Fullscreen HTML5Med Res (??MB)HTML4

AUDIO OUT AUDIO OUT L R By Jake Rothman Discrete audio op amp – Part 1 +15V 2.5V R2 1.2kΩ R1 1.2kΩ From TR1/2 2.5V IC1 NE5534 1 7 3 + 7 2 – 8 Comp 4 2mA 2mA 13kΩ 13kΩ V– 1 Trim i Non-inverting input *Optional To rest of internal circuit TR2 –V +Vi TR1 Inverting input RE* RE* TR1/2 Low-noise devices eg, 2SC2240 4mA 3+ 2– Op amp LTP input transistors biased off Alternative wiring for using an LM318 op amp V+ 5 8 Output 6 20pF 1.2kΩ V– 6 Output Typically 0.4mA 4 –15V Fig.1. Putting discrete transistors in front of an IC op amp to get lower noise. The internal input transistors are biased off by connecting them to the negative rail. The new low-noise transistors are connected to the next stage in the IC via the compensation pins. The LM318 (shown inset) with external input transistors was the only way to get a top-quality audio op amp in the early 1970s. T he simple economics of semiconductors means that the vast majority of modern electronic products are based on ICs (integrated circuits). It’s quite rare to see circuits made from individual, separate transistors – what are called ‘discrete’ designs. Most analogue signal processing systems use IC op amps, which are almost always better than discrete versions, especially for precision instrumentation where the inherent advantages of transistor parameter matching in monolithic devices, and the resulting low DC offset errors, are important. Audio priorities However, in audio systems, which are generally AC coupled, DC offset minimisation is less important. Also, very high frequency response is not required. So, a reasonable question is, why pay for a response up to 300MHz when 100kHz is plenty for audio? In fact, 50 output drive headroom, low distortion and low noise are much more important criteria for audio. Given this, the small production runs of top-quality audio equipment and the high price of professional audio op amp chips, it is well worth investigating if discrete design for audio offers any advantages. First a reminder that the OP and AD series of op amps from Texas Instruments are typically £2.50 to £10 each, and these single-source specialist audio ICs are often hard to source or become ‘obsolete’ just after you’ve finished a design! Regular readers of my column will know that there is one low-cost, multi-sourced exception to the problem of expensive, hard-to-find audio-friendly op amps: the NE5534/NE5532, which typically costs under £1. To make a discrete op amp design worth considering, this is definitely the chip to beat. This project will attempt to do just that. +15V R1 10kΩ IC1 NE5534 1.5mA 10mA (Adjust with R3 and R4) +0.75V TR1 BC337 + 3 – 7 2 + 4 C1 10µF 6 R3 15Ω R5 10Ω Output + C2 10µF R4 15Ω R6 10Ω D2 1N4148 –0.75V TR2 BC327 R2 10kΩ –15V Fig.2. Boosting the output of an IC op amp with emitter-follower transistors. This is an effective way to get decent drive capability from early audio op amps such as the TL071, which is only happy with loads above 2kΩ. Practical Electronics | September | 2023 Fig.3. Simple discrete op amp for educational use from mitchelectronics.co.uk. Not Hi-Fi, but excellent for getting started. Hybrid designs I often use a hybrid approach – sticking discrete transistors on the front end of an IC op amp for lower noise (Fig.1) or adding transistors to the output for higher drive, as shown in Fig.2. While this is often a useful design technique, in the end I thought why not go the whole hog and try to make a purely discrete design to make a proper comparison with integrated versions. The main bottleneck with the ‘hybrid’ approach is that the supply voltage is limited by the op amp, typically to ±18V. In fact, with some new op amps, the voltage rating is even less, such as the LM4562 (±17V). This is caused by process developments shrinking the die size, and with it the maximum voltage rating. Discrete, small-signal bipolar transistors such as the BC556 only cost pennies (they haven’t suffered the recent price inflation of ICs) but are conservatively rated at 65V. This makes it possible to make a discrete audio op amp with ±24V to ±30V rails for a lower cost than the chip equivalent. True, this price does not factor in labour, but for hobbyists this is much less of a priority – it’s a ‘labour of love’! Somewhat surprisingly, with automated surface-mount assembly techniques, a discrete design can now