be crossed over and note the orange sleeving. 5 2 3 D2 S2 Top view 3 XY 4 G1 2 source ‘Y’ for IDSS = 1.2-3mA Fig.49. Dual JFET pinouts Practical Electronics | November | 2023 Fig.50. An adaptor PCB is needed to mount the SMT 2SK2145 JFET on the board. Note the gates are the two pins on the left side of the 5-pin pack. 61 crete op amp described in Wireless World, September 1971. I substituted an MPSA63 PNP Darlington for TR3 and was surprised at the improvement, giving the lowest distortion of all devices tried. I think the Hfe of 5000 and the extra Vbe voltage drop across the current mirror output accounts for this. The voltage rating of this device is only 30V, thus limiting the power rails to an absolute maximum of ±15V, but that is fine for the low-impedance version of the op amp. Two BC556B transistors connected in a Darlington configuration produced slightly better results without this voltage limitation. Optimum source impedance (OSI) The tail current for the LTP pair (TR1 and TR2) can be changed by adjusting one resistor – R5. Of course, the current is divided in two between the devices. If the tail current is increased the current mirror matching resistors (R3,R4) must be reduced, since there is only 0.7V operating voltage available for this stage. Table 2 shows a a list of noise voltages, optimum currents and source impedances for popular low-noise input transistors. Some of these figures are from Baxandall’s chapter eight in the Microphone Engineering Handbook edited by Michael Gayford (Focal Press 1994). Output transistors The BD139/40 transistors can be usefully upgraded. Into 100Ω the BD139/40 gave 0.004% THD. I made some boards, shown in Fig.51 with Molex 0.1-inch sockets which provide excellent transistor holders for substitutions and testing. The 2SD669A and 2SB649A (PNP) complementary pair from UTC are a superior replacement with equal values of Ft of 140MHz and a higher voltage of 160V. However, their maximum current is limited to 1.5A. Unlike most audiophile drop-in replacements, I was gratified to see the distortion was slightly better at 0.0015% into 180Ω. If larger powers, say over 2Wrms are needed for lower impedances (<50Ω), you have to use higher-current devices with heatsinks. The MJE253G (PNP) and MJE243G from OnSemi are also good, with 40MHz 15W ratings. The MJE182/172 were used in Samuel Groner’s op amp (rated 3A 80V 50MHz). None of these transistors needed C5 for stability. Generally, you have to change from fast planar to traditional power epitaxial-base transistors to get even higher power and current ratings. These devices can have an Ft as low as 3MHz, such as the TIP31A/32A and BD435/6. In theory, this means the high-frequency distortion is going to be higher and the compensation arrangements will have to be redone. In practice, this did not come to pass and surprisingly they worked fine, giving 0.002% THD into 180Ω. This complex op amp topology gives excellent results with average transistors because all the conditions are held rigidly by high feedback, current sinks and voltage references. Simpler audio amplifiers show up the differences more. Profusion sent me some samples of triple-diffused Sanken 2SA1725 and 2SC4511 30W 20MHz devices (although Hfe was low at around 50). These would work very well using the op amp as a small Hi-Fi power amp. For example, driving a tweeter in an active speaker. Note the metal tab faces outwards or downwards for the bigger TO220 devices, the opposite for the Table 2. Low-noise input transistors Device Polarity nV/√Hz* NE5534A NPN 4 BC549C NPN BC559C PNP BC550C NPN BC560* PNP BC546 NPN BC556 PNP BC143 PNP 2SC2240** NPN 1 2SA970A** PNP 0.9 2SC2362 NPN 2N4401 NPN 2N4403 PNP 2N5564 N-chan 3 2SK2145-Y N-chan J230 N-chan 2 HN3C51F** NPN HN3A51F** PNP OSI 8kΩ 5.5kΩ 5kΩ 5kΩ 5kΩ 2kΩ 2kΩ 260 Ω 1kΩ 1kΩ 10kΩ 420Ω 473Ω >10kΩ >50kΩ >100k 10kΩ 10kΩ Ic 0.13mA 0.1mA 0.1mA 0.1mA 0.1mA 1mA 1mA 1mA 1mA 1mA 0.1mA 1mA 1mA 2mA 1mA 0.5mA 0.1mA 0.1mA Notes and noise factor IC op amp reference 1-3dB specified 1-3dB specified 2-4dB specified 2-4dB specified Unspecified ~4dB Unspecified ~4dB 0.65dB 0.5dB 0.5dB 1dB 0.61dB 0.5dB 1dB JFET dual 1dB JFET dual 1dB JFET 1dB dual NPN 120V 1dB dual PNP 120V Fig.51. Right angle Molex sockets make good power transistor sockets for testing. Fig.52. Orientation of power transistors TR13 and TR14 using TO220 devices. smaller TO126 cases. See Fig.52. Table 3 details some output transistor options. Other applications There are almost as many applications for discrete op amps as for the integrated variety. But we’ll have to leave off here with just a few ideas. Phono amplifier *At 1kHz **Obsolete – sadly, the best audio transistors are disappearing fast. Single op amp phono amplifiers are very demanding, needing very low noise and high gain. A rather odd requirement is that they also need to drive an RIAA feedback network that often has an impedance down to 220Ω at high frequencies. High headroom is also needed to cope with sudden scratches. If