Fullscreen HTML5Med Res (??MB)HTML4

Harold S. Black, Negative Feedback and the History of Operational Amplifiers Op amps and negative feedback circuits are ubiquitous today, and you would be forgiven for thinking that they have been around forever. But there was a time when electronics was still developing, and such devices had not yet been invented. That changed in 1927 with the bright idea of one clever man… by Roderick Wall & Nicholas Vinen O Fig.1: Harold Black’s original hand-written notes on the principle of using negative feedback for distortion cancellation. 38  Silicon Chip Australia’s electronics magazine ne of the most significant early circuit ideas was Harold Steven Black’s invention of negative feedback. In 1927, Harold S. Black (18981983) was on a ferry heading towards his office in the West Street Labs of Western Electric, the forerunner of Bell Telephone Laboratories in New York City. An idea popped into his head that would dramatically change electronic communications, which continues to be used to the present day. His idea was for a negative feedback amplifier, where the gain is accurately set and distortion limited by feeding part of the output signal back into the amplifier. Black sketched his idea on a misprinted page of his copy of the New York Times, the only paper that he had on him. When Black got to his office, he had a colleague witness and sign it – see Fig.1. Black’s job had been trying to improve three- and four-channel telephone amplifiers based on carrier telephony for the last six years. For long-distance telephone calls, repeaters had to be added to cover the distance. But these repeaters had too much distortion, so by the time the audio signal reached its destination, it was unintelligible. Black realised that amplifier distortion and noise could be reduced using negative feedback, at the expense of reduced amplifier gain. He later said that he did not know what made his idea pop into his head; it just came. siliconchip.com.au Fig.2: a page from one of Harold Black’s many patents regarding negative feedback. This one is from patent 2,102,671, showing some possible ways of building an amplifier with negative feedback using valve(s). Black used his new idea to design low-distortion broadband repeater amplifiers that were finally suitable for long-distance telephone calls. That allowed more channels over a pair of wires. His patents Harold S. Black was granted 62 patents during his career, 18 of which relate to negative feedback; these are listed in Table 1. His most famous patent is number 2,102,671, which you can view at https://patents.google.com/ patent/US2102671A If you replace the number in that link with the other patent numbers (plus the “US” prefix), you can view the relevant PDF. This patent, titled “WAVE TRANSLATION SYSTEM”, was filed in 1932 and granted in 1937. It comes to 87 pages and includes many detailed drawings (including circuits and plots) and plenty of explanatory text. One of the most important sets of circuit diagrams (but far from the only one!) in this patent, appearing on page four, is reproduced in Fig.2. It shows four different ways of implementing his idea using ‘tubes’ or valves, the technology of the day. Other important plots in the patent include gain curves, stability criteria, equivalent circuits and several practical implementations of the technique. control, battery monitoring, instrumentation and sometimes RF too. The principle is used in TVs, radios, computers, medical equipment, control circuits, measuring instruments and mobile phones. You would find it very hard to find an electronic appliance that does not use negative feedback. You will see negative feedback being used with operational amplifiers and in discrete circuits in most issues of Silicon Chip. Operational amplifiers This paved the way to the development of operational amplifiers (op amps); essentially, a monolithic implementation of a circuit which applies negative feedback. Thousands of different types of op amps are available to suit just about any application; low-power types, highspeed types, high-gain types, precision types, singles, doubles, quads etc. The term “operational amplifier” goes back to about 1943, when this name was mentioned in a paper written by R. Ragazinni with the title “Analysis of Problems in Dynamics”. The paper was the work of the US National Defence Research Council (1940), was published by the IRE in May 1947 and is considered a classic work in electronics literature. George A. Philbrick Researches introduced the K2-W valve-based generalpurpose op amp in 1952, more than a decade before the first transistorised version appeared (Figs.3 & 4). The first solid-state transistor was successfully demonstrated on December 23, 1947, but it took a while before transistors were in widespread use. The first series of solid-state op amps were introduced by Burr-Brown Research Corporation and GA Philbrick Researches Inc in 1962. Fig.3: a popular early valve-based op amp, the Philbrick Research K2-W. The importance of negative feedback Almost all analog equipment manufactured today uses negative feedback. This includes circuits that handle audio signals, analog video, motor siliconchip.com.au Australia’s electronics magazine August 2021  39 Table.1: Harold S. Black’s patents relating to negative feedback (patent numbers are hyperlinks) When filed UNKNOWN 8 August 1928 3 December 1929 3 December 1929 26 March 1930 3 April 1931 22 April 1932 30 September 1932 29 December 1932 29 March 1933 29 March 1933 25 September 1934 6 October 1934 5 December 1936 5 December 1936 23 March 1937 10 November 1937 27 May 1938 20 December 1938 30 July 1940 28 February 1942 Serial number UNKNOWN 298,155 411,223 411,224 439,205 527,371 606,871 635,525 649,252 663,316 663,317 745,420 747,117 114,391 114,390 132,559 173,749 210,333 246,791 348,433 432,860 The first solid-state monolithic op amp IC, designed by Bob Widlar and offered to the public in 1963, was the uA702 manufactured by Fairchild Semiconductors. But it required strange supply voltages such as +12V and -6V and had a tendency to burn out. Still, it was the best in its day, and sold for about US$300 (a fortune today!). It was used mainly by the US military due to its high cost. Then the uA709 from Fairchild Semiconductor was released in 1965. It was introduced at about US$70, and was the first to break the $10 barrier, then not much later, the $5 barrier. By 1969, op amps were selling for around $2 each. From then on, multiple manufactures produced op amps in When issued Patent number Title 7 February 1928 CA277770A Wave signalling system Split into serial numbers 411223 & 411224 below 21 December 1937 2,102,670 Wave Translation System 4 June 1935 2,003,282 Wave Translation System NA Not granted Wave Translation System 1 August 1933 1,920,238 Wave Translating System 21 December 1937 2,102,671 Wave Translation System 28 May 1935 2,002,499 Wave Translation System 20 August 1935 2,011,566 Wave Translation System 9 July 1935 2,007,172 Wave Translation System 27 September 1938 2,131,365 Wave Translation System 16 November 1937 2,098,950 Vacuum Tube Circuit 17 March 1936 2,033,917 Electric Wave Amplifying System 27 September 1938 2,131,366 Electric Wave Amplifying System 6 August 1940 2,209,955 Wave Translation System 18 April 1939 2,154,888 Wave Translation System 3 December 1940 2,223,506 Wave Amplification 17 June 1941 2,245,565 Wave Translating System 7 October 1941 2,258,128 Wave Translating System 26 May 1942 2,284,555 Signaling System 20 July 1943 2,324,815 Stabilized Feedback System many varieties, up to the present day. One particularly popular model was the uA741, which has been improved since it was first introduced in 1968. Some variants of it, such as the LM741, are still being produced today! Its equivalent circuit is shown in Fig.5. Modern op amps mostly use the same principles, but differ in some implementation details, such as the method of internal frequency compensation. One big benefit of the op amp is its flexibility. It can perform a wide range of analog ‘functions’ with the addition of a few passive components. These functions include signal mixing, amplification, filtering (low-pass, high-pass, bandpass, notch etc), integration, differentiation, multiplication, simulated inductance and more. Fig.4: the K2-W uses a similar configuration to transistor-based op amps, with an input pair (one 12AX7 twin triode) followed by a voltage amplification/ buffering stage made from another 12AX7 twin triode plus two neon lamps. 40  Silicon Chip Australia’s electronics magazine Pages NA NA 21 12 NA 17 87 10 7 6 12 5 5 5 29 5 7 5 9 8 7 You can think of op amps as the building blocks for most analog circuits. Negative feedback So how is negative feedback used to control an op amp to reduce the distortion and set a fixed gain? The output voltage of an op amp is the non-inverting input voltage minus the inverting input voltage times a large factor (in some cases, over one million). If we say the gain is exactly one million, this means that: • If the + input is 100μV and the − input is 99μV, the output will be +1V. • If the + input is 100μV and the − input is 100μV, the output will be 0V. • If the + input is 100μV and the − input is 101μV, the output will be -1V. From this, you can see that if the difference