even save money compared to higher-end IC designs. Other advantages discrete has compared to conventional IC fabrication are access to high-gain PNP transistors, low-noise metal thin-film resistors, large-value Fig.4. Sparkos Labs, the ultimate discrete op amp? good-quality capacitors as well as ease of modification and upgradability. These last two factors are particularly relevant to hobbyists – a discrete design offers endless opportunities to tinker, fiddle with and improve a basic design. Discrete problems That said, most discrete op amp designs I have built have measured worse than standard chips and I have mainly only used them for educational purposes, such as the kit in Fig.3. It is a real testament to the NE5534 that I have found it very challenging to produce a discrete design that is just as good. The circuit design presented here has similar distortion levels, (but with no high-frequency distortion rise above 10kHz), half the noise, twice the load capacity and ±25V output. There are plenty of kits online available for discrete op amps, mainly designs based on the venerable Jensen 990 from the US. While these are excellent, they are also expensive. Price aside, a nice aspect to them is that they can be improved upon – for example, the Samuel Groner SGA-SOA-1 upgrade. The Automated Processes API 2520 modules are also very popular, and I’ve built many Gar2520 kits from Classic Audio Products. The API pinout seems to have become a de-facto standard for discrete op amps. Fig.4 shows a discrete SMT-built audio op amp from Sparkos Labs. The SS2590 must be one of the highest spec audio op amps ever made – it should be, it uses 40 transistors! I tried to import some, but the distributor couldn’t do it for one-offs. (They cost $90 each!) The Elektor Prelude amplifier from 1983 (shown in Fig.5) and Codd’s Wireless World October 1979 design were both excellent discrete op amps. I suspect these designs formed the basis of Douglas Self’s ubiquitous ‘Blameless power amplifier’. It’s well worth reading Self’s detailed analysis of this topology in his Wireless World articles on power amp design (1996). This twogain-stage system (see Fig.6) has formed the basis for several PE/ EPE/Silicon Chip designs, such as the Hifi Stereo Headphone Amplifier from PE, October 2014. It’s worth pointing out that this sort of power amp is effectively a power op amp, and it formed an ideal basis for my design. Most specialist IC audio op amps use a three-stage topology, which gets rather complex to implement in discrete circuitry, although the API2520 does it. The NE5534 for example, uses 29 transistors. Not so discrete Fig.5. Elektor magazine has now moved from audio to Arduino projects. The Prelude was an excellent discrete audio op amp. Mine lasted over 30 years. Practical Electronics | September | 2023 The main practical problem with an all-discrete op amp design is its parts count, with typically a dozen transistors required to get decent performance, as shown in the initial breadboard design in Fig.7. All these connections and complexity do 51 Total gain ≈ 100,000 Difference amplifier input stage long-tailed pair (LTP) + Input CComp Output stage gain = 0.96 and stage boosts output current by a factor of 100 – Fig.6. Basic structure of a two-gain-stage op amp with difference amplifier, main voltage amplifier and output stage. Note the compensation capacitor, which defines the highfrequency roll off. This has to dominate the roll off since the roll off provided by the transistors is poorly defined. have an impact on reliability, and I would estimate the mean time between failure of a discrete op amp is over ten-times worse than for an equivalent IC version. Therefore, I wouldn’t recommend using a discrete op amp for any system-critical application such as avionics, but elsewhere, say in a recording studio environment, they are certainly viable, providing they use standard commodity parts. (Interestingly, the first UK solid-state Hi-Fi amplifier from Toby Dinsdale (1964) started off as an avionic servo op amp.) Another discrete-design problem is that discrete transistors have not improved much in the last 30 years, which means old Mullard/Philips workhorses, like the BC549 and BD139 from the 1970s are still used. After Europe gave up transistor development, Japanese companies, particularly Toshiba and Sanyo, continued improving their