the pick-up is a moving-magnet cartridge then an optimum source impedance is needed that is fairly high, in the region of 3kΩ to 9kΩ. For this, standard BC549C input transistors running at 0.1mA to 0.2mA work well. The NE5534 op amp, which was designed for moving-magnet cartridges, runs its input transistors at 0.19mA. Conversely, moving-coil cartridges are very low impedance, sometimes only 22Ω, and low base-spreading Rbb resistance devices such as the 2SB737 and BFW16 are used at high currents. A suitable 62 Practical Electronics | November | 2023 Fig.55. (left) Basic Discrete Audio Op Amp set up for drop-in replacement of an integrated op amp. Note R20, R21 and C9 feedback, and the input and output components are left out. The op amp connections are 25 swg tinned copper wire soldered directly into an 8-pin DIL turned-pin socket which plugs into a chip socket to replace an IC op amp. Note the earth pin on the PCB will have to go to the 0V on the IC PCB. There’s normally no earth pin on the chip socket. Fig.53. Fitting SMT output transistors using the SMT adaptors. Note the bent over TR8 for temperature sensing. I used a ZTX651 here for its flat pack. This made an excellent 80Ω ±12V headphone amp. low-impedance equalisation network is shown in Fig.54. High-power discrete op amp Input + +25V Output – –25V 180Ω 1% 2.7kΩ 1% 20kΩ 1% 15kΩ 1% 24nF 1% 90.9nF Suflex 0.5% 121Ω 1% 2nF 1% For unusual parts see AO Shop on p.64 220µF Non-polar RIAA network 50Ω signal generator booster Many audio signal generators don’t have high output swing. My Rapid Pintek FG-32 is limited to ±10V. The solution is to have an add-on amplifier with a gain of 3, running on ±25V. A signal generator is expected to have a 50Ω output impedance and be able to drive 50Ω. An amplifier with a capability to drive 100Ω is needed, via a series resistor of 50Ω 2W. The design would have to be tweaked to have a very high maximum frequency. 48V single-rail working Fig.56. Stop press. There will be a dedicated dual JFET adaptor board for the 2SK2145. (It will be supplied with the main PCB.) biased to half-rail to obtain symmetrical output swing on a single power rail. This gives us the tantalising prospect of designing systems using the audio/ telephone standard of +48V rail. This makes a single-rail microphone amplifier with phantom power (which is +48V) a possibility. Normally, microphone amplifiers need three rails, +17V, –17V and +48V. A single-rail design reduces power supply complexity and cost as a result. Note that polarised capacitors C7, C9 and C10 will have to be reversed. Going straight During audio product development it is useful to have a discrete op amp that can plug into a standard 8-pin DIL socket to substitute for a chip. In conjunction with the uniTable 3. Output transistor options versal op amp board (Audio TR13 (PNP) TR14 (NPN) C5 Vce Out, December 2022) one Ic Notes can easily verify a system BC327 BC337 NO 45V 600mA Higher current than BC546/56, driver. before committing to a final BD136 BD135 NO 45V 1.5A Low-voltage (±12V) 80Ω PCB. An ‘umbilical cord’ BD140 BD139 YES 80V 1.5A High-voltage (±30V max) can be made to an 8-pin DIL BD436 BD435 NO 45V 7A Low-voltage, high-current 7A 32Ω plug as shown in Fig.55. TIP32A TIP31A NO 60V 6A Makes small power amp, TO220 case The leads are connected ZTX751 ZTX651 NO 60V 2A Small, fits well on thermal pad, driver to the op amp pins under MJE253G MJE243G NO 100V 4A High-frequency small power amp the board and the audio 2SB649AL 2SD669AL NO 160V 1.5A Higher quality BD139/40 input/output components 2SA1725 2SC4511 NO 80V 6A High-frequency power amp TO220 are omitted. You have to 2SA2039-TL* 2SC5706-TL* YES 50V 5A SMT SOT-223 outline. 50V Vce, so limit be careful to keep the input rails to ±22V max. leads short to avoid pick-up and instability. * I wanted to test these devices for a possible SMT discrete op amp. I was disappointed to find All this goes to show the they initially gave more distortion (0.01% THD into low impedance 100Ω loads) compared to the huge versatility and diverothers in the CFP output. The 5A maximum current and Hfe of 300 suggested they should have sity of tweaks that can be been better. It turned out the problem was very high-frequency oscillation. Their very high 300MHz applied to a discrete operaFt was so good it may have caused the problem. It was fixed by making C5 270pF and adding tional amplifier. I hope you an additional capacitor of 150pF across the base-collector junction of TR9. A Zobel network, come up with some unique connected to ground from the op amp output of 15Ω and 47nF, was also required. Using these designs and applications of transistors in the straightforward emitter-follower output (TR9 and TR10) gave excellent results, so a your own. good SMT op amp is possible. A prototype taking shape is shown in Fig.53. Fig.54. Suitable RIAA feedback network for the Discrete Audio Op Amp. Practical Electronics | November | 2023 As with all dual-rail op amps, this circuit can be 63