between the input voltages is more than a few microvolts, the output voltage will be ‘pegged’ at one supply rail or the other. So unless we are using the op amp like a comparator (a possible op amp function), the inputs will almost always be at a very similar voltage. The negative feedback is typically configured to ensure that this is the case. Let’s say we feed 10% of the output voltage back to the inverting input and apply 1V to the non-inverting input. siliconchip.com.au Fig.5: the internal circuitry of perhaps the most ubiquitous op amp, the uA741 (actually, National Semiconductor’s equivalent). It contains 20 transistors, 12 resistors and one ‘Miller’ compensation capacitor for stability. When the output is less than 10V, the voltage difference between the inputs will be positive, so the output voltage increases. When the output is greater than 10V, the voltage difference between the inputs will be negative, so the output voltage will decrease. Thus, the output voltage will tend towards 10V. The only real sources of error in a DC context are the input offset voltage (the output not being exactly 0V with both inputs at the same voltage) and the finite gain, which adds a few additional microvolts of error. But that’s just one part per million or so. So it is pretty close to an ideal amplifier with fixed gain; that is certainly not the case with a typical single-transistor or single-valve amplifier! Due to manufacturing tolerances, it is challenging to set up (bias) a single transistor or valve to provide an exact gain. Even if you achieve it (eg, by trimming), it will likely change with temperature and over time. Note how the exact gain of the op amp is not important; it only affects the (tiny) gain error. The overall gain is set by the feedback divider, usually made of resistors (and sometimes capacitors), so it’s easy to set it close to the desired value. It can be trimmed to be almost exact if required, and it’s unlikely to drift. Negative feedback also gives close siliconchip.com.au to ideal results for AC signals, as long as they are well below the op amp’s bandwidth (usually specified as gain bandwidth, which must be divided by the configured gain). Thus, an op ampbased amplifier can give an essentially flat gain curve across a range of frequencies, whereas a transistor or valve will typically be far from flat unless it is a special type. Here are some basic op amp circuits: 1) Unity-gain buffer Fig.6 shows an op amp arranged as a unity-gain buffer. The output is fed back to the inverting input, so the output voltage tracks the noninverting input. As the output of an op amp has near-zero impedance (due to feedback), but the input has a relatively high impedance, this configuration is useful to avoid the circuit feeding the input from being loaded Fig.6: using an op amp to buffer a signal can be as simple as connecting its output to its inverting input. However, resistor Rf is a good idea to balance the input currents if the source impedance for the noninverting input is relatively high. Australia’s electronics magazine by the circuit the output is driving. Often, the output will be connected directly to the inverting input. But in some cases, the resulting source impedance mismatch between the inputs can cause temperature drift and other problems. Resistor Rf can be chosen to match the non-inverting source impedance to avoid this. 2) Non-inverting amplifier Fig.7 shows an op amp providing non-inverting gain. The output voltage is an AC signal with the same shape as the input signal but an increased magnitude, by a factor of Rf ÷ R1 + 1. As with the buffer, this circuit can be connected to a signal source that has a high impedance, but it still provides a low-impedance output. Capacitor C1 may be omitted, but it’s usually a good idea to keep it. It reduces the circuit’s gain at higher frequencies, thereby increasing stability and preventing the amplification of unwanted high-frequency signals. You might see a high-value capacitor at the bottom of the feedback divider, between the bottom end of R1 and ground, shown as an alternative connection for R1 in Fig.7. This sets the circuit’s DC gain to unity regardless of the AC gain, so it is mostly used when amplifying AC signals; also refer to Fig.19. By reducing the DC gain of the circuit, it prevents the output from pegging at the positive rail on positive signal excursions, and also reduces the amplification of the input offset error voltage. The practical gain limit depends on the op amp’s gain bandwidth and the maximum signal frequency. For example, an op amp with a gain bandwidth Fig.7: you only need two resistors to set up an op amp