through-hole bipolar discrete transistors until the end of the 1990s. Their new 2SA/B/C devices, which were developed for the large Japanese Hi-Fi amplifier industry became highly sought after. Sadly, this progress has also stopped, and further process improvements were directed towards ICs, giving chips a further edge. Excellent versions of Japanese-designed transistors are still being produced in Korea by Unisonic Technologies Co. (UTC) with the prefix ‘KT’ rather than ‘2S’. Some of these can still be obtained from Profusion Plc and Tayda, but these are gradually being dropped. This has led to the usual eBay situation of exorbitant prices and fake Toshiba devices for the unwary. Closer to home, some discrete development continued for a while in the UK with the Ferranti ‘Zetex’ or ZTX transistor range, which have been popular with audiophiles. These are now made by Diodes Incorporated. The overall small-signal transistor development plateau led John Little from Little Labs to use surface-mount output driver chips in his Monotor headphone amplifier (Fig.8). Unfortunately, I can’t find out which SMD devices he used – very frustrating! As you can tell from this somewhat extended introduction, the discrete vs integrated debate is complicated. For home constructors though, I firmly believe a (well-designed) discrete op-amp-based design is still the best option for a top quality Hi-Fi headphone, phono or microphone amp. Fig.7. A busy breadboard – 12 transistors is about the limit for me before jumpy connections become a problem. same. Thus, we will have made an amplifier with a gain of 10. For this system to work predictably we have to assume ‘ideal’ op amp characteristics – ie, the open-loop gain is infinite, no current is drawn by the inputs and the output has zero source resistance. I’ll now work my way through each of the stages of a conventional op amp. The difference amplifier The fundamental architectural feature of an operational amplifier is its differential input. That is, it has two inputs called ‘non-inverting’ and ‘inverting’, where the difference between the two is the voltage to be amplified. A differential input means the circuit can be configured to perform many types of operations – for example, multiplication (amplifying), summation (mixing), subtraction (input balancing) and lots of other analogue functions or mathematical operations. The difference amplifier is usually based on Blumlein’s famous long-tailed pair (LTP) configuration, which he patented in 1936. He originally used triode valves, but today the most common arrangement is a pair of NPN bipolar transistors (or N-channel FETs) as shown in Fig.9. To minimise DC offset and distortion, the two devices need to be well matched in terms of transconductance. That is, how much collector current flows for a given base-emitter voltage Op amp 101 Building a discrete op amp is a great way to learn about the internal operation of all op amps and see what is really going on under the hood. Let’s look at the internal stages and get a feel for how they operate and some possible optimisations. When it comes to explaining basic op amp negative feedback linear operation, I always say, ‘the output voltage steers itself to make the voltage difference between the two input terminals zero’. The network in the feedback path from the output to the inverting input then dictates what the system does. For example, if the attenuation in the feedback network is a factor of 10, then the output will have to be 10-times bigger than the input to keep the two input terminals the 52 Fig.8. The Little Labs Monotor headphone amplifier – a wonderful design, but I can’t fix it due to single-sourced SMT output chips Practical Electronics | September | 2023 Long-tailed pair 0.6V +Vi C1 1nF TR1 BC546 +15V TR3 BC556 R2 620Ω 1mA 1mA TR2 BC546 –0.7V C3 39pF R4 220kΩ –1.3V Non-inverting input R1 13.8V 220kΩ R3 7.5kΩ 2.7µA DC bias current (varies) 0V 2mA Inverting input C2 220µF NP R5 3.3kΩ 4.5mA –15V high input impedance, fulfilling one of the key criteria for an ideal op amp. In audio applications, the bias currents from bipolar transistor inputs (see R1, Fig.9) mean that potentiometers have to be AC coupled to prevent scratching. However, normal JFETs have about 40-times less transconductance than bi-polar transistors, which means twice as much distortion. There are esoteric audio JFETs available, such as the 2SK170 and 2SJ74 which achieve high transconductance, at a high cost. Also, there are good Toshiba SMT devices, such as the 2SK2145. This device has its two sources connected together and brought out on one pin, which is fine for our LTP purposes. As an aside, it’s worth noting that it’s difficult to process FETs at the same time as bipolar devices on the same IC, which makes low-noise audio FET op amps expensive. On the other hand, with discrete op amps, you can just solder them in. Constant-current sources and sinks Fig.9. The Blumlein long-tailed pair (LTP), outline dotted, originally designed to replace transformers so that much higher bandwidths were possible. It can use bipolar transistors, FETs or the original triodes (the ECC88 works well at 90V) for the input devices. The circuit here is the simplest ‘op amp’ one can make. It’s open-loop gain is 3000. The feedback resistor (R4) is completely decoupled by C2 so that the openloop gain can be measured. +15V +15V TR3 BC556 Transistor current sink From LTP R4 220kΩ To LTP 4.5mA 10kΩ TR7 BC546 4.5mA Set I = 1/R8 4.5mA current-regulator diode + AJ Semitec E-452-E-562 (Rapid 47-2608) 1.7V –15V 1V R8 220Ω –15V Red LED, low-current, Vf = 1.7V L7113 SRD-D (Rapid 55-0136) Fig.10. Adding a constant-current load (sink) to the voltage amplifier stage TR3 bumps the gain up to 12,500. Note ‘upside-down’ PNP arrangement. For comparative purposes, I swapped out the resistor for a current-regulator diode before replacing that with a biased transistor TR7. expressed in mA/V. The current gain (Hfe) is of less importance, unless high source resistances are used. The transistors can be hand matched with a multimeter or analyser, such as those made by Peak Electronic Design (see back cover!). The best approach is to use a dual transistor, such as the SSM2210 – popular but pricey – so not appropriate for this high-quality, low-price design. Toshiba introduced some surface-mount audio dual-transistors in 2001, such as the 50V HN1C01FY. We will have provision on the PCB for these. They are available from Mouser for around 30p. One historical detail worth mentioning is that in the 1970s, PNP transistors were preferred over NPNs for LTP circuits since they have a slightly lower base spreading resistance, resulting in a lower noise level. This difference is much smaller today, so we’ll use NPN types for our LTP since there is much more device choice. FET inputs Fabricating JFETs instead of bipolar transistors in LTPs has been used in op amp IC designs. It offers the advantage of having a very Practical Electronics | September | 2023 For audio work, the current needed through each LTP input device generally has to be higher than 1mA for low noise from low source impedances. (There are however higher impedance applications, such as moving magnet pick-up cartridges, that may need lower currents). This means the collector load resistors (R3 and R5, Fig.9) have to be low value, resulting in low gain. Replacing the resistors with current sources gives the current required but with a high effective dynamic impedance, greatly increasing gain. Using a current source has an additional benefit of an increased power supply rejection ratio (PSRR) since the current remains constant with power supply fluctuations. Current regulator diodes are the simplest way of doing it, needing no bias supply. For R&D use they are excellent since they just fit in a standard resistor position. However, as a carefully selected gate-source coupled JFET, they have become rather expensive. It’s much cheaper to use the standard transistor design. I found replacing the tail resistor (R3) with a current source in Fig.9 only increased the gain to 3500, but it did make the tail current independent of supply voltage. (With just a resistor, the current is proportional to the supply voltage, an unwanted variable.) The noise from the tail current source, although higher than a resistor, is not a problem since it is common mode – it’s applied to both LTP input transistors equally and thus cancelled. One of the great advantages of a discrete op amp is that bias voltages for the current sources can be easily accessed and decoupled with large capacitors for low noise. Voltage amplifying stage (VAS) The bulk of our system voltage gain is provided by the voltage amplifying stage (VAS), which is based on a common-emitter