as a fixed gain voltage amplifier. As the signal is fed directly into the non-inverting input, the input impedance is high. Optional capacitor C1 limits the bandwidth for stability, while C2 can be used to reduce the DC gain to unity while having a higher AC gain. August 2021  41 Fig.8: the inverting amplifier configuration also uses two resistors and one optional capacitor. While it has the advantage that the gain can be less than unity, the disadvantage is that the input impedance is equal to Rin, rather than the usually much higher figure for the op amp’s inputs. Fig.9: the virtual ground mixer is an inverting amplifier with multiple signal sources. As both op amp inputs are held very close to 0V, there is no way that the signals being fed in can interact with each other, except at the output of the mixer. Fig.10: the basic differential amplifier calculates the difference between two voltages, multiplied by a constant, plus an offset. It needs good resistor matching. of 3MHz has a maximum practical gain of 30 times for signals up to 100kHz (3MHz ÷ 100kHz). Noise and distortion in the output increase with gain, as there is less feedback (closed-loop bandwidth) for the op amp to work within. 3) Inverting amplifier By feeding a signal into the inverting input rather than the non-inverting input, via a resistor, the signal is inverted and gain can still be applied, as shown in Fig.8. The gain is -Rf ÷ Rin, so unlike the non-inverting version, gain values less than unity (ie, attenuation) are possible without a separate input attenuator. An unfortunate consequence of this configuration is that the typically high input impedance of the op amp is reduced to the value of Rin, so the circuit feeding the input is loaded more heavily. This can be solved by adding a unity-gain buffer between the signal source and the inverting amplifier. One advantage of this configuration is that both op amp inputs are held at a constant voltage (Vbias), so there is no common-mode signal and therefore no common-mode distortion (often the dominant distortion mechanism). In this circuit, capacitor C1 performs a similar role as in Fig.7, although it is arguably more effective here since it reduces the gain at very high frequencies to zero rather than unity. 4) Virtual ground mixer Fig.9 shows a circuit that is basically an inverting amplifier with multiple resistors feeding different signals into the inverting input. As the inverting input is held at a fixed DC voltage by the negative feedback, there is no possibility of cross-talk between the signals (which might be significant in a mixing console, where they are fed to multiple locations). 5) Differential amplifier This is a very useful circuit used in many different forms. While you can build it using regular op amps, it is probably more widely used in monolithic instrumentation amplifiers (albeit in modified form), difference amplifiers and current shunt monitors. Fig.10 shows the basic version of this circuit. It provides an extremely useful function; it takes the difference between two voltages, multiplies it by a constant (determined by the resistor values) and then possibly adds a positive or negative offset voltage. However, Vref is often set to 0V so the output voltage is referenced to ground. This circuit needs precise resistor matching for a good common-mode rejection ratio (CMRR). Even with 0.1% tolerance resistors, a CMRR of more than 60dB is difficult to guarantee. Trimming can give good results, although the procedure can be tricky. It’s generally better to use lasertrimmed monolithic devices like instrumentation amplifiers (‘inamps’) that can have CMRRs over 100dB. Most instrumentation amplifiers use a slightly different internal circuit that includes three op amps; besides having a very good CMRR, this has the advantage that the gain can be set using a single external resistor. However, the basic principle is the same. A difference amplifier is basically an instrumentation amplifier where the input voltages can be well outside (usually above) the device’s supply range. A current shunt monitor is a specialised version of an instrumentation amplifier. All are based internally on op amps or similar circuits. A shunt monitor allows you to place a low-value shunt resistor in the positive supply to a section of the circuit, Fig.11: this full-wave rectifier circuit uses op amps to effectively cancel out the forward voltage of the diodes. As a result, for positive voltages at Vin, Vout tracks very closely (within microvolts, given sufficiently high precision op amps) while for negative voltages at Vin, Vout = −Vin (again, within microvolts). This is ideal for circuits that need to sense peak signal levels, such as audio clipping meters. ► Fig.12: this Sallen-Key low-pass filter provides ► a reduction in amplitude at -12dB/octave above its -3dB frequency, and multiple stages can be cascaded for an even steeper slope. Changing the resistors to capacitors and capacitors to resistors makes it a high-pass filter instead. 