strapped between the two power rails, as shown in Fig.10. The term ‘VAS’, as coined by Douglas Self, is really a misnomer since it has a current input. Again, a constant current load here can provide higher gain and better DC stability. In Fig.9, replacing the collector load resistor R5 with a 4.5mA current-regulating diode (CRD) increased the gain to 12,500 (from 3500). Sometimes, for higher output voltage swings, a bootstrap capacitor is used. A current source typically drops around 1.5V, so it is worth bootstrapping it in low voltage applications. The load impedance on the VAS stage output is important. If it is too low the open-loop gain will drop and the gain advantage of the constant-current sink will 53 Compensation Current mirror +15V I reflected 0.3V 0.6V R3 300Ω 0.3V TR4 BC556 I set R4 300Ω TR5* BC556 *TR5 wired as diode (transdiode) TR1 BC546 To VAS Inverting input Non-inverting input 1mA 1mA TR2 BC546 2mA 2mA CRD E-202-E562 (Rapid 47-2602) (or 6.2kΩ) –15V Fig.11. A current mirror on the LTP equalises the current and increases gain further to around 150,000 (without emitter resistors). be lost, bringing us back to the position of just using a load resistor again. This problem also applies to a lesser extent to the LTP current mirror output loading. This effect can be minimised by using a high Hfe transistor (>450) for TR3. Unfortunately, we need a Vce rating of at least 50V, and most normal high-gain transistors such as the BC559C are around 30V. It’s all done with mirrors A current mirror comprising TR4 and TR5 can be used to enforce the same current through both LTP transistors, reducing the need for close matching of TR1 and TR2, see Fig.11. This circuit has a pushpull action, with the current on one side reflected in the other, so the transconductance is doubled for a given current. This and the high dynamic load results in a gain increase of around 10 compared to a resistive load. Putting the current mirror into Fig.9 increased the gain to around 150,000 at 1kHz. We are now in the realms of ‘proper’ op amps with an open-loop gain of around 100dB. The gain does roll off towards high frequencies due to Miller effect in the transistors, (the magnification of base-collector capacitance). Of course, the current mirror can itself be mismatched. The inclusion of emitter resistors in the mirror (R3 and R4 in Fig.13) reduces the effect of transistor mismatching. A voltage drop across these resistors of 0.3V provides sufficient negative feedback to equalise the mirror currents. This is also a place to insert a DC offset trimmer to set the quiescent voltage of the op amp output to zero. When using discrete transistors for TR1, TR2, TR4 and TR5, this offset is worse due to transistor mismatching, so adjustment is often needed. 54 +15V All negative feedback circuits Iq = 6mA TR3 need compensation to prevent Input BC556 from LTP TR9 (VAS) high-frequency oscillation. This BC546 limits the closed-loop gain to unity at the frequency where 180° phase shift occurs, preventing R1 R3 70mV 6.8kΩ 12Ω for 6mA positive feedback. This phase+ C1 shift is a result of high-frequency Output 10µF 10V losses in the transistors, mainly R4 R2* TR8 12Ω 6.8kΩ the slower output transistors. The BC546 *Can be replaced simplest way to fix this problem with trimmer for Iq is to set a dominant defined roll4mA off by wiring a small NP0 ceramic TR10 Sink Thermal link BC556 (TR7) capacitor (C3) of around 22pF to 100pF across the base-collector –15V junction of the VAS transistor TR3, depending on the closedFig.12. Emitter-follower output stage. In this loop gain. It is difficult to determine the case a fixed bias is used in conjunction with optimum value for this com- large emitter resistors R3 and R4. Often R2 ponent. Too much, and your has a series trimmer to allow quiescent current high-frequency distortion is adjustment. This will drive loads down to 300Ω. higher than it needs to be. Too If lower impedance 100Ω loads are to be driven, low, and the odd amplifier may the output transistors will have to be upgraded oscillate in production. One ad- to ZTX651/751 and provided with heatsinks. vantage of discrete over chips Loads below this, such as 32Ω headphones, will is that you can optimise this need extra output transistors (see next month). capacitor to suit your particular transistors and layout. For this design, of 10 enables the compensation capacitor 39pF was used for 10-times gain and to be reduced by 10. This then presents a 82pF for unity gain. Another trick is to higher impedance load at high frequencies use two capacitors to get second-order (>10kHz), reducing the possibility of the compensation (This is not an option on input stage overloading. This slewing ICs because the bigger second capacitor distortion is an issue when high output would use too much die area.) voltage at high frequencies is required. It Baxandall found it useful to add an manifests itself when a 20kHz sinewave additional compensation feed from the becomes a lopsided triangle wave. output stage. He called it ‘inclusive A problem with these emitter resistors Miller compensation’. This did not (R1 and R2) is that they add Johnson noise. make much difference with a low-power This becomes an issue where high closedop amp, possibly because the small loop gain is required at low impedances, output transistors did not make such such as in a mic or moving-coil phono a big crossover glitch. However, it does pre-amplifier. The input signal here is make a difference at higher powers and low, therefore distortion is minimal, so will therefore be included on the PCB it’s best to leave the resistors out. With as an option. JFETs these resistors are not used, since the transconductance is low and we need all the gain available. Emitter resistors A low-noise upgrade for the input stage The maximum input into a bipolar LTP LTP transistors (TR1, TR2, TR4 and TR5) before noticeable soft-clipping distortion is to use the 2SA970 and its complement occurs, is around 50mVpk-pk. (Of course, the 2SC2240. These are rated at 120V and the actual signal, the error voltage, is have a noise factor of 0.5dB. only a couple of mV in most negative feedback situations). The way round this is to install emitter resistors (R1 and Output stage R2) to provide local negative feedback. The output stage is a voltage follower or A good analysis of bipolar transistor current amplifier. In most op amps this distortion was given by WT Cocking is a class AB push-pull emitter follower, in Wireless World, May 1972. These as shown in Fig.12. The VAS stage can resistors reduce the gain by around comfortably drive this for low powers 10x, but this can be compensated to down to 300Ω. For higher power, extra a degree by increasing the operating high-current output transistors in a Darcurrent of the LTP. lington stage could be used. This is the This gain reduction is useful where best option for a high power op amp, the current mirror is used as a collector such as a transformer or headphone load for the LTP, since there is generally driver. These high power output stages too much. Reducing the gain by a factor add another couple of output transistors, Practical Electronics | September | 2023 and the original output transistors become driver transistors. Thermal stability The output stage normally has a preset quiescent current, set to typically 3-15mA to reduce crossover distortion. In discrete circuits we have the option to increase this to 100mA to ensure class A operation for minimum distortion. This current will increase as the output stage heats up, and failure due to thermal runaway is likely unless effective heatsinks and stabilisation is used. Normally, a Vbe-multiplier bias transistor (TR8) is thermally coupled to track the temperature. It’s also bypassed by a capacitor (C4) to ensure equal drive to both output transistors. This set-up is easier to arrange with discrete circuits. On integrated circuits the whole chip gets cooked. However, in a high-power design using the CFP output stage, only the drivers need be coupled to the bias generator. Protection racket All output stages need output short-circuit protection in the form of current R3 330Ω R4 330Ω DC Offset TR5 BC556 1mA R9 6.8kΩ 1mA R1 180Ω PR2 5kΩ R2 180Ω R7 10kΩ R6 2.2kΩ TR6 BC546 R5 470Ω + C2 1µF 10V 6mA R8 220Ω + +15V – Suggested test circuit R14 1.5kΩ –15V R18 22kΩ R11 12Ω C5 100nF 70mV Output R12 12Ω Thermal link Discrete R19 1kΩ* C4 10µF TR11 10V BC556 Iq set 4.5mA TR12 BC546 + TR8 BC546 TR7 BC546 LED red low I R13 2.7kΩ R15 82kΩ R16 5.6kΩ R17 47kΩ + TR10 BC556 Mute (Pull to 0V or V–) C6 1µF 10V V– –6 to –25V L1 10µH** C10 100µF + Input + C7 4.7µF R10 3.9kΩ Inverting input V+ 6 to 25V (symmetrical) TR9 BC546 C3 39pF (x10 gain) 82pF (unity gain) TR2 BC546 Note: resistors and capacitors have been renumbered from Fig.9 and Fig.12 Last, an unconventional approach shown in Fig.14, which we will not be using on the circuit’s PCB. This design shows how audio op amps can depart from the more conventional design shown in Fig.13. It is possible to make an op amp with a difference amplifier comprising one transistor, which can save cost if an expensive device for TR1, such as a 2SJ74 FET is needed. Single-ended stages have higher even-order distortion since there isn’t the cancellation of two curved transfer characteristics (as with the LTP). There is also a bad DC offset due to the Vbe drop of the transistor and drop across the feedback resistor, (since the full stage current has to flow through it). This is not a problem in single-rail audio applications with low-value feedback resistors. It has the significant advantage of giving a 3dB lower noise level than the differential pair. TR3 BC556 TR1 BC546 Non-inverting input Odd-ball op amp PR1 5kΩ TR4 BC556 C1 220nF impractical to provide the full protection circuitry available to the IC designer, such as thermal shutdown. limiting. A dodgy lead or slipped probe is all it takes to destroy the VAS and output transistors. In its simplest form, this can be high-value output emitter resistors of a few tens of ohms or a series output resistor. This of course raises the output impedance and causes loss of headroom. A better approach is to provide current sensing on the emitter resistors or power rails. One quirk of the circuit developed here is that shorting out the LED voltage reference for the current sources turns the whole thing off. This allows it to have a ‘shutdown pin’ for protection purposes. It also turns off if either power rail is lost since the LED is wired between the two rails. Using this approach, a rather crude protection circuit has been devised which allows 300Ω to be driven while shutting it off below 100Ω. The final discrete op amp circuit is shown in Fig.13. It uses low-cost BC546/56 devices. A nice feature of discrete circuits is that if you blow them up it’s usually only a few cheap transistors and resistors that are easily replaced; with an expensive chip, that’s it. However, it’s easier to damage the discrete circuit because it’s Output R22 47Ω** Output isolator 150pF* C8 R21 22kΩ R20 2.2kΩ (gain set x11) + *RF filtering on input **Isolation from cable capacitance C9 100µF (non-polar preferred) 0V Fig.13. Nearly there now, the full op amp circuit. Emitter resistors R1 and R2 have been added to the LTP, reducing the stage gain and linearising it, making it easier to prevent HF oscillation. Also, DC offset PR1 and Iq trimmers PR2 have been added. The final PCB will have some minor enhancements to make it more versatile. There will also be provision for feedback components for basic non-inverting and inverting configurations, as shown in the test circuit. Practical Electronics | September | 2023 55 Cascoding +25V 2.2mA C6 100nF *R3 sets symmetrical clipping on output R3 220kΩ + Input R5 2kΩ C2 68µF 15V R2 2.2kΩ TR3 ZTX651 R9 1kΩ D3 100mV 1N4148 R7 1kΩ 1.3V –25V Set gain 8mA Op amp output R11 470Ω C10 6.8pF (comp) TR7 BC546 TR2/7 Cascode 23V R1 110kΩ R13 120Ω +0.7V TR5 BC556 Inverting input Non-inverting 23V input 0V TR4 ZTX751 R6 10kΩ TR1 BC559C C4 100nF R10 4.7kΩ C9 100pF C3 330pF D1 1N4001 D2 1N4001 0.12mA offset* C1 220nF R8 22kΩ 0.13mA 0.6mA TR2 BC549C R4 5.6kΩ Audio output TR6 ZTX651 8mA Bias R12 100Ω Iq sense D4 2.7V 1.4V C5 22µF + This unusual op amp design also uses a cascode in the VAS in Fig.14. This is a special technique that places a common-base amplifier (TR7) in series with a normal common-emitter (TR2) to make a ‘super’ transistor from two cheap devices. This topology improves high-frequency response by eliminating modulation from the Miller effect (high frequency attenuation) and Early effect (distortion), giving much lower overall high-frequency distortion. It allows a low-voltage high-gain device to be used for the lower transistor (TR2) and a higher-voltage general-purpose device (TR7) for the upper transistor. It does require a bias voltage of around 2.7V to 4.7V provided by Zener diode D4. Cascoding can also be used with LTPs, enabling the use of low-voltage FETs for the input devices. This was done on the phono pre-amp on the Pioneer SX-1980 amp using ±34V rails. Cascoding could be a later upgrade for