42  Silicon Chip Australia’s electronics magazine siliconchip.com.au ► Fig.14: this active bandpass filter blocks signals outside of a given frequency range, although the slopes are only -6dB/octave. For steeper slopes (eg, -12dB/ octave), one of the active lowpass filters described above can be connected in series with a similar high-pass filter. ► ► Fig.13: this multiple feedback filter does the same job as the Sallen-Key filter, but is more effective at higher frequencies. That’s important for low-pass filters as otherwise, it can pass signals that the filter is supposed to block. As only one extra resistor is needed, it’s a worthwhile upgrade, and the gain can be set without any more resistors (although it does invert the signal). Fig.15: this Twin-T active notch filter attenuates signals at a specific frequency. Both that frequency and the steepness/depth of the notch can be controlled by careful selection of the passive component values. and obtain a ground-referenced voltage to feed to an analog-to-digital converter (ADC) or similar. They have a high CMRR to reject supply ripple. 6) Precision rectifiers A precision rectifier acts like a diode or bridge rectifier, but without the forward voltage drop. This is important for rectifying low-level signals (too low to forward-bias a diode), or for accurately rectifying AC signals in order to measure their magnitude etc. They are commonly employed in devices like VU meters or AC current monitors. Fig.11 shows the full-wave version, similar to a bridge rectifier. The halfwave version is basically just one of the op amp/diode/resistor sections. The op amps reduce the effective forward voltage of the diodes by a factor of their open-loop gain, meaning the ~0.7V drop of a standard silicon diode is effectively less than 1μV for an open-loop gain of around one million. The resistor values shown result in unity gain. This circuit originally came from National Semiconductor who specified R = 100kW, although other values can be used. The values could be changed to give a fixed gain if necessary. 7) Active low-pass filter The simplest way to implement a low-pass filter with an op amp is to combine a basic RC low-pass filter with a unity-gain buffer. However, a more economical arrangement is the Sallen-Key low-pass filter shown in Fig.12. This has a -12dB/octave slope, compared to -6dB/octave for the RC filter, using just one op amp. It also allows gain to be applied. siliconchip.com.au Fig.13 shows a multiple-feedback low-pass filter. This provides precisely the same function as the Sallen-Key filter, but it is less prone to signal feedthrough, which means it performs much closer to an ideal filter at frequencies approaching the op amp’s bandwidth. The only disadvantage is the use of one more resistor. To calculate the required resistor and capacitor values for a given cutoff frequency, go to siliconchip.com. au/link/aajq Note that it is possible to build a third-order Sallen-Key active low-pass filter using a single op amp. This will give you an 18dB/octave roll-off with one op amp, 30dB/octave with two etc. This is shown at siliconchip.com. au/link/ab8v 8) Active high-pass filter To convert the low-pass filters shown in Figs.12 & 13 into high-pass filters, simply transpose the resistors and capacitors. As with the low-pass filters, these will provide a 12dB/ octave slope per op amp. For both the low-pass and high-pass filters, by adjusting the resistances and capacitances, it is possible to design filters with characteristics other than Butterworth. Butterworth has minimal (essentially no) ripple in the passband, but different filter types such as Chebyshev trade off increased passband ripple for a steeper roll-off beyond it. To calculate the required component values, see siliconchip.com.au/ link/ab8w 9) Active bandpass filter A second-order bandpass filter can be created by combining active secondAustralia’s electronics magazine order low-pass and high-pass filters. Alternatively, you can use the configuration shown in Fig.14, where a single op amp can act as a first-order bandpass filter with adjustable gain and a Q of up to 25. This configuration inverts the signal phase; however, if chaining multiple filters, it can be re-inverted by another stage. 