Fig.13. Next month, we’ll make a full discrete through-hole op amp with construction details and test results. R15 22kΩ R14 220Ω Total current on each power rail: 10.5mA –25V C7 4.7µF+ 35V Overall gain of amplifier = 6x C8 100nF 0V Fig.14. Single-input transistor op amp derived from a design originally developed for Avondale Audio, who eschew LTP inputs. The VAS stage is a cascode which brings HF distortion down to 0.001%. The Taylor configuration output stage has self-regulating quiescent current, See Audio Out, Aug 2016. STEWART OF READING Fluke/Philips PM3092 Oscilloscope 2+2 Channel 200MHz Delay TB, Autoset etc – £250 LAMBDA GENESYS LAMBDA GENESYS IFR 2025 IFR 2948B IFR 6843 R&S APN62 Agilent 8712ET HP8903A/B HP8757D HP3325A HP3561A HP6032A HP6622A HP6624A HP6632B HP6644A HP6654A HP8341A HP83630A HP83624A HP8484A HP8560E HP8563A HP8566B HP8662A Marconi 2022E Marconi 2024 Marconi 2030 Marconi 2023A HP 54600B Oscilloscope Analogue/Digital Dual Trace 100MHz Only £75, with accessories £125 (ALL PRICES PLUS CARRIAGE & VAT) Please check availability before ordering or calling in PSU GEN100-15 100V 15A Boxed As New £400 PSU GEN50-30 50V 30A £400 Signal Generator 9kHz – 2.51GHz Opt 04/11 £900 Communication Service Monitor Opts 03/25 Avionics POA Microwave Systems Analyser 10MHz – 20GHz POA Syn Function Generator 1Hz – 260kHz £295 RF Network Analyser 300kHz – 1300MHz POA Audio Analyser £750 – £950 Scaler Network Analyser POA Synthesised Function Generator £195 Dynamic Signal Analyser £650 PSU 0-60V 0-50A 1000W £750 PSU 0-20V 4A Twice or 0-50V 2A Twice £350 PSU 4 Outputs £400 PSU 0-20V 0-5A £195 PSU 0-60V 3.5A £400 PSU 0-60V 0-9A £500 Synthesised Sweep Generator 10MHz – 20GHz £2,000 Synthesised Sweeper 10MHz – 26.5 GHz POA Synthesised Sweeper 2 – 20GHz POA Power Sensor 0.01-18GHz 3nW-10µW £75 Spectrum Analyser Synthesised 30Hz – 2.9GHz £1,750 Spectrum Analyser Synthesised 9kHz – 22GHz £2,250 Spectrum Analsyer 100Hz – 22GHz £1,200 RF Generator 10kHz – 1280MHz £750 Synthesised AM/FM Signal Generator 10kHz – 1.01GHz £325 Synthesised Signal Generator 9kHz – 2.4GHz £800 Synthesised Signal Generator 10kHz – 1.35GHz £750 Signal Generator 9kHz – 1.2GHz £700 HP/Agilent HP 34401A Digital Multimeter 6½ Digit £325 – £375 56 17A King Street, Mortimer, near Reading, RG7 3RS Telephone: 0118 933 1111 Fax: 0118 933 2375 USED ELECTRONIC TEST EQUIPMENT Check website www.stewart-of-reading.co.uk HP33120A HP53131A HP53131A Audio Precision Datron 4708 Druck DPI 515 Datron 1081 ENI 325LA Keithley 228 Time 9818 Marconi 2305 Marconi 2440 Marconi 2945/A/B Marconi 2955 Marconi 2955A Marconi 2955B Marconi 6200 Marconi 6200A Marconi 6200B Marconi 6960B Tektronix TDS3052B Tektronix TDS3032 Tektronix TDS3012 Tektronix 2430A Tektronix 2465B Farnell AP60/50 Farnell XA35/2T Farnell AP100-90 Farnell LF1 Racal 1991 Racal 2101 Racal 9300 Racal 9300B Solartron 7150/PLUS Solatron 1253 Solartron SI 1255 Tasakago TM035-2 Thurlby PL320QMD Thurlby TG210 Modulation Meter £250 Counter 20GHz £295 Communications Test Set Various Options POA Radio Communications Test Set £595 Radio Communications Test Set £725 Radio Communications Test Set £800 Microwave Test Set £1,500 Microwave Test Set 10MHz – 20GHz £1,950 Microwave Test Set £2,300 Power Meter with 6910 sensor £295 Oscilloscope 500MHz 2.5GS/s £1,250 Oscilloscope 300MHz 2.5GS/s £995 Oscilloscope 2 Channel 100MHz 1.25GS/s £450 Oscilloscope Dual Trace 150MHz 100MS/s £350 Oscilloscope 4 Channel 400MHz £600 PSU 0-60V 0-50A 1kW Switch Mode £300 PSU 0-35V 0-2A Twice Digital £75 Power Supply 100V 90A £900 Sine/Sq Oscillator 10Hz – 1MHz £45 Counter/Timer 160MHz 9 Digit £150 Counter 20GHz LED £295 True RMS Millivoltmeter 5Hz – 20MHz etc £45 As 9300 £75 6½ Digit DMM True RMS IEEE £65/£75 Gain Phase Analyser 1mHz – 20kHz £600 HF Frequency Response Analyser POA PSU 0-35V 0-2A 2 Meters £30 PSU 0-30V 0-2A Twice £160 – £200 Function Generator 0.002-2MHz TTL etc Kenwood Badged £65 Function Generator 100 microHz – 15MHz Universal Counter 3GHz Boxed unused Universal Counter 225MHz SYS2712 Audio Analyser – in original box Autocal Multifunction Standard Pressure Calibrator/Controller Autocal Standards Multimeter RF Power Amplifier 250kHz – 150MHz 25W 50dB Voltage/Current Source DC Current & Voltage Calibrator £350 £600 £350 POA POA £400 POA POA POA POA Marconi 2955B Radio Communications Test Set – £800 Practical Electronics | September | 2023