10) Active notch filter Fig.15 shows a “Twin-T” active notch filter. One interesting aspect of this design is that the Q, and thus the depth of the notch, changes based on the resistor and capacitor values selected. See the online calculator at siliconchip.com.au/link/ab8x 11) Gyrator Fig.16 shows a ‘gyrator’, an active element that behaves similarly to an ideal inductor at low current values. It does this by using the op amp’s negative feedback to effectively invert the behaviour of capacitor C. This is useful in circuits like graphic Fig.16: the gyrator is a particularly clever circuit. It uses negative feedback to make a capacitor behave like an inductor. It is superior to an actual inductor in many signal processing applications. August 2021  43 equalisers, where resonant (LC) elements are needed with accurate resonance frequencies, low distortion and small size. Inductor tolerances are typically much wider than capacitors, and high-value inductors can be very bulky, so in signal-processing circuits, the gyrator is almost always better than a resonant circuit based on an actual inductor. 12) Baxandall active filter Fig.17 shows a basic version of the widely-used Baxandall active tone control. It has many good properties, such as the ability to have as many or as few bands as you want, with no interaction between the controls, and no special requirements for the potentiometers. This one shows bass and treble pots, but one or two midrange pots can easily be added. Fig.18 is the Baxandall active volume control. The traditional volume control method is a logarithmic potentiometer, but dual versions usually have poor tracking at the low end, so they are not great for stereo circuits. The Baxandall active circuit provides logarithmic-like control with a linear potentiometer for superior tracking. It can also offer significantly better noise performance as the pot adjusts the gain over a wide range, from zero up to many times (as set by the fixed resistors). 13) Audio amplifiers Fig.19 is a simplified version of the circuit from our SC200 audio amplifier. It is essentially a high-power op amp with large output transistors that can source and sink plenty of current (and that are well heatsinked). Most Class-A, Class-AB and similar amplifiers are variations on this theme. Even Class-D amplifiers typically use some form of negative feedback to avoid gross distortion. 14) Other uses for op amps An op amps can be used as a basic comparator by operating it in openloop mode, or with positive feedback (hysteresis). A comparator IC is essentially just an op amp with the frequency compensation component(s) removed for a faster swing at the output. An op amp can also be used to build an ‘integrator’ or ‘differentiator’. An integrator produces an output ramp proportional to its input voltage, while a differentiator produces an output voltage that’s proportional to its input ramp (rate of change). A log amp takes the exponential nature of a bipolar transistor and turns it on its head using negative feedback to provide a logarithmic transfer function. As a result, its output voltage is a constant multiple of the natural logarithm of its input voltage. This can be used as the basis of a multiplier circuit; by taking the natural loge(x) of several voltages, summing or averaging them, then exponentiating the result, the output voltage is the product of the input voltages. Other mathematical functions can be applied to voltages by an op amp, including addition, subtraction, division and inverse logarithm (the exponentiation mentioned above). Op amps can also be used to build controlled current sources/sinks, including constant loads, by combining op amps with large transistors that can handle lots of power with sufficient heatsinking. The generalised impedance con- Fig.17: the Baxandall tone control was initially designed with a valve or transistor as the active element, but it works even better with an op amp. It is elegant and expandable, with virtually no interaction between the stages (in this case, two: bass and treble adjustments). No matter how many bands it has, only one op amp is required per channel (ie, two for stereo). 44  Silicon Chip verter uses two op amps to present a load impedance proportional to another impedance. The ratio can be set using fixed or variable resistors (or even other impedances!). Many op amps are designed to drive relatively low load impedances, such as 600W. These work quite well as basic headphone drivers, with relatively low distortion figures driving typical headphone loads, even as low as 16W. They can’t supply a tremendous amount of power, but enough for most headphones to deliver decent volume, using one low-cost IC. An op amp can also be used as an error amplifier in feedback control. For example, to adjust the drive to a motor to maintain a constant speed despite a varying load. An op amp can form the basis of various oscillators, to generate waveforms at fixed or variable frequencies; primarily sinewaves, but also triangle waves or sawtooth waveforms. An op amp (especially a CMOS type) can be used as a high inputimpedance buffer amplifier or guard ring for monitoring sensors that cannot handle any loading, such as narrowband oxygen sensors and pH sensors. CMOS op amps can have input impedances in the terohms range (more than one trillion ohms)! CMOS op amp buffers can also be combined with analog switches and low-leakage capacitors to form sampleand-hold circuits, often used for sampling voltages over small time windows to feed an ADC or similar. Signal swing limitations For a very long time, the signals at the inputs and outputs of an op amp Fig.18: the Baxandall volume control also places the potentiometer in the negative feedback loop. This gives exponential gain control with a linear potentiometer and a wide range of gain settings with a reasonably constant noise level. Australia’s electronics magazine siliconchip.com.au could only have a considerably smaller swing than the supply range of the op amp. For example, if you had an op amp running from 12V, the inputs and outputs might be limited to a range of 3-9V. Or, with a dual supply like ±15V, you might be limited to a signal swing of ±12V. That’s because the op amp’s internal circuitry needs some voltage ‘headroom’ to operate. But more recently, single-supply and rail-to-rail output op amps started to become available. Single-supply op amps typically allow the inputs and outputs to go down to the negative rail (eg, 0V). So a single-supply op amp running from 12V can handle signals of say 0-9V. Rail-to-rail output op amps generally have the same input limitations as standard op amps, but their output can swing over virtually the entire supply range. This is especially useful when applying gain to AC signals, as in that case, the input swing will never reach the rails anyway (at least, not without ‘saturating’ the op amp). These days, rail-to-rail input/output (RRIO) op amps are very common. Some can even run down to very low supply voltages, like 1.8V! These op amps essentially remove the above limitations, with input and output signals that can range anywhere between the supply rails. Some will even handle input signals outside the rails, although usually only in one direction (eg, positive) and by a limited number of volts. Note that RRIO op amps sometimes compromise performance in other ways, such as having higher noise or distortion, or just costing more than ‘regular’ op amps. Multiple op amps As op amps became cheaper and more versatile, dual and quad op amps became popular. These save money and space; a quad op amp IC often costs less than twice what a single one does, and only requires two power tracks to be routed and one bypass capacitor. Most dual (8-pin) and quad (14-pin) op amp ICs use the same pinout so they can be interchanged. Single op amps are not quite so interchangeable, as these usually come in an 8-pin package. After accounting for the two supply rails, two inputs and one output, the remaining three pins can be used for trimming/balancing, external compensation capacitors or various other functions. Some are interchangeable (even if they don’t have exactly the same features), but not all. These days, single op amps are also available in tiny 5-pin SMD packages for where space is at a premium. Conclusion The op amp is an incredibly flexible device, available these days at very low cost and in a vast range of different versions, optimised for different tasks. While it is possible to process analog signals without op amps, generally, the results will be worse. So most analog designers make extensive use of op amps in their circuitry. They are an essential building brick that most designers would have difficulty doing without. We have Harold S. Black to thank for making our lives SC a lot easier! Fig.19: a slightly simplified version of our SC200 power amplifier circuit. It’s essentially a big op amp; transistors Q1 & Q2 are the differential input pair (the inputs are at their bases), Q8 is the voltage amplification stage, Q11 & Q12 are the output drivers and Q13 and Q15 are the power output transistors. The components highlighted in red form the negative feedback path, from the output at the emitter resistors of Q13 & Q15 back to the base of Q2, which is the inverting input. siliconchip.com.au Australia’s electronics magazine August 2021  45